Sei sulla pagina 1di 11

Nonlinear finite element

analysis of latticed
transmission towers
F. G. A. AI-Bermani and S. K i t i p o r n c h a i

Centre for Transmission Line Structures, Department of Civil Engineering, The University
of Queensland, St. Lucia, Queensland, 4072 Australia

Current design practices for transmission tower structures are based


on 3D linear elastic truss analyses and on full-scale testing
experience. This paper reviews current practices and presents a
nonlinear analytical technique for accurate simulation and prediction
of the ultimate strength and behaviour of transmission towers under
static load conditions. Both geometric and material nonlinearities are
accounted for in the analysis. A formex formulation is used for the
automatic generation of data that is necessary for the analysis. The
behaviour of four different full-scale towers is described and the
predicted results are compared with tests.

Keywords: angle member, elasto-plastic, finite element, formex


algebra, full-scale testing, geometric nonlinearity, nonlinear analysis,
transmission towers, ultimate strength

Transmission towers are normally classified as self- Design practices for transmission towers are different
supporting (free standing) or guyed structures. This from those for other steel structures in that stresses are
paper concentrates on the self-supporting transmission permitted to be higher because towers are tested to their
tower which is the more conventional form. A tower ultimate design strength and designs incorporate
structure can be subjected to a multitude of load com- modification based on test results. The two most widely
binations. These include the weight of conductors and used design specifications for the design of axially
tower, ice load, wind load, transverse load resulting loaded angle members in self-supporting transmission
from an angle in the line, longitudinal loads caused by towers are the ASCE Manual No. 52: 'Guide for Design
unbalanced forces in the conductor tensions, loads of Steel Transmission Towers '4,5 and the 'ECCS
imposed during the stringing operation, torsional loads Recommendations for Angles in Lattice Transmission
resulting from broken conductors and dynamic loading Towers '6. The lattice tower structure is considered to
from galloping conductors. All of these loads must be consist of members supported by stress-carrying bracing
considered to act in various combinations as specified by and redundant members which are nominally
the electricity safety codes, statutory regulation and unstressed. The design manuals specify limiting
industry standards ~2. slenderness ratios for different member types to account
Transmission tower structures are widely regarded as for partial end restraint and joint eccentricity. In the
one of the most difficult lattice structures to analyse. ASCE Manuals ~'5, the ultimate maximum stress is
This difficulty stems from the fact that these structures determined using the SSRC 7 basic column curves
are generally composed of asymmetric thin-walled angle (Curves 1 or 4 as appropriate), whereas in the ECCS
section members eccentrically connected. For this Manual 6 the ultimate maximum stress is based on the
reason proof-loading or full-scale testing of the structure ECCS multiple column curves (Curve ao) 8.
has traditionally formed an integral part of the develop- A study by the Electric Power Research Institute
ment of tower design. Stress calculations in the structure (EPRI) 9 indicated that current design practices have,
are normally obtained from a linear elastic analysis for the most part, served the industry well. However,
where members are assumed to be axially loaded and, data from full-scale tests have shown that the behaviour
for the majority of cases, pin-connected. In practice, of transmission towers under complex load situations
such conditions do not exist and members are detailed cannot be consistently predicted using present techni-
minimize bending stresses. Despite this, full-scale ques. The investigation by EPRI also revealed that out
testing of transmission towers shows that bending of the 57 structure load cases conducted, 23%
stresses in the members are often as significant as axial experienced premature failure. On average, failure
stresses 3. occured at 95.4% of the design load, but at unexpected
0141-0296/93/040259-11
1993 Butterworth-Heinemann Ltd

Engng. Struct. 1993, Volume 15, Number 4 259


Nonlinear FE analysis of latticed transmission towers: F. G. A. AI-Bermani and S. Kitipornchai

