Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

OPTIMAL DESIGN OF STEEL TRANSMISSION POLES

By Fatma Y. Kocer1 and Jasbir S. Arora,z Member, ASCE

ABSTRACT: Design of steel transmission poles is fonnulated as an optimization problem by identifying design
variables, a cost function, and constraints. Nonlinearities in structural response calculations due to large deflec-
tions are included in the fonnulation. An example problem available in the literature is solved and the solution
is compared with the published one. Several other cases of the example are solved to investigate the effect of
certain parameters on optimal design as well as to show the flexibility of the optimal design process. Based on
the results of this study, it is concluded that the optimal design process can give more economic designs compared
to the conventional design process. Once the optimal design problem is fonnulated and implemented into a
software, many variations in the problem conditions and parameters can be studied in a relatively short time,
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/09/15. Copyright ASCE. For personal use only; all rights reserved.

which can lead to better designs.

INTRODUCTION Additional cases are solved to investigate the effect of certain


parameters on the design as well as to show the flexibility of
According to Randle (1985), the use of steel poles in trans- the process. Finally, analyses and conclusions are presented.
mission lines is a relatively recent trend compared to the use
of latticed towers, going back only about 35 years. Design PROBLEM DESCRIPTION
procedures for these poles have evolved over this period. In
this paper, an optimal design process for steel transmission Poles can be classified as dead-end and tangent-type de-
poles which can lead to more economic designs is presented. pending on their location in the transmission line. Dead-end
There is a significant amount of literature on the design of poles are located at line termination; therefore, they are de-
transmission poles and towers but little study has been done signed to withstand unbalanced loads from the conductors in
for their optimal design. Ghannoum and Yaacoub (1989) have one direction together with the wind load and other conductor
studied the optimization of transmission towers and founda- loads. As a result, dead-end pole conductors induce loads in
tions; however, that study is a comparison of the cost of two vertical, transverse, and longitudinal directions as opposed to
tower designs rather than obtaining designs through a system- tangent-type poles that have only vertical and transverse com-
atic optimization procedure. ponents. Tangent-type poles should also be designed for un-
The basic purpose of this paper is to introduce a practical balanced loads that result from ice or a broken conductor. H-
formulation for the optimal design of steel poles and solve an frame poles are two or three poles connected together to carry
example problem. Also, advantages and disadvantages of the the wires whereas the others are classified as single poles.
optimal design process over the conventional design process Single poles are divided into two groups, namely self-sup-
will be discussed. Another purpose of the paper is to illustrate porting and guyed poles, depending on how they withstand
the process of transcribing a design problem into an optimi- the loads (Design 1990; Guidelines 1991). The pole cross sec-
zation problem. It is observed that the optimal design process tion can range from a round to a dodecagonal (12-sided pol-
also needs all the basic steps of the conventional design pro- ygon) depending on the material for the poles.
cess. The design loads need to be determined, the structure To demonstrate the optimal design problem formulation and
needs to be analyzed for each loading condition, and safety comparisons between optimal and conventional designs, only
and service requirements need to be met. The optimal design the problem presented in Design (1990) is considered in this
process, however, is quite fonnal and all its requirements and
conditions need to be precisely predefined and mathematically
formulated. The design variables must be clearly identified, a
function that needs to be optimized must be defined, and all
design requirements must be precisely formulated. The quality
of the optimal solution depends heavily on the problem for-
mulation.
In the following section, the geometry, material properties,
and design loads for the poles are described. That section is
followed by the analysis of the pole structure, where nonli-
nearities due to large deflections are included in calculations.
The optimal design process-identification of design varia- H
bles, and formulation of cost and constraint functions in terms
of the design variables-is presented in the problem formu-
lation section. Next, the optimal design for an example prob-
lem is presented and compared to the conventional design.
IGrad. Res. Asst., Dept. of Civ. and Envir. Engrg., The Univ. of Iowa,
Iowa City, IA 52242.
'Prof., Dept. of Civ. and Envir. Engrg., The Univ. of Iowa, Iowa City,
IA.
Note. Associate Editor: Scott A. Burns. Discussion open until April I,
1997. To extend the closing date one month, a written request must be
filed with the ASCE Manager of Journals. The manuscript for this paper
was submitted for review and possible publication on February 20, 1996.
This paper is part of the ]ourlUll of Structural Engineering, Vol. 122,
No. 11, November, 1996. CASCE, ISSN 0733-9445/96/0011-1347- FIG. 1. Self-Supporting Single Transmission Pole (O••/gn
1356/$4.00 + $.50 per page. Paper No. 12680. 1990)

JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1996/1347

J. Struct. Eng. 1996.122:1347-1356.


z z
work. However, other types of pole structures and towers can
be formulated and treated in similar ways. The pole in this
problem is a self-supporting single pole as shown in Fig. 1 (aj
is the vertical distance from the tip of the pole to the ith con-
ductor, hi is the horizontal distance from the undeflected pole
axis to the tip of the ith conductor, and H is the height of the
pole above the ground). The cross section is a dodecagonal.

Geometry of Poles
(a) (b)
Overall Geometry
FIG. 3. Cross Sections (Design 1990): (a) Round; (b) Dodecag-
Steel transmission poles are designed and manufactured in onal
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/09/15. Copyright ASCE. For personal use only; all rights reserved.

pieces depending on the safety, manufacturing, and transpor-


tation requirements. The loads acting on the pole increase to- TABLE 1. Geometrical Properties of Round and Dodecagonal
wards the butt of the pole. Therefore, to satisfy the safety Cross Sections (Design 1990)
requirements, it may be necessary to manufacture the lower Property Round Dodecagonal
pieces from thicker plates. As the manufacturing cost increases (1 ) (2) (3)
significantly with the increasing number of pieces, it is com- A, 3.14(Do - tit 3.22(Do - t)t
mon to use only two pieces (Randle 1985). These pieces can I, =I, 0.393(Do - t)'t 0.41 I (Do - t)'t
be assembled by a slip or a flange joint. The tapering is the C, 0.500Do cos(a), a = 45' =
0.518Do cos(a), a 15',45',75'
same in different pieces but there is a discontinuity in the C, 0.500Do sin(a), a = 45' =
0.518Do sin (a), a 15',45',75'
r 0.354(D. - t) 0.358(Do - t)
joints as can be seen in Fig. 2 where D to = tip outside diameter, Max.QI(lt) (O.637)1[(D. - tit] (0.631)/[(Do - tit]
m (ft); tk = thickness of the kth piece, m (ft) k = I, 2, ... , m; Max.CIi (O.637D.)I[(D. - t)'t] (0.622Do )/[(Do - t}'t]
m = total number of pieces; S = tapering, mlm (ft/ft); x = w 0.268(Do - 2t - 2BR)

distance measured from the tip of the pole, m (ft); and Do =


outside diameter for round sections, across flats for polygonal
sections, m (ft) (across flat is the distance between two op-
posing parallel lines in polygonal members, as shown in Fig.
3).
The discontinuity that occurs in the joint should be taken
into account when Do is calculated. This can be done by fol-
lowing the fabricator's joint detailing that gives the locations FIG. 4. Definitions of Cross-Sectional Terms (Design 1990)
of the beginning and ending points of the joint. Using this
information, Do can be calculated accurately. In the example sign of Steel Transmission Pole Structures (1990). In that man-
solved, Do has been calculated using the values given in De-

