Engineering Structures 99 (2015) 492501

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Geometric nonlinear static and dynamic analysis of guyed towers using

fully nonlinear element formulations
H. Shi a,, H. Salim b
a
Pepco Holdings Inc., Washington, DC 20068, USA
b
Department of Civil and Environmental Engineering, University of Missouri, Columbia, MO 65211, USA

a r t i c l e i n f o a b s t r a c t

Article history: The nonlinear responses of guyed towers under static loads and under dynamic loads are investigated by
Received 11 February 2013 using the developed nonlinear Finite Element Method (FEM). Through the use of the developed fully non-
Revised 13 May 2015 linear element formulations for structural components such as trusses, cables, and beams, it is possible to
Accepted 13 May 2015
capture the complete geometric nonlinear response with the assistance of iterative algorithms such as
Criselds method and the Newmark beta method. One 50 ft high guyed tower subjected to a constant
top-concentrated load and one 328 ft guyed tower subjected to an EI Centro earthquake are used as
Keywords:
demonstration. The results of the static analysis are compared with the results from SAP2000, ANSYS,
Geometric nonlinearity
Finite Element Method (FEM)
and a linear equivalent column method. The results of the SAP2000, ANSYS, and the developed nonlinear
Guyed tower FEM algorithm match well when the deection is relatively small. In the Linear Equivalent Method, the
Nonlinear static analysis equivalent spring-supported beam method provides the most accurate estimation when the deection is
Nonlinear dynamic analysis relatively small. The complete time history of the guyed tower is also obtained for the nonlinear dynamic
Linear Equivalent Method analysis.
2015 Elsevier Ltd. All rights reserved.

1. Introduction
Guyed towers are one of the most efcient structures that can
use a relatively small amount of materials to sustain relative large
loads at signicant heights. Therefore, guyed towers are often used
as communication structures such as transmission, cellular, and
antenna radio towers. It is composed of the mast and cables, which
combines the advantage of combining a stiff structure (the mast)
with a exible structure (the cables). The vertical loads are trans-
ferred by the relatively stiff mast to the foundation, and the lateral
movement of the mast is constrained by the exible cables. The
guy clusters are located at different heights along the mast and
there can be as much as nine guy clusters in a high guyed tower.
The mast base can be pinned or clamped to the ground.
The analysis and design of guyed towers are explored in detail
in a number of documents and building codes, such as [14].
Some efcient ways to assess the responses of guyed towers are
available in the literature [58]. Madugula [9] and Smith [10] stud-
ied the responses of guyed towers subjected to wind loads,
whereas Irvine [11] has expanded the simplied dynamic analysis
models under wind loads by using modal decomposition. Gerstoft
Corresponding author. Tel.: +1 202 872 3210.
E-mail address: shihaijian2002@yahoo.com (H. Shi).
and Davenport [12] has devised separate the guyed mast into
high-frequency and low-frequency region using patching loads
and dynamic magnication factor. Using a stiffness generation pro-
cedure, Desai et al. [13] analyzed a guyed tower subjected to a
head load. Desai and Punde [14] proposed a nine-degree-of-
freedom cable model applicable to cable-stayed structures sub-
jected to gusty wind. Meshmesha et al. [15] developed empirical
methods and formulas to estimate the seismic responses of guyed
towers, and, a number of other studies [16,7,17,18] have addressed
the same issue.
Various approaches to obtain the guyed towers complex struc-
tural response have also been used by researchers. Ekhande and
Madugula [19] conducted three dimensional guyed towerss geo-
metric nonlinear analysis using linear Lagrangian coordinate
framework. Kewaisy [20] adopted the combination of nite differ-
ence method and nite element method within total Lagrangian
coordinate framework to analyze the dynamic response of guyed
tower under wind induced forces. Amiri and McClure [21] used
nite element computer program ADINA to analyzed the seismic
response of guyed tower under three types of earthquakes. Jorge
and Marta [22] employed nite element software ALGOR to model
and study guyed towers dynamic response.
However, compared to the state-of-the-art analysis and design
techniques used for buildings and bridges, the corresponding tech-
niques for guyed towers with large deformation are

http://dx.doi.org/10.1016/j.engstruct.2015.05.023
0141-0296/ 2015 Elsevier Ltd. All rights reserved.
 H. Shi, H. Salim / Engineering Structures 99 (2015) 492501 493