locations. Further, available data showed considerable considering the geometric nonlinearity due to large
discrepancies between member forces computed from tower displacements 3.
linear elastic truss analyses and those measured from When a truss type model is used to analyse a transmis-
full-scale tests. The EPRI study concluded that, in the sion tower, the structure should be free of planar joints
case of transmission towers, linear elastic truss analysis which cause local instability. Significant effort is
should be used with extreme care. required on the part of the designer to remove planar
joints, a process which requires the addition of stabiliz-
ing members. Identifying and correcting such
Procedure for full-scale load testing
instabilities may require a few additional computer runs.
Full-scale testing of transmission tower structures plays For these reasons, there is a need for a method of
an important and integral role in the development of the analysis which can predict the ultimate structural
designs. Guidelines for transmission tower testing are behaviour of transmission towers more accurately than
available 4~. The test is generally set up to simulate the the linear elastic truss approach. Such a refined techni-
most critical design conditions. Loads are normally que would provide the designer with a better understan-
incremented to 50%, 90%, 95% and 100% of the max- ding of tower behaviour, which would undoubtedly lead
imum specified loads. Typically, each load increment is to a more economical structural design. Any saving in
held for one or two minutes. When a premature failure the design of one tower is magnified many times over
occurs, corrective measures are taken and all failed because large numbers of towers of the same design are
members are replaced. The load case which caused the usually constructed.
failure is repeated until the tower is able to support the
ultimate design load. Sources of nonlinearity
Although the ultimate load testing will, to some
extent, verify the adequacy of the tower in withstanding In any structural system, many sources of nonlinearity
the specified static design loads, it cannot predict exactly influence the structural response. Their effect and order
how the structure will behave in practice under alter- depend on the structural system, the loading and the
native load conditions. boundary conditions. For a latticed transmission tower
structure one can identify three major sources of
nonlinearity: geometric nonlinearity, material
Methods of analysis and design nonlinearity, and joint flexibility and slippage.
Stress calculations in a transmission tower structure are Several techniques can be used to account tbr these
generally based on a linear elastic analysis, normally nonlinear effects. The geometric nonlinearity can be
assuming that members are axially loaded and pin- accounted for by incorporating the effect of initial
connected, with the stiffer main leg members considered stresses as well as the geometrical variations in the struc-
as continuous beams. Forces or stresses in the members ture during the loading process. Such an approach is
are usually determined using a computer-aided method able to predict the transient response of the structure in
of analysis. the pre- and postcritical ranges ~3. Predicting the full
Two basic approaches have been used to develop com- response is essential in identifying the ability of the
puter programs for analysing transmission towers. The structure to sustain loads at larger displacements. Fur-
first approach translates the logic of conventional ther, the characteristics of the postcritical response
methods into routines to carry out the analysis of the reveal the sensitivity of the structure to imperfections.
structure. The second approach uses structural analysis The material nonlinearity can be incorporated by
methods such as the stiffness method. using a fibre type model ~4"~5 or a lumped plasticity
Most of the computer programs available are based on model ~6. In the former model the element cross-section
a linear 3D elastic truss approach using the stiffness is discretized into a number of elementary areas and a
method, for example the BPA TOWER program ~ and record of the strain history of each elementary area is
the TRANTOWER program 3'~2. The BPA TOWER kept and updated during the loading process. The effec-
program is a linear elastic truss analysis program tive section properties of the element are obtained and
adjusted to handle long, slender, tension-only bracing used to formulate the tangent stiffness of the element. In
members. The analysis requires a certain number of the lumped plasticity model, any plastic behaviour is
iterations to determine which bracing members are deemed to be concentrated at the two extremities of an
loaded beyond their compression capacity and to remove element. The stress resultants in the cross-section
such members from the model, thus forcing the remain- interact with each other to produce yielding for the sec-
ing bracing members to carry the tensile load. The tion. This interaction is achieved through the concept of
member response is determined via the use of a member a yield surface in force space.
performance curve obtained from a member perfor- The structural joint is a medium through which forces
mance data base gathered from available test results for and moments are transmitted from one member to
single members. another, hence the joint behaviour is bound to influence
In the TRANTOWER program, members are the global structural response. Due to the complex
assumed to be fully active when in tension and are nature of joints, their behaviour is usually nonlinear
capable of sustaining only a certain compression. The from the very onset of loading ~7. The effect of joint
compression members are characterized as having a flexibility can be incorporated by modifying the tangent
bilinear force-displacement relationship where the stiffness of the element using an appropriate
member buckling load is obtained through the use of moment-rotation (M-Or) relation for the joint t8. Joint
appropriate design formulae recommended by codes or slippage is another aspect of the joint behaviour which
design manuals a. Secondary effects are incorporated by has not so far received much attention. A preliminary