T~i-r----ril
ual, pole section properties at certain heights have been pro-
vided in a table. These properties include the diameter,
thickness, moment of inertia, and cross-sectional area. The
heights where the slip joint starts and ends are some of the
locations where these data are given. An equation relating the
diameter below the slip joint to the diameter at the tip is ob-
tained using these values. The diameter of the pole below the
joint is calculated using this equation.
x(m)
Cross-Sectional Geometry
The steel transmission poles have tubular cross sections var-
ying from a round section to a dodecagonal section. With the
changing geometry the cross-sectional properties such as area,
moment of inertia, and radius of gyration change. In Table 1,
the geometrical properties of the round and dodecagonal sec-
tions are given (Design 1990). Some of the terms used in Table
1 and Fig. 4 are defined as follows: Do = outside diameter for
round sections, across flats for polygonal sections, m (ft); As
= gross area, m2 (sq ft); t = thickness, m (ft); C = distance
from neutral axis to extreme fiber, m (ft); Cy = y-coordinate
of the extreme fiber, m (ft); C, = z-coordinate of the extreme
fiber, m (ft); I = moment of inertia, m 4 (ft4); Q = moment of
section about center, m 3 (cu ft); w = flat width of a side of a
polygon, m (ft); BR = effective bend radius, m (ft) = actual
BR, if BR < 4t = 4t, if BR > 4t; Max.Qllt = formula for de-
termining the maximum flexural shear stress, 11m2 (lIsq ft);
Max.CIl = formula for determining the maximum torsional
shear stress, 11m2 (llcu ft); r = radius of gyration, m (ft); and
a = angle between the x-axis and the corner of the polygon
(degrees).
The effective bend radius, according to the foregoing defi-
FIG. 2. Pole Section along Its Longitudinal Axis nition, cannot exceed 4t; however, the Manual of Steel Con-
1348/ JOURNAL OF STRUCTURAL ENGINEERING I NOVEMBER 1996

J. Struct. Eng. 1996.122:1347-1356.


L
struction (1980) limits the minimum bend radius to 4t for the
steel used in the example problem. From the data given in
Design of Steel Transmission Pole Structures (1990), it is seen
... (TO
Top view e=-f-TI
~
rTo
that BR = 4t has been used in all calculations. As the bend
... °r
VOfLTI
TI
radius is a material preparation detail that cannot be calculated, ~
VI VI
4t is used in the optimal design calculations.
r~
\2
r12
\2
Volume of Pole
As observed earlier, the steel poles are manufactured in
r
V3
T3
r
V
3
T3

pieces. In practice, these pieces are usually fabricated in mul-


tiples of 15.24 m (50 ft). If the total length has a remainder
from the multiples of 15.24 m (50 ft), the top piece is fabri-
cated with a length equal to the remainder. If the total height
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/09/15. Copyright ASCE. For personal use only; all rights reserved.

is assumed to be between 15.24 m (50 ft) and 30.48 m (100


ft), the volume of the pole is the addition of the volumes of (8) (b)
two pieces as FIG. 5. Load Components In: (a) NESC Light Loading, ASCE
Vs = ct\[(D,o - t,)(H - 15.24) + S(H - 15.24)2] Ice and Wind Loading; (b) Broken Conductor Loading (Design
1990)
+ ct2 [(D 1 - t2 )15.24 + S(l5.24i] (1)
effect on them besides their own weight. The longitudinal
where c = area constant ('Tl" for round sections, and 3.22 for loads due to unbalanced conditions are not considered in this
dodecagonal sections); D 1 = outside diameter or across flat at case; therefore, the conductor loads have only vertical (VI) and
the connection point of the pieces, m (ft); t l = thickness of the transversal (Ti ) components. The wind load acting on the pole
first piece, m (ft); and t2 = thickness of the second piece, m is calculated due to a uniform wind pressure q [Fig. 5(a)].
(ft).
Loading Case 2: ASCE lee and Wind Loading
Material Properties This loading case has the same components as the NESC
Combination of tensile stress, temperature, strain rate, and light loading but their magnitudes are different. Icing is con-
geometrical discontinuities as well as other design and fabri- sidered in this case, therefore the vertical components of the
cation factors may cause brittle fracture in structural steels conductor loads are larger. On the other hand, since the wind
(Design 1990). As the relationships between these factors are pressure is not too high, the transverse loads and the wind
not known, their effects cannot be calculated. Therefore, brittle load acting on the pole are smaller [Fig. 5(a)].
fracture can be avoided only by the selection of appropriate
steel. Steel used in these poles is manufactured from structural Loading Case 3: Broken Conductor Loading
grade plates with the yield strengths up to 448 MPa (65 ksi) In this case, the first conductor is assumed to be broken. As
and a modulus of elasticity of 199,948 MPa (29,000 ksi). The a result, there is a longitudinal load (L) acting on that con-
yield strength F y , the ultimate tensile strength F u , and the mod- ductor in addition to the vertical and horizontal loading. There
ulus of elasticity E are determined according to ASTM A370. is also a wind load acting along the pole [Fig. 5(b)].
The mass density of the structural steel is 7,849 kg/m 3 (490 All three loading cases also have dead load due to the
Ib'm/cu ft). weight of the pole. The first two loading cases, NESC light
and ASCE ice and wind, represent the normal operating en-
Design Loads vironment for the pole structures. To increase longitudinal ca-
pacity of the pole, broken conductor loading, which has lon-
Loads acting on a transmission line are mainly due to the gitudinal loads, is also imposed (Design 1990).
forces applied on the wires andlor forces applied directly to The design approach presented in the Design of Steel Trans-
the supporting structure. Loads for poles are divided into two mission Pole Structures (1990) is based on the ultimate
main groups as weather-related loads and loads such as acci- strength method where the loads include the desired factors of
dental loads, and construction and maintenance loads (Guide- safety. The basis for the approach presented in the foregoing
lines 1991). For a complete discussion of these loads, Design guide is the specifications of the American Institute of Steel
(1990), Krauthemmer (1987), and Guidelines (1991) should Construction (AISC) and American Iron and Steel Institute
be consulted. (AISI). However, these specifications are based on working
The transmission lines have been designed primarily based loads and an allowable stress design approach. To account for
on the National Electrical Standards Code (NESC) (National the difference, all allowable stresses from AISC specification
1977). More recently, however, a reliability-based load and have been multiplied by a factor ranging from 1.67 to 2.0
resistance factor (LRFD) design procedure has been developed (Design 1990). These adjustments are valid for the tubular
and accepted as an alternative or a supplement to the NESC members; the AISC specification should be consulted for non-
(Peyrot and Dagher 1984). The manual Guidelines for Trans- tubular members.
mission Line Structural Loading (1991) has been prepared ac-
cording to LRFD. STRUCTURAL ANALYSIS
The example problem solved in this paper is taken from the
Design of Steel Transmission Pole Structures (1990). In this The design of steel transmission poles requires calculation
example, a pole is to be designed such that it can withstand of quantities such as the design moment, stresses, and deflec-
the following three loading cases: tion. This section explains calculation of the design moment,
compressive stress, bending stress, shear stress, and deflection.
Loading Case 1: NESC Light Loading Determination of Design Moment
The loads are due to conductors and the transverse wind The design moment is determined for three loading cases
acting on the poles. The conductor loads are due to the wind given in the "Design Loads" section. The first two loading
JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1996/1349

