Professional Documents
Culture Documents
1993
(T, 1994ElsevierScienceLtd
Printed in Gnat Britain.
0045.7949/93$6.00+ 0.00
Abstract--The inherent nonlinearity in the structural behavior of guyed towers leads to difficulties in their
structural analysis, and prevents the formulation of a general-purpose design methodology. As a result,
simplifying analysis assumptions regarding the loading and the modeling of structural behavior have to
be made, and approximate design methods are used, that are often unjustified, and can lead to disastrous
failures.
In this paper, the authors fhst summarize the results of an investigation they carried out on the collapse
of a 1900 ft tall guyed tower under ice and wind loads. Based on this investigation, they then proceed to
present some structural analysis recommendations relating to loading and modeling concerns. Special
emphasis is placed on the importance of ice loading, and on the level of accuracy required in modeling
the nonlinear response behavior. Finally, the conclusions drawn from this study are used to formulate
preliminary design guidelines. This facilitates a systematic approach for the design of tall guyed towers.
797
798 C. GANTFS et al.
Fig. I. Perspective view of the collapsed tower. Fig. 3. Influence of ice distribution on deflected shape.
Modeling considerations of tall guyed towers 799
a
A
initiated in a vertical member on the leeward side
between the sixth and seventh guy levels. Once this
member buckled, the guys at the seventh and eighth
level, highly tensioned due to the combined loading,
pulled the upper tower section down causing the d 1
d
I
aed%
I
lower five guy levels to behave out-of-phase with one
t
another. A push-pull mechanism instantly developed 2
ment w in the directions shown in the figure, is given where mg is the dead weight of the cable per unit
by length.
This approximate approach is well known and
2
provides satisfactory results, especially for high initial
F,=3T, ; +;@A,-T,) ;
0 0 pretensioning and small additional in-service dis-
placements [lo]. Note that in reality the tension along
a guy is not constant due to sag and dead weight
x (;){I -$)(;)}. (4)
effects. Use of an average initial guy tension T,, is
recommended. Then, the expressions for the horizon-
This result is based on the assumptions that w is of
tal component of tension becomes
order u2, and that terms of order u3 or higher are
negligible. For the opposite horizontal displacement
direction we have
U
F, = N _
(12)
F,=3T, : +$A,-T,) ; ’ z 0c
0 00
k,, = N
(6)
The vertical load is given by eqn (10).
This model is adequate for initial approximations
For small displacements the terms of order (u/c)~ can and for low in-service loads in comparison to the
be neglected. In addition, it is assumed that T, < EA,. pretensioning forces. For a more refined analysis, and
Then, we obtain the same horizontal force com- given the lack of cable elements, the guys can be
ponent regardless of displacement direction modeled as a linkage of nonlinear truss elements. To
avoid having an unstable stiffness matrix due to zero
stiffness of all degrees of freedom associated with
displacement perpendicular to the cable, the initial
pretensioning strains have to be introduced. A
The vertical component is
method that has been used in the past consists of
modeling the cable as a straight series of truss
Fw=3Tp b . elements with the initial pretensioning, and then
0C applying the dead weight of the cable incrementally
Taking advantage of the isotropic nature of the to obtain the deformed configuration at the beginning
results for small displacements we can generalize of the in-service phase. This model however, over-
expressions (7) and (8) for the case of a cluster of N estimates the lateral stiffness provided by the guys,
guys arranged symmetrically around the mast because the stresses due to dead weight are added to
those due to mechanical pretensioning, which is not
an accurate simulation of the real pretensioning
process.
Alternatively, one can use this model of a straight
and for the vertical component truss linkage with the initial pretensioning, but use a
reduced axial stiffness EA, according to Dishinger’s
F,= NT, b . formula (11). This accommodates for softening of the
0C cables due to their dead weight. Then, the service
loads are applied directly on the configuration with
These results can be modified using Dishinger’s for- straight cables. This approach gives satisfactory re-
mula to take into account the nonlinearity due to sults for cases of high initial pretensioning.
sag [31 A more general approach is to calculate analyti-
cally the sagged cable shape, and to start the finite
(EA& = EAg (11) element analysis from that deformed configuration.
,+y2E&’
( > P P
The sagged geometry can be modeled with a series of
straight truss elements (Fig. 7). Either a continuous
802 C. GANTFS et al.
i(J-4
d=a
’ 0p
H 2J2 ’
4. LOADING CONSIDERATIONS
Table 1
Level Truss model Spring model
I 1215.1 1215.5
2 575.9 579.1
3 399.4 411.7
4 252.1 263.3
5 180.4 193.3
6 158.2 175.0
7 178.9 203.3
Fig. 8. Comparison of P-u graphs for truss and spring
8 101.2 117.5
model.
Modeling considerations of tall guyed towers 803
condition triggered by self-excited vibrations that tation of reality. The critical ice formation will be of
result in a single degree of freedom motion [21]. ‘in cloud icing, and the deposits on guy wires will be
As long as the designer can avoid unstable behav- influenced not only by altitude, but also by wind
ior of both the tower and the cables, the dynamic direction relative to guy cable direction. Significantly
nature of the wind load can be accounted for by different cable tensions at each of the three guys at a
applying a gust factor on the equivalent static loads. level can be expected because of irregular ice deposits.