260 Engng. Struct. 1993, Volume 15, Number 4


Nonlinear FE analysis of latticed transmission towers: F. G. A. AI-Bermani and S. Kitipornchai

study investigating the effect of joint slippage on the before deformation, and are parallel to the principal
global structural response has been conducted t9 and and ~ axes of the cross-section. A parallel set of coor-
revealed that although slippage may influence joint dinates .~,)7,~ passes through the end cross-section cen-
deflection, it has negligible effect on the ultimate troids C and C' of the element. Neglecting the effect of
response of the structure. warping, there are six possible actions (Fx, F,, F z, Mx,
The three sources of nonlinearities discussed in this My and Mz) with corresponding displacement com-
section can be viewed in a local sense and in a global ponents (u, v, w, 0x, 0y and Oz) that can be applied at
sense. However, since latticed transmission tower struc- each end of the thin-walled element.
tures are, in general, of a large scale and highly redun- The following assumptions have been made: (i) the
dant with a multiplicity of load paths, the global effect element, but not necessarily the member, is prismatic
of the nonlinearity on the structural response is of most and straight; (ii) cross-sections are rigid and do not
interest. distort; (iii) shear deformations are negligible; (iv) the
material is homogeneous, isotropic and elastic-perfectly
Proposed nonlinear analytical technique plastic; (v) strains are small but displacements and rota-
tions can be large; (vi) warping of the cross-section is
negligible and (vii) loads are conservative.
General
A nonlinear finite element analytical technique will be Geometric nonlinearity
described. In the method, the tower is modelled as an
Figure 2 shows the deformation of an element in the pro-
assembly of general thin-walled beam-column elements.
jected YX and ZX planes of the global X, Y and Z coor-
Since most of the tower connections are multiple-bolted
dinate system. The element deformation may be
end connections offering some degree of restraint, it is
described using three different configurations, Co, C~
assumed that the restraint offered by a connection
and C2. These configurations represent, respectively,
relative to the moments induced in the tower members
the initial undeformed state, the current (known)
is large enough to regard the connection as rigid. The
effect of joint flexibility can also be handled provided
information on the joint flexibility is available ~8

Basic assumptions (a) YX plane C 2


Figure 1 shows an element of general thin-walled open
section. The right-hand orthogonal coordinate system ~'~'\ / I-~--._~_.-~Z2
x,y,z has been chosen such that y and z pass through the
end cross-section shear centres S and S' of the element
\ \\~12- ,

1
t "Xz Co
Lo
_ ~..----~X0
2 X

/
Zo

(b) ZX plane py -C2

*i

Z. Oyl~py I ~ ~ 11"

r 1
+

Figure 1 Element generalized forces and displacements and Figure 2 Element deformations in projected global YX and ZX
referenceaxes planes

Engng. Struct. 1993, Volume 15, Number 4 261


Nonlinear FE analysis of latticed transmission towers: F. G. A. AI-Bermani and S. Kitipornchai

deformed state and a neighbouring (desired) deformed generalized plastic hinges located at the two extremities
state. The incremental equation of equilibium can be of an element. Plastic hinges are assumed to be elastic
stated as prior to the full plastification so that the initial stiffness
of the complete element corresponds to that of the elastic
[KT] [ A r l r = {AR} (1) beam. As the stress resultants at the ends of the element
increase, the hinges yield resulting in a reduction in the
in which Kr is the tangent stiffness matrix, 2w is the element stiffness. For a steel section the hinges are
incremental nodal displacement and ~ is the increment assumed to become fully elastic again upon unloading.
of the applied load. When solving equation (1), the The solution to the nonlinear response is obtained as
objective is to minimize the discretization process a sequence of linearized solutions in which either the
implied in this equation since we are dealing with large- load is modified by a residual force to maintain
scale structures. This can be achieved by adopting a ten- equilibrium, or the tangent stiffness is modified to
sor definition for the stress and strain that follows the reflect the extent of the development of plastic flow. In
particle in its movement in the updated Lagrangian this case the tangent stiffness matrix has to be augmented
frame of reference. Following this and using the princi- by a plastic reduction matrix, Ke. This matrix can be
ple of virtual displacement, one can write derived using the associated flow theory of plasticity
16
as

