Professional Documents
Culture Documents
(1999) WAHBA - Static and Dynamic Analyses of Guyed Antenna Towers PDF
(1999) WAHBA - Static and Dynamic Analyses of Guyed Antenna Towers PDF
ANTENNA TOWERS
YOHANNA M. F. WAHBA
B.Sc.(EIons), M.A.Sc., P.Eag., P.E.
A Dissertation
subrnitted to Collcge of Graduate Studies and Research through
Civil and Envitonmental Engineering Program
in partial Nfilment of the requirements for the
degree of Doctor of Philosophy at the
University of Windsor
The author has granted a non- L'auteur a accordé une licence non
exclusive licence allowing the exclusive permettant à la
National Library of Canada to Bibliothèque nationale du Canada de
reproduce, loan, distribute or sel1 reproduire, prêter, distribuer ou
copies of this thesis in microfonn, vendre des copies de cette thèse sous
paper or electronic formats. la fome de microfiche/film, de
reproduction sur papier ou sur fomat
électronique.
Canada
Yobanna M.F. Wahba
1999
AI1 Rights Reserved
1 hereby declare that 1am the sole author of this document.
Yohanna M. F. Wahba
Yohanna M. F. Wahba
THE UNIVERSITY OF WiNDSOR requires the signatures of al1 persons ushg or
photocopying this document.
investigated using two different finite element models (three-dimensional ûuss model.
cornparison between the analytical techniques is presented for different loading Ievels.
Results from the anaiytical models are verified by testing four scale model towers to
collapse. The analytical techniques are also extended to prototype towers and
conclusions are drawn regarding the suitability of the analfical models to the static
towers, was designed, built, and instrumented at the Structural Engineering Laboratory at
the University of Windsor. This study presents the experimental investigation, testing
facility set-up, construction of models, materials and procedures used for the static, fiee
vibration, fmed vibration, and ultimate load tests of guyed tower models. The facility was
used to test five scaled guyed tower models under fke vibrations and foiwd base motion.
investigate the main parameters influencing the fhx vibration of these structures. Based on
this parametric study, three empirical equations were derived to estimate the fundamental
establishg the dynamic characteristics of the towers for use by design engineers.
ACKNOWLEDGEMENTS
The author wishes to express his deep appreciation and gratitude to his advisors Dr.
M. K. S. Madugula and Dr. G. R Monforton, for their constant support and valuable
supervision during the development of this research. To them 1 say, "1 am gratefhi - thank
you".
The author wishes to thank Dr. G. Abdel-Sayed, for his help and encouragement
Many Thanks are also due to Mr. D. Marshall, for his encouragement and for the
pictues of towers that he supplied me. Also, th& to Messrs.G. Patton, K. DeBelser, R.
Sullivan, of LeBlanc Ltd. for their support and encouragement to complete this dissertation.
The author wishes to acknowledge the financial support provided by the Naturai
Sciences and Engineering Research Council of Canada, The Ministry of Colleges and
Finally, the author is deeply grateful to his parents and his wife for their great
support, understanàiig, and patience throughout the course of this research; to them 1 Say
''Thanks".
viii
TABLE OF CONTENTS
5.1 Introduction............................................................................................................. 98
5 -7Discussioa ................... .
............................................................................... 112
,
5.8 summq ..........................*......... ........*...*............................................*..................
113
CHAPTER VI RESULTS FROM 'TYPICAL PROTOTYPE TOWERS ~..~.~~.~~..
172
Table 6-4: EfTective Mass Compoaents for the Fist 100 Modes of Tower PV 199
Table 6-6. Denvation of Empirical Equation for Tower Naturai Frequency 205
Table 6-8. Naturd Frequencies and Mode Shapes for Tower V Under Different Ice
Accretions 207
Table 6-10. Effective Mass for the First 10 Vibrational Modes of Mast (Tuwer V- Guy
Table 6-12. Cornparison of the Mast Frequencies for Various Initial Tensions 21 1
xiv
List of Figures
Figure 1-1.550 m Guyed Tower 7
Figure 3-4. Typical Mast Section Sub-Mode1 Used to Detemine Equivalent Bearn
Properties 46
Figure 4-2. Typical Mast of Guyed Tower used for Heights up to 200 m 68
Figure 4-21. Fundamental Natutal Frequency and Mode Shape of Shake Table
Figure 4-27. The Actuator Controller and the Data Acquisition System
Figure 5-1. Deflection of Model Tower 1 under Load Case 1 (45 N at top loading
point) 130
Figure 5-2. Deflection of Model Tower I under Load Case II (27 N at al1 loading
points) 131
Figure 5-3. Deflection of Model Tower II under Load Case 1 (45 N at top loading
point) 132
Figure 5-4. Deflection of Model Tower II under Load Case II (27 N at al1 loading
points) 133
Figure 5-5. Deflection of Model Tower III under Load Case 1 (45 N at top loading
point) 134
Figure 5-6. Deflection of Model Tower III under Load Case II (27 N at al1 loading
points) 135
Figure 5-7. Deflection of Model Tower N under Load Case 1 (45 N at top loading
point) 136
Figure 5-8. Deflection of Model Tower IV under Load Case II (27 N at al1 loading
points) 137
Figure 5-9. Deflection of Model Tower V under Load Case 1 (27 N at the top three
Figure 5-10. Deflection of Model Tower V under Load Case II (27 N at al1 loading
points) 139
Figure 5-1 1. Acceleration Response for Free Vibration of Model Tower I 140
5-1 1 140
Figure 5- 13. Acceleration Response for Free Vibration of Mode1 Tower II 141
5-13 141
Figure 5-15. Acceleration Response for Free Vibration of Model Tower III 142
5-15 142
5-17 143
Figure 5- 19. Acceleration Response for Free Vibration of Model Tower V 144
5-19 L 44
Figure 5-21. First Nahiral Frequency and Mode Shape of the Top Guy (Model
Tower V) 145
Figure 5-22. First Nahinil Frequency and Flexural Mode Shape of the Mast (Model
Tower V) 146
Figure 5-23. Second Naairal Frequency and Flexural Mode Shape of the Mast
Figure 5-24. First Torsional Frequency and Mode Shape of the Mast (Model
Tower V) 148
Figure 5-25. First Natural Frequency and Mode shape of the Bottom Guy (Model
Tower V) 149
Figure 5-29. Forced Vibration Results For Northridge Time History (Tower IV
Figure 5-30. Forced Vibration Results For Northndge Time History (Tower N
xix
Figure 5-31. Forced Vibration Results For Nanbu T h e History (Tower III
Figure 5-32. Forced Vibration Results For Nanbu Time History (Tower III
Figure 5-33. Shake Table Acceleration History for Ground Motion modeled after
Figure 5-34. Measured and Calculated Acceleration Histories for Shake Table
Figure 5-35. Measured and Calculated Acceleration Histories for Shake Table
Figure 5-36. Measured and Calculated Displacement Histories for Shake Table
Figure 5-37. Measured and Calculated Displacement Histories for Shake Table
Figure 5-38. Load versus Displacement at the top Guy Level (Model Tower 1) 162
Figure 5-39. Load versus Displacement at the top Guy Level (Model Tower III)163
Figure 5-40. Finite Element Displaced Shape of Model Tower III at Failure 164
Figure 5-42. Load versus Displacement at the top Guy Level (Model Tower IV)166
Figure 5-43. Finite Element Displaced Shape of Model Tower IV at Failure 167
Figure 5-44. Load versus Displacement at the top Guy Level (Model Tower V) 168
Figure 5-45. Finite Element Displaced Shape of Model Tower V at Failure 169
Figure 6-9. Cornparison of Leg Loads and Face Shear under Design Loads for
Prototype Tower P 1 22 1
Figure 6- 10. Cornparison of Deflections under Design Loads for Prototype Tower
PI 222
Figure 6-1 1. Cornparison of Leg Loads and Face Shears under Design Loads for
PV 224
Figure 6- 13. Comparison of Leg Loads and Face Shears under Design Loads for
Figwe 6- 14. Comparison of Deflections under Design Loads for Prototype Tower
PVIII 226
Figure 6-16. Load as a Ratio of Design Loads Vs. Deflection at the Top Guy Level
Figure 6- 17. Failure Shape as Predicted by the Finite Element Model for Prototype
Figure 6-18. Load as a Ratio of Design Loads Vs. Deflection at the Top Guy Level
Figure 6-19. Failure Shape as Predicted by the Finite Element Model for Prototype
Tower PV 23 1
Figure 6-20. Load as a Ratio of Design Loads Vs. Deflection at the third Guy
Figure 6-21. Failure Shape as Predicted by the Finite Element Model for Prototype
xxii
Tower PVIII 233
Figure 6-22. Failure Shape as Predicted by the Finite Element Mode1 for Prototype
Figure 6-23. First Twenty Mode Shapes of Prototype Tower PV Modes (1-4) 235
Figure 6-24. First Twenty Mode Shapes of Prototype Tower PV Modes (5-8) 236
Figure 6-25. First Twenty Mode Shapes of Prototype Tower PV Modes (9-12) 237
Figure 6-26. First Twenty Mode Shapes of Prototype Tower PV Modes (13-16)238
Figure 6-27. First Twenty Mode Shapes of Prototype Tower PV Modes ( 17-20)239
Figure 6-28. Variation of Fust Natural Frequency with Height of Tower 240
Figure 6-30. First Flexural Frequency and Mode Shape of Prototype Tower PI
Figure 6-3 1. First Flexural Frequency and Mode Shape of Prototype Tower PI1
Figure 6-32. First Flemiral Frequency and Mode Shape of Prototype Tower PlII
Figure 6-33. First Flexural Frequency and Mode Shape of Prototype Tower P N
Figure 6-34. First Flexural Frequency and Mode Shape of Prototype Tower PV
(Guy Modes Suppressed) 246
Figure 6-35. First Flexural Frequency and Mode Shape of Prototype Tower PVI
Figure 6-36.First Flexural Frequency and Mode Shape of Prototype Tower PVII
Figure 6-3 7. First Flexwal Frequency and Mode Shape of Prototype Tower PVIII
Figure 6-38. Variation of First Flexural Frequency of the Tower (Guy Modes
Figure 6-39. Time History of Deflections of Mast for Tower PIV Subjected to El-
Figure 6-40. Time Histocy of Guy Stresses for Tower PIV Subjected to El-Centro
Figure 6-4 1. Time History of the Top Guy Stresses for Tower PVi Subjected to
Figure 6-42. Time History of the Top Guy Stresses for Tower PVI Subjected to
Figure 6 4 3 . Time History of the Deflection at the Top of Tower PVI Subjected to
atea of guy
displacements
modulus of Elasticity
shear modulus
guy radius
torsionai constant
length of guy
densiiy
Poissons's ratio
angle of twist
CHAPTER I
INTRODUCTION
1.1 General
demand for reliable communications is growing. Due to the ongoing expansion of the
information highway, and the introduction of new technologies such as digital television,
there is a growing need to better utilize the capacity of existing towea and to optimize the
capacity of new structures. Furthemore, the demand for more towers to be erected in
urbanized areas has not only made it dificult to obtain building pennits but also created
the need for far more reliable structures. This also has made the existing ''vertical real
Most of the communication failities built today f d into one of two types of
applications: the fitst for use by the communications indusûy (mostly for wireless
types: (i) monopoles, which are cantilevered tubes with heights up to 70 m. (ii) self-
supporting lattice towers, wbich are commonly used for heights up to 120 m, although for
urban sites where the pnce of the land is more valuable, seKsupporting towers for heights
up to 300 m have k e n used, and (iii) guyed towers, which have been utilised for taller
structures up to 620 m. Figure 1-1 shows a pichue of a 550 rn guyed tower, while Fig. 1-2
shows a close up of a tall guyed tower mast and Fig. 1-3 shows the tower base. These
figures represent typical tall (broadcast) tower applications. Figure 1-4 is the photograph of
a typical short (100 m) guyed tower norrnally used for wireless and telecornmunications
applications.
A typical guyed tower consists of a mast that is usually of a constant cross section
and one or more levels of guy cables (that provide lateral support to the structure) anchored
at the ground level. Masts with triangular cross sections are most commonly used in North
America; however, square cross section rnasts are popular in Europe and other parts of the
world. The mast n o d l y consists of a latticed structure made fkom angles, solid rounds or
pipes. Guy cables are made of braided high strength cable +S.
tension due to applied loads or even original design pre-tensions, the non linear force
deformation relationship of the structure, and the large displacements experienced even
ALthough ment advancements in cornputers have allowed fat more larger and
sophisticated techniques to be used, the normal anaiyticai methods used in the design of
these towers are still very simple. However, there has been very little experirnental data that
While the cumnt design practices in North America recommend the minimum
static loads and analysis methods for the design of guyed towers, very few, if any, towers
are checked under dynamic loads. Practicai considerations in the design process require the
designer to understand the effect of different key parameters on the dynamic characteristics
structure would be helpful in understanding the dynamic behaviour of the structure and in
dynamic anaiysis is usually tequireâ, eqecially for transient loads. Dynamic non-linear
analyses are computationally expensive and redts must be examined catefhily. That is, the
numeric solution may be stable, yet the mathematical mode1 may not represent the acnial
vibrations or loaded to failure are unavailable. Therefore, ceseaich in this area is needed.
1. Compare the finite elernent models (beam and tms models) with the analytical methods
communication towers.
communication tower models under static service loads, dynamic loads, and failure
loads.
4. Verification of the analyticai models in item (1) above through the experimental results.
5. hvestigate the parameters influencing the nahual frequencies and mode shapes of guyed
towers.
seems to be the only approach capable of including al1 the important factors uifluencing the
structural behaviour. The main panuneters examined in the present research are: total
height, span-to-face width ratio, number of guy levels, guy systems, height of cantilever,
Design, fabrication, and instrumentation of a new dynarnic testing facility (shake table)
nurnber of guy levels, and guy systems under static and dynamic loading to mesure
techniques.
Verification of the analytical techniques used in the static, ultimate failure, and dynamic
A study of the important panuneters that determine the dynamic properties of guyed
towers.
Recommendations with respect to the analysis procedures to be used for the static and
dynamic analyses.
The literature review and previous work on guyed towers is ~ufnmarizedin Chapter
II. Chapter iïI describes the various numerical modelling techniques used in the analficd
study including the idealkation and modelling of guyed towers as well as the non-linear
static analysis, pushover static analysis, fiee-vibration analysis, and forced vibration
analysis. In Chapter IV, the experimental work conducted on five scale-mode1 guyed
towers is described. This includes the details of the models, instrumentation, loading
system, and the test procedure. Chapter V pmsents the discussion of the results obtained
from the experimentai investigation as well as fiom the numerical analyses. Chapter VI
deals with the application of this study to actual prototype towers. Chapter W gives a
2.1 Geaeral
The structurai complexity of guyed towers has attracted many researchers to study
the behaviour of such structures under various conditions and extensive related research is
available. A recent litemture review included in the Draft Guide for Dynarnic Response of
Lanice Towea (ASCE 1999) has cited over 500 publications on that subject alone,
including guyed and self-supporthg towers. This interest maybe attributed to the fact that
these structures are the tallest in the world as they stand over 600 m above ground, they
have a relatively high rate of failure, and their behaviour is generally non-linear.
This chapter presents the background for the present investigation which is divided
mainly into four distinct parts. The first part is concerned with the static analysis of guyed
towers; the second part with the stability and uitimate load carrying capacity; the third with
the previous experimental work; and the fourth with the dynamic response of guyed towers.
23 Static Analysis of Guyed Towers
The first step in the design or aaalysis of a guyed tower is a static analysis. Early
research in this area was concemed with the elementary design methods. Schott and
Thunton (1956) applied a simplified approach for the analysis of guyed antenna towers. It
was assumed that al1 loads are symmetrical with no twisting induced on the tower. Another
simplification was the assumption that the tower's deflected shape under wind loads would
be a straight line. This later simplification was important for the subsequent application of
the three-moment equation for the analysis of the mast which was treated as a continuous
beam. Graphical methods were later used to account for the different support conditions.
Also, Pocock (1956) performed low speed wind tunnel tests to study the effect of wind
loads on antennas and mast sections and to determine the drag and cross-wind forces.
Cohen and Perrin (1957a) carried out extensive research to ascertain the primary
loads on a guyed tower and to study the effect of the shape and orientation of the tower on
wind resistance. Also, empirical formulas and charts were derived to calculate the drag
coefficients, lift coefficients, and shape factors depending on the individual member and
mast section. In a continuation of the same work, the authors published another paper
(Cohen and Penh 195%) outlinhg the use of the three moment equation for the analysis of
guyed towers considering the mast as a bearn-coIrmui supporteci on linear elastic springs.
Rowe (1958) included the effect of changes in geometry on the analysis of guyed towers; a
simplified non-linear method was introduced, but it was not considered effective as it fded
Hull (1960) presented a study on the bending analysis and stability of ta11 multi-level
guyed towers. Several approximations were used to simpliQ the problem: the cable was
assumed to be parabolic with only d o m vertical loads acting on it and cable tensions
remained constant tbroughout the length of the cable. The moment distribution method was
also used in the bending analysis. Furthemore, the deflected shape was assumed to be in
Dean (1961) has presented a rather accurate formulation of a cable mode1 that is
suitable for both static and dyaamk analyses. nie differential equations of the cables were
formulated based on a perfectly flexible behaviour. This allowed accounting for the change
in cable tensions due to the movement of the rnast. Later research (Meyers 1963) studied
guyed towers as non-linear sûuctural systems. This research emphasised the stability of the
structure. Chord length of the cable, instead of the arc length, was used in the calculations.
A standard solution was prepared, then a parametric study illusûated the effect of the
various factors on the solution. Some of the secondary eRects such as shortening of the
mast due to bending and axial loads, the forcedisplacement, and the moment-rotation
characteristics of rnast and mast rotation at guy attachment points were included in the
analysis. This was considered to be a comprehensive approach that Iooked into the non-
this approach, bending was considered in the two principal planes of the mast and the
analysis of the guy cabie was based on the parabolic approximation. Axial movement of the
mast was neglected but the effect of axial loads on the flexural stiffness was taken into
account. Aiso, in this appmach, guy eccentricities at the attachment points were included.
The tower movement was considered to be dong the line of action of the forces.
based on an iterative approach that was readily suitable for cornputer programming. This
method analyzed the mast and the guy systems at each guy level separately and an iterative
approach based on the compatibility of the displacements at the guy points was introduced.
Thus, secondary moments at the guy attachent points were included. The readiness of this
approach for programming made it the choice of several commercial programs developed
for the analysis of guyed towers. This model also considered axial loads on the mast as well
as wind and ice loads on the guys. This model was based on an elastic beam supported by
non-linear springs. The sprhg constant at any deflection point was denved based on the
Marshall (1966). This also included some experimental venfication of the analysis method.
This methd simplified the guy models and restricted its application to shorter towers.
Another method was developed in the same year and it used interactive diagrams to design
in s e v d stages (Müdofslq and Abegg 1966). The aaalysis was carried out in two stages.
First, the tower was anaiysed as a continuous beam on elastic supports and then the tower
was analysed as a beam-column to include the amplification of stresses resulting fiom the
axial loads. The maximum stresses at the guy points and mid spans were presented on the
diagrams for different face widths of towers. This method neglected the displacements at
the support points due to the small displacement-span ratio. The other major drawback of
this method: it was not versatile to account for many special conditions and it could not be
programmeci effectively.