All available preliminary design approaches, includ- The overall tower safety will ultimately be ensured
ing the methodology proposed in this paper, are only if these unsymmetric loading cases are taken into
based on the behavior of towers under equivalent consideration. Further research on statistical aspects
static loads using height and gust factors. of these irregular deposits is necessary before rec-
The combination of the wind loads with the accu- ommendations for their consideration on design
mulated ice loads complicates the response further. practice can be formulated. Until then application of
As discussed earlier, accumulated ice formation on higher factors of safety is recommended.
the tower and the cables has a multiple effect. The
dead load increases considerably. The projected area 5. PRELIMINARY DESIGN GUIDELINES
of the members increases, and therefore, the wind
loads become larger. The sag in the cables also Due to the complicated behavior of guyed towers,
increases and, as a result, their lateral stiffness de- their design is today still a trial-and-error procedure.
creases. In addition, the distribution of ice along the Our conclusion, both from an extensive literature
height of the tower is not uniform. Usually, there is survey, and from discussions with experienced de-
more ice at the top than at the bottom. This worsens signers, is that the selection of initial trial sizes for
the effect of the combined ice and wind action on the member cross-sections and level of pretensioning is
tower. This effect is illustrated by both our studies largely based on past experience. To our knowledge,
(Fig. 3), and those of Williamson [22]. no systematic procedure exists in the literature that
The most recent guidelines for the design of guyed actually comes up with recommendations for initial
towers are the ones given by the ANSI/EIA-222-D dimensioning. In this section a simple but systematic
code [16], and our studies have been carried out in methodology for this initial selection is proposed. The
accordance with them. The design wind load on the procedure is based on the observations made in the
tower and the guys is determined using the ex- preceding sections.
pressions provided by this code. The expressions for Several simplifying assumptions have been adopted
the drag and lift forces on the guys involve the angle in order to obtain analytical expressions for the
between the wind direction and the cable direction. response. The main assumption is that the mast will
Three angles of attack for the wind relative to the remain approximately straight in the deformed pos-
vertical plane of a series of guys have been taken into ition and will just rotate about its base by an angle
consideration. These angles are O”, 60”, and 90”, with 4. This assumption is based on the so-called ‘straight-
respect to one of the three symmetry axes of the ness constraint’[6], according to which the defor-
triangular mast basis. mations at all guy attachment points must be less
The load combinations used are than 6 in from a line joining the tower base to its top.
The physical meaning of this constraint is to reduce
D + W, individual member deformations and overall buck-
ling probability.
D +0.75W,+Z, The deformed configuration of the model used for
preliminary design is shown in Fig. 9. The equivalent
where D is the dead load, W, is the wind load on the spring model is used for the guys and the equivalent
structure without ice, Wi is the wind load on the beam model for the mast. The tower is subjected to
structure with ice, and I is the ice load.
Again, the effective area taken into account for
wind loads on the structure without ice, is smaller 3 (Y)
than the one used to calculate wind loads on the
1
structure with ice. As a genera1 rule, the projected
area of the members is increased by 6A = 2tL, where
t is the accumulated ice thickness and L is the length
of the member. Based on weather reports, it is also
reasonable to assume a linearly varying ice thickness
from a low value at the bottom of the tower to a
maximum value at the top.
It should be noted, however, that the assumption
of triangular ice distribution along the height of the
tower, although an improvement in comparison to
uniform distribution, is not a satisfactory represen- Fig. 9. Model used for preliminary design.
804 C. GANTFS et al.
+O(u2), (15)
where T,, is the initial pretension, &,, is calculated Hence, the initial trial sizes for the required cross-
from Dishinger’s formula, A, is the cross-section area sectional areas of the guys can also be calculated.
of the guy, and a,, h,, L’,, u, refer to the notation of
Fig. 6. For the working stresses we have 5.2. Preliminary design oj’ the mast
The design of the mast is quite straightforward.
Knowing the guy reactions, equilibrium consider-
ations at any desired mast level can provide the axial
force F and the bending moment M for the corre-
sponding cross-section. Keeping eqn (2) in mind, and
where y is the density of the material of the cable. choosing d, we can obtain the required section area
Assuming that the geometry of the guys is known, cr, A of the columns
is equal to the allowable stress call, and 4 is specified
as the maximum allowable rotation &,,, we can
solve the above equation numerically for o,,,. The
method of successive iterations can be used for the
numerical solution starting with the trial value It should be stressed here again that the procedure
CP, = 0.40,,, . as recommended in [ 181. Hence, the proposed in this section can serve only as a method-
initial trial values for the pretensioning stresses can be ology for the preliminary initial sizing of the basic
calculated. It is interesting to observe that these components of a guyed tower. Using these results as
values depend only on geometry and material proper- a starting point the designer should proceed and carry
ties. and not on the intensity of applied external out a more refined analysis taking nonlinear and
loads. Variations in this intensity is accommodated dynamic effects into account.