1 I,
IDIeL~leLIdV +
l 1,
17"bleNtdV [Kp] = - [KT] [G] [ [G] r [ K r
+ K, ] [G] ] -~ [G] T[Kr] (3)
+ i' IOleL6teNIdV = tW (2)
,Iv
in which G is the gradient matrix of the yield surface and
Kh is a matrix containing the plastic moduli coefficients
in which eL and N are the incremental linear and of the material. For elastic perfectly plastic material,
nonlinear strain components of the Green-Lagrange this matrix becomes a null matrix.
incremental strain tensor, ~ denotes the Cartesian com- A solution method suitable for elasto-plastic nonlinear
analysis of large-scale structures has been presented 16.
ponent of the Cauchy stress tensor, D is the material
matrix, W is the virtual work increment and V is the ele- Since transmission towers are almost invariably con-
structed of angle sections, a single-equation representing
ment volume. In equation (2) the left superscript denotes
the configuration in which the quantity occurs while a the stress-resultant yield surface for angle sections under
a combination of axial force and biaxial moments
left subscript denotes the configuration from which the
presented by the authors 22 may be used. Approximate
quantity is measured, the absence of any superscript
yield surfaces for angle sections are shown in Figure 3
indicates an increment between configurations C~ and
for varying normalized axial (compressive and tensile)
C~. force values. A single-equation describing the yield sur-
The linear, geometric and deformation stiffness
matrices, KL, Kc and KD can be determined from the faces can be expressed
first, second and the third integrals, respectively, on the
left-hand-side of equation (2). Hence the tangent stiff- 4
ness obtained takes into account not only the stress state (p, m,,, m:) = 27 ,3(~ _ 1)
embedded in Kc but also the deformation state of the
element as reflected in Ko. In this approach the ele- + ([2 + /x) 3 sign(l, p)
ment deformations are imposed as an initial imperfec-
tion on the line element, hence preserving the
integration over a straight configuration but at the same - 3([2 + #) sign(l, p)
time introducing the necessary coupling between the
membrane and flexural deformations. This greatly
reduces the number of elements needed to accurately + J(O + ~)2 + 4, = 0 (4)
model the tower structure in a nonlinear large displace-
ment analysis. The linear stiffness matrix KL is
available in standard texts 2. The geometric and defor- in which p, my and mz are the normalized axial and
mation stiffness matrices, Kc and Ko, for a general bending moments about the centroidal axes parallel to
thin-walled beam-column element have been presented the legs, and the coefficients q~, ?, if, #, fL and ~ are
by Kitipornchai and Chan 2~ and by AI-Bermani and expressed in terms of p, mr and m=22.
Kitipornchai ~3, respectively.

Material nonlinearity Solution procedure


For large-scale structures such as transmission towers, For a given load increment {AR}, the incremental
modelling of material nonlinearity based on the assump- equilibrium equation (1) is solved for the corresponding
tion of lumped plasticity, coupled with the concept of a displacement increment [Ar }. Using the nodal
yield surface in force space, provide a compact and displacements and the incremental constitutive law, the
practical method for modelling nonlinear global struc- incremental resisting forces of the structure can be
tural behaviour. The stress resultants in the cross-section obtained. These are then compared with the externally
interact to produce yielding for the section. Any plastic applied forces to obtain the out-of-balance forces,
behaviour is deemed to be concentrated at the familiar {AR, }, which must be dissipated through an iterative

262 Engng. Struct. 1993, Volume 15, Number 4


Nonlinear FE analysis of latticed transmission towers." F. G. A. AI-Bermani and S. Kitipornchai

Y convergence criteria has been adopted using the Eucli-

1.0
pFz my
....
,Compression
p -ve
Tension
D +ve
dean norm measure with a convergence tolerance set to
5%.
When analysing a large-scale structure such as a
transmission tower, the self-weight of the structure has
to be considered. In the present analysis the tower's self-
weight is generated automatically and applied incremen-
0.8
tally on the tower prior to the incremental application of
0.6 external loads.

0.4 Configuration processing


In this work the formex algebra approach 24 has been
0.2 used to generate the necessary data for describing the
topology and geometry of the transmission tower struc-
ture. This approach has been implemented as a
-0.2
preprocessor for the automatic generation of the tower
topology, geometry, loading and supporting conditions.
-0.4 A detailed description of the formex formulation of
transmission towers has been given elsewhere 25 and
-0.6 will not be repeated here.

