Professional Documents
Culture Documents
4, OCTOBER 2010
2997
I. INTRODUCTION
where
is the antinode amplitude of the sine-shaped deflecis the distance from the
tion curve, the wavelength, and
fixed support beyond which the sine-shaped deflection curve
is defined. The fixed support is assumed to be a square-faced
bushing.
The bending curvature of the conductor is assumed to be
given by the usual solid beam equation (Fig. 3)
(2)
where is the departure of conductor centerline (Fig. 3) from
in the vicinity of the bushing. It will
the sine-shaped loop
2998
may be expressed as
(11)
(3)
From (6)
With the boundary condition that approaches zero for large
, the solution to the differential equation is given by
(4)
The slope of the conductor relative to the
(13)
and, combined with (4), (12) becomes
(14)
axis is
(5)
and at
0,
is equal to
(Fig. 2). The curvature
of the conductor at the mouth of the square-faced bushing is
(6)
is assumed to be equal
As a first approximation, the angle
of the sine-shaped deflection
to the maximum value of
curve and it is given by
(7)
(16)
This last equation is known as the Poffenberger & Swart stress
(PS ).
III. PARAMETERS INFLUENCING THE COMPUTED
Both (11) and (16) seem to indicate that stresses are directly
proportional to wire diameter . So it would be interesting to
check how sensitive these equations are with respect to .
Assume: 1) that a multilayer conductor is made of wires
of the same material (aluminium), which is the case in ACAR,
AAC, AAAC conductors; 2) that all wires are of the same diameter ; 3) that all wires act independently in bending (thus
2999
TABLE I
GEOMETRICAL AND MECHANICAL PROPERTIES OF ACSR CONDUCTORS
(17)
where
is the density of aluminium kg/m .
It is found that no transverse dimension of the wire section
appears in the stress equation meaning that the last expression is
independent of wire diameter and, thus, the equation of stress
is proportional to a constant which depends only on material
properties. In the case of aluminium
(18)
In fact, for aluminium-conductor steel-reinforced (ACSR)
multilayer conductors, [2] and [3] indicate that the constant
may vary from 0.171 to 0.200 and in the same
reports, it is suggested to use a mean value of 0.186 (MPa
is
s/mm). It should be remembered that the computation of
based on the assumption of independent bending of wires.
Now, with the other limit hypothesis that all wires act as
if they were welded together (no slip), the bending rigidity is
and is calculated by considering the conductor as a solid
beam. In this case, the overall diameter of the conductor is
used instead of the wire diameter in (9a). The resulting computation of
of conductors tested to build the S-N
diagram shown in Figs. 4 and 5 is summarized in Table I. The
differs by a factor of, at most, 15%
relationship versus
. It should be noted, however,
from the one computed with
that the ACSR conductors in this table do not exactly meet the
previous hypotheses, as they include a 7 steel-wire core.
is used as a fatigue severity
The alternating stress
indicator in order to regroup all fatigue results of different conductors in a single S-N diagram. In Figs. 4 and 5, most data
points are from [2, Figs. 3.213a and b], with some additional
data points [4], [5]. Each data point is computed with its own
. In Figs. 4 and 5,
proportionality factor (Table I) by using
Fig. 4. Fatigue tests of a two-layer ACSR. This is from [2, Fig. 3.213a] updated with some additional data points.
Fig. 5. Fatigue tests of a three-layer ACSR. This is from [2, Fig. 3.213b],
updated with some additional data points.
some data points deviate slightly from those shown in [2, Figs.
3.213a and 3.213b] since those data coming from [3] computed with a average value of the proportionality factor of 0.186
3000
were used again in [2] and some raw data from the GREMCA
laboratory have been slightly adjusted. Also, some data points of
the ACSR Drake conductor [10] have been modified according
to [4]. Looking at Figs. 4 and 5, the fatigue results expressed
as , which are practically directly proportional to a value of
, are very consistent. As reported in [2], the S-N
diagram for three-layer ACSR conductors differs from the one
obtained for the two-layer ACSR conductors; their endurance
limits , defined as the highest alternating stress amplitude
with a zero wire break at 500 Mc, are about 22 MPa and somewhat less than 30 MPa, respectively. These S-N diagrams are
for ACSR conductors fitted in short commercial metallic suspensions clamps and in metallic bell mouth clamps.