underdeveloped. When the guyed towers undergo large deforma- The tangent stiffness matrix can be obtained by taking the dif-
tion, the global stiffness of the guyed towers has been signicantly ferentiate of Eq. (3) with respect to the displacement vector
changed and the response becomes very different with the guyed ^ @kfug @e @f/g
towers without large deformation. The cables behave in a highly k  EAlf/g EAle
@fug @fug @fug
nonlinear way, the response of which is hard to predict using linear
cable theories. The mast and cluster of cables also interact, which @2e
EAlf/gf/gT EAle 4
further increases the complexity of the analysis. @fug2
The objective of this paper is to present a developed nonlinear
FEM that can fully accommodate the highly nonlinear behavior The mass matrix is the same as that of the linear truss element
of the cables and the mast whether subjected to a static load or in [24],
to a dynamic load. The FEM element formulation capable of simu-
2 3
2 0 0 1 0 0
lating the large deformation is introduced rst for the truss, cable, 60 2 0 0 1 07
6 7
and beam elements. Then the iterative algorithm for the nonlinear 6 7
static and dynamic analysis is presented. Finally, two cases are
qAl 6
6
0 0 2 0 0 17
7
m 5
studied to illustrate the ability of the developed FEM system to 6 661 0 0 2 0 07
7
6 7
predict both the nonlinear static response and the dynamic 40 1 0 0 2 05
response of the guyed tower. 0 0 1 0 0 2
where q is the mass density. Parametric study using this truss ele-
2. Nonlinear FEM element formulations ment can be found in [25].

To capture the complex nonlinear behavior of the structure, 2.2. Fully nonlinear FEM cable element
fully nonlinear nite element formulations for truss elements,
cable elements, and beam elements [23] have been validated with It is assumed that the static load is applied before the dynamic
experimental study and are adopted. loads are applied. From Fig. 2, the axial strain [23] is
q
 0     
2.1. Fully nonlinear FEM truss elements e a1 u01 2 a02 u02 2 a03 u03 2  1 6

where fa1 ; a2 ; a3 g is the deformation vector under a static load,

The undeformed length of the truss element (Fig. 1) is
fu1 ; u2 ; u3 g is the deformation vector under a dynamic load, and
q
0 @@s. From the variation of energy, it can be found that the pro-
l x2  x1 2 y2  y1 2 z2  z1 2 1
duct of the stiffness matrix and the displacement vector is
where (x1 ; y1 ; z1 ) and (x2 ; y2 ; z2 ) are the position coordinates of Z l
points P1 and P 2 , respectively. Assume (u1 ; v 1 ; w1 ) and (u2 ; v 2 ; w2 ) kfug EA0 1  te2 eDT f/gds 7
0
are the deformation coordinates of points P1 and P2 , respectively,
such that the total Lagrange strain e along the deformed axis where the initial strain caused by the prestress in the bar is e0 , the
becomes Poissons ratio is t, the original cross-section area is A0 ; and the
q Youngs modulus is E,
u2  u1 x2  x1 2 v 2  v 1 y2  y1 2 w2  w1 z2  z1 2  l  
e : a01 u01 a02 u02 a03 u03
l f/g ; ; ; and
2
1e 1e 1e
2 3
From the variation of elastic energy, Pai [23] found that the pro-  1l 0 0 1
0 0
l
duct of the stiffness matrix and the displacement vector is 6 7
D 4 0  1l 0 0 1
l
0 5:
kfug AlEef/g 3 0 0  1l 0 0 1
l

where E is the Youngs modulus, A is the cross-section area, l is the

n oT
@e @e @e @e @e @e
length of the truss element, and f/g @u 1
; @ v 1
; @w 1
; @u 2
; @ v 2
; @w 2
.