-0.8

-1.o1
Practical applications
The predicted nonlinear response of four different full-
scale towers is presented using the developed software,
Figure 3 Yield surfaces for angle section AK TOWER. Full-scale testing of each of these towers
was performed in Australia by various contractors. For
the first two towers, the nonlinear analysis was per-
procedure subjected to some imposed constraint condi- formed first and the full-scale test was conducted with
tions depending on the solution strategy selected. due consideration of the analytical results, while for the
In the present method, the effect of nonlinearity is remaining two towers the full-scale test was performed
treated as an effective load in conjunction with the before the nonlinear analysis. These four towers are
applied loads for general equilibrium. This reduces the shown in Figure 4. No record of member forces was
need for continuous updating of the tangent stiffness taken during any of the full-scale tests. For this reason,
matrix and results in a substantial saving in computation the comparison will be mainly based on the ultimate
time, especially for large-scale structures. Further, a load. The load applied on the tower is presented as a
more stable algorithm can be achieved when the predic- load factor X. The specified design ultimate load cor-
tor part of the algorithm (i.e. the tangent stiffness) is responds to X = 1.0. The formex formulation has been
kept as simple as possible while a strict force recovery used to generate the tower topology, geometry, loading
procedure is employed. and boundary condition for all the towers presented in
The solution strategy chosen was the arc-length this section.
method 23. In order to avoid the case where the force-
point for a certain element jumps from within the yield Karana - Rocklea 2 75 kV double circuit tower
surface to a point outside the surface during a loading As shown in Figure 4(a), this tower has a height of
cycle, and so as to avoid the case where the force-point 36.7 m with a square base of 5.5 m 5.5 m. It was
deviates excessively from the curved yield surface, a designed by Queensland Electricity Commission (QEC)
simple solution advancement control method has been and detailed by Transfield Pty Ltd 26. One special
used in conjunction with the arc-length method. The feature is that it has a rotating crossarm attachment. Six
solution advancement control has been achieved using loading cases were specified by QEC, these are respec-
the arc-distance from the previous cycle and the max- tively: (i) intact/maximum wind condition; (ii) broken
imum value of the yield function in the last two cycles maintenance on the earthwire peak; (iii) broken ea-r--
to extrapolate a maximum arc-distance for the present thwire condition; (iv) broken bottom phase; (v) broken
cycle. This brings the force-point gradually to the yield middle phase; and (vi) broken upper phase.
surface and guards against excessive deviation from the The tower was modelled using 992 beam-column
surface. elements and 446 nodes. Due to the rotating nature of
The solution method is a nonlinear incremental the crossarm attachments, an equilibrium configuration
method. Within each loading cycle several iterations are for the attachments under the specified load case had to
performed to account for the effect of the variation in be determined prior to the analysis. The nonlinear
stresses and strains during the cycle. This iteration pro- analysis has shown that the tower will sustain 100% (i.e.
cess is terminated whenever equilibrium, as defined by load factor X = 1.0) of the design ultimate load for load
a certain convergence criteria, is satisfied and the force- cases 1 - 5 with no problem. Under load case 6 (see
point for any element which has entered the plastic stage Figure 5), the nonlinear analysis predicted an ultimate
returns to the yield surface. An out-of-balance force load of h = 0.97 with a substantial number of plastic

Engng. Struct. 1993, V o l u m e 15, Number 4 263


Nonfinear FE analysis of latticed transmission towers." F. G. A. AI-Bermani and S. Kitipornchai

Ross-Chalumbin
Marulan

Nebo-Ross

Karana-
Rocklea

0
o
0
o

l
v

(a) (b) (c) (d)


Figure 4 Towers used in practical applications

3.4 3.4
I0.71 . - ~ . ~ . = _ ~L--10.71
1.0, I ! I / I I ~-="

4, ; , 6 L -'~ /'236
21 26.3
6.5 [ 30.4 ~ 2.3 0.8 I

1120,
8.5 -
17.6

I
~!2, ~._ 30.4 ~ 65.6
2,'-'76!
,,<
t-
O 0.6
o
17.6 ~17.6 a~
~1 21
7- { 30.7 - ~f["~ 65.6 "o
~ 0 . 4
o
6.4 I .J
9.3 - I
0.2 Load case 6
7.4 '
11.4=
l f l j y (L)
I 9" x (T) I I I I I I
" / i / / / / , ' / / , ' / , v - 7/11/11/~
0 50 100 150 200 250 300 350