The fact that (11) is practically independent of diameter and
of the bending rigidity hypothesis seems to indicate that drawing
taken as the ordinate is as good as
the S-N diagram with
(11).
using
By comparison, it will now be shown that the value of
yielded by (16) is not directly proportional to the wire diameter.
The mathematical development is based on the same assumpbe the normal traction stress computed
tions as before. Let
according to (19) (lay angles are neglected)
Fig. 6. Fatigue tests of two-layer ACSR. From Fig. 3.224 [2] with some additional data points.
(19)
and
becomes
(20)
It is possible to deduce approximately the influence of wire diameter using (21). It appears that for a given static traction stress
, the alternating bending stress
is not proportional to wire
diameter .
In Figs. 6 and 7, while most data points have already been reported in [2, Figs. 3.224 and 3.225], there are some additional
data points [4], [5]. Considering all of the tested ACSR conductors (with the associated testing conditions) in Figs. 6 and 7, the
computed by using (16) varies from
corresponding ratio
27.2 to 46.3 (MPa/mm) (Table I). Using (21),
differs by,
at most, 6% from the one computed using (16).
, it is now assumed that the prevailing
Instead of using
and that the distance from the neutral
bending rigidity is
axis to the outer fiber is
. In this case,
is at least 2.6
(Table I).
times the value computed with
Comparing Figs. 6 and 7, one sees that the S-N diagram for
three-layer ACSR conductors differs from the one obtained for
the two-layer ACSR conductors; the vertical axis is the altercomputed by using
. The endurance
nating stress
for three-layer ACSR conductors is about 8.5 MPa.
limit
Howewer, for the two-layer conductor case, it is noted that at
the 15-MPa level, the number of cycles of the first wire break
Fig. 7. Fatigue tests of three-layer ACSR. From Fig. 3.225 [2] with some
additional data points.
3001
TABLE II
EXPERIMENTAL DATA
All data were collected from original papers. Regression lines through experimental data were computed. Data appearing in [2] and [3] are indicated by .
There are some minor differences between the computed ratio in [2] and the ratio given in [3].
Experimental conditions:
obtained from experimental data f , y
;
1) regression line through five data points; strain gauges on wires of the outer layer; f y
obtained from , H and through (7) and (10);
2) regression line through two data points; strain gauges on wires of the outer layer; f y
3) regression line through 16 data points; the 16 data points come from curves in the original paper; curves obtained from strain gauge data measured on the
deduced from , H and through (7) and (10);
two uppermost and the two lowermost wires of the outer layer; f y
obtained from experimental data y
, , H ,
4) regression line through 10 and more data points; strain gauges on wires of the outer layer; f y
and through (10);
obtained from experimental data f and y
;
5) mean of four data points; one strain gauge on the two adjacent uppermost wires of the outer layer; f y
obtained from experimental data f and y
;
6) mean of two data points; one strain gauge on the two adjacent uppermost wires of the outer layer; f y
7) regression line through 4 data points; the nearest strain gauge from the Keeper Edge (KE) on the uppermost wire of the outer layer (Drake, wire 15;
Bersfort, wire 17) and the nearest strain gauge from the Last Point of Contact (LPC) on the lowermost wire of the outer layer (Drake, wire 113; Bersfort,
obtained from experimental data f and y
. In the calculation of the theoretical f y
, EI
is computed with 4 aluminium
wire 17); f y
wires removed from the outer layer and m is computed with no aluminium wire removed.
using (22), will vary according to the square root of the traction
force and the variation of the alternating stress , computed by
using (21), is smaller. For example, consider the AAAC Aster
570. It is made of 61 wires (wire dia. 3.45 mm) and its construction is 24/18/12/6/1. Doubling the traction force from 20%
(22) by a factor
RTS to 40% RTS multiplies the ratio
. The ratio
given by (16) and (21) changes by a factor
of 1.265 and of 1.245, respectively. Sepp [7] and Poffenberger
and Komenda (discussion in [8]) reported that the strain measured in the clamp vicinity of a conductor varies as the square
root of the applied tension force.