Livesley (1970) formdated an automated design method for guyed towers mbjccted to
towers. Later Bell (1972) developed a structural optimisation system for the design of guyed
towers. The stiffiess maûix method with stability fhctions was used to mode1 the mast,
while guys were modelled as springs. The technique was limited to a certain class of towea
ranging fkom 30.5 to 152.4 m. Torsional effects were neglected and bending in one
Skop (1979) developed a new method for the detennination of the cable spring
constants. The method was more general in that it accounted for arbitrarily loaded guys
containing any n u m k of disrrete masses. Later Rosenthal and Skop (1980) published an
analysis method where the mast was idealised as a two-dimensional structure and the cables
were analysed by the method of unaginary reactions as formulated by Skop and O'Hara
(1970, 1972). No predehed shape was asmmed for the cables. A h , this method
modelled the base as mounted on rotational springs, which allowed a wide range of
formulation for a cable element. A twodimensional example was iilustrated in that paper;
however. the method can be generally used to mode1 any guyed tower system. Later Issa
and Avent (1 991) presented a method to obtain dkctly the forces in al1 individuai members
The current trend with standards including the Canadian Standard for antenna
towen CSA S37-1994 (Canadian Standards Association 1994) has changed to limit States
design in which the stnictwes are analyzed under factored loads or near failure loads
(Wahba et al. 1994). It is also expected that the US Standard for Towea (ANSIIIIAIEIA-
2 2 2 3 will switch to load and resistance factor design based on the Arnerican Mtute of
Steel Construction design specincation (ABC 1994). Also, with the introduction of new
technologies, existing structures are analyzed for the addition of new loads (antenoas).
Generally, the industly is ushg bearn-on-Mngs models (Odley 1966) to analyse ever-
The stability analysis of guyed towers is not a simple problem. As the governing
equilibrium equations are not linear homogeneous equations, instability is not a
bifurcation problem. Several design approaches use the member effective length factors
based on unsupported length of the mast between the guy attachment points. What
complicates the matter further is that under horizontal loads, these supports (guy
attachent points) are also moving. One of the advantages of using Limit States Design
is that the tower is analysed under factored loads. The andysis of the structure under
factored loads (at failure) helps the designer to examine any signs of instability at the
ultimate load level, provided that the analysis method is sophisticated enough to account
for large deformations and detect instability. The tower is considered unstable when large
towers. in this research, a method for the analysis of towers was formulated and the effect
of parameters afEecting the stability of towers was presented. This study Iooked into the
effect of initial tensions, guy sizes, and spans on the stability. Also, this snidy considered
the stability of the towers under increased horizontal loads. increashg the applied
horizontal loads h m wind would redt in an increase of the applied axial load fiom the
supporthg guys that may evenhially lead to overall buckling of the tower. In their analysis,
buckling of the tower was considered as the point at which no M e r horizontal loads could
be applied to the structure. Later Cbjes and Chen (1979) and Chajes and Ling (1981)
studied the critical buckling load of towers. Formulas and curves were presented for the
detemination of the critical load for guyed towers. However, this research was lllnited to
Williamson (1973) examined the effect of king on the stability of a special taIl
guyed tower. In this tower the uppemost guy level consisted of an anay of twenty-four
conducting cables senhg as a radiahg element of the antenna system. Also, the effect of
increasing the stiffiess of mast and guys on the relative cost was studied. Later Costello
and Phillips (1983) looked hto the p s t buckling behaviour of guyed towers.
towers or scale mode1 towers especially under transient loads (seismic response) or for
free vibration. Marshall (1966) developed a procedure to detennine the deflections and
guy tensions of guyed towers subjected to direct loads and torsional loads. This study,
which included experimental research, was geared to observing the torsional distortions
Novak et al. (1978) studied the vibration of towers due to gailoping of iced cables.
The study included: (i) measurements of the aerodynamic properties of the ice-covered
guys on stationary sectional models of cables in the wind tunnel, (ii) theoretical prediction
of the galloping oscillations of the sectional models and their experimental venfication on
vibrating sectional models in the wind tunnel, and (iii) theoretical prediction of vibration
of a tower due to galloping oscillation of its guy.
guyed tower. The tower had four guy levels. Accelerometers and anemorneters were
placed on the tower. Results fiom full-scale wind velocity and structural response data
fkom this tower were analysed to determine (i) the power-law exponent for wind profile,
and (ii) estimates of the resonant fkequencies and ratios of the critical damping.
perfomed on 1.9 rn models of guyed towers for wind turbines. Tests were perfomed on
single level guyed towers of a tubular cross section. Also, the effect of the top mass and
pertain exclusively to wind response. Davenport (1959) and Davenport and Steels (1965)
proposed a linear mode1 to describe the vibration of the guys. This work was later used to
study the dynarnic response of the CFPL tower in London, Ontario. Most of that work
was restricted to the prediction of the dynamic respoase of the structures where the wind
loads are random in nature. McC&y and Hartmann (1972) have published a study on
the dynamics of guyed towers and the parameters used in determination of the
mathematical mode1 used for the anaiysis of guyed towers.
Earlier research published in this area was focused on the dynamic response of
towea due to wind loads. This research questioned the application of a static gust factor
that is applied throughout the height of the structure (Vellozzi 1975). Modal dynamic
analysis using available wind spectra was perforrned to simulate the effect of gusting
winds*
special analytical skills made it very rare to apply these procedures in a design office.
Later research by Davenport and his collaborators (Davenport and Allsop 1983; Gerstoft
and Davenport 1986; Davenport and Sparling 1992) concentrated on the development of
patch loading methods that approximate more realistically the dynamic eKects of
turbulence, using a series of patch load patterns. Since these methods require only non-
linear static analysis tools which are readily available, different standards (BSI 1994;
CEN 1997) have adopted different variations of patch loading methods. More recently,
even simpler analysis methods that account for the gusting effects of wind (Gress and
Sparling 1998) were developed to avoid the large number of loading cases required in the
dimensionai dynamic response due to wind turbulence. In this study, the non-linear
a model of 200 m guyed mast with three guying levels using equivalent elastic linear
springs for the guy cables. The stifbess of the springs varies with the frequency of
oscillation; however inertia effects of the cables were aot modelled. Later, Augusti et al.
(1990) have presented a model of guyed towers in which the cables were presented by a
Other sources of dynamic loads of guyed towers arise from seismic loads, sudden
ice shedding of guy wires, and sudden rupture of wires. Until recently very little research
was done to investigate these areas. Generally guyed towers have behaved reasonably
well under seismic loads although there have been reports of local failures, and serious
misalignment of towers as a direct result of the Loma Prieta and Northridge earthquakes.
Also, there have k e n reports of catastrophic failures of towea due to sudden ice
shedding and also due to rupture of guyed wires (usudly involving sabotage).
Recently, McClure and collaborators (McClure et al. 1993, McClure and Guevara
1994) have published tesults on the non-linear seismic response due to two horizontal
m.
Sudden ice shedding, (where a major chunk of ice on a guy wire is forced to shed
almost instantaneously) induces the tower to respond dynamically. McClure and Lin
(1994) published results of an anaiysis performed on three guyed towen (24.4, 60.7, and
213 m) with two, four and seven guy levels. Results indicated that the effect of ice
shedding should be m e r examined. especially on ta11 towers (more than 100 m).
CHAPTER III
3.1 General
The available anaiysis tools for guyed towers were reviewed in Chapter II.
However, in order to substantiate and assess the different modelling and analysis
techniques, this study has used two approaches for the d y s i s . The first, which is the most
commonly used in the industry, is the treatment of the tower as a conthuous beam resting
equations. The second approach is a finite element procedure and within this approach three
different models have been suggested. This Chapter describes the various models used in
this research and the analysis methods applied. The different methods of analysis described
herein were applied to the experimental mode! towers of Chapter IV as explained in detail
Most of the available commercial programs that are mitien specificaiiy for the
analysis and design of guyed communication towers use the beam on non-linear elastic
Consultants Inc., Downsview, Ontario, Canada (Guymast 1987). This model is primarily
3.2.1 Modelling
In this approach the tower is broken into three different numerical models as
follows: (i) every set of guys connected to the tower at the same level is analysed
independently, (ii) a continuous beamîolumn model is used to analyse the mast in the
two orthogonal vertical planes, and (iii) the shaft subjected to twisting moments is
3.2.2 Analysis
loads to be applied to the mast models in order to calculate a new set of mast
displacements which are then fed back into the guy model to obtain a better
approximation of guy stiffhess and loads. This process is repeated until the
displacements calcuiated by the mast model match those used by the guy model in
of the horizontal displacement Hgat the guy level is determined as follows (Odley 1966):
A, =-
Ha
(sec' 0+-)
w,'
AgE, 12~:
L, =L-A,
Where, A, is the cross-sectional area of the guy, Eg is the guy elastic modulus.
Since the unstretched length, Lt, is invariant with the load, the originally calculated length is
compared to that given by Eq. (3-3) and if the values do not agree, a new value is assumed
for H, the horizontal displacement, and the calculations repeated until satisfactory
agreement is reached.
The above procedure is repeated for al1 guys at the same attachment point fiom
which the force at the guy level redting h m a ceriain displacement, A, is calculated.
As mentioned before, the nomal indusîry practice does use commercial programs
based on the above a p p m h and specifically written for guyed communication towers. The
simplicity of these programs imposes some lhitations on the type of cross-section used,
panel configuration, similarity in ail mast faces, mast pmperties, etc. This approach is
applicable only for static loadhg where mernbers rernain in the elastic region.
flexible tool. Using a finite-element method for analysis makes it possible to model a tower
in a more redistic rnanner and to provide a full description of its structural members and
thus expect a more accunite structural response within the elastic and postslastic stages up
to collapse. The most important advantage of this method is its capability to model various
general f i t e element package named ABAQUS (HKS 1995) was used throughout this
study to determine both the static and dynarnic response of guyed towea. A general
description of the program is presented herein dong with the modelling of the different
discretized into a nnite numbei of regions (elements) that are interco~ectedat certain
denved and the global stiftiiess matrix of the entire structure can be formulated by the direct
stifiess method. The assembly of these elements to form the whole tower superstructure
result is a large set of simultaneous equations that can be solved using cornputers.
Nodal forces act at each nodal point, which result in displacements and rotations.
quantities. From the potentid energy formulation, the following equation is obtained in a
matrix form:
where npis the potential energy of the system; {U} is the global displacement vector; (P}
is the global load vector; [Ke] is the global elastic stifl'hess matrix; and b]is the global
geometnc stiffiess matrix. The geometric stiffness matrix is included in the analysis to
account for the defomed geometry of the elements and the effect of initial conditions on
the structure. This will account for the non-hear load deflection behaviour.
Differentiating the potential energy of the system with respect to the displacement
where: K = & + KG
The above equation is in tum solved and thus the displacements and forces can be
determined.
In a iinear elastic problem, loads are applied to a mode1 and the response is obtahed
directiy in one step. This single step analysis c m o t be used for guyed tower as the effect of
the initial conditions resulting h m the pre-stressed guys needs to be included in the
analysis. Thus, even for an analysis in which the response is in the elastic range of a tower, a
incrementation scheme. There are several methods to solve non-linear problems. One of
these methods is the well-known Newton's method, which is a numerical technique for
solving the non-linear equiliirium equations. By solviag a series of hear problems, the
non-linear solution of the problem is iteratively obtained. The non-linear response of a
structure to a srnall load increment, AP, is shown in Figure 3-2. Let [Po]be the initial load
at a certain time increment, vol be the initial displacement, and [&] be the initial tangent
stiffaess at Po] and [Po]. Using the stmctwe's tangent s-ess, Ko and AP, the
displacement correction, ci, can be calculated for the structure. Using ch the structure's
configuration is updated to y. Then the structure's intemal forces, Io, are calculated in this
updated configuration. The difference between the total applied load, P, and L can now be
calculated as:
Ra = P-1, (3-7)
if F&, is zero at every degree of fieedom in the model, then point "a" in Figure 3-2
would lie on the load deflection c w e and the structure would be in equilibrium. However,
in a non-linear aiiysis R will never be exactly zero, so an acceptable tolerance value has to
be set. If & is less than this force residual tolerance at al1 nodes, the solution would be
considered in equilibriurn and P, Ii ,and ii. would be a valid equilibcium configuration for
the structure under the applied load. However, before accepting the solution, the analysis
should also check that the last displacement correction is small relative to the total
incremental displacement. If the ratio is p a t e r than a prescribed percentage (e.g., l%), the
cornputer program performs another iteration by forming a new stifiess ma&, with a
value of tangent stifbess other than Ko, based on the updated configuration, Ui,and repeats
the two checks again. Both convergence checks must be satisfieâ before a solution is said to
have converged for that 1 0 4 incrernent
forces become nul1 or sufficiently close to null. Inctements should be small to ensure correct
computational efficiency. If the increments are too large. more iterations will be required. In
Newton's method, a large increment can prevent any solution fiom king obtained because
numerical modelling of structural response in linear and non-linear static and dynarnic
analysis. This computer program runs as a batch application to assemble a data deck bat
describes a problem so that it cm analyse the structure. A data deck for this computer
program contaùis mode1 data and history data. Model data defines a finiteelement model:
the elements. nodes, element properties, material definitions, nodal constraints, and any data
that specify the model itseIf. History data define what happens to the model, the sequence of
towers. Three different finite element rnodels weie explored for this analysis. Also,
different types of analyses were applied. h this section, element types used for different
models as weil as material modelling in both the elastic and pst-elastic loading stages are
presented. The results from the mudel presented herein were compared with results
obtained fiom testing five scale-mode1 guyed towers subjected to several loading cases
presented in Chapter N.
This mode1 represents a detailed and accurate modelling of the tower. Each
member (legs, diagonals and honzontals) forming the latticed mast was modelled as two
node three-dimensional truss elements with three degrees of freedom at each node.
Element T3D2 fiom ABAQUS library was chosen. This element had three degrees of
W o m at each node, namely the three displacements (ü1, U2, U3). The element used was
based on updated Lagrangian formulation and does not include higher-order non Iinear
t m o s . A plot of the nnite element mode1 of the mast for Model Tower-I is s h o w in Figure
3-3. The general p d c e for the detailing of these structures is that applied antenna loads
are concentric at the panel points and linle fixity is provideci at the co~ectionsbetween the
bracuig and the main tower legs. Thus it can be reasonably assumed that a three-
dimensional tms mode1 is applicable. Due to the large number of members in a guyed
tower, this mode1 would resuit in large number of degrees of fitedom, and in tum requires
(b) 3D-Beam M W
The legs of a guyed mast are continwus over the panel points and typically are
in ABAQUS library, was adopted to model the mast legs. The element had two nodes with
six degrees of fieedom at each node, three displacements (Ul, U2, U3) and three rotations
t u s ) elements. This model was chosen to study the eEect of the secondary moments on
members forces.
In this model, two-node bearn-column elements (B3 1) with six degrees at each
node were used to model the mast. One equivalent element was used for the entire mast
cross section. To account for any changes in the beam pmperties, an element has been
In this model it is important to fïnd the beam properties that closely define the
mast behaviour. General bearn elements that accouat for bending stiffness (EI), shear
rigidity (GA), torsional rigidity (GJ), and axial stiflhess (EA) were used. Therefore, a
shown in Figure 3-4, was created. The beam properties were calculated fiorn the anaiysis
1) A moment was applied at the top of the submodel (through tende and
compressive axiai forces applied at the opposite legs), and the deflection at the
top, A, was calculated. Since this mode1 represents a cantilever beam, the
A=- M L ~ (3-8)
2EI
where L is the height of the section, and M is the applied moment. Thus EI was
detemiined
2) A horizontal load, P, was applied at the top of the tower and the sub-rnodel is
anaiysed again for the top horizontal deflection. The calculated defiection was
used to detemine the shear rigidity (GA) h m the following formula using EI as
3) Torsional Moment,Mt, was applied and the torsional rigiâity GJ was estimated
To account for the eccentricity at the guy connection with the mast,MPC (multi-
point constraint) type, BEAM, was used. This option enforces a constraint between two
different nodes by introducing a rigid beam between these two points. This ensured full
interaction between the guys and the mast and kept the guy attachments eccenhic with
Each cable was discretized into several 3-Dcable elements. These elements have
the same formulation as the 3D ûuss elements with the exception that the cable material
used in these models, these elements were capable of handling applied loads even
perpendicular to their axial direction. A convergence test was applied to determine the
optimum number of elements, and it was noted that the overall behaviour was not
elements per cable, dependhg on the cable length, was adequate to mode1 the sag.
3.323 Boundary Conditions
Three typical arrangements at the base of the mast are show in Figures 4-3, 4 4 ,
and 4-5. In these support arrangements, dl three displacement components at the base were
resûained; however, only one rotational component, the twisthg angle about the vertical
axis, could be considered restrained. At the guy anchor points, dl three displacement
in order to ensure that the validity of the results, the materiai must be suficiently
For the elastic andysis, a purely elastic material option was used. In this case the
only input required was the elastic modulus of the material. In the dynamic analysis, mass
properties were usecl and a h damping properties were Uitroduced as determined for the
free vibration test perfomed on the model towers. However, with up-toîollapse anaiysis
the simple plasticity model, i r . perfect plasticity, was used In this model, the yield surface
Guy cable materials were modelled in the same manner as the mast material;
however, '"NO COMPRESSION" option was introduced which limits the material to
tension only. Also, in the material definition, the mass of cables was calculated based on
anaiysis of the beam model and to determine the mast properties. It also generated the
model for ABAQUS format including nodes, elernents, materials. boundary conditions, etc.