Load case 6 Deflection L E R , mm

Figure 5 Loading condition and theoretical l o a d - d e f l e c t i o n curve for Karana-Rocklea tower

264 Engng. Struct. 1993, Volume 15, Number 4


Nonlinear FE analysis of latticed transmission towers: F. G. A. AI-Bermani and S. Kitipornchai

hinges spreading in the bracing of the common body of Marulan 500 kV tension tower
the tower.
In view of the nonlinear analysis results, it was This tower, as shown in Figure 4(b), has a height of
decided that the test should go ahead with the design 57.3 m with a square base of 12.1 m 12.1 m. The
ultimate load for load case 6 reduced to 95 % of the tower was designed and fabricated by Electric Power
originally specified load. The tower performed well Transmission (EPT) Pty Ltd. Seven load cases were
under these six tests confirming the nonlinear analysis specified for the full-scale test 27. The nonlinear analysis
predictions. One minor failure occured at the very was performed prior to the full-scale test to ensure the
beginning of the first test at the middle crossarm attach- adequacy of the tower.
ment due to a misfabrication which was corrected. The The tower was modelled using 1616 elements and 648
predicted load deflection curve for load case 6 is shown nodes. The nonlinear analysis predicted that the tower
in Figure 5. The deflected shape of the tower under this was strong enough to withstand the specified seven load
test with X = 0.97 is shown in Figure 6. cases. The lowest load factor obtained was X = 1.1 for
load case 7. The design ultimate load and the predicted
load deflection curve for this load case are shown in
Figure 7. The deflected shape of the tower under this
load case with X = 1.1 is shown in Figure 8. The
analysis predicted that plastic hinges would start
developing at ~, = 0.75. However, these hinges remain
confined to the earthwire arm until the ultimate load of
2 = 1.1 is reached. The analysis also revealed that a
Load c a s e 6 plastic hinge develops at the main bracing member of the
leg assembly at an early stage of the loading. This hap-
;k= 0.97 pened under all load cases. This hinge unloaded at a later
stage of each test. It was recommended that this bracing
member be upgraded, and this recommendation was
implemented. The full-scale test results agreed with the
nonlinear analysis that the tower is strong enough to
withstand the specified load cases.

Ross-Chalumbin 275 kV double circuit heavy


suspension tower
This tower is shown in Figure 4(c). It has a height of
59.6 m with a rectangular base of 7.7 m x 5.7 m. The
tower was designed and detailed by EPT Pty Ltd. Six
loading cases were specified. Details of the test pro-
cedure and results have been reported by EPT 28.
The tower performed adequately under the first five
load cases. However, when tested under the sixth load
case which represented the intact/maximum wind condi-
tion the tower collapsed at 90% of the design ultimate
load shown in Figure 9. The failure was described in the
test report 28 as 'an unusual failure which involved a
general collapse of the compressed face of the tower'.
Figure lO(a) shows a photograph of the failed tower
after the test.
This tower was modelled using 1732 elements and 768
nodes. The nonlinear analysis predicted that, under load
case 6, the first plastic hinge is formed at a load factor
X = 0.7 in the compression leg just below the common
body of the tower. Subsequent plastic hinges spread
rapidly to other parts of the tower below the common
body accompanied by a rapid increase in displacement.
A maximum load factor X = 0.91 was reached followed
by a breakdown of the numerical method indicating col-
lapse of the tower. This compares extremely well with
the 90% failure load reported in the test 28. The load
deflection curve for the tower under this load case is
shown in Figure 9. The deflected shape of the tower at
collapse is shown in Figure lO(b). The shape resembles
closely the actual collapse mode obtained during the test
as depicted by the photograph taken after failure in
Figure 6 Predicted deflected shape of Karana-Rocklea tower Figure lO(a).