Fig. 1. Undeformed and deformed geometry of a truss element. Fig. 2. Undeformed and deformed geometry of a cable element.
494 H. Shi, H. Salim / Engineering Structures 99 (2015) 492501

Taking the differentiation of Eq. (7) with respect to the displace- where
ment vector, the tangent stiffness matrix becomes 2 3
0 0 0 0 0 0
Z " # 6 0 EAe  2EI v 00 q 7
^ l
@2e 6 33 3 0 0 2EI33 v 0 q3 0 7
k  EA0 1  4te 3t2 e2  DT f/gf/gT D 1  te2 eDT D ds 6
60 0 EAe 2EI22 w00 q2 0 0
7
2EI22 w q2 7
0
0 @fx0 g2 C 6 7:
60 0 0 0 0 0 7
8 6 7
6 7
40 0 0 0 0 0 5
The mass matrix is the same as Eq. (5). 0 0 0 0 0 0

2.3. FEM beam element with von Karman nonlinearity The mass matrix is the same as the mass matrix for the linear beam
elements.
To account for the large deformation effect, both torsion and
stretch effect has been included in the nite element formulation 3. Nonlinear algorithms
in addition to the bending effect. The strains in the element can
be expressed as Either Criselds method or the arc length method can be used
e11 e zq2  yq3 9a for nonlinear static analysis. The Newmark-beta method is adopted
for the nonlinear dynamic analysis.
e12 zq1 9b
3.1. Criselds method
e13 yq1 9c
Though the NewtonRaphson algorithm is often used in nonlin-
e u0 v 02 =2 w02 =2 9d ear static analysis, it cannot capture certain special phenomena
including the snap through of exible truss structures. This is
q1 /0 9e because the tangent stiffness matrix becomes singular at some
special points, which makes the solution of the global equilibrium
q2 w00 1  w02 9f unattainable. To overcome this problem, Riks suggested a proce-
dure [26] that traces the intersection of the normal to tangent
q3 v 00 1  v 02 9g line with the equilibrium path, as shown in Fig. 3. Similarly,
Criselds method uses an arc (or a circular plane in
where eij is the tensorial engineering strain, e is the axial strain, y multi-dimension) instead of a perpendicular line (or a perpendicu-
and z are coordinates in the cross-section, and qi is the deformed lar plane in multi-dimension) to search for a solution. The incre-
curvatures with respect to the axes x; y, and z. In addition, u, v , ment of load factor becomes an unknown that needs to be solved
and w are the displacements on the cross-section. during the iteration.
When the variation of energy is applied, the product of the stiff- Under the assumption that fFg kf b
F g,where k is a load param-
ness matrix and the displacement vector becomes
Z eter and f b
F g is a preselected load vector, the residual force vector
T T for the i th step can be
kfdg B W Ufwgdx 10 
l
@R 
RfUgi ; ki RfUgi1 ; ki1 dfUgi1
where @fUgfUgi1 ;ki1

@N @R
B ; dki 8
@s @k fUgi1 ;ki1 1
2 3
EA 0 0 0 where fUg is the displacement vector, RfUgi1 ; ki1
6 0 GI 0 0 7 i1 b
6 11 7 KfUgjfUgi1 ;ki1  k f F g. Set RfUgi ; ki 0,
U 6 7;
4 0 0 EI22 0 5
f0g KfUgjfUgi1 ;ki1  ki1 f b
F g K T jfUgi1 ;ki1 dfUgi1 f b
F gdki1 9
0 0 0 EI33
2 3 where dfUgi1 is the rst increment of the displacement vector for
1 v0 w0 0 0 0
60 7 the i th step and dki1 is the rst increment of the load parameter
6 0 0 1 0 0 7
W 6 7;
40 0 2w00 w0 0 0 02 5
1  w F
0 2v 00
v 0
0 0 1  v 02 0
8 9 F1
> e >
>
> > F2
<q > =
1
fwg ;
> >
> q2 >
>
: >
;
q3
0 @
@s
and N is the shape function.
 fDd g with the appli-
Under the assumption that fd g fdg
i i

cation of the Taylor expansion without high-order terms, the tan-

gent stiffness matrix is U
^
Z
k   T UW CBdx
BT W U1 U2
l
Fig. 3. Scheme for Criselds method.
 H. Shi, H. Salim / Engineering Structures 99 (2015) 492501 495