This prograrn allows the analysis of a wide range of towea that define a spectrum in
industry practice.
the computer program offers two approaches to obtain a convergent solution at minimum
cost Direct user control of increment size is one choice, whereby the user specifies the
incrementation scheme. This is particularly usehl in repetitive analysis where the user has a
very good feel of the problem. Automatic control is an aiternative choice: the user defines a
period of history (a '%tep"in the tenninology of the program) and at the same time specifies
certain tolerances or emr measures. The computer program then autornaticaily selects the
increments to mode1 the step. Genedy, the automatic control is more efficient thm the
scheme. In addition, automatic control is extremely valuable in cases where the tirne or load
This analysis is performed in iwo steps. In the first step the gravity loads resulting
fiom the self-weight of the tower, in addition to the forces due to the initiai pre-stressing
in the cables, are applied and equilibrium is achieved. Then, in the second step,
environmentai loads (wind and ice loads) are applied. In this step the load is applied
wananted. However, the nanual îkquencies of the tower determine the dynamic behaviour
of îhe structure. Although the naturai muencies are not used in a direct integration
analysis, yet it is used in the modal analysis, which have been used in the predication of the
behaviour of towers under dynamic wind loads (Vellozzi 1975). It was also recommended
that for the modal analysis under wind loads, the nanual frequencies of the structure be
determined in its displaced position due to mean wind speed. A fke-vibration analysis,
which is a simple eigenvalue extraction problem, is based upon the following generai
equation:
( - W 2 ~ ~ + s +~ K* U ) # =O (3-1 1)
is neglected by the computer program during the eigenvaiue extraction); K'' is the stiffiess
mairix; 4 is the eigenvector (the mode of vibration); a> is the fkquency value; and i and j
are degrees of fieedom. This is a linear perturbation problem and depends on the initial
conditions of the structure. Therefore, the stiffhess of the structure as determined at the end
of the previous analysis step is included Ui the eigenvalue extraction. This is important in
order to detect the natural frequencies of the structure under different displaced positions, or
due to various loadings resulting fiom wind or ice, or wind and ice combined.
eigenvaiue techaique to extract the kquencies of the current system and the correspondhg
mode shapes. The users need to specify the number of eigenvalues required. in the
computer program, the eigenvectors are normaiised so that the largest displacement entry in
each vector is unity. If the displacements are negligible, as in a torsional mode, the
eigenvectors are nomalised so that the largest rotation entry in each vector is UN@. In
addition to extracting the naturai fkquencies and mode shapes, the computer program
automatically calculates the participation fector (which indicates the strength of the motion
in the global x-, y-, and z-directions), and the effective mass (which if added for al1 modes
shouid sum to the total mass of the model). Thus, if the effective masses of the modes used
in the analysis add up to a value that is significantly less than the mociel's total mass, this
suggests that modes that have siBnificant participation in a certain direction have not been
extracted. For that purpose, Standards such as W C 1995) requires that at least 90%of the
participating mass of the structure must be accounted for in the modes used in a modal
analysis for each p~cipalhorizontal direction. For guyed towers, the number of mode
shqes required to get that mass participation ratio was found to be hi&, as discussed in
The d y s i s was performed in two steps. in the first step, the gravity loads
resulting fiom the self-weight o f the tower, in addition to the forces due to the initial pre-
stressing in the cables are applied and equilibrium is achieved. Also, in this step
additional loads resulting fiom ice or wind are applied if the response is required to be
investigated in these pre-set conditions. Then, in the second step, the naturai fiequency
analysis is performed.
analysis of structures in which inertia cffects are important. Direct integration of the system
must be used when non-linear dynamic response is king studied. In this method, the global
expensive and tixne-collsuming solution. This method was used to ver@ the analysis
against the m d dynamic response for the tested tower models. The method used by
the program is an extension of the trapemidal rule. Also, a numerical damping parameter,
a,needs to be defined. This is purely numenc and it varies for O to a maximum level of 6%
when the time incmnent is 40% of the period of oscillation. Therefore, this artincial
damping is never ~bstantialfor this application. T i e increments used in this analysis
The applied loads in this step were d e t e d e d for the measured base motion
accelerometer mounted on the shake table for the mode1 towers tested. This motion was
meanued at a rate of 400 rradings per second and read by the data acquisition system and
then transferred into an ASCII file that is read by the ABAQUS andysis. The AMPLITUDE
option in the cornputer program allows arbitrary time variations of load given throughout
the step. The default of this option is to give the magnitude of the force as a multiple
(hction) of the reference magnitude given on the data liM of the loading. For prototype
towers, actual ground acceleration histories were used for the analysis. The output of the
dynamic analysis includes the dynamic response of the structure due to the prescribed load-
time history in the fom of the shauiing action-tirne histories. These straining actions
This analysis was also perfomwd in two steps: fïrst under static loads of initial
tensions, and the second under dynamic transient loads.
Up-to-collapse analysis was performed to determine the ultimate behaviour and
also the uitimate load capacity of guyed towers. This d y s i s is of particular significance
especially today were most Standards have adopted Limit States Design, which requires
that towers be analysed under loading conditions that are very close to failure levels.
modified Rücs method (HKS 1995). with automatic incrementation for solving such cases.
By using this method, the cornputer program prints out a load proportionality factor, h, at
each increment during the step, then the current magnitude for the load component, Pm,can
be defined as:
where PO is the magnitude of the load component at the start of the step; Plcf is the
magnitude of this load component as dehed by the user for the step. For dead load, Prcf =
Po, so that the load magnitude remains constant. The Riks algorithm attempts to step dong
the equilibrium path (the load-displacement cwe) by prescribing the path length dong the
curve to be îraversed in each increment. This means that the load magnitude is determined
as part of the solution and the user must oniy specify the fmishing conditions of the step.
This can be done through defïning either a maximum vaiue of h. beyond which the solution
value. If no finishing condition is specified, the analysis will continue up to the number of
increments specifieâ by the user at the beginaing of the analysis in the non-linear step.
In this anaiysis, the nist step is for initial conditions and the second step is used to
determine the dtimate load as a ratio of the applied loads under service load conditions.
A convergence study was conducted to choose the finite-element mesh. Pilot runs
were used to detemine the number of cable elements used for each guy. As for the tnrss
model, it would not be of any additional value to break the elements any M e r . For
bearn models, a k a m element was used for every panel (which is a very refined mesh).
However, it was verified that a beam element for every section would result in the same
level of accuracy as the refmed model provided that there is no change in material
The two fuiite-element models were verified and substantiated by results nom
testing five tower models as described in Chapter IV. Later, a wide range of guyed towers
was examinai as described in Chapter VI. Appendix Al shows a list of the fnite-element
Displacement
4.1 Introduction
This chapter is concemed with the testing set-up, construction of models, materials
and procedures used for the static, fke vibration, forced vibration, and ultimate load tests of
five scale-mode1 towers. in order to conduct this experimental investigation, a new testing
facility was designed and buiit by the author. The sizes of the scaled guyed tower models,
mnge of accelerations, guy forces, displacements were first estimated so that the range and
sensitivity of the sensors would be of the same order. A shake table was buiit and
undertaken. The main objectives of the experimental sndy were: (i) to establish accurate
experimental data to venfy and substantiate the structural response predicted by the
analytical modelling of guyed towcrs; (ii) to obtain data related to the dynamic response of
guyed towers subjected to ground excitations and to ver@ the numerical technique for
dynamic d y s i s ; (iii) to obtain data related to the k-vibration response of such towers;
and (iv) to examine the non-linear up-to-failure response of this type of tower. in order to
achieve these goals and obtain the most meaningfid experimental da@ five 1/20 scale
models of towers were built. This scale was chosen with due consideration to the space
available in the Laboratory, the size of the shake table, and the guy force levels as discussed
The experimental program was c d e d out on five 1/20 linear-scale guyed tower
dictated that the model study should be of the bbdùecttype". In a "direct" model test, the
model is built such that al1 its properties are as similar as possible to those of the prototype
but on a reduced d e . Although it was not intended in this study to model a particular
design and develop a o n e - t ~ n ecomspondence between the model and the prototype
response, it was possible to closely simulate the construction of typical welded prototype
guyed towers. The deteminhg factor used in the choice of model materials was the
practicai aspects involved in fabrication, instrumentation, and testing technique. Mso, a
small scale bad to be used to accomodate the model within the laboratory space; this made it
dificuit with respect to the choice of sizes of sections and materials available.
The five models had the same mast configuration. However, the parameters changed
in the models were tower heights, span between guys, number of guy levels. and the guying
system. The guy anchor radius was kept constant, thus the ratio between the guy d u s and
The towers tested had a ûiangular cross-section made fiom welded steel rods. The
legs of the tower mast were made of 3.1 mm (118 in) solid round bars and the bracing
system was made of 1.S mm (1/16 in) round bars. The mast had constant properties i.e., face
width, bracing pattern, panel heights, and member sizes. The face width was consaicted as
30.5 mm (1.2 in) and the panels had a constant height of 38 mm (1.5 in). Figure 4- 1 shows
the typical mast of model towers. Figure 4-2 shows a section of an AM radiator tower, in
which the structure is utilised as the radiating antema. These two figures show the
Bases of typical modem guyed towers are aaiculated to allow rotations in the
bending planes. Figure 4-3 shows one of the methods used to provide such articulation by
bringing the tower legs into one point and thus reducing the moment resistance of the mast
base. Figure 4-4 shows a more elaborate and expensive detaii where the tower base is M y
articulated through the concave and convex plates. This detail is normally used on taller
and heavier towers. Figure 4-5 shows a mast base in which the mast legs are connected
through a base of heavy beams (star base) bearing on a smaller cylinder. This detail is
typically used for middle range heights of towers and a similar detail has been adopted for
tower models as shown in Figure 4-6. The mast legs are connected to the upper plates and a
solid cylinder is in-between the plates, allowing rotations about the horizontal axes. The
fiction on the cylinder, resulting fiom the self weight of the mast and the additional axial
loads fiom the prestressed guys, provides partial resistance to rotation about the vertical
axis.
Three guy levels supported the mast and each level had three guys. The tower had a
total height of 2204 mm and a guy radius (distance tiom the centre of the tower base to the
anchor point) of 1400 mm.The guy levels were connected to the mast legs at 570 mm, 1330
mm, and 2090 mm elevations. A profile of tower model4 is shown in Fig. 4-7. The guys at
the top guy level had a diameter of 0.5 mm (0.018 in) while the lower two levels had a guy
This tower model had a total height of 2432 mm (96 in) and supported by three guy
levels of the same sizes as model I. The guy levels were comected to the mast at 570 mm,
1292 mm, and 2280 mm elevations, thus leaving a top cantilever of 152 mm (6 in). A
@ile of model-II is show in Figure 4-8.
This model was coasmicted to study the effect of torsion-resistoa, which are anns
made to extend the tower face. Figure 4-9 shows a typical torsion resistor instaiîed on a
ûiarigular tower. Figure 4-10 shows the actual torsion resistor coastructed on tower Model-
m. The tower was identical to Model 11 in al1 other aspects with the exception of the torsion
resistor that was used on the top guy level. Figure 4-1 1 shows a piciure of the model as
This tower model had a total height of 3048 mm (1 20 in) and mpported by four guy
levels at the elevations shown in tower profile (Fig.4-13). The top two guy levels were
made of 0.53 mm (0.021 in) size guys and 0.45 mm (0.018 in)guys were used for the lower
two guys. The top cantilever was 229 m (9 in). A photograph of tower model-IV is shown
in Figure 4-1 4.
This is the tallest tower rnodel. It had a total height of 3658 mm (144 in) and
supporteci by four guy levels at the elevations shown in the tower profile (Fig.4-15). The
laboratory ceiling as weli as the dimensions of the shake table limited the height. On the
top two guy levels 0.61 mm (0.027 in) guys were used and 0.51 mm (0.021 in) guys were
used for the other three levels. A photograph of the tallest tower tested ( d e l - V ) is shown
4.4 Materials
Material, sirnilar to the prototypes, were used to fabncate the tower models. The
mast steel was solid round bars that were cut to size. The bars were of steel grade CSA
The guys were made of corrosion resistant Type 302 stainless steel wire ropes
(aircraft cables). Al1 the guys used were 1x7 single strand construction. The ropes were
preformed by the manufacturer, to remove intemal stresses and provide easier handling
while minimizing hying. The ropes were dso pre-stretched to provide uniformity in
strength and elongation properties. The ultirnate breaking strengths from the
manufacturer's published data were used in the analysis; however, tests were made to
coafum these numben and also to detemine the equivalent modulus of elasticity for the
cables. The tested modulus of elasticity for the guys was 165 000 MPa.
complex structure such as a ta11 guyed tower within the space and performance limitations
of a shake table can prove clifficuit. Some of these difficulties are discussed in the design of
the shaking table (Section 4 3 . However, the model as constnicted and describeci above
would have a h e m scale 1/20 - 1/30 of typical towers. Thus the prototype towers would
range in heights h m 45 m (150 A) to 110 m (360 ft). The member sizes for the mast
components would range fiom 3 1 mm (1. î S in) to 95 mm (3.75 in) solid round bars. The
guy sizes would range h m 6 mm (0.25 in) to 18 mm (0.75 in). Considering the above, the
medium-height guyed towers used for wireless applications. Taller towers with heights up
to 600 m used prllnarily for broadcast purposes would be even far more difficult to model.
However, one of the purposes of this experirnental program is to establish confidence in the
analytical procedures, which in tum cm be used to analyse taller structures that are difficult
The steel rods for the bracings (horizontals and diagonals) were cut into length. The
main legs were cut into long pieces, and each piece spans the total tower height. The panels
were marked on the legs and a form mis made to hold the legs and the bracing as they were
welded togethet. This helped to d u c e the deformations and twist in the tower due to the
heat h m the welding. However, it was not feasible to entkly eliminate any twist in the
tower construction and a close look at Figure 4-17 show signs of twist in the towea.
The tower was then comeeted to the top plate of the base and held vertically while
adjusting the tensions in the three guys simultaneously with the aid of the length adjustment
tools shown in Figure 4-18. The tensions were monitored through the data acquisition
system and bmught up to the initial tension levels which are nomally specified as 1O- 15%
of the ultirnate guy tension.
There were severai challenges that had to be overcome in order to design and
construct a vibration excitation system that cm be used for light structures in gened. Some
1- the size of the structure: Guyed towers are very taU, theu heights may vary anywhere
fiom 30-600 m. Considering that the prefened economical anchor radius for a typical
guyed tower is about 0.6-0.7 of the height of the specific tower (and c m be as small as 0.3),
a relatively large shake table is required to model the tower with the anchonng guys.
2- the natural fiequencies of the towers: The structure is flexible, lightweight, and has
relatively low natural ûequencies. Therefore, the shake table design involved selecting an
optimal stiflhess-to-mass ratio while avoiding frequency interference between the shake
3- memkr sizes: Guyed towers are usually made of a d o m face width and due to the
lateral support of the guys, smail members are used for the legs and the bracing of such
towers. For example, a typical 100 rn guyed tower may have soüd round legs of 51 mm (2
in) diameter. In order to use a (l/20) sale model of such tower, a mast of 5 m height wodd
be nquired with an anchor radius of 3.0 m using legs of oniy 2.5 mm diameter which is
Thus, the shake table had to be large enough to accommodate the anchorhg points
of the guys and to enable the use of reasonable size scaie models, rigid enough in order to
minimise the interference effects from a vibrating mode1 on the table, and with minimum
The similitude laws of a dynarnic model constnicted fiom the same material as the
prototype dictate the scaling factors for the different physicai quantities of the model. The
length scale Si ,which is the ratio between the length of the model to that of the prototype
determines the dimensions, displacements, weight, and the naturai fiequency of the model
(Sabnis et al. 1983). Therefore, the resdting kquency of the scaled model is (s$' times
the natumi h q e n c y of the prototype. Thus for a typical guyed tower with a natural
fiequency of 1.0 Hz. the resdting 1/20 d e model of mch tower would have a fkquency
of 20 H z This requUes that the shake table be able to perform with frequencies up to 30
Hz
4.7.2 Configuration
Therefore, the maximum anchor radius of the guys that can fit within the shake table (1.4
m) and the ceiling of the Structures Laboratory limited the height of the test towes to 3.6 m.
In order to simplify the table construction, the choice of the supporting material for the
table was limited to either spring supports or to mller bearings. Nine spring supports were
originally used at the support points: however. that proved to be inpracticai as the horizontal
prevent it fiom snapping out of position. The final layout of the table is shown in Fig. 4-19
while the details of its constmction are shown in Fig. 4-20. As shown in Fig. 4-20, the table
was made of a sandwich plate consisting of two 6.5 mm (114 in) steel plates bonded to a
relatively thick 220 mm styrofoarn (plystyrene) core. The two plates were connected
together by strips of steel plates welded dong the perimeter of the table and through bolts
that tie the two plates every 500 mm (20 in). The use of a sandwich plate allowed for high
rigidity while minimishg the rnoving mass thus allowing the use of an available 90 kN (20
The shake table is supported at nine points by low Wction roller bearings to
minimise distortion of the desired table response. The b e a ~ g sare mounted on three rails
*
allowing a maximum travel of 75 mm (3.0 in). Each rail is connected to a W360x45 steel
A finite element mode1 of the table was nrst used to predict the naturai fkquencies
of the system, the required dnving force of the actuator, and the performance of the table
before its fabrication. in order to ensule that the table wouid perfonn for frequencies up to
30 Hz,the first bending fkquency of the table must be higher than 30 Hz Figure 4-21
shows the finite element mode1 used for the analysis of the shake table and its hdamentai
The performance of the table was measured through different sinusoidal waves for
the required range of fkquencies. Three accelerometers mounted on the table at the
opposite end to the actuator were used to verify the performance of the table. The purpose
of these accelerometers was to measure the motion of the table to ensure that accelerations
at the different points are uniform with minhum distortion. The readings of these
accelerometers were compared to the expected theoretical waves. Acceptable perfomiance
of the table was verified for fkquencies up to 35 Hz. Figure 4-22 shows the reading fiom
4.8 Instrumentation
Mechanicd dial gauges havhg 0.01 mm mvel sensitivity were used to measure the
tower deflections at the mid-span between guy leveis and at the top two guy attachments
levels of the tower models. The did gauge readings were manualiy taken at each hcrement
of the static load throughout the test procedure. Figure 4-23 shows the dia1 gauges mounted
five locations dong the height of the towers. Figure 4-1 1 shows the locations of the LVDTs
for the tested tower model III. Mount Details of LVDTs and accelerometers are shown in
Fig. 4-24. The plate attached to the tower leg was only to provide contact surface for the
LVDT,
4.8.3 Accelerometers
locations dong the tower height. Five measured the accelerations in the direction of the
motion and the sixth m e d the acceleration in the perpendicular direction. Aiso, a
seventh accelerometer was mounted on the shake table at the tower base to measure the
input motion to the tower. This base accelerometer is shown in Fig. 4-25. The
circuit, with a meamhg range of IO. 1 to *50 g's, a fkquency response up to 1400 Hz. One
of the important characteristics of these accelerometers was theV lightweight, which made it
possible to meamre the acceleratiom without distorthg the mass-stiffhess properties of the
system. Each accelerometer had a mass of ody 13 grams. A view of the accelerometer
The locations of the instruments used on each of the tower model are shown on their
A Gilmore 20-kip (89 kN)model 433-20 hydraulic actuator was used as the âriving
force for the forced vibration tests of the tower modeis. One end of the actuator was fixed
to the testing h e while the other end was bolted to the shake table (Fig. 4-26). The
hydraulic actuator system consisted of the actuator, which included a position transducer,
servo valve, and an intemal LVDT. A 76 litres per minute hydraulic pump powered the
system and the signal wes controlled through a digital controller manufactured by
Interlaken The servo-amplifier controller controlled the servo-valve of the acniator, the
rate of flow of oil into the acniator, the static set point, gain and damping. This controller
was also capable of using a predefhed set of hamonic displacements or a userdefined set
of displacement hctiom. It was decided that the input fùnctiom would be displacement
but the ndting accelerations at the base of the tower would be measured for input
accuracy into the numerid model. Figure 4-27 shows the acnuitor controller computer to
the left and the data acquisition system to the right.