Engng. Struct. 1993, Volume 15, Number 4 265


Nonlinear FE analysis of latticed transmission towers." F. G. A. AI-Bermani and S. Kitipornchai

1.2 I I i I
10.83 10.83
6.83 -~ J-11.37
1.0
23.86--|45.39 4 5 . 3 9 r 32.32 X=0.91
23"86~145.39 45.391 " 3 5 " 7 8
23.86 "145.39 45.39~--40.66 .< 0.8
20.5 t,. Hinges_ start at / -
--15.4 0
X
--35 ~ 0.6
--35 10
0
-- 0.4

0.2 Load case


V L

I I I I
Load case 6 50 100 150 200 250
Vertical deflection, VER (mm)

Figure 9 Loading condition and theoretical load-deflection curve for Ross-Chalumbin tower

ld case 6
X=0.91

Figure I0 Photograph of failed Ross-Chalumbin tower and predicted tower deflected shape

Engng. Struct. 1993, Volume 15, Number 4 267


Nonfinear FE analysis of latticed transmission towers. F. G. A. AI-Bermani and S. Kitipornchai

164 is6
1.0 . . . . . . X_= 0 . 9
2.2Q - 3 ~
-< 0.8 - ~'=0-75~ t _
16.5
lO2.2/159.1 0
101.9J157.4 o 0.6 -
9.7 0~ Theory
3.6 Nm
102.7 J~ 53,5
5.3~ "10
6.1Q ~ 0.4 -

Test no.2
9.6Q 0
10.9Q .._1 Vertical deflection
0.2- at X l

0 1 1
-0.5 0 5.0 10
Deflection, VX 1 (ram)
(a)
Figure 1 1 Load condition and theoretical l o a d - d e f l e c t i o n curve for N e b o - Ross t o w e r

the tower in order to achieve the specified load level.


When the load increment is applied to a particular point,
however, loads at other locations are relaxed and they in
turn must be re-adjusted to the same percentage level.
The sequence of loading and re-adjustment every time a
certain load increment is applied to the structure cannot
be simulated easily in the analysis. This effect becomes
more significant as the structure begins to yield. Hence,
some discrepencies between the analysis results and the
test can be expected.
! case 2
Conclusions
= 0.75 Accurate structural analysis of transmission towers is
complicated because the structure is three-dimensional
and comprised of asymmetric angle section members
eccentrically connected. The influences of geometric
and material nonlinearities play a very important role in
determining the ultimate behaviour of the structure.
This paper describes a nonlinear analytical method in
which all factors affecting the ultimate behaviour of the
tower structure can be incorporated. These include
geometric and material nonlinearities, joint flexibility
and the effects of large deflection. The developed soft-
ware, the AK TOWER program, has been used to
predict the ultimate structural behaviour of four different
electric transmission towers tested in Australia. The
nonlinear analysis was performed prior to the full-scale
testing for the first two towers, to check the design ade-
quacy and avoid any possible collapse condition. For the
remaining two towers, the nonlinear analysis was per-
formed after full-scale testing to investigate the unex-
pected collapse encountered in the test. Predictions of
the ultimate loads and the failure deflected shapes have
generally been very good, considering the complexity of
this type of structure. No comparison of calculated
member forces and actual member forces has been made
because of the lack of such data in the test.

Acknowledgments
The work in this project has been supported by funds
Figure 12 Predicted deflected shape of N e b o - R o s s to w e r from the Australian Electricity Supply. Industry

268 Engng. Struct. 1993, Volume 15, Number 4


Nonlinear FE analysis of latticed transmission towers. F. G. A. AI-Bermani and S. Kitipornchai