for the i th step. If dki1 is assumed, dfUgi1 can be solved accordingly. can then be expanded and substituted into the global differential
Update the displacement vector and the load parameter by equation,
fUgi1 fUgi1 dfUgi1 and ki1 ki1 dki1 . Because the rst-round C  fUg
_ K  fUg fFg
M  fUg 15
estimation is usually not the exact solution, more iterations are
needed. When fUgi1 is updated and ki1 , Eq. (9) becomes where M is the global mass matrix, C is the damping ratio matrix,
K is the stiffness matrix, and fFg is the load vector. After substitu-
f0g KfUgjfUgi ;ki  ki1 f b
F g K T jfUgi ;ki dfUgi2 f b
F gdki2 10 tion and manipulation [25], the solution of Eq. (15) can be solved
1 1 1 1
numerically by the incremental of the displacement vector. As the
When the following incremental searching path is set perpen- tangent stiffness and the damping matrix are computed at the
dicular to the normal of the rst incremental path, beginning of each time step, the tangent stiffness matrix may be
not right for the entire time step. Therefore, NewtonRaphsons
f0g dfUgi1  dfUgi2 f b
F gdki1 f b
F gdki2 11
iteration is used to obtain the exact solution. The iterations are ter-
Solving Eqs. (10) and (11) for dfUgi2 and dki2 yields minated when the convergence criteria is met. The displacement,
 velocity, and acceleration vectors will be constantly updated until
dfUgi2 K T jfUgi ;ki 1 KfUgjfUgi ;ki  ki1 f b
F g fb
F gdki2 12 the desired time steps have been achieved.
1 1 1 1


4. Case study of a 50 ft guyed tower
dfUgi1 K T jfUgi ;ki 1 KfUgjfUgi ;ki  ki1 f b
Fg
dki2 1 1 1 1
13
f b
F gdki1 f b
F g  dU i1 K T jfUgi ;ki 1 f b
F g 4.1. Geometric nonlinear static analysis of a 50 ft guyed tower
1 1

The above steps are repeated until convergence is achieved. The The geometry of the guyed tower is as shown in Fig. 4. The mast
converged displacement vector and the converged load parameter arm (Fig 5(a)) is composed of steel pipes with a Youngs modulus of
are fUgi fUgi1 dfUgi1 dfUgi2 dfUgi3    and ki ki1 29  106 psi. The outside rim diameter of the pipe is 1.25 in. and
dki1 dki2 dki3   , respectively. the inside diameter is 1.04 in. The lateral and oblique struts
between the poles are made of round steel bars with a radius of
3.2. Newmark beta method 7/16 in. The steel cables have two layers, which are anchored at
different elevations along the mast. The radius of the lower-layer
General integration methods trace the change of system proper- cables is 5/32 in., and the radius of the upper-layer cables is
ties by integration using small time steps. Therefore, these are the
only methods suitable for solving nonlinear dynamic problems.
Commonly used methods are the Newmark Beta method and the
constant average acceleration method (or the EulerGauss proce-
dure). As the EulerGauss procedure is just a special case of the
Newmark Beta method, the Newmark Beta method is adopted in
this paper.
The basic integration formulas [26] for the velocity and dis-
placement of the i th step are expressed, respectively, as

_ i fUg
fUg _ i1 1  cdtfUg
i1 cdtfUg
i 14a


_ i1 1  b dt 2 fUg
fUgi fUgi1 dtfUg i1 bdt 2 fUg
i 14b
2

_ i , and fUg
where fUgi ; fUg i are the displacement vector, the velocity (a) (b)
vector, and the acceleration vector, respectively, of the ith step. The
displacement vector, the velocity vector, and the acceleration vector Fig. 5. (a) Cross-section of the mast, and (b) top plan view of the applied head load.

600

500

400

300

200

100

0
500 500
400
300
200
0 100
0
-100
-200
-500 -300

Fig. 4. Geometry of a 50 ft guyed tower.

496 H. Shi, H. Salim / Engineering Structures 99 (2015) 492501

600

500

400

300

200

100 500

0
-400 0
-300
-200
-100
0
100
200
300
400 -500
500

Fig. 6. Deformation of the guyed tower subjected to a head load.

Fig. 7. Deformed shape using ANSYS for a 50 ft tower subjected to a head load.

Fig. 8. Deformed shape using SAP2000 for a 50 ft tower subjected to a head load.