J. Struct. Eng. 1996.122:1347-1356.


Determination of Design Stresses
Axial Compressive Stress
a The design axial compressive stress is found by dividing
x the axial load by the cross-sectional area. In self-supporting
b single poles axial loads are due to the vertical loads from the
q
conductors and the self-weight of the structure, and can be
I====f~ expressed as follows:
v p= Vo + Vt + ... + Vn + VsPw (4)
3
where Pw = unit weight of the steel, 76,973 N/m (490 Ib/cu
Y---I---~------+y ft) and
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/09/15. Copyright ASCE. For personal use only; all rights reserved.

(8)
Vs=c f(D,o+2S~-tt)ttd~, ifx«H-15.24) (5a)
z
V.=ct1[(D,o - t)(H - 15.24) + S(H - 15.24f]
FIG. 6. Design Moments In Pole

cases-NESC light and ASCE ice and wind-have transver- +c r


JH-15.24
[D t + 2S(~ - H + 15.24) - t2]t2d~,
sal and vertical loads from the conductors, besides the weight
of the pole and the wind acting on the pole itself. Therefore, ifx ~ (H - 15.24) (5b)
they cause a bending moment only around the z-axis (Fig. 6).
The third loading case (broken conductor loading) has a lon- with c and D 1 as given in (1). As a result design compressive
gitudinal load from the first conductor in addition to the afore- stress is.
mentioned loads. As a result, it causes moments around the x- p
and y-axis as well as the z-axis. The equation for the bending !o=- (6)
moment about the z-axis is given first. Following that, equa-
Ag
tions for the moments about the x- and y-axes are derived. where Ag is given in Table 1.
The bending moment about the z-axis results from the trans-
versal and vertical components of the conductor loads and the Design Bending Stress
wind load on the pole, and is given at any point x as (Fig. 6) The design bending stress, due to the design moment in the
M,= -[To(x - ao) +Tt(x - at) + ... + Tn(x - an)] - (Vob o + Vtb t cross section, is given as

+ ... + Vnb n) - [q(O.50D,ox 2 + 2/3Sx3 )], ifx«H- 15.24) M,Cy MyC,


!b=--+-- (7)
(2a)
I, Iy
where C and I are listed in Table 1, and M, and My are given
M,= -[To(x - ao) + Tt(x - a,) + ... + Tn(x - an)] - (Vob o
in (2) and (3).
+ V1b t + V2b 2 + '" + Vnb n) - {q[0.50D,o(H - 15.24)2
Design Shear Stress
+ 2/3(H - 15.24)3]} - {q[0.50D,o(x - H + 15.24f The shear stress is due to both shear force and torsion. flex-
+ 2/3(x - H + 15.24)3]}, if x ~ (H - 15.24) (2b) ural shear stress is due to the transverse loads from the con-
ductors and the wind load on the structure, and torsional shear
where n = number of conductors; and (y) = y if y > 0 and stress is due to the twisting moment. The loads causing flex-
(y) = 0 if y :5 O. In (2) if any conductor i is located below x, ural shear stress are as follows:
Vjb j is not included in the expression. Attention must be paid
to the sign of the moment, as the location of V may cause j
V= To + Tt + ... + Tn + q(D,o + Sx)x. ifx< (H - 15.24)
moment opposite to others. (8a)
The moment from the vertical concentrated loads is calcu- V= To + Tt + ... + Tn + q[Dro + S(H - 15.24)](H - 15.24)
lated using appropriate lever arms.
As the pole deflects the distance between the point of load + q[D t + S(x - H + 15.24)](x - H + 15.24),
and the longitudinal axis of the undeflected pole increases,
ifx~(H-15.24) (8b)
inducing secondary moments. An iterative method used in the
calculation of secondary moments is explained in the •'Nu- Total shear stress is calculated as
merical Examples" section. The center of gravity of the pole
shifts due to deflection, resulting in moments caused by the f,v
=VQ
It
+ MxC
J (9)
weight of the pole. Therefore, once the pole deflects the mo-
ment due to self-weight should also be included in the moment where max[Q/(It)] and Max(C/J) are given in Table 1 and M x
calculations. In calculating this moment, the centroid of a is given in (3).
piece is assumed to be located at the midpoint.
The design moment in the broken conductor case should Design Compressive Stress
include the effect of the longitudinal load resulting in two
more components The design compressive stress is due to the axial loads and
the bending moment. It is simply the addition of (6) and (7).
Mx=-Lb; My=-L(x-at) (3a,b)
Determination of Deflection
where L = longitudinal load from the broken conductor; N (lb);
and M x = torsional moment. If the conductor is located below Expressions for deflection of the pole in the y- and z-direc-
x, b should be taken as 0 in (3). tions are
1350 I JOURNAL OF STRUCTURAL ENGINEERING I NOVEMBER 1996