by appropriate scaling of cross-sections as explained
below. 6. SUMMARY AND CONCLUSIONS
The calculation of the cross-sections of the guys is
based on the second criterion. In order to have equal The first part of this paper investigates the collapse
lateral resistance by all springs, it is required that of a tall guyed tower under ice and wind loads. The
observations of previous investigators about the
F,,=F,-k,u,=k,u,-k,h,tan&=k,b,tan$. (17) importance of ice loads are verifed. Then, several
Modeling considerations of tall guyed towers 805
approaches for the modeling of the mast and the guys 8. H. B. Jayaraman and W. C. Knudson, A curved element
for the analysis of cable structures. Comput. Struct. 14,
are evaluated. An equivalent beam model appears to
325-333 (1981).
be a simple and acceptable solution for the mast, 9. J. W. Leonard and J. H. Nath, Comparison of finite
while equivalent springs are satisfactory for modeling element and lumped parameter methods for oceanic
of the guys for preliminary analysis. A nonlinear truss cables. Engng Struct. 3, 153-167 (1981).
representation in the sagged configuration is possible lo. D. Bruno, F. Maceri and R. S. Olivito, Analysis of the
elastic response of stays and stayed systems. IABSE
for a more exact finite element analysis when cable Periodica i, 29-44 (19%). _ _
elements are not available. Considerations about 11. A. Chaies and W.-S. Chen. Stabilitv of euved towers.
the calculation of wind and ice loads are also J. Struit. Div., ASCE 105,‘163-174-(197$.-
presented. 12. N. V. Raman, G. V. Surya Kumar and V. V. Sreedhara
Rao, Large displacement analysis of guyed towers.
Finally, a methodology for preliminary design is
Comput. Struct. 28, 93-104 (1988).
proposed as a first step towards a more systematic *3. F. Rosenthal and R. A. Skop, Guyed towers under
approach for the design of guyed towers. It is believed arbitrary loads. J. Struct. Div., ASCE 106, 679492
that this methodology constitutes an improvement to (1980).
today’s state of the art by introducing a clear design 14. F. Rosenthal and R. A. Skop, Method for analysis
of guyed towers. J. Struct. Div. ASCE 108, 543-558
philosophy as outlined in Sec. 5.1, and by recom- (1982).
mending specific initial trial values for member cross- 15. B. A. Schrefler, S. Odorizzi and R. D. Wood, A total
sections and guy pretensioning. This approach is Lagrangian geometrically nonlinear analysis of com-
intended to improve the current trial and error tower bined beam and cable structures. Comput. Struct. 17,
115-127 (1983).
design practice.
16. Electronic Industries Association, Structural standards
for steel antenna towers and antenna supporting struc-
REFERENCES tures. American National Standard, ANSI/EIA-222-D-
1986 (1986).
1. H. M. Irvine, Cable Structures. MIT Press (1981). 17. DIN 4131, Antennentragwerke aus Stahl (1988).
2. J. W. Leonard, Tension Structures: Behavior and Analy- 18. W. H. Greene, Minimum weight sizing of guyed an-
sis. McGraw-Hill (1988). tenna towers. J. Struct. Engng, ASCE 111, 2121-2137
3. A. G. Davenport and G. Steels, Dynamic behavior of (1985).
massive guy cables. J. Struct. Div., AXE 91, 43-70 19. R. T. Nakamoto and A. N. L. Chiu, Investigation of
(1965). wind effects on tall guyed tower. J. Struct. Engng, ASCE
4. A. J. Wilson and R. J. Wheen, Inclined cables under 111, 232&2332 (1985).
load-design expressions. J. Struct. Div., ASCE 103, 20. J. W. Vellozzi, Tall guyed tower response to wind
1061-1078 (1977). loading. Proceedings of the 4th International Con-
5. M.-C. Tang, Analysis of cable-stayed girder bridges. ference on Wind Effects on Buildings and Structures,
J. Struct. Div., ASCE 97, 1481-1496 (1971). (Edited by K. J. Eaton) pp. 735-743, Heathrow,
6. Task Committee on Cable-Suspended Structures of the London (1975).
Committee on Special Structures of the Committee on 21. M. Novak, A. G. Davenport and H. Tanaka, Vibration
Metals of the Structural Division of ASCE, Commen- of towers due to galloping of iced cables. J. Engng
tary on the tentative recommendations for cable-stayed Mech. Div. ASCE 104, 457473 (1978).
bridge structures. J. Struct. Div. ASCE 103, 941-959 22. R. A. Williamson, Stability study of guyed towers under
(1977). ice loads. J. Struct. Div., ASCE 99, 2391-2408 (1973).
I. R. L. Huston and J. W. Kamman, Validation of finite 23. R. R. A. Issa and R. R. Avent, Microcomputer analysis
segment cable models. Comput. Struct. 15, 653-660 of guyed towers as lattices. J. Struct. Engng, ASCE 117,
(1982). 1238-1256 (1991).