3/16 in. The initial prestress for the lower-layer cables is 900 lbs
and the initial prestress for the upper-layer cables is 100 lbs.
Additional details can be found in [27].
The point load is applied on the top of the mast (Fig. 5(b)). To
7000
trace the nonlinear response, the load is increased gradually with
small increments. Once a converged solution for the current load Geometrical nonlinear analysis
ANSYS
step is achieved via the iteration process, the load level is updated 6000 SAP

and then a new iteration process begins and continues until the
desired criteria, such as maximum displacement on a specic part 5000
or the maximum load, is reached.
The nal deformation shape is as shown in Fig. 6. The tip
Force (lb)

4000
deects as much as 83.8 in. under a concentrated load of
6584 lbs. But the lateral drift for the upper cable cluster connection
3000
point is much smaller than that for the tip deection. There is
almost no deection for the lower cable cluster connection point.
2000
Therefore, the deformation for the top unrestrained mast is similar
to the deformation of a cantilever beam under tip point load. But
the entire mast behaves more like a multiple supported beam 1000

under tip load than a cantilever beam.

The loaddeection curve in Fig. 9 clearly shows that the tower 0
0 10 20 30 40 50 60 70 80 90 100
is softened after the rst 10 in. deection. When the deection
Deflection (inch)
reaches 45 in., the guyed tower becomes stiffer with an increase
in deection, which is due to the stretching of the already taut Fig. 9. Comparison of load deection curves.
 H. Shi, H. Salim / Engineering Structures 99 (2015) 492501 497

cables on the opposite side of the top load direction. It can also be
seen that the global stiffness is still less than the original elastic
stiffness because the slope of the loaddeection curve is smaller
than the initial slope of the curve.
4.2. Comparison of results by full geometric nonlinear FEM and
commercial FEM software
To validate the proposed approach, the same guyed tower is
also modeled using commercial FEM software ANSYS and
SAP2000 for comparison. Link 8 is used to model the struts in
ANSYS. Link 10 (tension-only) is applied to the modeling of the
cables. The pole is simulated by beam 4. The large static analysis
option is turned on in order to capture the large deformation.
The nal deformed geometry, shown in Fig. 7, is very similar to
the nal deformed geometry as shown in Fig. 6. Under a 6000 lbf
external tip load, the maximum deection is 97.4 in. The loadde-
ection curve is shown in Fig. 9. Though the analysis is nonlinear,
the results appear linear because the deection is linearly propor-
tional to the loads.
In the SAP2000 analysis, the nonlinear and large deformation
option in analysis is also chosen. The deection shape under
6000 lb external tip loads by SAP2000 is shown in Fig. 8. The
deection is less than the deection shown in Fig. 6 and less than
that shown in Fig. 7, which indicates that the guyed tower behaves
stiffer in SAP2000 than in ANSYS. This is veried in the load deec-
tion curve, as shown in Fig. 9, because the slope of the loaddeec-
tion curve obtained by using SAP2000 is steeper than the curve
obtained by using ANSYS. The SAP2000 results provide a good
match with the results by geometrical nonlinear analysis when
the deection is small. But as the deection increases, the stiffness
of the guyed tower or the slope of load deection curve provides a
good match with the ANSYS results.
From the comparison, it can be seen that results of the large
deection analysis of the guyed tower vary depending on which
FEM software is used. Though the analysis is based on a nonlinear
algorithm, the results appear to be linear for both ANSYS and
SAP2000. The response by the adopted nonlinear FEM approach
is fully nonlinear.
4.3. Comparison of results by full geometric nonlinear FEM and the
Linear Equivalent Method

To further validate the developed FEM approach, Linear

Equivalent Method is adopted. When the pole is treated as a beam

7000

Fixed top Simply Spring Nonlinear

6000 supported supported FEM

5000
Force (lb)

4000

Fig. 10. Equivalent cross-section of the guyed tower. 3000

2000
F F F
1000

0
0 10 20 30 40 50 60 70 80 90
Deflection (inch)

Fig. 12. Loaddeection curves of the mast tip using different approaches.