J. Struct. Eng. 1996.122:1347-1356.


6. y= i ~IM.Xdx; i ~IMyxdx
H

6..=
H

(1Oa,b)
decagonal sections are given. For other cross sections, Design
of Steel Transmission Pole Structures (1990) can be consulted.
Round members subjected to axial compression and bend-
These equations can be integrated by any numerical method, ing should satisfy the following condition:
such as Simpson's rule (Chapra and Canale 1985). The final
deflection is calculated as f.+fb SI (13)
Fa Fb
(11)
wheref. andfb are given in (6) and (7) and Fa, MPa (ksi) and
OPTIMAL DESIGN PROBLEM FORMULATION F b , MPa (ksi) are calculated as follows [the units for Fy are
MPa (ksi) in the following expressions]:
In this section, the problem of designing steel poles is for-
mulated as an optimization problem. Following the procedure
given in Arora (1989), the design variables are identified, and when Da S 26,200 (Do S 3,800): Fa = Fy (14a)
t Fy t Fy
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/09/15. Copyright ASCE. For personal use only; all rights reserved.

a cost function and design constraints are defined.


when 26,200 S Do S 82,737 (3,800 S Do S 12,000):
Design Variables Fy t Fy Fy t Fy
The design variables for a problem are the ones that define
the system (Arora 1989). For steel poles made of two pieces 6,550 ( 950)
Fa=O.75Fy + - - Fa=0.75Fy + - (14b)
they are identified as D ro = outside diameter at the tip of the Doll Doil
pole, m (ft); t 1 = thickness of the first piece, m (ft); S = taper,
m/m (ft/ft); and t 2 = thickness of the second piece, m (ft). when Do S 41,369 (Do S 6,000): Fb =Fy (15a)
The design variables are chosen as the geometrical proper- t Fy t Fy
ties of the poles; however, the material properties could be
chosen as design variables as well. To compare the optimal when 41,369 S Do S 82,737 (6,000 S Do S 12,000):
design with the conventional design, material properties are Fy t Fy Fy t Fy
not treated as the design variables. The diameter D ro is treated
12,411 ( 1,800)
as a continuous design variable. The variables t 1 and t2 are =
Fb 0.70Fy + -D - Fb =0.70Fy + -
olr D
-
olr
(15b)
discrete, and their allowable values are given by the available
plate thicknesses. The structural plates are manufactured with The allowable compressive stress Fa [MPa (ksi)], in polyg-
a thickness that can vary between 0.0 and 1.27 cm (0.5 in.). onal tubular members is given as
The increment in the thickness is 0.08 cm (1132 in.). S is
treated as a continuous variable. The type of design variables,
discrete/continuous, affects selection of the optimization Fa=1.45Fy (1.0-4.91 X 1O-4~~)
method (Huang and Arora 1995).

Cost Function
. [Fa =1.45Fy ( 1.0 - 1.29 X 1O-3~~) ] (16)
The cost of steel poles consists of material, manufacturing, According to the requirements, polygonal tubular members for
transportation, assembly, and maintenance costs. The manu- which the design compressive stress is less than the yield stress
facturing, transportation, assembly, and maintenance costs de- F y , should be proportioned so that
pend on the manufacturer, and the distances between the plant
and pole locations. As these factors are specific to a case, the -Ws -630(W
- s .240)
r.::- for dodecagonal members (bend angle =30)
cost function for the present formulation is taken as the ma- t Fy t V Fy
terial cost only. For specific situations, other costs can be for-
mulated in terms of the design variables and added to the (17)
material cost. They may also be estimated as a fraction of
structural steel cost. In that case the optimum design will not If the design compressive stress on the extreme fiber does not
change, but the cost function will be multiplied by a factor. exceed the allowable compressive stress, these limits on wit
Thus, the cost function for the problem is expressed as may be exceeded as follows:

Cost =cost of steel ($) ~


t
s 958
Fy
(~s
t
365)
~
(18)
(12)
For a dodecagonal section, the compressive stress constraint
where V. = volume of steel, given in (1); p. = unit price for is written for two different conditions as follows [Fa is cal-
steel (0.77 $/kg, 0.35 $/lb); and P. = mass density of steel culated using (16)]:
(7,849 kg/m 3, 490 lb'm/cu ft).
Iff. + f" S Fy , constraint of (17) should be imposed.
Design Constraints
Iff. + fb S Fa, constraint of (18) should be imposed.
The design constraints are formulated according to the De-
sign (1990) requirements. First, the constraint for compression
members, including local buckling effects, is defined. Shear Shear Stress Constraint
stress, bending stress, combined stress, and deflection con- The shear stress caused by the shear force and torsion
straints are then formulated. should be less than the allowable shear stress at any point x
Compressive Stress Constraint fv S Fv (19)

Requirements for compression members differ for different where fo = as given in (10); and F o = allowable shear stress
cross sections. In the sequel, requirements for round and do- (0.58Fy).
JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1996/1351

J. Struct. Eng. 1996.122:1347-1356.


Bending Stress Constraint shield wire (ao) and the horizontal distance from the unde-
fleeted pole axis to the tip of the shield wire (bo) are not given
The bending stress must be less than or equal to the allow- in the manual; these are assumed as 0.15 m (0.5 ft) and 0.0
able tensile stress and allowable compressive stress m. The pole is made of two pieces, 15.24 m (50 ft) and 13.72
(20,21) m (45 ft) long. Slip joint is used to connect the two pieces
and is located at 15.24 m (50 ft) above ground. Steel with a
where F, = allowable tensile stress (O.83Fu); and Fa = allow- yield strength (Fy) of 448 MPa (65.0 ksi) is used. Its ultimate
able compressive stress based on local buckling. tensile strength (Fu ) is 552 MPa (80 ksi) and modulus of elas-
ticity is 199,948 MPa (29,000 ksi). It has a unit weight of
Combined Stress Constraint 76,973 N/m3 (490 lb/cu ft).
Combined stress is evaluated by the distortion energy The pole is designed to withstand any of the three loading
(Hencky-Mises) yield criterion (Design 1990). The maximum cases described earlier. The loads for these cases are as follows
combined stress is not just an addition of maximum shear (Fig. 5):
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/09/15. Copyright ASCE. For personal use only; all rights reserved.