Special load cells were custom-made for this testing. These consisted of aluminium
rings with two strain gauges fixed on the inside and outside of the ring to form a half-bridge
configuration. Calibrated weights were used to calibrate the readhg from the load cells and
these were used with an amplification of 2.0 on the data acquisition system to convert the
strain readings into forces in Newtons. Figure 4-28 shows the four load cells at one of the
anchor points and each ~g is connected to one of the cables. These load cells had a
maximum capacity of 400 N and were used to continuously monitor the guy forces through
During both the static and dynamic tests, the data h m the sensors (LVDTs and
accelerators) were captured by a test control softwarr (TCS) using a MEGADAC 3000
Senes data acquisition unit. The test control software, TCS, is a powerful tool developed
especially for acquiring, reducing, and anaiyzing the dynamic analog data captured using the
MEGADAC. It simplifies the process for collecting, converting, monitoring, plotting, and
rwiewing the MEGADAC test data For dynamic tests, the TCS was adjusted to sarnple the
data at a rate of 400 readings pet second per sensot*As for the static tests, readings were
oniy taken at each load increment The MEGADAC 3000 Series is designed to measure at
rates up to 25,000 samples per second. The OPTIM's Differential Input Modules used with
MEGADAC accept signais directly h m vimially al1 active and passive sensors. In the h t
model tower, the filtet used in the module captured fnquencies up to 250 Hz, while for the
Each tower model was guyed through adjustable length devices (tumbuckles) and
the load cells monitored the forces in the guys. It was possible to obtain useful data with the
simplest set of instrumentation, described in Section 4.9, when appropriately deployed. The
scope of the measurement program was decided at the beginning. Deflections, accelerations
dong various points of the mast, and the forces in the guys were measured. Stresses in the
mast codd be derived nom the numerical models if it agreed with the measured deflections
secured to the main testing fisune and the accelerometersmounted on the moving mast. The
test procedure comprised of static, free vibration, forced vibration, and up-to-collapse tests
During this stage of loadllig, the model towers were tested elasticaiiy under various
loading conditions representing the applied horizontai wind loads. These static
concentrated load tests were important to confinn the existing anaiysis tools commonly
used in the industry to design these structures. Also, it was important to evaluate the finite
element model used. The loads were appBed through horizontal cables attacheci to the mast
at various points dong the height; each cable was connected to a weight pan over a pulley
that is attached to the test fiame. Figure 4-30 shows these cables with the weight pans
attached to them. The mcdels were tested in the elastic range while deflections and guy
forces were measund. Three -tic loading conditions were applied: (1) one concentrated
load at the top; (2) two concenûated loads applied at the top two loading points; (3)
The loads were applied through standard caiibrated steel weights placed in the
The fixe-vibration tests on the five tower models were perfomed by first mounting
the accelemmeters at the different elevations on the mast. Due to the flexibility and
relatively light weight nature of the structure, it was feasible to caphue at least the nrst five
modes by either appiying small impact loads (hammer test) ot by imposing initiai
displacements that are suddenly released. These loads or displacements were applied at the
top of the mast. in either case the structure was fke to vibrate and the accelerations were
caphired at a rate of 400 readings per second. The test control software (TCS) using the
MEGADAC 3000 &ta acquisition unit capRued the data h m the sensors. The output fiom
the test cases was then fed into a data analysis and display software, DADiSP, @SP
Development Co. 1991) for a fast Fourier transfomi (FFT). The FFT analyzer gave the
spectrum respome of the towers in the fiequency domain from which the naturai
tiequencies of the models were deduced. The DADiSP software also yielded the
magnitudes and phase angles for the mode shape. The naturai fkequencies of the structure
were determined under two conditions (i) initial conditions (undisplaced shape) and (ii)
under a set of horizontal loads resultîng in a displaced condition. This second loading
condition simdated the condition of the structure under static wind or mean wind speed.
During this stage of testing, two forced vibrations test were canied out:
1. Sweepsine wave test: This was perfomd by conducting a set of tests that varied the
displacement magnitude was kept constant while the frequency was varied. The data
acquisition system continuously ncorded and the tests were visually observed. During
programmed to the hydraulic actuator and the towers were tested under various
intensities of these displacements. The ground motions were obtained fkom the data
published by Iwan (1997) for Landen Earthquake with a maximum velocity direction
(N80W) and for NorthRdge Earthquake N-S direction, and Nanbu Earthquake;
maximum velocity direction (N49W). The displacement histories were used as the
input to the actuator's controller. It should be noted that these displacements did not
closely simulate the above mentioned earthquakes as only twenty discrete input
fiinctions could be used. However, the resulting applied mast base accelerations were
recorded by the accelerometer mounted on the table and fed to the analysis model as the
The primary purpose of these tests was to test the analytical procedures in the prediction
of mast displacements and accelerations under ground motion or any other applied transient
loads.
Finally, each model was tested to failure using a simulated static wind load that
varied with the height simüar to the CSA S37-94 (Canadian Standards Association 1994)
wind profile. Six horizontal loading points were used as show in mode1 tower profiles
(Figures 4-7'4-8,442,443, and 4-15). For al1 models tested, the load was applied in equal
increments. Mer each increment, the load was maintained constant while mast deflections,
and guy forces were recorded. The loads were applied until collapse resulting either fiom
broken cables or fiom mast failms. Figure 4-3 1 shows tower mode1 V at failure.
Table 4-1 Details of Experimental Towers
No of Guy
Mode1 Tower Height(m) Comrnents
Levels
I 2.2 1 3
II 2.44 3
A
Cross Section
Cross Section
80
4 Al- Aï
Cross Section
81
Figure 4-16. Mode1 Tower V
82
Figure 4-17. inherent Twist as a Result of the Manufacturing of the Models
Figure 4-18. Guy Tension Adjusters
1 MRECTKJN OF
MOTION
85
6 mm STEEL PLATE 300 mm STYROFOAM
///A I
W SEC
95
Figure 4-30. Static Load Application Set-up
Figure 4-3 1. Mode1 Tower V at Collapse
CHAPTER V
5.1 Introduction
The results of the static and dynamic tests obtained fiom the five mode1 towers
described in Chapter IV are discussed in this chapter. The main objective of the experimental
program was to evduate and demonstrate the accuracy of the various nm&cal anaîyticai
produres used for the anaiysis of guyed towers. In doing so, the data fiom the tests were
also usefiil to study the efféçt of height on the behaviour of guyed communication towers, to
study the influence ofaumber of guy levels torsion resistors,and base supports on the dynamic
In this chapter, the experimental and theoretical results for the five tower models are
presented and compareci. The r d t s presented include deflections, guy forces, and failure
loads of the tower modds under static horizontal loads. A h , tbe r d t s presented herein
indude the deflections and acceleration histories under severai dymmic loading conditions.
Furthemore, the naturai fiequenues, mode shapes, and damping characteristics of the tower
models are presented hereia Because of the large amount of data coliected, for brwity, only
In order to study the elastic response of guyed towers due to static loads, five guyed
model towers were tested aud d y s e d using the dinerent numerid techniques. Concentrated
horizontal loads were applied at several locations dong the tower height wMe deflections and
guy forces were measured. These horizontal loads simulate wind loads on the structure.
Figures 4-7, 4-8, 4-12, 4-13, and 4-15 of the tower profiles show the locations of the load
applications points and the measured defledon points for tower models I, II, IiI, IV, and V
respectively.
In this load case, the model towers 1to IV were loaded at the top (Ioadtog point L1) of
the towers. Results in Figures 5-1, 5-3, 5-5, and 5-7 show the calculateci and the measured
deflections dong the heigbt of the tower under a concentrated load of45 N (10 Ib) for model
towers 1to [V rrspedively. Note that this load on the model towers represents a concentrated
load of 17.8 kN (4.0 kips) on a 20:l prototype structure. Figure 5-9 shows the redts for
model tower V under thra concentrated loads of 26.7 N (6 Ib) each applied at the top three
loadhg points (L1 to L3 in Fig. 4- 15). Tables 5-1,5-3,505, 5-7, and 5-9 compare the different
a d y t i d models with the measured cable tensions for this load case. In ail cases, very good
agreement between the experhental and the Werent theoretical models can be observed.
Also, it can be noticed that the results for the two finite element methods show here produce
virtually identical resuits in terms of deflections and guy forces. The beam on non-hear elastic
supports adysis results are in general agreement to the two finite element numencal modeh.
Cornparhg the deflectiom of tower modds 1 and II, both towers have identical mast
configuration and the same guy sizes and initial tensions but the oniy ditfierence is that tower II
is 10% taiier than tower I and the deflections under the same loads are 35% more for tower II
(Figs. 5-1 and 5-3). In cornparison, deflections of tower üX, (which is identical to towers 1 and
II with the exception of the torsion resistor at the top level) are less than half of those of towers
1 and TI (compare Fig. 5-5 with Fig. 5-1 and 5-3).
Mts in Figs. 5-2, 54, 5 6 , 5-8, and 5-10 dustrate the d d a t e d and measured
deflections at the deflection measwing points dong the tower. in this load case, concentrateci
loads of 26.7 N (6 Ib) were applied at each of the loadhg points show on the tower profiles.
Tables 5-2, 5-4, 5-6, 5-8, and 5-10 compare the mea~u~ed
guy tensions with the caiculated
tensions f?om the three anafysis techniques. In a l cases, good correlation between the
experimental and the theoreticai results can be observeci. However, it should be wted that the
experimental deflections are generaiiy higher than the theoretical deflection by an average of
6%. Also, it can be notiad that the r d t s nom both Wte element procedures are nearly
identical while the beam on non-linear springs anaiysis r d t s are comparable but not with the
same lwel of close agreement as that noticed for load case 1. Furthemore, it can be noticed
that the defledoas for mode1 tower Vary by up to 1% kom both the FE techniques and the
beam on s p ~ analysis.
g in this loading case, the cables are stress& up to 500/0 of tfieir uitimate
From an engineering point of view, the experirnental redts are in good agreement
with the analytical models. Any of the three analysis techniques can be saiisfactorily used to
determine the static behaviour of the structures at these load levels. Furthermore, modeiling
the mast as an equivalent beam, whose properties are based on various submodels of the truss
mast, give r d t s that are nearly identicai to that ofthe 3-dimensional ûuss analysis.
In order to detennine the natuml âequencies of the tested towers, severai impact loads
were applied to the tower models, and then the towers were lefl to vibrate fkely. The towers
Before analyshg the results, the adequacy and quality of the data were verified by visual
examination of the aderation titane histories on the monitor. Figures 5-1 1, 5- 13, 5- 15, 5- 17,
and 5-19 show typical acceieration t h e histones for tower models I, 4 IV, and V,
respectivdy. The extraction of usefui infomtion invaxiably requires the computation ushg a
Fourier T d o r m . Therefore, the data captured fiom the sensors were fed into a Fast Fourier
Traosform adyser to give the response of the tower in a ne<ruency domah instead of a
tirne domain. Figures 5-12, 5-14, 5-16, 5-1 8, and 5-20 show the fiequency spectra for the
acceleration the histories shown in Figures 5- 1 1, 5-13, 5-1 5,s-17 and 5-19 for the f i e tower
models. The naturai fiequencies of the towers were determineci by the examiaation of the
frequency spectra rdting fiom the acceleration time histories captured from the five
accelerometersrnounted on the tower. Tables 5-1 1 to 5-1 5 summarise the natural fiequencies
and the mode shapes of the five tower rnodels. Good correlation between the experirnental and
theoreticai hdings can be observed. A h , it should be noted that both finite element analyticai
For the shortest tower, Mode1 Tower I, the computed &st natural fiequency of
vibration using the finiteelement andysis was 15.6 Hz which is quite close to the fiequency of
16.5 Hz rdting from the experimental data coilected (Wahba et al. 1996). Also, the
correspondîngmode of vibration was torsional mode. Furthemore, it should be noted that this
first tower mode1 was @al, as the base support did not provide any restrain agaiast twist of
the rnast. A diierent tower base design was later used for the other towers as shown in Fig. 4-
26. A numerid analysis used to check the effect of the support conditions and restrallriag the
base against twist d t e d in a first flexml mode shape with a natural fkpency of 25.8 Hz
A h , it can be noticed nom Tabie 5-1 1 that the tirst four mast modes wae exgted fiom
dinerent impact tests, and the modes were identi6ied on the &eqpency spectni. The fiequency
spectrum was usefùi in predicting the mainil fiequencies; however the mode shapes were
predicted from the sine-sweep tests and with the help of the numerid analysis as explained in
the sequel.
The experhental resuits shown in Table 5-1 2 show that the value of the fùndarnental
Iiequency for model tower [Iis 22 Hz measured versus 22.3 tiom the FE bearn model and 22.7
Eom the tniss model. This 15% reduction in the natural tieqyency compared to model tower
I., is a direct result of the change in height as this is the oniy parameter that was changed
between models II and 1. A h , it can be noticeâ that the natural fiequency of the fint torsional
mode was much higher and it came in order after the second bending mode. This is a direct
result of the fwty of the base against twist. This f i t y is a r d t of the ûiction on the pin
c o h g fiom the weight of the structure and the pre-stressing force from the initial tensions of
the guys.
Table 5-13 shows the r d t s for model tower JII, and masistent results between the
numerical FE models and the experirnental r d t s are evident; however, wmparing the
fundamental natuml 6equency of model tower KiI with that of model tower II, there is a 15%
increase. This increase may be attriiuted to the increase in sMbess of the top guy level as six
guys of the same slle were used instead of only three used on tower II. Aiso, a sign<ôcant
increase in the torsional mode tkequency can be noticed (the torsionai mode frequency is more
than two times that of model II), which is a direct result of the presence of a torsion-resistoron
Resuits for model tower IV are shown in Table 5-14. h addition to the agreement
between the theoretical and experimental resdts, it can be also noticed that the lowea mast
bencihg fiequency is higher than that of tower LI and comparable to that of tower 1. Although
this tower is 33% tder than that of tower l,the mast spans between guy leveh are comparable-
This dong with the increase in guy stühess due to the use of larger guy sizes on tower N,
Table 5-15 shows the results of the naniral fkequencies for model Tower V. This
tower was chosen to show the typical mode shapes of the test specimens. Figure 5-21 shows
the fundamental mode shape of the tower, which involves mast and guy motion. Figure 5-22
shows the est f l d mode shape of the mast and it is the seventh mode. The second flexural
mode of the mast is shown in Fig. 5-23 whiie the torsionai mode is shown in Fig. 5-24. Figure
5-25 shows a typical guy mode for the lowest set of guys. It aui be noticPd that this mode
does not include any mast motion. Cornparkg the dominant fieqyaicies with those of model
tower IV, it cm be n o t i d that the fiequency of the fùndarnental rnast mode shape is 22%
lower than that of tower IV. Thme two towers share the same mast properties, number of guy
The objective ofconducthg the sweep-sine wave tests on the five tower models was to
obtain the vibration signature of the acceleration tirne histories at different locations on each
tower. The forcing fùnction tbat was applied to the sbake table, and consequently to the tower
base and anchors, was a sinusoidal wave as shown in Fig. 5-26. The towers were subjected to
the same hction over a range of âequencies mging from 10 to 32 Hz A typical set of
results for one ofthese tests is s h o w in Figures 5-27 and 5-28. The displacement histories and
the fiequency spectra are show for the six accelerometers that are mounted on the tower, and
the seventh, which is mounted on the shake table. The set of results presented in these two
figures is for Tower IV under a forcing sinusoidal wave of 22 Hz. These tests were used to
help iden* the mode Sapes of the tower models after detennining their naturd frequencies.
The aeady state vibration at the first few modes of vibration was measured at the six locations
dong the entire beight of each tower. The identification of the associated mode shape required
the determination of the phase between the signal fiom the six measured locations Simple
addition and subtraction of signais and obsenritlg how the amplitude of the Fourier Transfonn
peaks changed could achieve this, as peaks of signais in-phase wodd krease and those of out
of phase wodd decrease. However, this method is only applicable if the real part of the
fiequency components are either in-phase or outsf-phase. Also, studying the magnitude of the
response of the different semn in the âequency domaui helped in identifying the dominant
mode dupe at that location. Tbis uui be noticed fiom Fig. 5-27, where for accelerometer #3
(A3 in Fig. 4-13), higber fr#luencies d t e d in a kger magnitude than the fiindamentai
fiequency. This is opposite to accelerometer #1 (Al in Fig. 4-13) where the dominant
vibration mode is the fiuidamental fiequency. It should be noted that these tests also helped to
CO* the predicted oahiral fiequencies, as the viirations of the modeis at resonance
fiequencies where observed to be very vigorous and, in s e v d cases, this was strong enough
The determination of the mode shap is a very difficuit ta& to extract theoretidy and
even more so experimentally. This is due to (i) close spacing of the Merent nequencies, (ii)
the interaction between the guy modes and the mast modes, (üi) the large number of guy
modes that corne in with the free vibration analysis, and (iv) very small deflections of the mast
as comparecl to the guys (shown on the normalisai mode shape diagrams). For example, Fig.
5-21 shows the first mode shape of model tower V, from which it is clear that this mode is an
interaction between the top guy level and the mast's fjrst bending mode; however, the second
mode shape, shown in Fig. 5-22, is pureiy the first mast bending mode. In order to identify the
mast mode shapes without the Uiterence fiom the Merent guy modes, a model consisting of
one element for each guy was constructeci. This suppressed the guy mode shapes only. Using
this analysis, it was usetiil to go back to the tiill model and pick the rnast modes that would
have similar frequencies to those predicted fiom the singie elernent guy models. Also, the
information regarding the effective mass associateci with each mode helped in establishg the
The model towers were tested under different sets of transient forced vibration
fûnctions. This dernoristrates the use of the analytical procedure in a seismic analysis,
where the time history of acceleration at the base and the anchor points of the tower give
the forcing fùnction. As explained earlier in Chapter IV, the forcing functions were taken
From three earthquake displacement time histories that are fed to the shake table-forcing
actuator. The acceleration time history at the tower base is measured from accelerometer
No. 7, which is mounted on the table at the base of the structure (Figure 4-25). Figures 5-
29 and 5-30 show typical r e d t s for one of these tests. The results show in these figures
are for model tower IV under a simulation of the Northridge earthquake scaled to a
Al in Figure 5-29) are about 23 g, which is close to five times the maximum input at the
Figures 5-31 and 5-32 show another set of results for an acceleration history
the base acceleration time history (Figure 5-33). Figures 5-34 and 5-35 compare the
damping had to be introduced to the model and detemined by calibration of results to the
measured response.
The concept of structural damping as a fiaction of the critical damping associated
motion of the system are integrated directly, and the fiequencies are aot part of the
solution. Furthemore, the natural fiequencies of the system are constantly changing
practical choice of damping mode[. From the analysis of the experùnental results under
proportional Rayleigh damping P = SE-04 were evaluated from the caiibration of the
The displacement response for the acceleration history shown in Figure 5-33 is
shown in Figures 5-36 and 5-37. The good agreement between the measured and the
Findy, each modd was tested to fdure to examine its non-linear response upto-
collapse. In the ôrst three modd towers, five horizontal wncentrated loads were applied dong
the height of the tower pulling it towards one of the anchors. In the remahhg two towers, six
concentrated loads wen used. For al1 the rnodels, the load was incremented at a constant rate.
Afier eech incremeat, the load was maintained constant whiie recording deflections and guy
tensions. After Wure of each tower modei, the appüed load was released slowly. A towa
model was considerd to have M e d d e r when one or more of the cables fàüed (and
wnsequently, the mast) or when the m a t Med to a point that it could not cany any m e r
loads.
For tower I, equd loads were applied at each of the fbe loadhg points. These loads
were increased just above 36 N at each of the loading points. At that point, the second guy
lwel cable ruptureci; immediately der, the lowest guy lewl also broke and the mast became
unstable because ofthe shear at the base and the tower fded. The second guy level f?om the
top broke at a tension of 82 N (181b). This was a sudden and unexpected failure as this load
was about 25% lower than the pubiished breakhg strength of 112 N for the cable. This
reduction in the cable strength can be attriiuted to the method of guy termination (using a
closed loop for the guy t e e t i o n typically reduces the load capacity by 25%). Figure 5-38
compares the load applied at each of the loading points versus the defleztion at sensor Dl.
Also, the analytid resuits are wmpared to the expetimental data. From this cornparison, it
can be noticed that both arialyticai models are in agreement with the experimental r d t s .