F1 F1

F2 F2

(a) (b) (c)

Fig. 11. Linear equivalent methods: (a) a clamped beam with xed support at the
upper cable cluster location, (b) a simply supported beam with supports at cable
cluster locations, and (c) a beam with springs at cable cluster locations. Fig. 13. Details of the 328 ft guyed tower.
498 H. Shi, H. Salim / Engineering Structures 99 (2015) 492501

with a constant cross-section, as shown in Fig. 10, the equivalent deection of the cable clusters is relatively small compared to that
bending stiffness is of the tip deection. Therefore, the rst simplication approach is
to simulate the top fraction of the tower as if a cantilever beam
p were xed at the top cable cluster location (Fig. 11(a)). The theoret-
I 0:6254  0:52044  3 p0:6252  0:52042 FL3
4 ical solution for tip deection is 3EI1 . The results can be seen in
p !2
3 Fig. 12. It can be seen that this approach predicts a very stiff
  16:75  2 p0:6252  0:52042 response.
6
Although the lateral deection of the second cluster point is
p !2 p !2
3 3 negligible, the rotation at this point greatly affects the tip deec-
2
  16:75 p  0:2188   16:75 p tion. Therefore, the second simplication releases the rotation
3 6
restraint at the guy cluster points (Fig. 11(b)). When the force
p! !2
3 method [28] is used to analyze the indeterminate beam, two sup-
2
 0:2188  0:5   16:75  2 port reaction forces are assumed and the deections at these two
6
supports must comply with the constraint conditions.
4
60:2599 in: 16
F 1 L21 F 2 L21 FL2
D1 L1  3L1  L1  3L2 1 L1  3L 0
Three different scenarios, described next, are used to simplify 6EI 6EI 6EI
the analysis. Because neither the height nor the span of the cables
is large, the cables are taut with 10% prestress. The lateral F 1 L21 F 2 L22 FL2
D2 L1  3L2  L2  3L2 2 L2  3L 0
6EI 6EI 6EI
where L1 is the distance from the tip to the upper cluster point, L2 is
the distance from the tip to the lower cluster point, and L is the
height of the tower. The tip defection is

Shaking direction F 1 L21 F 2 L22 FL2

Dtip L1  3L  L2  3L L  3L
6EI 6EI 6EI
The results from this approach are plotted in Fig. 12. It can be seen
that the response of the mast tip is softer in this approach in than
the response in the rst approach. However, the deection in this
approach still appears to be stiffer than the deection in the nonlin-
ear response.
When the lateral stiffness of the guy cluster is taken into con-
sideration as springs with constant stiffness (Fig. 11(c)), the deec-
tions at the low cable cluster points and the high cable cluster
points become

F 1 L21 F 2 L21 FL2 F1

D1 L1  3L1  L1  3L2 1 L1  3L
6EI 6EI 6EI K1
Fig. 14. Plan view of the guyed tower with the input direction of an EI Centro
earthquake.

100

80

60

40

20

0
40

20

-20
30 40
10 20
-40 -10 0
-30 -20
-40
5

-5
0 5 10 15 20 25 30 35

Fig. 15. Response of the guyed tower at t = 3.5 s with the corresponding ground acceleration time history.
 H. Shi, H. Salim / Engineering Structures 99 (2015) 492501 499

F 1 L21 F 2 L22 FL2 F2 It can be seen that the response of this approach match well
D2 L1  3L2  L2  3L2 2 L2  3L with the response from nonlinear FEM approach when the deec-
6EI 6EI 6EI K2
tion is less than 10 in. The complete results from the three simpli-
where the equivalent lateral stiffness of the upper taut guy cluster ed approaches are compared in Fig. 12. It is clear that with the
[9] can be obtained by using release of constraints, the equivalent beam becomes softer and clo-
ser to the nonlinear response. The spring-supported beam
X EAi approach matches with the nonlinear FEM response very well in
Kj  cos2 hi  cos2 bi the initial stage when the deection is relatively small. But as the
li
deection increases, none of three simplied approaches can pro-
where li is the chord length of the cables, Ai is the cross-section of vide a precise estimation for the large deformation of the mast. It
the cables, hi is the vertical angle between the chord line and the should also be noticed from the comparison that the nonlinear
horizontal reference, and bi is the horizontal angle between the FEM approach and Linear Equivalent Method can provide similar
cable and the direction of the mast displacement. results when structural deformation is relative small.

100

80

60

40

20

0
40

20

-20

20 30 40
-40 -10 0 10
-40 -30 -20
5

-5
0 5 10 15 20 25 30 35

Fig. 16. Response of the guyed tower at t = 4.5 s with the corresponding ground acceleration time history.