stress and maximum normal stress as these two stresses do not NESC light loading: To = 9,991 N (2,246 lb), T 1 = T2 = T3 =
occur at the same point. The maximum normal stress and max- 25,586 N (5,752 Ib)
imum torsional shear occur at the extreme fibers, whereas the
maximum shear stress occurs at the neutral axis. As a result, Vo = 1,459 N (328 Ib), VI = V2 = V3 = 8,960 N (2,014 Ib)
only normal stress and torsional shear are taken into consid-
eration for the calculation of the maximum combined stress.
q = 1,077 Pa (22.5 psf)
Although this is the case, the Design of Steel Transmission ASCE ice and wind loading: To = 8,025 N (1,804 Ib), T I = T2
Pole Structures (1990) has included the effect of both shear = T3 = 15,858 N (3,565 lb)
stresses in their formulation. Contradicting this formulation,
the example problem given there considers only maximum Vo =2,811 N (632 Ib), VI = V = V 2 3 = 10,872 N (2,444 Ib)
normal stress and maximum torsional stress while formulating
q = 570 Pa (11.9 psf)
the combined stress constraint, as follows:
2 Broken conductor loading: L = 26,689 N (6,000 Ib), To =8,025
(fa + fb) 2
+3 (MxC)
J ::s; F or Fa
2
y
2
for polygonal members N (1,804 lb)
T1 =4,942 N (1,111 Ib), T2 = T = 15,858 N (3,565 Ib)
3
(22a)
2 Vo =2,811 N (632 Ib), VI = V = V = 10,872 N (2,444 Ib)
2 3

(fa + fb)2 + 3 (M;C) ::s; F; or F; for round members q = 570 Pa (11.9 psf)
(22b) These loads include the load factors. While calculating the
where Fa = allowable compressive stress [(14) or (16)]; and dead load due to the weight of the pole, the load factor for
=
Fb [(15)] allowable bending stress. In (22), the bending and NESC is 1.5, and for ASCE it is 1.0.
shear stresses are always positive. For a single pole structure,
as the axial force causes only compressive stress, it should be Conventional Design
taken as negative when checking the tensile stresses and pos-
In the conventional design a nonlinear, finite-element-based
itive when checking the compressive stresses.
computer program was used for the analysis of the pole struc-
ture (Design 1990). Design of Steel Transmission Pole Struc-
Deflection Constraint tures (1990) reports the results of this analysis at certain
Deflection at the tip of the pole should satisfy both safety heights of the pole where the design requirements are checked.
and aesthetics requirements. Therefore, the deflection con-
straint is imposed as Optimization Techniques
(23) The optimal design problem is solved with /DESIGN 4.2
(Arora et al. 1993). This program has many algorithms. Dif-
where a = calculated using (11); and a. = an allowable de-
1I ferent combination of algorithms have been used to solve the
flection. problem. As the least expensive designs are obtained with two
of these approaches, only the results with them are reported
NUMERICAL EXAMPLES here. Also, since details of the approaches are given in Huang
In this section, an example problem that has been solved in and Arora (1995; in press, 1996), only their basic ideas are
the Design of Steel Transmission Pole Structures (1990) with explained.
the conventional design method is solved by the optimal de- The optimization techniques automatically improve an ini-
sign method. Using the results, comparisons are made about tial design estimate until a final feasible design is obtained that
the performance of the optimal and conventional design pro- minimizes the cost function for the problem. The first of the
cesses. Furthermore, effects of certain factors on optimal de- two approaches used in the present work is based on sequential
signs are investigated. quadratic programming (SQP) for continuous variables and the
branch and bound method for discrete variables. In the ap-
proach, all the variables are considered to be continuous ini-
Description of Problem
tially. The SQP method is used to obtain a continuous opti-
The example problem presented in the Design of Steel mum solution. The method assumes all functions of the
Transmission Pole Structures consists of a 28.96 m (95 ft) tall problem to be continuous and needs their gradients during its
structure having a shield wire and three conductors attached calculations (Arora 1989). In the present problem, these gra-
to it. The data for the conductors are: al = 3.05 m (10 ft), a2 dients are calculated using the central finite difference method.
= 7.62 m (25 ft), a3 = 12.19 m (40 ft); b l = b2 = b3 = 1.83 m During the optimization process, the method automatically
(6 ft). The vertical distance from the tip of the pole to the generates and uses second-order derivatives of the Lagrangian
1352/ JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1996

J. Struct. Eng. 1996.122:1347-1356.


function. Once a continuous solution is obtained using the TABLE 2. Comparison of Analysis Results for Present Pro-
SQP method, the branch and bound method is used to obtain gram with Conventional Design Program at 28.96 m (95 ft) for
NESC Light Loading Case
discrete values for the design variables. This method defines
many optimization subproblems and solves them using the Conventional Present
SQP method to "force out" a discrete solution. For each sub- Analysis design program program Difference
variable results results (%)
problem, the upper and lower bounds for design variables are
(1) (2) (3) (4)
defined in such a way that their final values correspond to the
available discrete values. Deflection. m 2.48 (8.13)" 2.61 (8.57)" +5.24
y-moment, leN· m 2,268 (1,673)b 2,165 (1,597)' -4.54
The second approach uses a genetic algorithm for all the Axial load, N 73,236 (16,464)' 71,398 (16,051)' -2.51
design variables. This method is designed specifically for dis- Shear load, N 106,633 (23,972)' 105,040 (23,614)' -1.49
crete optimization problems. It does not require differentiabil- Bending stress, Pa 403,136,468 385,485.889 -4.38
(8.419,680)· (8,051,040)"
ity of the problem functions; it only assumes that the problem Axial stress, Pa 3,240,536 (67,680)" 3,171,588 (66,240)" -2.13
functions can be evaluated at a given design. The method uses Shear stress, Pa 9,514,765 (198,720)" 9,514,765 (198,720)" 0.00
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/09/15. Copyright ASCE. For personal use only; all rights reserved.