Also, both fiaite element models r d t s were the same. The b m n - ~ n - ~ model
p ~ g ~r d t s are
consistent with the FE moâeb up to the guy t'ail~e-Howwer, the bearn-on-springs analysis
fdeû to predict the dtimate cap- of the structure and the discrepuicy between its results
and that âom the FE analysis inmeases as the load approaches the faim level as ihstrated in
range and presented by load case II for this model. Later, the tower mast was salvaged and
modifieci by adding the tomkm resïstor to it. A new set of guy cables was used and the tower
ln loading tower iil, the horizontal loads were increased with the height of the
structure to simulate the increase in wind pressure with height. The loads applied to the
loading points were incceased with the foIîowirig ratio âom bottom to top: 1, 1.25, 1.25, 1.5,
1.5.2. To dustrate that, if2 newtons were appiied at the lowest level (L6), 2.5 N would be
applied at load points 5 and 4, and 3 N at points 3 and 2, etc. Figure 5-39 shows the deflection
at point D2 versus the load applied at point L2. As the ioad at that point reached 40 N, the
taught guy at the second guy level fiom the top has failed, however the tower was capable of
resisting more loads, another two load increments were applied before the tower was
considered fded. The tower fidure starteci with the second guy level but at ultimate load it
was an overall bucwiog failure between the top two spans. Figure 5-40 shows the deflected
shape of the FE model, whiie Figure 541 shows a photograph of the tower at fdure.
Cornparison of Figs. 5-40 and 5 4 1 shows how the finite element model predicted the faiied
shape. A h , Fig. 5-39 shows that the ânite element model has predicted the fdure in the
second guy level fkom the top as ihstrated in the grapb It can be also noticed tiom the same
graph that the beam-on-springs anaiysis is in agreement with the tinite element models at lower
load levels As the load increases, the Merence between the two di&rent anaiyses is more
apparent and that dEerence reaches 28% at the guy breaking point. After tbat load level, as
the material non-linearities are not coDYdered in the bearn-on-springs analysis, the failure load
In Load testing tower model IV,qua1 loads were applieâ at each of the six loading
points. Figure 5-42 shows this load versus the displacements at the highest measured point Dl.
The fdwe was suddm due to the cable rupture at the second guy level. As won as the taught
cable fded, the mast snapped and it was not possible to accurately record the tower failure
load. The last load appüed at each loading level was 53.5 N. Figure 5-43 shows the finite
element model showing the deflected shape at failure. The load deflection diagram also
indiates the same characteristics as noticed in tower iII,but not as apparent since the failure
was primarily a guy failure and not a mast failure, and the maximum deflection reported is at
the guy point and not in the middie ofthe guy span
For the tallest tower tested, model tower V, small load increments of 4.5 N were
applied at the top three loading points. This aiîowed capturing the load deflection behaviow of
this tower. The measured f i e load was 62 N and the predicted fdwe loadsfiorn the finite
element models was 64 N. The chronologid mode of fdure can be summarised as follows:
(0 buckling in the mast top span b e e n the tbird and fourth guy level nsulting in the
formation ofa plastic hhge (Figuns 5-46 and 5-47) fonowed by, (ii)rupture of taught cable at
the top guy level.
Figure 5 4 4 shows the load displment airw for Tower V. Experimental readings in this
test were fiequent aad thus a good ihstration of the bebaviour was available. The redts
confirm the findings of tbe previous tested towers; a good agreement between the experimental
data and the finite element modek, with the beam-on-springs modd not haviog the same level
of agreement at higher load levels. At fdure load the Merence between the iinite element
5.7 Discussion
The r d t s presented inthis chapter indicate a good agreement, in geaeral, between the
expehental and fiaite-element results in the elastic, fk-vibration, forced vibration, and non-
iinear responses of the tower models. The r d t s indicate that the level of agreement in the
static analysis in both lineu and non-hear static response is vay good considering di types of
condition enors. It can be observeci that in the elastic range at very low load levels, the 6nite-
eiement solution overestunated the arperimental deflection. This maybe attriiuted to the smaii
resistance in the LVDTs seasors that may have impacted on the vexy mail measured
deflections.
element models. Howeverythe proximity of the natural fiequemies of the tower and al1 the
supeduous guy modes made it very difncuh to detect the higher mode shapes either in the
muneriai models or even in the measwed deta Therefon, a higher level of discrepancy
between the measured and cornputeci naniral fiequencies can be scpected at the higher mode
shapes.
The simple constniction of the shake table with even mal1 possibiüty of tiltingy or
yawing in the table movement could be attributed to another source of error that would appear
in the forced vibration response. Furthermore, the small sale used to create the models made
5.8 Summary
horizontal defidoos, guy forces, and horizontal acceierations. From this anaîysis, the ultirnate
load capacity of a structwe was predicted dong with its natural fiequencies, mode shapes and
transient dynamic response. The foIIowing is a sumrnaq of the findings that can be drawn î?om
this analysis:
1. Good agreement between the acpaimental and theoreticai results supports the reiiabiiity of
using the fite-eiement aaalysis presented to predkt the elastic response, fiee-viiration
response, forcd viiration response, ad dtimate load-caqhg capacity of guyed towers.
2. For symmetrical m m , the eSUn,alent beam finite element model, explaineci in detail in
Chapter IV, does predict the behaviour of the structure with the same level of accuracy as
3. The beam-on springs model, wbich is currently the most commonly aud widely used in the
industry, does provide a v a y good agreement with the euperimental r d t s at service load
levels. However, as the loads approach failure, this level of agreement deteriorates by 8% -
4. Height is the major parameter affecting the natural fiequency of the tower. The use of
stiffer guys improves the t'undamental fiwency and the application of torsion resistors
5. The simple shake table designed and constructed by the author as explained in Chapter IV,
was successtiily used in the excitation and testing of the model guyed towers and can be
r
Level
GUY
Number
FE Bearn
Model
N
FE Tmss
Model
0
B eam-On-
Springs
1
Table 5-2 Guy Forces for Model Tower 1 (Load Case-2)
- ---
"""'"""I
Level Number FE Beam
Model
FE Tmss
Model
Beam-On-
Springs
N (NI (NI
6.2
Table 5-6 Guy Forces for Model Tower III (Load Case-2)
GUY
Level
I*""'"
Number
FE Beam
Mode1
FE Tmss
Mode1
Beam-On-
Springs
Table 5-1 1 Natural Frequencies for Mode1 Tower I
Natural Frequency
1 Torsional
q TRUSS l
rn
Bending 1' (x-plane)
Bending 2* (x-plane)
Bending 2M (y-plane)
2M Guy modes
(Hz)
1 Bending (y-plane)
1 22
Top Guy 1" Modes 27.5 27.5
Bending 2nd&-plane)
Guy Modes
I 34w2
40 42.1 42.1
Torsional Modes 47 53 -6 53 .O
Bending 3m (x-plane) 63 74 74
Bending 3m (y-plane)
Natural Frequency
EXP
Guys Modes
Bending 2d Mode (x-plane)
Torsional
Bending 3m (x-plane)
w
Bending 3m (y-plane)
0.0 5.0 10.0 15.0
hflution (mm)
Figure 5-1. Defîection of Mode1 Tower 1under Load Case 1 (45 N at top
loading point)
2500
2000
1500
IWO
500
Figure 5-2. Deflection of Model Tower 1under Load Case II (27 N at al1
loading points)
5.0 10.0 15.0 20.0
Deflectlon (mm)
-
- -- -
kperimentai
Truss FE Model
Beam FE Model
'+
Beam-on-springs,
1 I I
a
I 1 I
Figure 5-6. Deflection of Model Tower III under Load Case iI (27 N at
al1 loading points)
-5.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0
Defiaction (mm)
Figure 5- 15. Acceleration Response for Free Vibration of Model Tower III
Figure 5-24. First Torsional Frequency and Mode Shape of the Mast
(Mode1 Tower V)
Figure 5-25. First Namal Frequency and Mode shape of the Bottom
Guy (Mode1 Tower V)
1.O0 2.00
Time (sec)
150
Figure 5-33. Shake Table Acceleration History for Ground Motion
modeled after Nanbu N-S Direction
I I
I I
I I
0.00 0.40 0.80
m e (s)
0.40
Time (s)
Broken
---
Cable
- -
--
bpenmentaî
-
r
FE Tmss Model
FE barn Madei
Rem-on-springs Model
40.00 80.00
Displacement (mm)
170
Figure 547. Close-up of Failed Mast of Mode1 Tower V
CHAPTER VI
6.1 General
The purpose of tbis chapter is to report on the application of the f i t e element models,
described in Chapter Ui and verified from the experimental results as discussed in Chapter V,
to a number of typical towers. This would also help generate information that is usefd to the
designer of guyed antenna towers regarding their structural behaviour. As the common
pradce in North America uses simple analytid models, descnied earlier as beam on elastic
supports, to analyse these towers, this study compares the various finite elernent models to the
beam-on s p ~ g analogy.
s Generaiiy, the m e n t North Amencan Standards (CSA S37-94,
ANSVEIA 222-F 1996) do not account for the dynamic charaderistics of the towers in the
design The main objectives of this chapter are: O to compare the finite elemmt models with
the m e n t adysis methods at âifkrent load levels, (ii)to investigate the influence of the major
parameters affecting the natural fiequemies of guyed towers, (ii to generate a database fiom
typical towers that can be used in the development of design methods to account for the
dynamic 104s on these towers, (iv) to deveiop empirical formulas for the prediction of the
n a d fiequencies of guyed towers that can be used in the design process and (v) to
demonstrate the application of the finite element analyticai techniques in the forced vibration
analysis of towers.
There are many parameters involved in the design of towers. Usually the end user
specifies the required height and the maximum allowed degree of tilt/twia at the antenna
locations that would satie certain swiceability requirements of the communications aetwork
or broadcast pattern The sûucturai designer wrmaüy has the fieedom to change the size of
the masr (face width), nimber of guy levels, radius of the guy anchors, location and number of
torsion resistors, initial guy tensions, etc. Based on the designer's choice for the above
parameters, sizes of structural members of mast and guys are determineci. It should be noted
that these parameters are closely inter-related so that any small change in one of these
parameten may very well require changes to the structural components of the tower. For
example, for the same design requirements a larger (face width) mast supponed by a s d
number of guys can be used instead of a d e r mast with more number of guys to s a w the
same requirements. Tûerefore, it was decided to perfionn this study on typical towers in order
Eight towers were chosen for the purpose of this anaiysis. These are existing
towers, the data for which were available from various manufacturers and consultants so
al1 the information regarding the designer, manufacturer, orner and location of the tower.
The towers range in height fiom 46 m (150 ft) to 600 m (2000 ft). The reference wind
pressures various from 306 Pa (lowest wind region in Canada) to 1002 Pa (highest wind
region). The ice thicknesses also represent the four ice classes o f 10, 25, 40, and 50 mm.
The tower profles (Figure 6-1 to 6-8) show the geometry including member sizes and
initial guy tensions, antenna loads, design wind speed or pressure, and nominal ice
thickness. The tower height, number of guys, and specified wind and ice loads are also
summuised in Table 6-1. It should be noted that the first five towers represent typical
towers used mainly in the wireless industry, and the other three towers represent the ta11
Various de@ trends were represented in thk group of towers. For example, tower
P W has relatively smd mast face width (2.45 m) and seven guy levels with a typical guy span
(distance between two consecutive guys) of 45 m compared to tower PVI, which bas a larger
mast face width (3 .O m) and but only four guy levels with typical span of 65 rn
A computer program was developed to create the tower models in the format of
the Finite Elernent package ABAQUS, thus generating nodes, elements and boundaty
conditions depending on the type of FE model chosen (tniss or beam model). This finite
element modelling dows for al1 geometcical and structural details of the tower (e.g.
174
changes in cross-section properties, torsion resiston, eccentnc locations of antenoas,
slope in cross section towards the base of the towers, etc). For the Finite Element beam
models, the program also created the sub-models as discussed in Chapter III fiom which
the equivalent beam properties were denved. The source code in Fortran for this program
is provided in Appendk M.
The main purpose of this analysis is to compare the fiaite element anaiysis with the
model commody used in Uidustry @eam on non-hear elastic supports). One of the avdable
commercial programs that utilises this d o g y and used in this analysis is GUYMAST
(Weisman Consultants lac. 1997). Both finite element analytical models and the beam-on-
springs model are explaineci in detail in Chapter III. This study was perfomied on the towers
under two different loading conditions: the design loading condition and the up-to-collapse
(dtimate) loading condition Towers PI (46 m),PV (122 m) and PVIII (600 m) were used in
Loads were d e t e h e d eccording to the design standard (CSA S37-94; ANSI EIA-
222F-1996)based on the geographical location ofthe tower. T h forces applied to the towers
due to wind and ice loads based on the exposeci wbd a m , drag factor, orientationof the tower
with respect to wind direction, etc were calculated. Six different load combinations were
considered: in the first three, design wind was considered blowing in three diffetent directions
on the bare tower and in the other three load combiitions, fi@ percent of the maximum wind
was considered blowing in the same three wind directions on fiilly iced towers (Wahba et al.
1998~).The loads also considered any eccentricities in the applied loads due to the location of
Figures 6-9 and 6-1 1 compare the envelope of forces in the mast (kg loads and face
shears) resulting fiom the three Werent models for Towers PI and PV respectively. It can be
shown âom these two figures, that the resulting leg loads Corn the beam-on-springs model are
within 8% of the Finite element models. Also, the equivalent beam FE model results are
almost identical to the FE truss model. Face shears (which determine the forces in the bracing
members) are shown to be in very close agreement between the three different models. Table
6-2 compares the guy tensions resulthg fiom the three models which demonstrate that they are
in close agreement as weii. However, the guy forces from the beam-on-springs model are
siightly underestimateci (up to 7%). That also CO& the aulier hdings of the analysis of the
Figure 6-13 shows the leg louis and fiice shear for the tallest tower (PMII). This
tower was onginally designed using the version C of the "EIA -222" code. However, the
loads for this study were prepared accordhg to the current revision (F) of the Standard. Also,
it is located in an area where ice accretion is of no wncaq thus only the bue condition was
considereâ. In this figure,the Ieg 104s and tace sûears are shown for the wind blowing paraHel
to the plane of one the guy sets. It cm be noticed that the d t s are consistent behueen the
Figures 6-10, 6-12, and 6-14 show a cornparison betweea the deflections due to the
three models. It cm be clearly seen that for displacemeats, the beamsn spRogs model is
consistent with the finite element models. However, the twist angle does not show the sarne
modelling of the equivaht sprhgs that replace the guy assemblies at each level. The angle of
twist is not show for Tower Vüi as the loads were mostly symmetncal and produced very
Little torsion.
Figure 6-15 shows an array of seven towers connected at the top through a set of
catenary cables that are tensioned. The analysis of this array was feasible only through a finite
element mode1 as it can account for the interaction between towers under loaded conditions.
As explained earlier, the significance ofthis loading condition is to test the ability of the
analytical models to predia the behaviour at the ultimate loading conditions (near failure). This
is of prime importance for structures such as guyed towers where non-linear behaviour is
exhibiteci. Furthemore, the recent cbange fiom Workhg Stress Design to Limit States Design
requires tbat the towers be anaiysed at or near Mure. By adysing the structure at that level,
and provided that the nght analytical tools are used, the designer should be able to predict any
sigas of Uistability of the sûucture. In order to detect instability of the mast, both P-delta
effkcts must be considered: (i) the chord rotations due to the displacement of the guy points,
and (i) the member m a t u r e due to the instabiility between two or more guy levels. ui the FE
models the foliowing procedure was appüed: (i) non-linear geometrical analysis was generally
applied in which large displacement theory was introduced, (i) the mast between guy levels
was M e r discreticized into large number of elements (one per panel) to account for the
member mature between guy points Also, matenal non-linearity was introduced in the
modei to account for guy rupture. For the beamsn-sp~gsmodel, the displacements a? guy
levels were accounted for by the eccentric application of the guy forces at the displaced
In this study, Towers PI, PV, and PVIII were used io demonstrate the up-to-collapse
behaviow of the structure. The applied loads that were used in the fust step (design loads)
Figure 6-16 shows the percentage of the design load versus the deflection at the top of
Tower PI. The load deflection cum is show for loadiig condition in which the wind is
blowing in the direction of paralle1 to one the guy sets on a bue tower. It cm be noticed that
ultimate load is about five times the design load level for this load case. This is an example
where the design of the structure is governeci more by the deflection cnteria than by the
strength. Therefore, the f w e load was much greater than the design load. The tower
behaves in a more hear marner once the guys go slack @active) and only the taught cable
provides the lateral support. Also, there is close agreement between the FE model and the
beam-on-sp~gsmodel at the design load; however, the difference hcreases as the load level
increases anci the tower reaches the Mure Iwel. Furthemore, Figure 6- 17 shows the deflezted
sbape for the tower at fidure. It can be noticed that the ddected shape is quite strajght. The
Wure in this case was in the top guy. F i one of the taught cables in the wind direction
fàüed and the point of M u e can be noticed on the load deflection m e (Figure 6-16). Then
the second cable at that top level aiso Medunder M e r inaease in load. From the deflected
shape and the load-deflection behaviour of this structure, it can be seen that the fiilure was only
Figure 6-18 illustrates the load deflecîion curve for Tower PV. Sidar cuncIusiow
made nom the analysis of Tower PI can be made for this tower except that the ultimate load is
only 7% higher than the design load. The deflected shape show in Figure 6-19 also shows
that the tower's deflected shape is close to a straight line. The failure also took place due to
the rupture of the taught cable at the top guy level, wbich experienced the highest deflection.
As the FE element mode1 was capable of introducing material non-linearities, it was feasible to
Figure 6-20 shows the load deflection m e for Tower PVm. This is the tallest tower
analyseci in this study and considered as one of the tdea towers ever built. It should be
pointed out that for these tail multi-Ievel guyed towers it is difncult to keep a straight deflected
shape for the tower. This is also evident fiom Figure 6-14, which shows the deflection of the
tower under design loads. Two types of analysis were perfomed: 6rst, material non-liaeacity
was included in both the mast and the guys, and the second where material non-linearities of
the guys oniy were considered. The percentage of the applied load with respect to design load
versus the deflection at the second guy I dGrom the top is show for both arialyses in Figure
6-20. It can be noticed âom the fidure (ddected) shape (Figure 6-21) that the Mure is in the
mast between the second âom the top to the fourth fiom the top guy levels. As the load
increases the deflection at the second and third guy level, fkom the top, increase rapidly thus
magnifjmg the load-âeflection eff'ects which in tum results in mast instability at the span
between these two guy levels. The ultimate load is only 50% higher than the design loads. in
the second case where only material non-hearity of the guys was coosidered, the mode1 was
able to resist higher loads until the third guy fiom the top reached fdwe and additionai
increase of the load caused the three guys immediately below to fd.Therefore, ththe mast was
completely unsupporteci for more than two-thirds the fidi tower height. The failure shape for
The main objective of this portion of the study was to investigate the key parameters
that may affect the natural fiquencies and mode dupes of prototype guyed antenna towers. A
sensitivity study was first undertaken to determine the various parameters that may influence
the vibration response (Wahba et al. 1998a). Based on this sensitivity d y s i s , an extensive
study was then wnducted to produce empirical equations for the determination of the lowest
frequency of the structure. A h , &ixt of ice ametion and initiai tensions on the natural
fiequencies of towers was investigated. In this study, the same eight towers (Figures 6-1 to 6-
8) were first used to develop the empincal equations. Later these equations were verified using
an additionai twenty-five prototype towers (Table 6-3)representing a wide range of height (46
m - 600 m). The mode sbapes of interest for the purpose of this study were: (i) the lowest
mode, which is n o d y the top guy wiration mode, c i the 6rst f l d mode of the mast,
and (iii) the first torsional mode of the mast.