100

80

60

40

20

0
40

20

-20

20 30 40
-40 -10 0 10
-40 -30 -20
5

-5
0 5 10 15 20 25 30 35

Fig. 17. Response of the guyed tower at t = 5.5 s with the corresponding ground acceleration time history.
500 H. Shi, H. Salim / Engineering Structures 99 (2015) 492501

100

80

60

40

20

0
40

20

-20

20 30 40
-40 0 10
-30 -20 -10
-40
5

-5
0 5 10 15 20 25 30 35

Fig. 18. Response of the guyed tower at t = 7 s with the corresponding ground acceleration time history.

0.3 If the tower were subjected to the EI Centro earthquake in the

X-direction, as shown in the plan view of Fig. 14, its responses at
0.2 different time instants would be as shown in Figs. 1518. Due to
the intensity of the computation, only the rst 11 s of the towers
response was captured. The mast appears to vibrate in its high
Output DOFs deflection (m)

0.1
modes. During this period, the tower gradually responds to the
ground motion (Fig. 19). In the rst two seconds, there is little
0
deformation. The mast shows signicant deection after 3 s. The
maximum tip deection exceeds 12 in. (0.3 m) in the rst 11 s.
-0.1 The cables movements during this period are also well captured.

-0.2 6. Conclusion

Full geometric nonlinear elements for trusses and cables have

-0.3
been developed with corresponding FEM formulations. When
these elements are paired with advanced algorithm, the approach
-0.4
0 5 10 15 20 25 30 can fully address the nonlinear static and dynamic responses of
Time (sec) guyed towers. The complex responses of guyed towers can be pre-
cisely traced even when there is a large deection.
Fig. 19. Time history of the mast tip deection with the corresponding ground The following detailed conclusions can be drawn:
movement time history.