stochastic concepts to select several trial designs and evaluates 'In ft.
their "fitness" using the values of cost and constraint func- "In kip-ft.
tions. At the end, a design with the best fitness value is taken "In lb.
dIn psf.
as the optimal design. Since the method requires discrete val-
ues for all the variables, the continuous variables are discre-
tized into 100 values in the present application. TABLE 3. Comparison of Analysis Results for Present Pro-
All solutions are obtained on an 100 MHz HP-UX715 gram with Conventional Design Program at 28.96 m (95 ft) for
Workstation. ASCE Ice and Wind Loading Case
Present Present
Comparison of Analysis Results Conventional program program
design results results
For the conventional design, Design (1990) gives values for Analysis program with load Difference with load Difference
deflection, bending moments, axial load, shear load, bending variable results factor 1.00 (%) factor 1.25 (%)
stress, axial stress, shear stress, and combined stress at certain (1 ) (2) (3) (4) (5) (6)
heights for all three loading cases. The design given in the Deflection,
foregoing publication is used to verify the optimal design for- m 1.73 (5.67)' 1.81 (5.93)" +4.59 1.81 (5.93)' +4.59
y-moment,
mulation and the present program. Before comparing the two kN'm 1,487 (1,097)' 1,427 (1,053)' -4.01 1,429 (1,054)' -3.92
results, the analysis procedure of the present program is ex- Axial load, 71,679 64,135 -10.53 71,314 -0.51
plained. N (16,114)' (14,418)' (16,032)'
Shear load, 65,184 65,273 -0.14 65,273 -0.14
As seen in (11), calculation of the deflection involves an N (14,654)' (14,674)' (14,674)'
integral that depends on the bending moment. The integration Bending 264,345,000 254,140,759 -3.86 254,347,602 -3.78
is approximated by a summation over 0.15 m (0.5 ft) intervals. stress, Pa (5,520,960)" (5,307,840)" (5,312.16O)d
Axial 3,171,588 2,826,851 -10.87 3,171,588 0.00
To include the secondary moments, an iteration method is stress, Pa (66,240)" (59,04O)d (66,240)"
used. In this method, the deflection is calculated using the Shear 5,860,544 5,860,544 0.00 5,860,544 0.00
original lever arms. Then these lever arms are adjusted due to stress, Pa (122,4OO)" (122.4OO)d (I 22,400)d
the deflection at the corresponding levels, and the moment and
deflection are recalculated. This iteration procedure is contin-
'nbIn ft.kip-ft.
ued until the difference between the current and the previous "In lb.
"In psf.
lever arms is less than 1%. The deflection calculated with this
method has to be less than the maximum allowable deflection
that is taken as 2.54 m (l00 in.). The moments at this deflec- values, the same cases have been solved by changing the dead
tion are the bending moments in the sections. The axial load load factor to 1.25 from 1.0. When the results with this factor
includes the vertical loads from the conductors and the weight are compared, the ASCE ice and wind loading deflection value
of the pole. The shear load is due to the wind on the structure is 4.59% greater than the corresponding conventional design
and the transverse loads in the conductors. In broken conduc- values. The bending moment is 3,92% smaller. The difference
tor loading, the longitudinal load is also considered. The shear in axial loads between the conventional design and present
load is calculated as the resultant of transverse and longitu- program is 0.51 % in the ASCE ice and wind loading case with
dinal loads. the dead load factor of 1.25, and 10.52% with the dead load
Table 2 summarizes the comparison of analysis results for factor of 1.0. The differences in shear loads are small, 0.14%.
the NESC light loading case. As given in this table, deflection Some of the differences in the analysis results could be due
with the present program is 5.24% larger than the deflection to the different models used in the two calculations. In the
with the conventional design program. There is a 4.54% dif- conventional design procedure, a nonlinear finite-element
ference in the bending moment, and a 2.51 % difference in the model and a computer program are used to analyze the pole.
axial load. The difference in shear load is smaller, 1.49%. Cor- In the present formulation, the pole is modeled as a simple
responding to these differences, bending stress differs by cantilever beam and an iterative procedure is used to treat the
4.38%, axial stress differs by 2.13%, and there is no difference secondary moments.
in the shear stress.
Table 3 summarizes the comparison of analysis for the Comparison of Optimal and Conventional Designs
ASCE ice and wind loading case. As mentioned, the dead load
factor for the ASCE ice and wind loading case is 1.0. When As observed earlier, there are four design variables. They
the analysis results are compared, a significant difference are: the outside diameter at the top, the thickness of the two
(10.53%) is observed in the axial load in the ASCE ice and pieces, and tapering of the pole. Among these variables, the
wind loading case. It is difficult to explain the reason for the diameter and tapering are continuous variables. The thickness
differences since details of the finite-element model and the is a discrete variable. Two different sets of values are used for
analysis program used in Design (1990) are not known. There- the thickness. Set I: 0.00476, 0.00556, 0.00635, 0.00715,
fore, to make axial load values closer to conventional design 0.00794,0.00873, and 0.00953 m (0.01563, 0.01823, 0.02083,
JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1996/1353

J. Struct. Eng. 1996.122:1347-1356.


TABLE 4. Conventional Design versus Optimal Design
Design Conventional Optimal Optimal Allowable ranges Starting
variables design design 1 design 2 for design variables point
(1 ) (2) (3) (4) (5) (6)
Dro.m 0.18440 (0.60500)" 0.17575 (0.57660)" 0.18293 (0.60017)" 0.15240-0.24130 (0.50000-0.79167)" 0.15240 (0.50000)-
th m 0.00635 (0.02083)" 0.00556 (0.01823)" 0.00556 (0.01823)" Set 1 0.00476 (0.01563)"
S. mlm 0.01413 0.01633 0.01545 0.01400-0.0175 0.01500
12 , m 0.00715 (0.02344)" 0.00635 (0.02083)" 0.00635 (0.02083)" Set 2 0.00556 (0.01823)"
Maximum vio.• % 7.80 0.00 7.67 - -
CPU. s - 225 262 - -
Cost, $ 2.259 2.199 2.130 - -
-In ft.
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/09/15. Copyright ASCE. For personal use only; all rights reserved.