It seems difficult to reconcile the two approaches based on
(11) and (16) since the traction force variation and the bending
rigidity variation affect the alternating stress differently. Obviously, the difference between (9b) and (16) comes from the
(23)
In (9b),
is obtained from the sine-shaped deflection curve
of the vibrating conductor through the following equation:
(24)
while in (16),
comes from (13) and it is expressed in term
of the amplitude A of the deflection curve. The value A is deter-
3002
(25)
It is easy to experimentally obtain an empirical relationship
between
and
at a given frequency of excitation but
and it does not provide
this does not give the exact value of
which position the equal slope between the deflection curve of
the local field near the bushing and the sine-shaped deflection
curve of vibrating conductor occurs. On the other hand, it should
be remembered that (23) and (25) are based on the assumption
of a uniform bending rigidity along the length of the conductor.
In fact, the bending rigidity is probably high (in the order of
) at a portion of the deflection curve very close to the fixed
end due to the clamping action of the bolting of the two halves
of the square-faced bushing (or of the bolting of the keeper for a
conductor-suspension clamp assembly) and in the asymptotic
portion of the deflection curve. Elsewhere, between these two
regions, the bending rigidity may be much lower, as slip between
wires is easier.
and in order to compare conductors of different sizes, the following ratio is used:
(27)
Based on
Based on
Equation (16) will be modified in order to facilitate the comparison between experimental and theoretical results; to compare measured strains, (16) is modified as follows:
(26)
Fig. 9. "
(central alternating) versus bending amplitude ACSR Drake with
the suspension clamp.
Fig. 10. " (central alternating) versus bending amplitude 583.2-kcmil ACSR
conductor with a suspension clamp.
by using the ratio of the vertical load to the tension load) is approximately 4.6 . The peak-peak bending amplitude is measured relative to the clamp. Strain gauges no. 2, 4, and 5 were
glued on the three uppermost wires on the conductor at the last
point of contact between the top of conductor and suspension
clamp. Gauge no. 3 was located on the same wire as no.4 at 12.7
mm from that gauge, on the clamp side. Gauge no. 1 was located
on the gauge no. 2 wire at 12.7 mm from that gauge, on the conductor span side. As seen in Fig. 10, the scatter of strain values
can be high; for example, at
0.5 mm, the strain amplitude
given by gauge no. 4 is 35% higher than the mean value obtained
from the remaining four gauges. The upper data point (partially
filled square) corresponds to an experimental point used by Poffenberger and Swart [13] to draw Fig. 11.
Fig. 11 shows the experimental data for seven conductors
reported by Poffenberger and Swart [13] and those obtained
for seven conductors by Claren and Diana [8]. Each point corresponds to the maximum strain value at the maximum displacement obtained in any set of data, where interstrand binding
would be at a minimum. The data (six of seven conductors) reported in [13] come from unpublished reports by Rulhman and
Swart and by Edwards and Boyd at the publication time of [13].
The data from Rulhman and Swart and Edwards and Boyd reports are used in [14] and it is reported that tests carried on
unarticulated clamps with various mouth radii showed that the
strain-bending-amplitude relationship was constant with radii
3003
3004
TABLE III
ANALYSIS RESULTS OF THE DATA REPORTED BY KOMENDA AND SWART [15]
studied but no specific information was given on how it was accomplished. All data were grouped together without regarding
the geometrical characteristics of the support (such as a squarefaced bushing) or a short commercial suspension clamp. The
prediction band (Table III), computed from the analysis based
on all data, gives an interval which probably characterizes the
effect of the geometrical parameters of the support, of the position where the strain gauges are glued, and of the variation from
one wire to the adjacent one if single strain gauge monitoring
is used; these effects can also be appreciated by the large variation of individual values of shown in the column identified as
.