In order to excite the viiration modes of the guys as wel as the mast, each cable was
discretized into 12 to 24 elements depending on its length. A free vibration analysis was
perfonned and the lowest twenty natural kequencies and their correspondhg mode shapes
were determined. It was noticed that for the towers analysecl, the lowest naturd fkequency
ahvays corresponded to the first vibrational mode of the top guys. The lowest twenty mode
shapes for tower PV are shown in Figures 6-23 to 6-27 as representative of the Srpical results
for this analysis. Table 6-4 shows the e f f i e mass in the t h e global directions. From this
table, it can be reaiised that the total effective mas in the two horizontal directions for the
Iowest hundred modes of vibration for this tower did not constitute more than 2 8% of the total
structural mass. Furthemore, the total effective mass in the vertical direction constituted less
than 12% of the structural mas. Therefore, fiom the vibration modes show in Figure 6-23 to
6-27 and the effkctive mass for these modes (Table 6-4). it can be concluded that Grst hundred
modes were ody guy modes, some of wbich involveci wupling motion with the rnast.
Results sbown in Table 6-5 show the lowest naturai fiequencies of the eight towers
under Merent sets of initiai tensions. From the sensitivity analysis pafomed on these towers
and d s e d in Table 6-5, the following conclusions were made with respect to the
noticed tom Table 6-5, that the height is the most direct factor in detemiiniag the
lowest naturai fiequency regardless of the desiga wind and ice Ioads, number of
Mast Snfiess to m a s ruth: Table 6-6 shows the "Wm" ratios for the différent
towers, and as can be noticed, it has linle infiuence on the natural fiequency of the
mat-guys system.
Torsion Resistors: These fixtures are usually used at heights just below the
microwave antennas in order to minimise the twist at these locations. Six guys
instead of three are comected to the mast (eight in case of square towers) at that
level. These torsion resistors help d u c e the twist of the mast but their efféct on
the lowest naturai fiequency of the tower is minimal (les than 5%).
IV. Inira/ Guy Tensi'ons: The initial guy tension has an effect on both the se~ceabiîity
and the safety of the towers (Wahba et al. 1996). In the current Canadian Standard
for towers CSA S37-94 (CSA 1994). the initial guy tension may Vary fiom 8 to
I 5% of the ultimate strength of the guy but it is usually specified to be 10%. The
initial guy tension of the eight towers ùicluded in this study were varied aom 8-
15% and the e f f i on the naairal fiequencies is show in Table 6-5. As seen in
Table 6-5, a change in the initial tension of the guys anéas the natural fhquencies
of such towers by as much as 35%. It is also clear that the height of the tower is
the prime parameter*which has the largest effect on the lowest natural fiequency of
these towers.
An empiricai formula to daennine the lowest oaturai fiequency of the tower, which
involves the guy motion, would be helpfiil in the initiai design phase to determine the expected
range of fiequencies of the structure. The devdopment of an empirical formula for the
The natural fresuencies of the eight prototype towers were calcuiated from the FE
model and presented in Table 6-6.It was concluded from the examination of the results that
the height of the tower, H, has the deteminhg effect on the lowest natural fiequency. Ln Table
6 6 the height of the tower is shown in second colwnn, the fourth column presents the
calculated frequency fiom the model and the fifth colurnn shows the calculated fiequency tiom
the empirical equation that was determined to be function of the height as shown in equation 6-
1.
where:
The constants (CI and C2)in the equation were d e t e d e c i by the minimisation of the
summation of square of the Merences. The lowest naturd tiequency, j; in cyc1edsec (Hz) of
the tower, based on initial design tension equal to 1û% of the uitimate guy capacity, can be
183
predicted f?om the following equation as hction of the tower height:
f =34.5*Ha9
where:
f =100*H"~
The maximum difference between the empirical equation (6-2) and the calculated values for the
eight prototype towers was found to be less than 6% as shown in Table 6-6. Thus, Equation
6-2 can be used even at the preliminary design stage, where the height would be the only
parameter.
Equations 6-2 was later applied to additional twenty-five prototype towen and the
results (Table 6-7) were consistent with those shown in Table 6-6. Furthemore, results of
the free vibration analysis of an existing 302 m tower (Saxena et al. 1989; McCafiey and
Hartmann 1970) show the lowest natural fiequency for îhat tower as 0.217 Hz. Applying
the simple equation (6-Za), the expected namal fiequency is 0.20 Hz (9% difference).
Figure 6-28 illustrates the effect of the structure height on the tower's lowest
natural fiequency. The graph shows good agreement between the predicted frequency
fiom the empincal equation 6-2 and with the computed values from the finite element
model.
Most of the previous research on free vibration of towen was aimed at the extraction
of the natural fiequemies and mode shapes on bue towers thus neglecting the effect of icing
on the dynamic properties of the towers (Wahba et al. 1998b). Also, design codes that deal
with the dynamic analysis of towers (e.g. the reconimendations for the dynamic d y s i s of
towers due to wind e f f i s uicluded in CSA-S37 do not include the dynamic &ect of wind on
iced towers.
Table 6-8 compares the tower naturd fkequencies and lower mode shapes, for Tower
PV, calcuiated for ice accretions of O (Le. no ice), 10, 25, 40 mm thichesses. From the
results, it is clear that these types of structures when iced exhibit significantiy reduced low
naturd fhquencies thus making them vuinerable to dynamic effkcts of wind even at low wind
speeds. What is even more signifiant is that for an ice accretion of 10 mm, a 19% reduction
in the nahiral fiequency cm result and thk reduction can reach 45% for the case of40 mm ice.
Although it is not anticipated tbat d o r m ice ametion of this magnitude would be formed
throughout the tower, Smaller ice accretions wodd stiii produce appreciable reductions of the
natural tieqwncies. Furthemore, for the natural fiequencies obtained under iced cases* it was
noticed that icing causes a considerabIe increase in the coupbg action between the mast modes
and the guy modes; this makes the towers more vulwrable to galloping of the guys. Also, in
the absence of ice, the motion in the mast was nepiigiile compared to that of the guys.
However, for the i d condition, not O@ did the number of coupled mast-guy modes increase
but also the efféct ofguy motion on the mast is greatly magmfied. This is shown in Figure 6-29
where the first and second bending modes coupled with guy motion for 40 mm ice are s h o w
This is particularly important for the serviceability of the tower under iced condition and
gusting winds.
It should also be noted that ice accretion on the tower is usudy accompanied by a decrease in
the temperature from the design temperature. This should also be accounted for in the
calculation of the t h e viirations of the towers under iced conditions. The reduction in the
temperature would result in an increase in the guy tensions which, in tum, inaease the naniral
frrquencies. As a r d t of these two opposite effects on the structure, the change in the
natural fiequencies of the tower may be lower than that s h o w in Table 6-8.
As stated in Section 1.3, another objective of this study is to identify the parameters
that affect the naturai frequencies and mode sbapes of the masts. Uniike the analysis perfonned
earlier to determine the lowest natural âequency of the tower, this d y s i s suppresses the
modes of vibration ofguys. This is achieved by using the same rnodels as explained in Chapter
Di, with the exception that the guys are modelied as one element per cable. This suppresses
the guy modes and oniy the mest fhdamental modes are excited.
The eight prototype towers under study were adysed using the modified model ofone
element per cable, and the lowest ten vibrational modes were extracted. The fundamental
flexu~aland torsiod modes for the eight towers, their order withui the extracted modes, and
the percentage of the effdve mass to the total mass of the structure were calcuiated and are
presented in Table 6-9. Cornparhg Tables 6-5 and 6-9, it can be concludeci that, for shorter
towers, the B e r n i d fiequency (with guy modes suppressed) is considerabiy higher than the
lowest fkquency of the structure. The différence is s d e r for taller towers. Cornparhg Table
6-4 and Table 6-9, it can be seen that the mast modes exhibit considerably higher effective mas
and the lowest ten naturai Eequencies excited 80 to 90% of total structure mass. From Table
6-9, it is obvious that the torsional modes are predominant for ta11 towers (taller than 300 m).
The fiindamental ûequencies of guyed towers (guy modes included) depends mostly on
the height of the tower. however, height is not the only factor innuencing the natural
frequencies of the towers with guy modes suppressed. For example, although prototype
Tower PV is 10% taller than tower PIV, it exhibits a higher natural fiequency. The lowest ten
modes with their respective effeaive masses in the three global directions are shown in Table
6-10 for Tower PV (as a representative of the typical results). It cm be noticed that the total
effèctive mas of these ten modes (81% of the total m m ) is much highei than the lowest
hundred modes dnven fiom the tiiU model (1 8% of the total mass) as shown in Table 6-4. Th
fact, both results complement each other: the effdve mass in the horizontal direction for the
guy modes constitutes 18% of the structure total mas, while 81% of the total mass is exciteci
tom the siagie cable dernent model. It can be aiso wtid k t the vertical effective mass
component in the single cabie elemmt modd is negügi1e (les that 0.01%). The same
conclusions apply to ail towers analysed. This means that the axial mast modes frequencies
were not excited as the fiequencies of mast axial modes are significantly higher than the
Figures 6-30 to 6-37 show the fundamental flexucal fiequencies and mode shapes of
towers PI to PVIII respectively. From these figures it can be noticed tbat the fuadamental
flmd mode shape varies largely corn one tower to another. On closer examination of these
mode shapes, it can be noticed that it is mainiy ulauenced by the relative horizontal component
of the axial stiffness of the guy, Khg, to the mast span flexural stiffness, Km,at the sarne span
where:
where Gris the guy radius (measured from the centre ofthe tower to the anchor point)
Closer examination of "IG,,JL" for the different guy levels for the eight towers,
shown in Table 6-1 1, explains the dEerent flemral mode shapes shown in Figures 6-30 to
6-37. For example, prototype tower PI has a (i(he/[<lll) for the top guy (28.6) that is more
than twice than that of the lowea guy (1 1.43). The higher stifbess of the mast support at
the top is reflected in the flexural mode shape of the tower shown in Figure 6- 30. The
same conclusions were h v e d at with respect to prototype Towers PU, PI11 and PV.
Prototype Tower P N has relatively constant "mJKM9ratios at the different guy levels
and therefore its fundamental flexural mode shape is quite similar to a free-standing
structure. Tower PVI has a relatively sMer lowest guy level(&#L= 35.0) compared to
the top level ( K A = 19.3) and that effect can be clearly seen on the mode shape. Also
note the relatively very flexible guy support at the seventh guy level in tower PWI and
towers (with guy modes suppressed), the initial tensions of the towers were increased
fkom the design values of 10% of the ultimate strength to 15%. The results are
sumrnarised in Table 6-12 where it can be noticed that the initial tension has negligible
effect (less than 1%) on the mast flexural modes. However, it has considerable effect on
the torsional modes of the structure. For example in Tower PVIII, by increasing the initial
tensions fiom 10 to 15%, the torsional mode natural frequency increased fkom 0.34 1 Hz
to 0.379 Hz and thus rnoving it from the first (lowest) mast frequency to the third
Another parameter that was examined was the guy cross-sectional area, which in
tum is translated into the stifiess of the mast supports. increasing the top guy area made
a sigdcant change to the mast natural frequency to Tower PIV only, where it is evident
fkom the mode shape and (Kt&&) for that guy that it is relatively flexible. Increasing the
area of the top guy by 50% increased the natural fiequency of the mast fiom 1.45 Hz to
1.67 Hz (1 5%).
Upon doser examination of the natural frequencies and mode shapes of these eight
prototype towers, it was possible to relate the following parameters as the most influentid
where:
empirical equation, which is a fhction of these two parameters, was Uiitiaiiy estimated
f =C, *(ZkCJC2
*(nt)' (6-7)
The constants G, C2, and G were determined through the foîîowing tterative
procedure:
Based on the above, the equation for the naturai fkquency of the flexu~almode of a
trianguiar guyed tower (guy modes suppressed) may be calculatecl f?om the followiag
empirical equation:
where:
As is the area ofeach guy at the ith guy level in mm' (in2)
when the equation was applied to a sample of twenty five additional towers (given in
naniral fiequency as dependent on the height of the tower only. Following the same
procedure used for Equation 6-8, but using the height as the only vanable, the
This equation wuld serve as a guide for the flexiity of the structure. It is particularly
usefbl at the initiai design stage where height is the O@ known parameter- Also, as
Standards generally consider structures to be flexible if they exhibit a natural fiequency
of l e s than 1 Hz. Equatioas 6-8 and 6-9 can be used to decide if the tower should be
considered fluaile in which case a more rigorous dynamic analysis would be justified.
Cornparison sbown in Fig. 6-38 shows a good agreement between the calculated
naturai fiequencies of the mast with the predicted fiequencies as per Eq.6-9.
The prototype towen were analysed under a forced vibration adysis for two different
loaâing cases: (0 the tower base and anchors were subjected to base motion using El Centro
KS acceleration history (HKS 1995), and ( i the top guys of the towers were subjected to
hannonic galioping motion in the vertical plane for two different frequencies, with the fira
coinciding with the guy natural fiequency and the second with the mast natural fiequency. In
order to capture the behaviour of the tower under these forced vibrations, the FE models with
eariier results reported by McClure and Guevara (1993 and 1994) that the deflectioos, guy
forces, mast axial loads are wd below their design values. Figures 6-39, and 6-40 show the
time histories of the 6rst five seconds of the deflections of the mast at the top, and the
compressive stresses in the legs of the tower base for Tower P N respectiveiy. It cm be
noticed that the axial compressive stresses are w d below the design strength value of
approximately 200 MPa It shouid be noted that the seismic analysis perfomed assumed the
same phase in the gmund motion between the base and the anchors. This may not be accurate,
as a change of phase between the base motion and the anchors is quite possible especiaiiy on
taler towers where the distance between the Merent supports may exceed 700 a
Galloping of guys on taii towers, in which the guys would wnnally undergo low-
(Novak et al. 1978) has indicated that taIl towers are more susceptible to gailoping guys.
Observations have confimeci this for taller towers (300 m and above). Furthemore, some of
these gaiioping episodes have been recordeci on tape in which the amplitude of vibration
reached 3 m This research studied the e f f i of the guy gdoping on the guy forces and the
mast loads. Two gdoping kequencies of vibrations, coinciding with the lowest guy mode and
mast mode, with the same amplitude of 3 m were used on the td Towers PM, PW, and
PMII. Figure 6 4 1 indicates the results for the axial stresses in the top guy stresses, under a
harmonic motion of 3 m with a fiequency o f 0.2 17 Hz ( k t natural frequency of the top guys).
Figures 6-42 aad 6-43 show the top guy stresses and the tower deflections at the top,
respectively, for the same amplitude of vibration but with a ûequency of 0.617 Hz (first mast
flexurai fiequency). It shodd be noted that in the second case where the fiequency of
oscillation of the guys coincided with that of the mast, the efféa on the tower is more serious
and the level of stresses in the guys exceeded their dtirnate strength which would in nim r d t
in fdure.
Table 6-1. Detds of Prototype Towers Used in the Study
(Pd (mm)
Torsion resistor at ievel3
No torsion resistors
-
No torsion resistors
No torsion resistors
No torsion resistors
No torsion resistors
Table 6-2.Guy Forces under Design Loads
8 1.12E-03 3S E - 0 3 8.18E-16
Effective mms
%
Tord mass
Table 6-5. Effect of Initial Tension on Natural Frequency of Towers
TT-
Tower 10 % 12 % 1s %
PIV
Table 6-6. Denvation of Empincd Equation for Tower Natural Frequency
-- -
Mode Number
O mm 2 5 m 40mm
Suppressed)
total mass
Table 6-10. Effective Mass for the First 10 Vibrational Modes of Mast
TOTAL
Total Mass
Effective -
mass
%
Total mass
Table 6-1 1. Ratios of Guy to Mast Stifhesses @Ch&)
Tower Ptl
G1 0.079 525
G2 0.1 12 117.5
G3 0.1 49 185
Tower Plll
G1 0.0379 59
G2 0.0653 119
G3 0.0653 179
G4 0.228 ns
Tower P1V
G1 0.19 66.6
G2 0.233 136.6
G3 0.233 199.1
04 0.284 266.6
G5 0.336 336.6
525
117.5
182.5
24s
312.5
367.5
Tower PVI
Gl 3.04 206.7
G2 3.75 420.8
G3 4.96 642.2
G4 6.83 856.3
147.63
295.27
$429
590.54
767.7
974.39
1196
Tower PVlll
G1 1.O3 154.2s
G2 1.97 368.32
G3 24 597.17
G4 135 826
G5 1-84 t ûS4.84
G6 287 tZï6.29
G7 1.47 1SOS. 14
G8 3.04 1733.97
W 1.97 1911.13
Table 6-12. Cornparison of the Mast Frequencies for Various Initial Tensions
Frequency Frequency
Tower Height
FEM
Name (m) (Eq.6-21
(Hz) (Hz)
A
Cross Section
214
-
1 SRI. 210-A4 O TOP T.L. f.ûF4-50A
2- PAL10-17C OlSH O 74.7m T L 1 5/8'
3- SRL 410-C4 O 70.15m T.L UF4-50A
4- PAL10-17C DlSH O 68.0m T.L 1 5/8"
5- SRL 210-A4 O 45.0m T L Wf4-SOA
6- SRL 210-A4 O 56.6m T.L. UW4-SOA
Cross Section
215
Y
Cross Section
216
A
Cross Section
Cross Section
Figure 6-9. Comparison of Leg Loads and Face Shear under Design
Loads for Prototype Tower P I
O 25 50 75 100 125 150 0.0 0.5 1.O 1.5
DefieCam(mm) Twisting of Tower (degrees)
Figure 6-1 1. Cornparison of Leg Loads and Face Shears under Design
Loads for Prototype Tower PV
O 200 400 600 800 0.0 0.5 1.0 1.5 2.0
Deflection (mm) Twisting of Tower (degrees)
Figure 6- 13. Cornparison of Leg Loads and Face Shears under Design
Loads for Prototype Tower PVIII
O 2 4 6
Leg Loads (khi)
Figure 6-16. Load as a Ratio of Design Loads Vs. Deflection at the Top
Guy Level for Prototype Tower PI
Figure 6- 17. Failure Shape as Predicted by the Finite Element Mode1 for
Prototype Tower PI
0.40 0.80 1.20
Displacement (m)
Figure 6-18. Load as a Ratio of Design Loads Vs. Deflection at the Top
Guy Levei for Prototype Tower PV
Figure 6-19. Failure Shape as Predicted by the Finite Element Mode1 for
Prototype Tower PV
I I I
0.00 4.00 8.00
Displaœment (m)
0.50 -
I
0.25 -
-
0.00
I
I T
I I
I I
I I
1 1
241
PACTOR 329.
lb S m 2 nmrT 1
QüEITC!?= 2.84
St'ERSIOlf: 5.7-7 DATE: 16-JAtf-1999
Figure 6-40. Time History of Guy Stresses for Tower PIV Subjected to
El-Centro N-SGround Motion
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00
Ttme (s)
Figure 6-41. T h e Histocy of the Top Guy Stresses for Tower PVI
Subjected to Top Guy Galloping (f= 0.2 17 Hz)
0.00 1.O0 2.00 3.00 4.00 5.00 6.00 7.00 8.00
Time (s)
Figure 642. The History of the Top Guy Stresses for Tower PM
Subjected to Top Guy Galloping (F0.6 17 Hz)
0.00 1.O0 2.00 3.00 4.00 5.00 6.00 7.00 8.00
Time (sec)
springs analogy and two finite element models: a three-dimensional tniss, and an equivaient
beam model. A cornparisonbetween the analytical rnodels was presented for different static
loading levels. The analyticai procedures were verified by experimental nsults obtained
fiom tests on five scaled guyed tower models and wbsequently applied to eight prototype
towers. Furthemore, the uitimate load carrying capacity was determined by testing of four
tower models and the resuits were verified against the predicted values fiom the fulte
element procedures. Conclusions were drawn with respect to the suitability of the anaiytical
designed, built, and performance tested. The table was used to test five scale-mode1 guyed
towers under fke vibrations and forced base motion. The test results substantiated the
theoretical finite element predictive procedures. ComparWn of the results was presented
subjected to fke vibration and eight towers subjected to forced vibrations. A study was
conducted to investigate the main parameters affecthg the al!-important lower naturai
frequencies of the mast-guy system and that of the mast. Based on this shidy, empirical
equations were developed to detemiine the fundamentai natural frequencies of the towers
(with and without guys modes suppression). The empincai equations should be valuable in
establishing the dynamic characteristics of the towers for use in design practice.