 The results from the available commercial FEM software pack-

5. Case study of a 328 ft guyed tower subjected to earthquake ages match well with those of the developed nonlinear FEM
loads approach when the deection is relatively small. However,
when a large deection is developed, the geometrical nonlinear-
With the increase in the height of the tower, the cables inevita- ity dominates the response. The developed nonlinear FEM
bly sag due to their own weight even when subjected to a rela- approach can well capture the transition between the softening
tively high level of prestress. To explore the nonlinearity that and the stiffening of the towers response.
was introduced by the sag of the cables, a 328 ft high guyed tower  Linear Equivalent Method is most effective with the considera-
is simulated. The poles and struts in the mast arm are composed of tion of the release of constraints in rotation and deection on
steel bars. The corresponding Youngs modulus is 28:9  106 psi. the supports. Linear Equivalent Method can provide similar
The steel pipes that constitute the poles have an outside rim diam- accurate solution when deformation is relatively small with
eter of 4 in. and an inside rim diameter of 1.6 in. The plane section much less computational effort compared with FEM. Using
of the mast is an equilateral triangle, the side length of which is spring supports to simulate the interaction between the masts
7.9 in. The prestress forces in all three layers of cables are set at and the cables works well with small deformation. But when
22.5 lb in order to ensure that the cables sag signicantly. All the the deformation is large, only the nonlinear FEM can precisely
struts and cables are assumed to have a diameter of 1.6 in. predict the guyed towers response.
Details are presented in Fig. 13.
 H. Shi, H. Salim / Engineering Structures 99 (2015) 492501
 For high and slim guyed towers subjected to earthquake loads,
the highly nonlinear behavior of the cables and the mast can
only be traced using the developed nonlinear FEM elements
and algorithm. The deformation of the structure at any particu-
lar moment can be captured precisely. It is found that at the
beginning of the earthquake, the structure is relatively soft
because the sagging prestressed cables do not provide sufcient
lateral support to the mast. But when the deection gradually
increases and the entire structure gradually agitated by the
ground motions, the cables opposite to the moving direction
of the mast get tighten. Therefore, these cables constrain the
motion of the mast and stiffen the entire structure.
References
[1] TIA/EIA-222-G.5. Structural standard for antenna supporting structures and
antennas. Arlington (VA): Telecommunications Industry Association; 2006.
[2] CSA. Antennas, towers, and antenna-supporting structures. CSA S37-94.
Etobicoke, Ontario, Canada: Canadian Standards Association; 1994.
[3] BSI. British standard: lattice towers and masts: Part 4, code of practice for
lattice masts. BS 8100 Part 4. London, UK: British Standards Institution; 1994.
[4] CEN. European pre-standard ENV 1993-3-1:1997: Eurocode 3: design of steel
structuresPart 3-1: towers, masts and chimneystowers and masts. Brussel:
Comit Europan de Normalisation; 1997.
[5] Galvez C. Static method for aseismic design of self-supporting towers. M. Eng.
Project Report G95-08. Department of civil engineering and applied
mechanics. Montreal, Quebec, Canada: McGill University; 1995.
[6] Sackmann V. Prediction of natural frequencies and mode shapes of self-
supporting lattice telecommunication towers. Project Report-Diplomarbeit-Nr.
76. Department of civil engineering. Montreal, Quebec, Canada: McGill
University; 1996.
[7] Amiri GG. Seismic sensitivity of tall guyed telecommunication towers [PhD
Dissertation]. Montreal: McGill University; 1997.
[8] Wahba YMF. Static and dynamic analysis of guyed antenna towers [PhD
dissertation]. Ontario: University of Windsor; 1999.
[9] Madugula MKS. Dynamic response of lattice towers and guyed
masts. Reston: American Society of Civil Engineers (ASCE); 2002.
501
[10] Smith BW. Communication structures. London: Thomas Telford; 2007.
[11] Irvine HM. Cable structures. Cambridge: MIT Press; 1981.
[12] Gerstoft P, Davenport AG. A simplied method for dynamic analysis of a guyed
mast. J Wind Eng Ind Aerodynam 1986;23:48799.
[13] Desai YM, Popplewell N, Shah AH, Buragohain DN. Geometric nonlinear static
analysis of cable supported structures. Comput Struct 1988;29(6):10019.
[14] Desai YM, Punde S. Simple model for dynamic analysis of cable supported
structures. Eng Struct 2001;23(3):2719.
[15] Meshmesha H, Sennah K, Kennedy JB. Simple method for static and dynamic
analyses of guyed towers. Struct Eng Mech 2006;23(6):63549.
[16] Ben KN. Dynamic analysis of guyed towers. Eng Struct 1994;16(4):293301.
[17] Hensley GM, Plaut RH. Three-dimensional analysis of the seismic response of
guyed masts. Eng Struct 2007;29(9):225461.
[18] Faridafshin F, McClure G. Seismic response of tall guyed masts to
asynchronous multiple-support and vertical ground motions. J Struct Eng
2008;134(8):137482.
[19] Ekhande SG, Madugula MKS. Geometric non-linear analysis of three-
dimensional guyed towers. Comput Struct 1988;29(5):8016.
[20] Kewaisy TH. Nonlinear dynamic interaction between cables and mast of
guyed-tower systems subjected to wind-induced forces [PhD Dissertation].
Texas: Texas Tech University; 2001.
[21] Amiri GG, McClure G. Seismic response of tall guyed telecommunication
towers. In: Proceedings of the eleventh world conference on earthquake
engineering. Acapulco, Mexico; 1996 [June 2328].
[22] Jorge SB, Marta BR. Parametric study of the dynamic along-wind response of a
guyed tower. La Plata, Bueneos Aires: Latin-American Congress Wind
Engineering, (CLIV 2012); 2012.
[23] Pai PF. Highly exible structures: modeling, computation, and
experimentation. Reston: American Institute of Aeronautics and Astronautics
(AIAA); 2007.
[24] Bathe KJ. Finite element procedures. New Jersey: Prentice Hall; 1996.
[25] Shi H, Salim H, Shi Y, Wei F. Geometric and material nonlinear static and
dynamic analysis of space truss structures. Mech Based Des Struct
2014;43:3856.
[26] Reddy JN. An introduction to nonlinear nite element analysis. New
York: Oxford University Press; 2004.
[27] Shi H. Nonlinear nite element modeling and characterization of guyed towers
under severe loading [PhD Dissertation]. Missouri: University of Missouri;
2007.
[28] Kassimali A. Structural analysis. Toronto: Thomson Canada; 2005.