TABLE 5. Conventional Design versus Optimal Design with As a result, the diameter at the butt of the pole has increased
Dto and S Fixed to Conventional Design Values by 11.8%. This can be a disadvantage in transportation and
Allowable assembly of the pole. The thicknesses of both the pieces are
ranges for smaller for optimal design, which results in a lighter pole. This
Design Conventional Optimal design Starting also leads to a smaller cost: 2,199 versus 2,259, which is a
variables design design variables point reduction of about 2.7%.
(1) (2) (3) (4) (5)
For the conventional design, the bending stress and com-
DIO,m 0.18440 0.18440 0.18440 0.18440 bined stress constraints at 7.62 m (25 ft), 13.72 m (45 ft), and
(0.60500)' (0.60500)' (0.60500)' (0.60500)'
111 m 0.00635 0.00715 Set 1 0.00476 14.60 m (47.9 ft) and the deflection for the NESC light loading
(0.02083)' (0.02344)' (0.01563)' case are active with the maximum violation of 7.8% for the
S.m/m 0.01413 0.01413 0.01413 0.01413 combined stress constraint at 7.62 m (25 ft). There are no
12, m 0.00715 0.00794 Set 2 0.00556
(0.02344)' (0.02604)' (0.01823)' violated constraints for the optimal design. This shows that the
Maximum optimal design process can impose the constraints very pre-
via., % 7.80 0.00 - - cisely and still have a smaller cost compared to the conven-
CPU, s - 184 - - tional design. There are 183 constraints to be checked. Ap-
Cost. $ 2,259 2,512 - -
'In ft.
parently, it is difficult to satisfy all the constraints precisely in
the conventional design process because design improvements
must be guessed based on past experience.
TABLE 6. Optimal Design with a Limit on Base Diameter To see the effect of a larger tolerance on constraint viola-
Optimal
tions, the problem is reoptimized by allowing the constraints
design to be violated by 7.8%. The final design is shown as "Optimal
with a Allowable Design 2" in Table 4. For this case, six constraints are active
limit on ranges for with a maximum violation of 7.67%, and the cost function is
Design Optimal base design Starting smaller by 5.7% compared to the conventional design.
variables design diameter variables point
(1 ) (2) (3) (4) (5)
Additional Optimal Designs
D""m 0.17575 0.20538 0.15240-0.24130 0.15240
(0.57660)' (0.67382)' (0.50000-0.79167)' (0.50000)' Effect of variations in certain parameters and conditions on
t., m 0.00556 0.00635 Set 1 0.00476
(0.01823)' (0.02083)' (0.01563)' the optimal solutions are investigated. These include thickness
S,m1m 0.01633 0.01400 0.01400-0.0175 0.01500 optimization and optimal design with round cross section.
0.00635 0.00715 Set 2 0.00556
'" m (0.02083)' (0.02344)' (0.01823)'
Maximum Optimization of Only Thickness
via., % 0.00 0.00 - - Although a limit in the diameter of the pole is not set, it is
CPU, s 225 323 - -
Cost, $ 2,199 2.325 - - an important factor in its fabrication, transportation, and as-
'In ft. sembly. To study the effect on the optimal solution with a
constraint on the pole diameter, the problem is solved where
the tip outside and tapering are kept fixed to the values of the
0.02344,0.02604,0.02865, and 0.03125 ft) and set 2: 0.00556, conventional design. Only the thickness is optimized to satisfy
0.00635,0.00715,0.00794,0.00873,0.00953, and 0.01032 m all the constraints. The optimal solution is given in Table 5.
(0.01823, 0.02083, 0.02344, 0.02604, 0.02865, 0.03125, and As seen there, both the thicknesses have increased. Conse-
0.03386 ft). The value for the thickness is taken from the Man- quently, there is an 11.2% increase in the cost function. No
ual of Steel Construction (1980). The height is the same as in constraint is violated for optimal design whereas the conven-
the conventional design procedure. tional design has a 7.8% violation.
The constraints are checked in 10 different cross sections.
There are 18 design constraints that have to be satisfied at each Limit on Base Diameter
section and there are three global design constraints that have
to be satisfied by the structure. Therefore, there are a total of It may be desirable to impose a limit on the base diameter
183 design constraints. The acceptable violation of constraints of the pole to satisfy transportation requirements. To see the
is set to 1%. Table 4 summarizes various data for the problem effect of this limitation on the optimal design, an upper limit
and the final results. The final design for this case is shown of 2.54 m (3.3 ft) is imposed on the base diameter. This value
as "Optimal Design I" ("Optimal Design 2" is explained is slightly larger than that for the conventional design. The
later). The optimal solution is obtained using the genetic al- results for this case are given in Table 6 along with the pre-
gorithm. vious optimal design.
As the results show, the pole tapering has increased whereas The optimal design with limit on the base diameter is 5.73%
other design variables have decreased for the optimal design. more expensive. As the results show, all the design variables
13541 JOURNAL OF STRUCTURAL ENGINEERING I NOVEMBER 1996

J. Struct. Eng. 1996.122:1347-1356.


TABLE 7. Round Section versus Dodecagonal Section
Design Dodecagonal Round Allowable ranges Starting
variables section section for design variables point
(1 ) (2) (3) (4) (5)
Dto,m 0.17575 (0.57660)" 0.21590 (0.70833)" 0.10160-0.24130 (0.33333-0.79167)" 0.19172 (0.62900)"
tit m 0.00556 (0.01823)" 0.00556 (0.01823)" Set 1 0.00556 (0.01823)"
S,m/m 0.01633 0.01719 0.0125-0.0175 0.01600
t2 , m 0.00635 (0.02083)" 0.00556 (0.01823)" Set 2 0.00556 (0.01823)"
Maximum vio., % 0.00 1.00 - -
CPU, s 225 303 - -
Cost, $ 2,199 2,145 - -
"In ft.
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/09/15. Copyright ASCE. For personal use only; all rights reserved.