7.2 Conclusions
The following conclusions are drawn based on the remlts obtained fiom the
a) Ana@icai
8 The finite element models presented (with geometric non lhearities included) in this
research are capable of predicting the guyed tower khaviour upto-collapse.
For symmetticaiiy b d masis, the quivalent beam finite element mode1 predicts the
behaviour of the structure within the same level of accuracy as the full miss model.
The beam-on springs model, which is cmntly the most comrnonly and widely used in
the industry, provides a g d agnement with the experimental results at s e ~ c load
e
levels. However, as the loads approaches fidure, this level of agreement deteriorates by
-
8% 25% depending on the type of failure.
6) Experimental
r A dynamic testing facility (shake table) that is suitable for testing of guyed towen was
r Good agreement between the experimental and theoretical results supports the reliability
of using the finite-element models presented to predict the elastic response, free-
guyed towers.
Heigbt is the major parameter affecthg the naturd frequency of the tower. The use of
M e r guys increases the f'undarnental frrquency and the application of torsion resistors
Mast fkquencies are siBnificantly highu than the lowest tower fmlwncies; however,
the ddynamic khaviour of the tower is largely dependent on the mast fiequency as it
towers, show good agreement with the caicuiated values ami c m be used to predict the
N a m frequencies and mode shapes of guyed towers are dependent on the relative
Generally, guyed towers less than 200 m in height need not be considered as flexible
structures and rigorous dynamic d y s i s is not waminted as the mast nanual fkequency
1- The seismic loads as well as wind gust factors need to be related to the naturai
fiequencies of the towers. The empkical equations derived fiom this study can be
2- Experimental research needs to be carried out on more towen. Dynamic models with
AISC, 1994, "Load and Resistance Factors Design- Second Edition", Amencan uistitute
Augusti, G., Bom, C., and Gusella, V., 1990, "Simulation of Wind Loading and
-
Anfennas", Sûuctural Safety, pp. 16 1 1 79.
Augusti, G., Borri, C., Manadi, L., and, Spinelli, P., 1986, "On Time-Domain Analysis of
BSI, 1994, "Lanice Tower and Masts; Code of Practice for Lattice Masts" British
CEN, 1997, "Design of Steel Structures- Part3-2: Towea, Masts and Chimney-
Chajes, A. and Chen, W. S., 1979, "Stability of Guyed Towers", J. of Structurai Division,
Structural Division, Proc. ASCE, Vol. 107, No. 12, pp, 23 13-2323.
Cohen, E. and Perrin, H., 1957% "Design of Multi-Leveled Guyed Towers : Wind
Loading", J. of Structural Division, P m . ASCE, Vol. 83, No. ST5, pp. 1355-1 - 1355-
29.
Cohen, E. and Perrin, H.,1957b,"Design of Multi-Leveled Guyed Towers : Structurai
halysis", J. of Structural Division, Proc. ASCE, Vol. 83, No. ST5, pp. 1356-29.
Costello, G.A. and Phillips, J.W., 1983, "Post-Buckling Behavior of Guyed Towers',
141.
Davenport, A. G., and Allsop, A., 1983, "The Dynarnic Response of a Guyed Mast to
Davenport, A. G., and Steels, O. N., 1965, "Dynamic Behavior of Massive Guy Cables",
I. o f the Stnichiral Division, ASCE, Vol. 91, ST2,Proc. Paper 4293, April, pp. 43-70.
Davenport, AG., and Sparling, B.F., 1992, "Dynamic Gust Response Factors for Guyed
Towers",L of Wind Engrg. and Ind. Aerodynarnics, vol. 44, pp. 223792248.
Dean, D.L, 1961, "Static and Dynamic Analysis of Guy Cables", ",J. of the Structurai
Division, Proc. ASCE, Vol. 87, ST1, Jan, pp. 1-21.
Ehande, S.G and Madugula. M. K. S.. 1988, "Geornetric Non-Linear Analysis of Three-
Gerstoft, P.,and Davenport, A. G.,1986, "A Simplified Method for Dynamic Analysis of
Gress, M., and Sparling, B., 1998, "A Simplet Dynamic Analysis Approximation for
Guyed Towers", CSCE Annual Conference, Halifax NS, Iune 10- 13.
GUYMAST, 1987, " User's Manuai", Weisman Consultants, Inc., Downsview, Ontario.
HKS,1995, ABAQUS v5.4 Gened Purpose Finite Element Code, Hibbit, Karlsson and
Sorenson, Inc., Providence, R L, U.S.A.
Hull, F. H., 1960, "Some Problems in the Analysis of Ta11 Multi-Level Guyed Towers",
international Association for Shell and Spatial Structures, 1981, "Recomrnendations for
Isheke, M. B., 1985, "Anaiysis of Guyed Towers", M. A. Sc. Thesis, Department of Civil
Lanices", Journal of Structural engineering, ASCE, Vol. 11 7, No. 4, pp. 1238- 1256.
Iwan, W. D., 1997, "Drift Spectrum: Measure of Demand for Earthquake Ground
Liveseley, K. K. and Poskitt, T. J., 1963, "Structural Analysis of Guyed Masts", Proc.
Constraints", International Journal for Numerical Methods in Engineering, Vol. 2, pp. 33-
Magwd M.H., Bumeau, M.,and Dryburgh, R.B., 1989,"Evolution of Design Standards
Marshall, D. G., 1966, "Deflectionand Guy Tensions of Guyed Masts Subjected to Direct
McCaBey, R., and Hartmann, A. J., 1972, "Dynamics of Guyed Towers", J. of the
McClure, G. and Guevara, E., and Lin, N., 1993, "Dynamic Analysis of Antenna
1994 on Spatial, Lanice, and Tension Structures, Atlanta, Georgia, pp. 259-268.
McClure, G.,Lin, N., 1994, "Transient Response o f Guyed TelecommuDication Towers
Meyers, V.J., 1963, "A Siudy of Guyed Towers Considered as Conservative Non-Linear
Diagrams". Journal of Structural Division, Proc. ASCE, Vol. 92, No.STI, pp. 245-266.
Nakamoto, R.T., and Chiu, A. N. L., L 985, "Investigation of Wind Effects on Ta11 Guyed
Novak M, Davenport A. G., and Tanaka H., 1978, "Vibration of Towers Due to
Galloping of Iced Cables" Journal of the Engineering Mechanics Division, ASCE, Vol.
Odley, E.G., 1966, "Analysis of High Guyed Towers" Journal of Structurai Division,
Company,pp.70-80.
Richelt, K.L., Brown, D.M, and Melin, J.W, 1971, "Tower-Design System for Guyed
Towers", Journal of the Structural Division, Proc. ASCE, Vol. 97, No. STl, pp. 237-25 1.
Rosenthal, F. and Skop, R.A., 1980, "Guyed Towers Under Arbitmy Loads", Joumal of
Rosenthal, F. and Skop, RA., 1982, "Method for Anaiysis of Guyed Towers",Journal of
the Structural Division, Proc. ASCE, Vol. 108, No. ST3, March, pp.543-559.
the Structural Division, Proc. ASCE, Vol. 84, No. ST6,October, pp.1821-1 - 1821-20.
Schon. G.J.and Thurston, F.R.. 1956, "The Analysis of the Structural Behaviour of
Guyed Antenna Masts Under Wind and Ice Loading. Part 1: Structural Analysis",
Skop, R.A. and O'Hara, G.J., 1970, T h e Method of Imaginary Reactions: A New
Technique for Analyzing Structural Cable System", Marine Technology Society Journal,
Skop, R.A. and O'Hara, G.J.. 1972, "A Method foe Analysis of Intemally Redundant
Structural Cable Arrays", Marine Technology Society Journal, Vo1.6, No. 1, Jan/Feb, pp.
6-18.
Skop, R.A., 1979, "Cable Spring Constants for Guyed Tower Analysis", Journal of the
Guyed Towers to Wind Turbulence", Cm. J. Civ. Eng. Vol. 25, pp 512- 525.
UBC, 1994, bbUnifomBuilding Code - Structurai Engineering Design Provisions",
Vellozzi, J. W., 1975,"Tail Guyed Tower Response to Wind Loading", Proceedings, 4'
735-743,
Vombatkere, S. G., and Radhaknshnan, R., 1986, "An Experimental Study of WECS
Tower Dynamics", Civil Engineering for Practicing and Design Engineers, Vol. 5, pp.
42944 1.
Wahba Y.M. F., Madugda M. K. S., and Monforton G.R, 1994, "Limit States Design
Wahba Y. M. F., Madugula M. K. S., and Monforton G. R., 1996, "Effect of initial
Wahba Y.M. F., Madugula M. K. S., and Monforton G.R., 1998a "Dynamic Response
ofguyed r n a s î ~Engineering
~~, Stnrctures, Vol. 20, No.12,pp. 1097-1011
Wahba Y. M.F., Madugula M. K. S., and Monforton G. R,1998b7"Effect of king on
the Free Vibration of guyed Antenna Towers",Atrnospheric Research, vol. 46, pp. 27-35.
Wahba Y.M. F., Madugula M. K. S., and Monforton G. R., 1998c, "Evaluation of Non-
Williamson, R.A., 1973, "Stability Study of Guyed Towers under Ice Loads", Journal of
+,XSAH(20),G(20)
REAL ELVG(20),GA(20), GD(20),GH(20), GR(20),MAR(20),AA(20), IT(20)
+lA1G(20),AiA(20),8S(20),GW(20),GAR(20), EMG(2O),FPL(300,7)
C W C T E R RL'15,F1LO'I5,HEAD'70,T1'70
WRITErl'(A18)')'ENTER INPUT FILE
READ(+,'(Af 5)')FIL
WRITE(*,'(A18)')'ENTER OUTPUT FlLE
READ(*,'(AlS)')FlLO
OPEN(1S,FILE=FIL)
READ(1SI1(A70)')HEAD
READ(15,'(A70)3Tt
DO 10 1 = 1,20
READ (151*lERR=100) ELVM(I)lTW(I),NL(t)~FW(I)l PH(I)JSAL(lO)
+,XSAD(1O),BW(10),EM(10),X,XSAH(lO)
NMS = I
10 CONTINUE
C READ GUY GEOMETRY
100 DO20 1 = 1,20
READ (15,',ERR=200)ELVG(I),GA(I),GD(I),GH(I),GR(I),MAR(I),AA(I)
+,IT(I),RG(I),AIG(I),AIA(I)
NG=I
20 CONTINUE
C READ GUY MATERIAL PROPERTIES
200 DO 30 l = 1,NG
READ (1S,')NNG(I), ELVG(I),GA(I), BS(I),GW(I), GAR(I), EMG(I),THCOF
+,UNSTRL,RG(I),AIG(I)
30 CONTINUE
REWIND 15
PH(J)=PH(I)
XSAL(J)=XSAL(l)
XSAD(J)=XSAD(I)
ew(J)=sw(i)
EM(J)=EM(I)
XSM(J)=XSAH(I)
CONTINUE
CONTINUE
READ GUY GEOMETRY
READ(15,'(A45)')Tl
DO70 1 = 1,NG
READ (15,')~(1),GA(I),GD(1),GH(I),GR(I),MAR(1),M(I)
+O IT(I)lRG(l)l~G(I),AIA(I)
DO 60 J =I,NG
ELVG(J)=ELVG(I)
GA(J)=GA(t)
GD(J)=GD(I)
GH(J]=GH(I)
GR(J)=GR(I)
MAR(J)=MAR(I)
WJ)=AA(I)
IT(J)=IT(I)
RG(J)=RG(I)
AIG(J)=AIG(I)
AIA(J)=AIA(1)
60 CONTINUE
70 CONTINUE
C RE30 GüY MATERIAL PROPERTES
REA0(15,'(A45)')Tl
DO 90 1 = 1,NG
READ (15,') NNG{l)lELVG(I)lGA(I)lBS(I)~GW(I)~GAR(I)lEMG(l)~THCOF
+,UNSTRL,RG(I), AIG(I)
DO 80 J=I,NG
es(J)=ss(r)
GW(J)=GW(I)
GAR(J)=GAR(I)
EMG(J)=EMG(I)
80 CONTlNUE
90 CONTINUE
NP=ELVM(NMS)IPH(I)
C REAO FORCES ON TOWER AND CABLES
c REAû(I5,'(A45)')Tl
c REAû(15,'(A45)')Tl
c DO 91 I = I,NP+I
c REAû (15,')FPL(I, 1),FPL(I12),FPL(l13)IFPC(I14)lFR(l,~lFPL(l16)
c +,FR(llT)
cg1 CONTiNUE
c READ(1S1'(A45)')T1
c READ(1S,*)AZI,SPD,REFP,TEMPC,RICElDICE,CON,PRFC,PRMI
DO 95 1=1,NMS
PH(I)=PH(1)'12.
ELVM(I)=ELVM(1)'12.
Fw(I)=MI(1)*12
95 CONTINUE
DO 190 I =I,NG
ELVG(I)=ELVG(I)*12
110 CONTINUE
NP=ELVM(NMS)IPH(l)
OPEN(2,FILE=flLO)
CLOSE(1S)
WR1TE(2,*)nHEADING'
WRIT€(2,'(A70)')HEAD
WRITE(2,')'"BEAM ELEMENTS FOR MAST'
WRITE(2,')"DATA CHECK'
WRITE(2,*)"PRE?RINT, HISTORY=NO, ECHO=NO, MODEL=NO'
NO(1)=1
DO 115 I=1,1500
CND(I,l)=û.O
CND(l,2)=O.O
CND(I,3)=O.O
115 CONTINUE
DO 120 I = 1,NP
NO(b1) = ND(I)+I
CND(I+l, 1) = CND(I,1)
CND(I+1,2) = CND(12)
CND(I+I,3) = CND(1,3) + PH(1)
120 CONTINUE
C WRlTlNG MAST NODES
WRlTE(2,*)"NODE ,NSET=MASf
WR1TE(2,'(l5,3F10.2)')(N0(I),(CN0(I, J),J=t ,3),t=,NP+l)
l
C WRfTlNG GUY NODES
ELN = 0.
DO 150 K = 1,NG
XT=MAR(K)'12WN(3.14159/ t ûû*AA(K))
M=MAR(K)'12COS(3. I415~18O*AA(K))
ZT=ELVG(K)
XB=GR(K)'12*SIN(3.14159/1BO'GA(K))
YB=GR(K)*lrCOS(3.14lWl ûûeGA(K))
ZB=O.O
X=(Xl-XB)M NG(K)
Y=(n-VB)/NNG(K)
Z=(rnB)INNG(K)
WRITE(2,'(At 5,ll)')nNODE ,NSET=GüY,K
M =(K-1 )V00 + 500
ND(M)=M
CND(M, 1)= XB
CND(M,2)= YB
CND(M,3)= ZB
DO 140 1 = M,M+NNG(K)
IF (CND(1,2).EQ.CNO(I-100,2)AND.CND(I,1).EQ.CND(l-t W,l).AND.CND
+(I,3).EQ,CND(l-100,3)) ND(I)=ND(I-100)
IF (ELVG(K).EQ.ELEV) NO(M+NNG(K))=ND(M+NNG(K~70)
WRJTE(2,'(I5,3F10.2)')ND(l),CND(I,1),CND(It2),CND(1,3)
ND(I+I)=I + 1
CNO(I+l, l)=CND(I,l) + X
CND(1+1,2)=CN0(1,2) + Y
CND(I+I ,3)=CND(1,3) + Z
140 CONTINUE
XT=MAR(K)*12'51 N(3.14 1591180'(AA(K)+120))
YT=MAR(K)' l2'COS(3.l4l59ll8O'(AA(K)+lZO))
A=ELVG(K)
XB=GR(K)'l2*SIN(3.14159118û~GA(K)+120))
Y B=GR(K)*12'COS(3.14159118ût(GA(K)+120))
ZB=O.O
X=(XT-XB)NNG(K)
Y=(YT-Y B)l(NNG(K))
Z=(ZT-ZB)/(NNG(K))
M =(K-1)'100 + 530
ND(M)=M
CND(M,l)= XB
CND(M,2)= YB
CND(M,3)= ZB
DO 160 I = M,M+NNG(K)
IF (CND(1,2).EQ.CNO(I-100,2).AND.CND(l,l).EQ.CND(l*~OO,l)~D.CND
+(i,3).EQCND(i-100,s)) ND(I)=ND(I-100)
IF (ELVG(K).EQ. ELEV) ND(M+NNG(K))=ND(M+NNG(K)Xl)
WRlTE(2,'(i5,3FlO.2)')ND(t),CND(t,1),CND(1,2),CN0(i13)
ND(l+l)=I + 1
CND(l+t ,I)=CND(t,l) + X
CND(I+I ,2)=CND(l,2) + Y
CND(I+1,3)=CND(1,3) + Z
160 CONTINUE
XT=MAR(K}'12°SlN(3.1 4159/180*(AA(K)+240))
YT=MAR(K)'l ZtCOS(3. f4159Ilôû'(AA(K)+24û))
A=€ViGK
()
XB=GR(K)'12%1 N(3.1415911 8Oe(GA(K)+24O))
YB=GR(K)ei2COS(3.141591180'(GA(K)+240))
284.0
X=(XTXBWNG(K)
Y=(YT-YB)r(NNG(K))
Z=(mZB)I(NNG(K))
M =(K-1)'100 + 560
ND(M)=M
CND(M, 1)= XE
CND(M,2)= YB
CND(Mt3)=ZB
DO 170 1 = MIM+NNG(K)
IF (CNO(I,2). EQ.CNO(I-t00,2).AND.CND(I,I).EQ.CND(I-100,1).AND.CND
+(ll3).EQ.CND(I-100,3)) ND(I)=ND(I-100)
IF (ELVG(K).EQ.ELEV) ND(M+NNG(K))=ND(M+NNG(K)-160)
WRITE(2,'(15,3f 10.2)')ND(I),CN0(Il 1),CND(It2),CND(I,3)
ND(I+l)=I + 1
CND(I+l,l)=CND(I, 1) + X
CND(I+1,2)=CND(It2) + Y
CND(I+1 ,3)=CND(1,3) + Z
170 CONTINUE
ELEV=ELVG(K)
150 CONTINUE
C ELEMENT DEFINITION FOR MAST
HSEC=O.O
00 190 I=I,NMS
ELNB=HSEC
HSEC=ELVM(I)
Nl=ELNB/PH(I) + 1
N2=ELVM(I)/PH(I) + 1
WRITE(2,'(A30,1 l)')"ELEMENTl VPE=B31, ELSET=MAST',l
DO 180 J=Nl,N2-1
ELQJ, l)=ND(J)
ELE(J,2)=ELE(J, 1)
ELE(Jt3)=ELE(J,l)+l
WRITE(2J315)') ELE(J,1),ELE(J,2),ELE(Jt3)
180 CONTINUE
190 CONTINUE
C ELEMENT DEFlNlTlON FOR GüYS
HSEC=O.O
DO 220 I=l,NG
M=(l-1)Y00 + 500
WRITE(2,'(A30,1 I)')'*ELEMENT, NPE=C1D2, ELSET=GUY,l
DO 205 J=M,M+NNG(I)-1
ELE(J,1)= J
ELE(J,2)=NO(J)
ELE(Jt3)=ND(J+1)
WRm(2,'(315)') ELE(J81),ELE(J,2),ELE(J13)
205 CONTINUE
0 0 210 J=M+3û1M*NNG(1)+29
ELE(J, 1)= J
ELE(J,2)=ND(J)
ELE(JI3)=ND(J+1)
WRlTE(2,'(3I5)') ELE(J,l), ELE(J12),ELE(J,3)
210 CONTINUE
00 215 J=M+6OlM+NNG(I)+59
ELE(J, t)=J
EtE(J,2)=ND(J)
UE(J,3)=ND(J+l)
WRITE(2,'(3l5)') ELQJ, 1), ELE(J,2),ELE(J,3)
215 CONTINUE
220 CONTINUE
C MPC DEFINITION FOR THE RIGERS
€LN=#.