Now, the precise mathematical meaning of the terms confidence and prediction band should be recalled. If 100 samples
from a population were drawn and a 99% confidence band of the
mean was computed for each sample, then 99% of these confidence bands would contain the true mean of the population. In
practice, usually one sample is drawn and one confidence band
is computed, so it cannot be said that this single confidence band
necessarily contains the true mean of the population. But it can
be said that with 99% confidence, the true mean of the population is contained in this single computed confidence band. This
last statement implies a notion of frequency: we do not know
whether this single confidence band contains the true mean of
the population but we do know that the method used to compute
the confidence band yields a true statement in 99% of the cases.
On the other hand, the prediction band shown in Table III gives
the interval in which the value of a single future observation is
expected to be in 99% of the cases.
Looking at the other results of Table III, which were obtained
following the same computational approach, several points
ought to be emphasized.
1) The field data probably correspond to a short metallic conventional suspension clamp and
is lower than 0.983.
2) Looking at the laboratory data of the ACSR Pheasant, the
resulting
is much less than 1.0 and the 99% confidence band does not include the value of
1; this indicates probably that
significantly departs from 1.0. Is
this behavior related to a specific conductor-support system
or to specific characteristics of the
measuring system?
(Is the
value identical in each data group?)
3) Considering the laboratory data of ACSR Drake, which has
a large number of data groups
22), the resulting
is higher than 1.0, and the 99% confidence band does not
3005
TABLE IV
EXPERIMENTAL DATA FROM [10]
Assumed value: 1.295 mm (0.051 in.). The Y value reported in [10] for this test is 0.38 mm (0.015 in.). Comparing the strain values of this test with the strain
values of the Y
= 1.09-mm (0.043 in.) test (line above in Table), it seems that the Y
value (0.015 in.) is very low. All tests were carried out under the
same testing conditions and should give approximately the same values of Y for the 12-in. radius clamp. Checking in [9] and [10], the reported values of
computed stresses, the ratio of these computed stresses is 1.1837 which is approximately equal to the following ratio of Y values 0.051 in./0.43 in. Thus, the
value of 0.015 in. seems to be a typographical error. The testing conditions were fy
mm), sag angle
The strain values of this test are not used in the computed mean of 54%.
V. CONCLUSION
Section II gives the theoretical development of an alternating
based either on vibration parameter or on the bending
stress
and
. In
amplitude, which have been noted
Section III, it is shown that it is difficult to reconcile both stress
values because the traction stress influences each value according differently and because the use of the two limit bending
and
) has a much stronger influence on
rigidities (
value than on the value
. In Sections IV-A
the PS
and B, the correlation between the reported experimental data
was examined. It was found that the
and the theoretical
scatter of experimental strain values is large and that in many
cases, the correlation between experimental strains and theoretis weak.
ical
Figs. 47 are the fatigue S-N diagrams for two-layer and
three-layer ACSR conductors fitted on short commercial
metallic suspensions clamps and on metallic bell mouth
clamps. In these diagrams, the vertical axis is the alternating
or
. The two-layer
stress amplitude
3006
[3] C. B. Rawlins, Chap. 2: Fatigue of overhead conductors, in Transmission Line Reference BookWind-Induced Conductor Motion. Palo
Alto, CA: EPRI, 1979, EPRI Res. Project 792.
[4] G. E. Ramey, Conductor fatigue life research Jul. 1981, EL-1946,
EPRI Res. Project 1278-1, Final Rep.
[5] GREMCA, Fatigue tests on the 48/7 Bersfort acsr conductor at Yb
varying from 0.3 mm to 0.35 mm (in French) Dept. Mechan. Eng.,
Laval Univ., Quebec City, QC, Canada, Tech. Rep. SM-2007-01, 2007.
[6] CIGRE Task Force B2.11.07, Fatigue endurance capability of conductor/clamp systemsUpdate of present knowledge, in Confrence
Internationale des Grands Rseaux lectriques, 2006, Tech. brochure
no TBD.