Research Board (AESIRB). The authors wish to thank thin-walled structures using least element/member', J. Struct. Engng,
the Queensland Electricity Commission (QEC) for mak- ASCE 1990, 116, (1), 215-234
ing the full-scale test data of tested towers available. In 14 Chan, S. L. and Kitipornchai, S. 'Inelastic post-buckling behaviour
of tubular struts', J. Struct. Engng, ASCE 1988, 114, (5),
particular, the authors wish to thank Mr Henry Hawes 1091 - 1105
of the QEC for his continued advice and technical sup- 15 Kitipornchai, S., Al-Bermani, F. and Chan, S. L. 'Geometric and
port. material nonlinear analysis of structures comprising rectangular
hollow sections', Engng Struct. 1988, 10, (1), 13-23
16 AI-Bermani, F. G. A. and Kitipornchai, S. 'Elasto-plastic large
References deformation analysis of thin-walled structures" Engng Struct. 1990,
12, (I), 2 8 - 3 6
1 American Society of Civil Engineers Committee on Electric 17 Jones, S. W., Kirby, P. A. and Nethercot, D. A. 'The analysis of
Transmission Structures. 'Loading of electrical transmission struc- frames with semi-rigid connections - a state of the art report', J.
tures', J. Struct. Engng, ASCE 1992, 108, (ST5), 1088-1105 Const. Steel Res. 1983, 3, 2 - 1 3
2 American Society of Civil Engineers Committee on Electric 18 Al-Bermani, F. and Kitipornchai, S. 'Elasto-plastic nonlinear
Transmission Structures. 'Guidelines for transmission line structural analysis of flexibly-jointed space frames', J. Struct. Engng, ASCE
loading', ASCE, New York, 1984 1992, 118, (1), 108-127
3 Roy, S., Fang, S. and Rossow, E. 'Secondary stresses on transmis- 19 Kitipornchai, S., AI-Bermani, F. and Peyrot, A. H. 'Effect of bolt
sion tower structures', J. Energy Engng, ASCE 1984, 110, (2), slippage on the ultimate behaviour of lattice structures' (to be
157- 172 published)
4 American Society of Civil Engineers. 'Guide for design of steel 20 Przemieniecki, J. S. Theory of matrix structural analysis, McGraw-
transmission towers', Manuals and Reports on Engineering Practice, Hill, New York, 1968
No. 52, ASCE, New York, 1988 21 Kitipornchai, S. and Chan, S. L. "Nonlinear finite element analysis
5 American Society of Civil Engineers. 'Guide for design of steel of angle and tee beam-columns', J. Struct. Engng, ASCE 1987, 113,
transmission towers', Manuals and Reports on Engineering Practice, (4), 721-739
No. 52, ASCE, New York, 1971 22 Kitipornchai, S., Zhu, K., Xiang, Y. and AI-Bermani, F. G. A.
6 European Convention for Constructional Steelwork (ECCS). 'Single-equation yield surfaces for monosymmetric and asymmetric
"Recommendations for angles in lattice transmission towers', January sections', Engng Struct. 1991, 13, (4), 366-370
1985 23 Crisfield, M. A. 'A fast incremental/iterative solution procedure that
7 Structural Stability Research Council: SSRC. Guide to stability handles snap-through', Camp. Struct. 1981, 13, (1), 5-62
criteria for metal structures, (4th edn) (ed. T. V. Galambos), John 24 Nooshin, H. Formex configuration processing in structural engineer-
Wiley, New York, 1988 ing, Elsevier Applied Science, London, 1984
8 Eurocode. "Common Unified Code of Practice for Steel Structures', 25 AI-Bermani, F. G. A., Kitipornchai, S. and Chan, S. L. 'Formex for-
Eurocode No. 3, 1984, Commission of the European Communities, mulation of transmission towers', Int. J. Space Struct. 1992, 7, (1),
Directorate-General, Brussels 1--10
9 Electric Power Research Institute. 'Structural development studies at 26 Transfield Pty Ltd, Test Report: 275 kV D.C. T/L structure type
the EPRI transmission line mechanical research facility', Interim D2S15F, Sydney, Australia, February 1992
Report No. 1: EPRI EL-4756, August 1986, Sverdrup Technology, 27 Uniquest, Marulan 500 kV Double Circuit Tension Tower Type
Inc., Tullahoma, Tennessee, USA VSD, Report 320213, March 1991, Queensland, Australia
10 International Electrotechnical Commission (IEC). "Loading tests on 28 Electric Power Transmission (EPT) Pry Ltd, Test Report No. 497.
overhead line towers', Publication No. 652, 1979 '275 kV T/L Ross-Chalumbin D.C. heavy suspension tower
11 Bonneville Power Administration. 'Elastic design program', D2S2D', Test Report No. 497, February 1988, NSW, Australia
Portland, Oregon, 1987 29 Transfield Pty Ltd, 'Test Report, The Queensland Electricity
12 Lo, D., Morcos, A. and Goel, S. 'Use of computer in transmission Generating Board, Northern Region, NEBO-ROSS Transmission
tower design', J. Struct. Div., ASCE 1975, 101, (ST7), 1443- 1453 Line, Tower D2T15 + 12M Queensland, Australia', November,
13 AI-Bermani, F. G. A. and Kitipornchai, S. 'Nonlinear analysis of 1983

Engng. Struct. 1993, Volume 15, Number 4 269

Potrebbero piacerti anche