WRiTE(2,'(A4)')'MPC'
DO 240 I=1,NG
IF (ELVG(I).EQ.ELEV) GOTO 230
NI =ELVG(I)IPH(l) + f
N2=NO((I-1)'lW + 500+NNGjl))
NEL=2ûûû +3'(1-1)
WRITE(2,'(A5,15,al ,i4)')'BEAM,',N2,',',N 1
WRITE(2,'(A5,15,al ,i4)')'BEAM,',N2+30,',',N I
WRlTE(2,'(A5,15,aI,i4)~BEAM,',N2+6O,',',Nl
230 ELEV=ELVG(t)
240 CONTINUE
C MATERIAL DEFINITION FOR THE MAS1
WRITE(2,'(A21)')"MATERIAL, NAME=STEEC
WRITE(2,'(A8)')'WASTlC'
WRITE(2,'(F10.2,F6.2)')EM(l),0.3
WRITE(2,'(A8)')"DENSITT
WRITE(2,'(A8)1)7.33€47'
C MATERIAL DEFINITION FOR THE CABLES
EtASMG=O.
DO 260 I=I,NG
IF (EMG(I).EQ.ELASMG) GOTO 250
WRITE(2,'(A21 ,II)')'*MATERIAL, NAME=CABLE',I
WRlTE(2,'(Aô)'~ELASTlC'
WRITE(2,'(F10.2,F6.2)')EMG(1),0.3
WRlTE(2,'(A8)')"DENSllY'
WRIT€(2,'(A8)')7.33E-O7'
WRITE(2,'(A15)')"NO COMPRESSION'
250 ELASMG=EMG(I)
260 CONTINUE
C SECTION DEFINITION FOR THE CABLES
EiASMG=O.
DO 300 I=l,NG
IF (DEMG) 270,280,270
270 K=l
GO TO 290
280 K=K
290 CONTINUE
WRlTE(2,'(A30,11 ,Al 1,I1)')"SOLID SECTION, MATERJAL=CABLF
+,K,,' ELSET=GüY,l
ARE(I)=GAR(I)
WRlTE(2,'(F53)')ARE(l)
ELASMG=EMG(I)
300 CONTINUE
C SECION DERNITION FOR THE RlGERS
C ELN=O.
C DO 310 I=I,NG
C IF (ELVG(I).EQ.ELR(J GOTO 305
C WRITE(2,'(A58,1 l)')'*BEAM SEC~ON,SECTlON=CIRCULAR,MATERIM=STEEL
C +,ELSET=RIGER',I
C WRITE(2,'(A3)') '5.0'
C W RITE(2,'(AS)')'0,0,11
C305 ELN=ELVG(I)
C3t0 CONTINUE
C
C SECTION DEFlNlTlON FOR THE MAST
EiASMG4.
DO 330 I=l,NMS
C
C CALCUiATION OF SECTION PROPERTIES
C
NM=39
NN=15
NDF=36
NLC=3
E=EM(l)
C DEFINITION FOR AREA
DO 305 K=1,12
AREA(K)=XSAL(I)
305 CONTINUE
DO 306 K=13,24
AREA(K)=XSAD(I)
306 CONTINUE
DO 307 K=25,39
AREA(K)=XSAH(I )
307 CONTINUE
CN(1,3) =0.0
00 308 K=t ,5
CN(K,l)=O.O
CN(K,2)=0.86603'FW(1)'2.B.
CN(K+1,3)=CN(K,3)+PH(l)
308 CONTINUE
CN(6,3) 4 . 0
DO 309 K=6,10
CN(K, l)=MI(l)IZ
CN(K,2)4.86603'MI(l )B.
CN(K+ll3)=CN(K,3)+PH(1)
309 CONTINUE
CN(11,3) =0.0
DO 310 K=lI,l5
CN(K, l)=FW(l)n.
CN(K,2)r0.86603'FW(l )B.
CN(K+l ,3)=CN(K13)+fH(1)
310 CONTINUE
DO 311 K=1,4
NCT(K, 1)=K
NCT(K,2)=NCT(K11) + 1
311 CONTINUE
DO 312 K=5,8
NCT(K, l)=KII
NCT(K,2)=NCT(K11)+ 1
312 CONTINUE
00 313 K=9,12
NCT(K, 1)=K+2
NCT(K.2)=NCT(Kl 1) + 1
313 CONTINUE
DO 314 K=l3,l5,2
NCT(K, 1)=K-7
-
NCT(K,Z)=NCT(K. 1) 4
314 CONTINUE
DO 315 K=14,16,2
NCT(K, l)=K-12
NCT(K,2)=NCT(K, 1) + 6
315 CONTINUE
0 0 316 K=l7,l9,2
NCT(K. 1)=K6
-
NCT(K,2)=NCT(K, 1) 4
316 CONllNUE
DO 317 K=18,20,2
NCT(KI l)=K-11
NCT(K,?)=NCT(K, 1) + 6
317 CONTINUE
00 318 K=21,23,2
NCT(K, l)=K-20
NCT(K,Z)=NCT(K, 1) + 11
318 CONTINUE
DO 319 K=22,24,2
NCT(K,1)=K-1O
-
NCT(K12)=NCT(K,1) 9
319 CONTINUE
00 320 K=25,34
NCT(K, 1)=K-24
NCT(KJ)=NCT(K,l) + 5
320 CONTINUE
DO 321 K=35,39
NCT(K, t)=K-24
NCT(K,2)=NCT(K11) - 10
321 CONINUE
DO 322 K=2,5
IVCZ(K, 1) =(K-2)3+1
IVCZ(K,2) =(K-2)3+2
IVC2(K13)=(K-2)3+3
322 CONïlNUE
00 323 K=7,10
IVC2(K11) =(K-3)9+1
IVCî(K,2) =(K-3)'3+2
IVC2(K,3) =(K3)'3+3
323 CONTINUE
DO 324 K=12,15
tVCZ(K, 1) =(K-4)3+1
1VC2(K12) =(K-4)%2
IVC2(K13)=(K-4)%3
324 CONTINUE
DO 325 K=t,11,5
lVC2(Kl 1)
IVC2(K12)=O
IVC2(K13)=O
325 CONTINUE
DO 326 K=1,39
IVC(K, 1) =IVC2(NCT(K11),1)
IVC(K,2) =IVC2(NCT(K11),2)
IVC(K,3) =IVCZ(NCT(K,1),3)
IVC(K,4) =IVC2(NCT(K12),1)
IVC(K,5) =IVC2(NCT(K12),2)
IVC(K,G) =IVC2(NCT(K,2),3)
326 CONTINUE
DO 328 K=llNDF
DO 328 M=l,NLC
P(K,M)=O.O
327 CONTINUE
328 CONTlNUE
P(12,1)=-10.
P(24,1)=5.
P(36,1)=5.
P(11,2)=10.
P(23,2)=10.
P(35,2)=10.
P(fOI3)=10.
P(22,3)=5.
P(34,3)=5.
P(l113)=0.
P(23,3)=8.66
P(35,3)28.66
c WRiTE(2,'(Fl0.4)')(AREA(K), K=1.NM)
c WRiTE(2,'(3F10.3)1)((CN(K,J), J=I
, W = I ,NN)
c WRITE(2,'(215J')((NCT(KlJ),J=i,2),K=l ,NM)
c WRlTE(2,'(6i5)l((iVC(K,J), J=l,G),K=I ,NM)
c WFUTE(2,'(F10.3)')(P(K, l), K=l ,NDF)
CAL DIRCOS(NM,NN,EL,CN,NCf,DC)
C CLEARING OF THE MASTER STJFFNESS MATRlX
DO 740 K=l,NDF
00 735 J=I,NDF
SK(K,J) =0.0
735 CONTINUE
740 CONTINUE
DO 745 M=I,NM
C U STlFF (MIE,AR~DCIEL,EK)
C A L GENK (MIIVCINM, EK,SKINDF)
745 CONTINUE
C
CAL1SOLVE (SKIUIP,NDFINLC)
C GALL STRESS (NM,NLC,E,U,DC,IVC,F,AR~EL)
FLM=10.'0.8660eFW(1)
Y=(U(lI, f)+U(23,1)+U(35,1))13.
SI11=FLM'(4.ePH(l))WI(2.'E'Y)
FP=(3û8(4.'PH(i))'3)/(3.'E'Sl 11)
Y2=(U(f 1,2)+U(23,2)4(35,2))13.
c wnte(2,32
GAREA=(N.V.'PH(l))I(YZ.FP)
C WRITE(2,870)
C870 FORMAT(II1, lOX,'RESULTS OF THE ANALYSIS')
C WRITE (2,875)
Ca75 FQRMAT(1~X,'===-Y-=========U-=' 1
C WRITE (2,880)
C88û FORMAT(11/13X,'D.0.F.'15Xt1DISPLACMENTS FOR CASES OF LOADING')
C IF (NLC-2) 881,883,885
C881 WRITE (2,882) (K,(U(KlJ),J=I ,NLC),K=1
C882 FORMAT (PX,12,5X,F10.4)
C GO 10 890
C883 WRITE (2,884) (K,(U(K,J), J=1,NLC),K=l ,NDF)
C884 FORMAT(SX,12,5X,F10.4,5X,F10.4)
C GOTOM
C885 WRITE (2,886) (K,(U(K,J),J=I ,NLC),K=l,NDF)
CM6 FORMAT (SX,l2,!%, F10.4,5XIF 1O.4,5X,FlO.4)
Ca90 WRITE (2,891)
C891 f ORMAT(1/1,3X,'MEMB. NO.',SX,'FORCES FOR CASES OF LOADING')
C IF (NLC-2) 892,894,896
C892 WRlTE (2,882) (Kl(F(Kl J),J=1 ,NLC),K=1 ,NM)
C GO TO 899
C894 WRITE (2,884) (Kl(F(Kl J), J=1,NLC),K=l ,NM)
C GO TO 899
C896 WRITE (2,886) (K,(F(K,J),J=l ,NLC),K=1 ,NM)
C GO TO 899
WRlTE(Z,'(A48,11,A9,E8.3)lHBEAMGENERAL SECTION,SECTlON=GENERAL
+,ELSET=MAST,I,',DENSIW=',DENS
WRITE(2,'(F9.3,al ,F9,2,Al,a101F9.2,A1,F9.2)~AREAMl',',1~04*S11
1
+,',',',', 1.04°S122,',',1.05'S133
G(I) = 2.1 7'G(I)
GOTO 887
ûû6 DLEN=SQRT(PH(I)"2+W(l)Y)
DRTO=(XSAL(I}'PH(I)+XSAH(I)'FW(I)+XSAû(i J'DLEN)/(XSAi(l)'PH(1))
OENS=7.33E-û7'ORTO
WRITE(2JA48, I l,A9,€8.3)')"BEAM GENERAL SECTION,SECTION=GENERAt
+,ELSET=MAST,I,',OENSIN=',DENS
WRITE(2,'(F9.3,aIlF9.2,A1,alO,F9.2,A1,F9.2)')AREAM,',',SIll,',',
+',',S122,',',SI33
887 WRITE(2,'(Aô)')'t. ,O.,O.'
c ELASMG=EMG(I)
330 CONTINUE
C BOUNDARY OEFlNlTiONS
WRlE(2,'(AS)')"BOUNDARY'
WRITE(2,'(A6)')'lI 1,3'
WRITE(2,'(A4)')'1 ,ô'
ND(400) = 400
DO 350 I=l,NG
1F (ND((1-1)'100+50O).EQ.ND((l-2)*100+500)) GOTO 350
WRITE(2,'(l4,A4)') NO((1-1).100 + 500),',1,3'
WRITE(2,'(MIA4)') ND((1-1)*100 + W),', 1,3
WRITE(2,'(14,A4)') ND((1-1)*IO0 + SO),', 1,s
350 CONTINUE
C INITIAL TENSION DEFINITION
WRITE(2,'(A32)')"1NtTlAL CONDITIONS, T(PE=STRESS'
DO 370 I=l,NG
STRS=IT(I)IARE(I)
WRITE(2,'(A3,11,F?.2)')'GUY,l,STRS
370 CONTINUE
C
C STEP DEFINITION
WRITE(2,'(A13)')"STEP, NLGEOM'
WRlTE(Zl'(Al5)r)"STAnC, DIRECT'
C
C GRAVIR LOAD ON CABLES
WRiTE(2,'(Aû)l)'DLOAD'
DO 380 I=l,NG
WRlTE(2,'(A3,I 1,A18)')'GUY,l,',GRAVl386.4,OIOl-1'
380 CONTINUE
DO 381 I=l,NMS
WRITE(2,'(AQ,Il ,A18)')'MAS~,l,',GRAV1386.4,0,0,-1 '
381 CONTINUE
C ELIMINATING PRiNTOUT FOR THIS STEP
WRITE(2,'(A28)')"EL PRINT, ELSET=GW1,FREQ=O'
C WRITE(2,'(A2)')'1E'
WRlTE(Z,'(A29)')"NODE PRiNT, NSET=GUY1,FREQ=O'
C WRITE(2,'(A2)')Y 1'
C
WRITE(2,'(A8)')I.€NDSTEP'
C
C FREQUENCY EXTRACTION
WRiTE(2,'(A13)')l.ST€P, NLGEOW
WRITE(Z,'(AlO)')"FREQUENCY
WRiTE(2,'(A10)')'1001,,,250'
WRlTE(2,'(A2l)')"EL PRINT, ELSET=GW 1'
WRITE(2,'(A3)')'S 11'
WRITE(2,'(A22)')"NODE PR1NT, NSET=MAST'
WRITE(2,'(AI)')'U'
WRITE(2JA1 S)')"RESTART, WRITE'
WRiTE(2,'(Aô)l)'ENDSTEP'
STOP
END
C SUBROUTINE DlRCOS TO CALCULATE DIRCOS OF MEMBERS
SUBROUTINE DIRCOS(NM,NN,EL,CN,NCT,DC)
DIMENSION EL(40),CN(30,3),NCT(4û12),DC(40~3)
00 600 M=l,NM
B(M) 4.0
DO 580 5=1,3
EL(M) =EL(M)+(CN(NCT(M,2), J)-C#(NCT(M, t)lJ))7
580 CONTINUE
EL(M)=SQRT(EL(M))
DO 590 J=1,3
DC(M,J)=(CN(NCT(M12),J)-CN(NCT(M,l),J))EL(M)
590 CONTINUE
600 CONTINUE
RETURN
END
C SUBROUTINE STlFF TO CALCUME THE ELEMENT STIFFNESS MATRlX
C IN THE GLOBAL DIRECTION
SUBROUTINE STlFF (MIEIAR&DCl&EK)
DIMENSION AR~40),DC(40,3),~40),EK(6,6)
DO 640 1=1,3
DO 620 J=1,3
K=1+3
L=J+3
EK(I,J)=E*AREA(MrDC(M,I)'DC(M,JyEL(M)
EK(I,L)=EK(t, J)
EK(KIJ)=EK(Il J)
EK(K,L)=EK(IlJ)
620 CONTINUE
640 CONTlNUE
RETURN
END
SUBROUTINE GENK TO GENERATETHE STlFFNESS MATRlX K
SUBROUTINE GENK (M,IVCINMIEK,SK,NDF)
DIMENSION IVC(40,6),EK(6,6),SK(40,40)
DO 690 1=1,6 \
N *IVC (Ml1)
IF(N.EQ.0) GO TO 690
DO 6ûû J=l,6
K = IVC (M,J)
IF ( K.EC2.0) GO TO 680
SK(N,K)= SK(N,K)+EK(I,J)
680 CONTINUE
690 CONTINUE
RETURN
END
SUBROUTINE SOLVE (SK,U,P,NDF,NLC)
DIMENSION SK(40,40),U(40,4),P(40,4)
K=O
480 K=K+1
A=SK(K,K)
00 500 J=l ,NDF
SK(K,J)=SK(K, J)IA
500 CONTINUE
00 515 J=f ,NLC
P(KlJ)=P(KlJ)IA
515 CONTINUE
N4
550 N=N+1
IF(N.EQ.W) GO TO 560
IF(N.GT.NDF) GO T0 575
0= SK(N,K)
DO 560 J=l,NDF
SK(N,J)=SK(N,J)-SK(K,J)'B
560 CONTINUE
DO 570 J=l ,NLC
P(N,J)=P(N,J)-P(K,J)'B
570 CONnNUE
ns IF(N.LT.NDF) GO TO 550
IF(K.LT.NDF) GO TO 48û
DO 620 I=I ,NDF
DO 610 J=1 ,NLC
u(IiJ)=p(I14
610 CONTINUE
620 CONTINUE
RETURN
END
StlBROüTiNE SIRESS(NM,NLC,E,U,DC,IVC8FIAREA,€L)
OlMENSlON U(4û,4),DC(4û,3),IVC(4û16),F(4Ot4)
DIMENSION AREA(4û),EL(40)
DO 990 I=1,NM
DO 985 N=1,NLC
F(I,N) =0,0
DO 980 J=1,3
A4.0
K=J+3
A=A+(DC(I,J)'U(IVC(I, K),N)*DC(I1J)'U(IVC(l, J), N))
1*E'AREA(I)IEL(I)
F(I,N)=F(I,N)+A
980 CONTINUE
985 CONTINUE
990 CONTINUE
RETURN
END
Yohanna M. F. Wahba
1967 Born on the 14* of March in Cairo, Egypt.
1989 Graduated with B.Sc. degree (Honour) in Civil Engineering fiom Cairo University,
Cairo, Egypt.
Assistant.
1992 Enroiied in the Faculty of Graduate Studies and Research, University of Windsor,
1995 Passed professional engineering examinationsI and II, for regisuation in State of
Michigan, U.S.A.
1996 Joined Jay Desai, Consuiting Engineen, West Bloomfield, MI, as a Project Design
Engineer.
1997 Joined LeBlanc & Royle Telcom inc., Oakville, Ontario, Canada, as the Chief
Engineer.