[7] T. O. Sepp, Self-damping measurements and energy balance of acsr
drake, in Proc. IEEE Winter Power Meeting, New York, Jan. 31Feb.
5, 1971, pp. 18.
[8] R. Claren and G. Diana, Dynamics strain distribution on loaded
stranded cables, IEEE Trans. Power App. Syst., vol. PAS-88, no. 11,
pp. 16781690, Nov. 1969.
[9] P. B. McGill and G. E. Ramey, Effect of suspension clamp geometry
on transmission line fatigue, J. Ener Eng., ASCE, vol. 112, no. 3, pp.
168184, 1986.
[10] G. E. Ramey, Conductor Fatigue life research eolian vibration of transmission lines, Jan. 1987, EL-4744, EPRI Res. Project 1278-1, Final
Rep.
[11] T. Sepp, Transmission Line Vibration IV, Fatigue Tests of ACSR Ibis
Conductor. Helsinki, Finland: Imatran Voima Osakeyhtio, 1968.
[12] A. R. Hard, Studies of conductor vibration in laboratory span, outdoor
test span and actual transmission lines, CIGRE, Rep. 404, 1958.
[13] J. C. Poffenberger and R. L. Swart, Differential displacement and dynamic conductor strain, IEEE Trans. Power App. Syst., vol. PAS-84,
no. 4, pp. 281289, Apr. 1965.
[14] IEEE comittee report, Standardization of conductor vibration measurements, IEEE Trans. Power App. Syst., vol. PAS-85, no. 1, pp.
1022, Jan. 1966.
[15] R. A. Komenda and R. L. Swart, Interpretation of field vibration data,
IEEE Trans. Power App. Syst., vol. PAS-87, no. 4, pp. 10661073, Apr.
1968.
[16] R. H. Scanlan and R. L. Swart, Bending stiffness and strain in stranded
cables, presented at the IEEE Winter Power Meeting, New York, 1967.
Sylvain Goudreau received the Bachelor and Master degrees in mechanical
engineering from cole Polytechnique de Montral, Montral, QC, Canada, in
1977 and 1980, respectively, and the Ph.D. degree in mechanical engineering
from Universit Laval in 1990.
He is a Professor at the Department of Mechanical Engineering at Universit Laval, Qubec City, QC, Canada, and Principal Researcher of Groupe de
REcherche en Mcanique des Conducteurs Ariens Research Group. He is a
Registered Professional Engineer in the Province of Qubec. Before beginning
his Ph.D. studies, he was with the National Research Council of Canada (Institut
du gnie des matriaux, Montral), where he was involved in the development
of a mechanical testing laboratory and in studies on composite materials. In
1988, he joined the Mechanical Engineering Department at Universit Laval.
His research activities are in the field of mechanical behavior of overhead line
conductors and their related fatigue problems. He is author or coauthor of many
technical reports and papers on these subjects.
Frdric Lvesque is a Postdoctoral Fellow at the Department of Civil Engineering, Sherbrooke University, Sherbrooke, QC, Canada. He received the
B.Sc., M.Sc., and Ph.D. degrees in mechanical engineering from the University Laval, Qubec City, QC, Canada .
His research interests are the mechanics involved in the fatigue of overhead
electrical conductors (contact and fracture mechanics, stress analysis, modelization, and mitigation of aeolian vibrations).
Alain Cardou received the M.Sc. and Ph.D. degrees in mechanics and materials
from the University of Minnesota at Minneapolis.
He is Adjunct Professor and, formerly, Head of the Department of Mechanical
Engineering at Universit Laval, Qubec City, QC, Canada. His general research
interests are stress and strength analysis, of which he is the author or coauthor
of many papers. For several years, in collaboration with some power utilities
and within the GREMCA research group, he has been working on overhead
electrical conductor fatigue problems.
Dr. Cardou is a Registered Professional Engineer in the Province of Qubec,
a Fellow of the Canadian Society for Mechanical Engineering, and a member
of the American Academy of Mechanics.