except slope have larger values than the previous optimal de- procedure. For example, an existing nonlinear finite-element
sign. This design is also 3% more expensive than the conven- analysis program can be used to analyze the pole and formu-
tional design. However, the maximum constraint violation in late the constraints using the corresponding response. The total
the conventional design is 7.8%, whereas no constraints are computational effort, however, is likely to increase substan-
violated with the present design. As seen earlier, a less expen- tially with the use of such finite-element programs.
sive design is obtained if the allowable constraint violation is The optimization process can be used to design transmission
relaxed in the optimal design process. poles using standard available sections. This topic will be in-
vestigated in the future.
Use of Round Cross Section
ACKNOWLEDGMENTS
To see the effect of the shape, the same problem is solved
with a round cross section. The geometrical constants, as well We would like to acknowledge partial support for this research under
as local buckling and combined stress constraints were the project, "Design Sensitivity Analysis and Optimization of Nonlinear
changed to correspond to a round section. The results for this Structures" from the National Science Foundation (NSF) of the United
case are given in Table 7. States of America, Grant No. CMS-93-01580. We would also like to
thank the reviewers of the paper for providing us with useful suggestions
The tip outside diameter and the tapering are bigger than to improve presentation of the material.
the corresponding values in the dodecagonal section. The
thickness in the second piece is smaller and it is the same for
the first piece. The cost of the pole with a round cross section APPENDIX I. CONVERSION TO SI UNITS
is $2,148, which is 2.4% smaller than that with the dodecag-
onal cross section. No constraints are violated. I in. = 2.54 X 10- 2 m; lib = 4.448222 N; 1 ksi = 6.894757
X 106 Pa; and 1 sq in. = 6.4516 X 10- 4 m 2 •
ANALYSES AND CONCLUSIONS
APPENDIX II. REFERENCES
The optimal design problem formulation for self-supporting,
single type steel transmission poles was presented. The design Arora, J. S. (1989). Introduction to optirrwl design. McGraw-Hili Book
variables for the problem were identified and the cost and con- Co., Inc., New York, N.Y.
Arora, J. S., Lin, T. C., Elwakeil, O. A., and Huang, M. (1993)./DESIGN
straint functions were expressed in terms of variables. An ex-
user's rrwnual 4.2, Tech. Rep. No. ODL-93.04. Optimal Des. Lab., Civ.
isting design example was optimized and results were com- and Envir. Engrg., The Univ. of Iowa, Iowa City, Iowa.
pared with the conventional design. Several additional cases Chapra, S. C., and Canale, R. P. (1985). Numerical methods for engineers,
were solved to investigate the effect of variations of the prob- 2nd Ed., McGraw-Hili Book Co., Inc., New York, N.Y.
lem conditions on the optimum solution, as well as to show Design of steel transmission pole structures. (1990). Task Committee on
flexibility of the optimal design process. Updating ASCE Des. Guide on Steel Transmission Pole Struct. of the
Struct. Div. of ASCE, New York, N.Y.
It is concluded that the optimal design process can lead to
Ghannoum, E., and Yaacoub, S. J. (1989). "Optimization of transmission
less expensive and safer designs compared to the conventional towers and foundations based on their minimum cost." IEEE Trans.
design process. All constraints of the problem are satisfied in on Power Delivery, 4(1), 614-621.
the optimal design process, whereas it is difficult to satisfy Guidelines for transmission line structural loading. (1991). Committee
them with the conventional design process. Also, once the on Electr. Transmission Struct. of the Committee on Anal. and Des. of
problem formulation is installed into an optimizer, solutions Struct. of the Struct. Div., ASCE, New York, N.Y.
for additional cases can be obtained quite easily. As a result, Huang, M. W. (1995). "Algorithms for mixed continuous-discrete vari-
able problems in structural optimization," PhD thesis, Civ. and Envir.
variations in the problem conditions can be studied in a shorter Engrg., Coli. of Engrg., The Univ. of Iowa, Iowa City, Iowa.
time, leading to, perhaps, a better final design. Huang, M. w., and Arora, J. S. (1995). "Engineering optimization with
In the present application, only the material cost was min- discrete variables." Proc., 36th AlAAIASME/ASCE/AHS Struct., Struct.
imized. To obtain more realistic designs, the cost function Dynamics, and Mat. Coni, Am. Inst. of Aeronautics and Astronautics,
should be modified to include the manufacturing, transporta- Washington, D.C., 1475-1485.
tion, assembly, and maintenance costs. Any of the foregoing Krauthernmer, T. (1987). "A numerical study of wind-induced tower vi-
brations." Compo and Struct., 26(112), 223-241.
costs that can be related directly to the material cost will not Manual of steel construction, 8th Ed. (1980). Am. Inst. of Steel Constr.
affect the optimal design; only the final value of the cost func- (AISC), Chicago, III.
tion will be changed by a factor. National electrical safety code. (1977). Nat. Bureau of Standards, Wash-
In the present formulation, a simple procedure was used to ington, D.C.
analyze the pole structure. Inclusion of secondary moment ef- Peyrot, A. H., and Dagher, H. J. (1984a). "Probabilistic design of trans-
fects made the problem nonlinear and an iterative procedure mission line structures." J. Struct. Engrg., 110(10), 2513 - 2531.
Peyrot, A. H., and Dagher, H. J. (1984b). "Reliability-based design of
was used to obtain the final response of the pole under the transmission lines." J. Struct. Engrg., 110(11), 2758-2778.
applied loads. The optimization procedure described here is Randle, E. R. (1985). Proc., ASCE Seminar on Innovations in The Des.
quite general because it can accommodate any other analysis of Electr. Transmission Struct., ASCE, New York, N.Y.

JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1996/1355

J. Struct. Eng. 1996.122:1347-1356.


APPENDIX III. NOTATION f, = tensile stress, MPa (ksi);
The following symbols are used in this paper:
f. = shear stress, MPa (ksi);
H = height of the pole above the ground, m (ft);
I = moment of inertia, m4 (ft4);
A, = gross area, m2 (sq ft); L = longitudinal load from the conductor, N (lb);
a = angle between the x-axis and the corner of the polygon M. = moment about x-axis, N· m (lb/ft);
(degrees); My = moment about y-axis, N· m (lb/ft);
a, = vertical distance from the tip of the pole to the ith con- M z = moment about z-axis, N· m (lb/ft);
ductor, m (ft);
m = total number of pieces that the pole is made of;
BR = effective bend radius, m (ft);
n = total number of conductors;
bl = horizontal distance from the undeflected pole axis to the
P = axial load, N (lb);
tip of the ith conductor, m (ft);
P, = unit price for steel, $/kg ($Ilb);
C = distance from neutral axis to extreme fiber, m (ft);
Q = moment of section about center, m3 (cu ft);
Cy = y-coordinate of the point, m (ft);
q = wind pressure, Pa (psi);
Cz = z-coordinate of the point, m (ft);
Downloaded from ascelibrary.org by Technische Universitat Munchen on 07/09/15. Copyright ASCE. For personal use only; all rights reserved.

c = area constant, 'IT for round sections and 3.22 for dodecag- r = radius of gyration, m (ft);
onal sections; S = tapering, m/m (ft/ft);
Do = outside diameter for round sections, across flats for polyg- TI = transverse concentrated load in the ith conductor, N (lb);
onal sections, m (ft); t1 = thickness of the kth piece, m (ft);
D,o = tip outside diameter, m (ft); V = shear load, N (lb);
D, = outside diameter or across flat at the joint, m (ft); VI = vertical concentrated load in the ith conductor, N (lb);
E = modulus of elasticity of steel, MPa (ksi); V, = volume of steel, m3 (cu ft);
Fa = allowable compressive stress, MPa (ksi); w = flat width of a side of a polygon, m (ft);
Fb = allowable bending stress, MPa (ksi); x = distance measured from the tip of the pole, m (ft);
F, = allowable tensile stress, MPa (ksi); A = deflection, m (ft);
F. = ultimate tensile stress, MPa (ksi); A'II = maximum allowable deflection, m (ft);
F. = allowable shear stress, MPa (ksi); Ay = deflection in y-direction, m (ft);
Fy = yield stress, MPa (ksi); Az = deflection in z-direction, m (ft);
fa = axial compressive stress, MPa (ksi); p, = mass density of steel, kglm3 (lb· m/cu ft); and
fb = bending stress, MPa (ksi); pw = unit weight of steel, N/m 3 (lb/cu ft).

1356/ JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 1996

J. Struct. Eng. 1996.122:1347-1356.

You might also like