850

Development of glass fiber reinforced plastic

poles for transmission and distribution lines
Sherif Ibrahim, Dimos Polyzois, and Sherif K. Hassan

Abstract: An extensive research project is currently being carried out at The University of Manitoba, Canada, to de-
velop lightweight glass fiber reinforced plastic (GFRP) poles for use in transmission and distribution networks. In this
paper, results from tests involving full-scale tapered GFRP poles with a hollow circular cross section subjected to canti-
lever bending are presented. The filament winding process was employed to produce the poles using polyester resin re-
inforced with E-glass fibers. Cantilever bending tests were conducted on twelve full-scale poles up to failure. Test
parameters included fiber orientation and number of layers. Extensive theoretical work preceded the test program and a
theoretical model was developed for evaluating the failure load. The results to date indicate that the developed theoreti-
cal model can predict quite well the ultimate capacity and behavior performance of GFRP poles. This theoretical model
was used in this investigation to determine the optimum cross-sectional dimensions for 6.1 m (20 ft) and 18.3 m
(60 ft) Class 1 GFRP poles.

Key words: transmission and distribution poles, filament winding, fiber-reinforced plastics.

Résumé : Un projet de recherche approfondie est présentement dirigé à l’Université du Manitoba, Canada, pour déve-
lopper des poteaux en plastique léger renforcé de fibres de verre (PRFV) pour une utilisation dans des réseaux de
transmission et de distribution. Dans cet article, des résultats de tests impliquant des poteaux PRFV en fuseau à échelle
grandeur nature avec une coupe transversale à ouverture circulaire et soumis à un fléchissement en porte-à-faux sont
présentés. Le processus d’enroulement des filaments fut employé pour produire les poteaux en utilisant de la résine en
polyester renforcée avec des fibres de verre de type E. Des tests de fléchissement en porte-à-faux furent conduits sur
douze poteaux à échelle grandeur nature jusqu’à la rupture. Les paramètres des tests ont inclus l’orientation des fibres
et le nombre de couches. Un large travail théorique a précédé le programme de tests et un modèle théorique fut déve-
loppé pour évaluer la charge de rupture. Les résultats à ce jour indiquent que le modèle théorique développé peut bien
prédire la capacité ultime et la performance du comportement des poteaux (PRFV). Ce modèle théorique a été utilisé
dans cette étude pour déterminer les dimensions optimales de la coupe transversale pour un poteau PRFV Classe 1 de
6,1 m (20 pd) et 18,3 m (60 pd).

Mots clés : poteaux de transmission et de distribution, enroulement des filaments, plastiques renforcés de fibres.
[Traduit par la Rédaction] Ibrahim et al. 858

Introduction The main disadvantage of concrete poles is their weight,

Traditional materials such as wood, steel, and concrete are
commonly used as poles in electrical transmission and distri-
bution aerial networks. However, the shortage of wooden
poles, their short life expectancy, and various environmental
concerns related to their treatment have promoted hydroelec-
tric utility companies to search for a cost-effective alterna-
tive. Wooden poles are continuously exposed to weather,
fungi, woodpeckers, etc., which result in a very significant
deterioration of their load bearing capacity with time. The
service life of wooden poles is approximately 20 years
(Vanderbuilt and Criswell 1988). Any extension of this ser-
vice life requires continuous inspection and follow-up care.
In a number of European countries, concrete poles are used.
Received July 28, 1999.
Revised manuscript accepted November 25, 1999.
S. Ibrahim, D. Polyzois, and S.K. Hassan. Department of
Civil and Geological Engineering, The University of
Manitoba, Winnipeg, MB R3T 5V6, Canada.
Written discussion of this article is welcomed and will be
received by the Editor until February 28, 2001.
Can. J. Civ. Eng. 27: 850–858 (2000)
which drastically increases transportation and erection costs.
Chemical influences on the concrete surfaces due to envi-
ronmental impact can also affect their long-term perfor-
mance. As in the case of other concrete structures, concrete
poles are subject to corrosion of the steel reinforcement, re-
sulting in further strength deterioration and expensive main-
tenance. Steel is the most common material for the
construction of transmission poles in North America. These
poles, however, are very expensive (Escher 1982). Corrosion
protection is of primary concern in steel poles, which must
be painted or galvanized, a process that does not always
guarantee long-term protection. Generally, traditional poles
made of wood, concrete, or steel are subject to deterioration
under environmental attacks. Regular maintenance is essen-
tial to prolonging serviceability of these poles.
On the other hand, glass fiber reinforced plastic (GFRP)
poles are lightweight and corrosion resistant. Being light-
weight is a major advantage of GFRP poles, making them
suitable for transportation and installation in mountain ter-
rains, marshes, and remote or inaccessible areas. Although
the initial cost of GFRP poles may be higher than that of tra-
ditional poles, the long-term benefits and the savings in
© 2000 NRC Canada
Ibrahim et al.
transportation and erection that these poles provide make
their selection attractive. The use of GFRP poles is not new.
A number of companies are already involved in the produc-
tion of such poles. Research in this area, however, is limited.
In 1988, Bell initiated its own investigation of GFRP poles
and conducted an experimental program at the Centre de Re-
cherche du Reseau Exterieur (CERRE) (McClure et al.
1992). The specimens were tapered with a hollow cross sec-
tion and were manufactured by centrifugal casting. The test
results indicated that GFRP poles could safely resist loads
comparable to those of wooden poles. The behavior of these
GFRP poles was found to be truly elastic even for large de-
flections.
An experimental investigation was also conducted by
Shakespeare Inc. on filament wound GFRP poles in 1993
(Derrick 1996). Class 4 GFRP poles were tested at the na-
tional test labs of Engineering Data Management (EDM) in
Colorado, U.S.A. The tests were successful, as each pole
met or exceeded the strength requirement specified for
wooden poles.
The objectives of the research program at The University
of Manitoba, Canada, were to (i) develop a theoretical model
for determining the ultimate strength and performance of
GFRP poles, (ii) evaluate the developed theoretical model
through a series of small- and full-scale testing, and (iii) de-
velop design guidelines for the use of GFRP.
To date, twelve small-scale specimens (2.5 m long) and
twelve full-scale specimens (6.25 m long) have been tested
under cantilever bending load up to failure. This paper pres-
ents the results from the full-scale tests.
Experimental program
Specimens and test setup
The poles were fabricated through the filament winding
wet process at the ISIS–Faroex Filament Winding Research
Facility in Gimli, Manitoba. E-glass fibers (1100 TEX) and
vinylester resin (DERAKANE 470-300) were used for man-
ufacturing these poles. The material properties and strength
for both the fiber and the resin were provided by the manu-
facturer (Dow Plastics Company 1997) and are presented in
Table 1.
The specimens were tapered hollow sections, 6250 mm in
length. The inner diameters at the base and at the top were
416 and 305 mm, respectively. The wall thickness for each
specimen varied depending on the number of layers, which
ranged from 4 to 8, in two layer increments, giving a total
thickness of between 2.75 and 5.5 mm at the base. Three fi-
ber angles with respect to the longitudinal axis of the pole
were used: ±5, ±10, and ±20. Hoop winding was also em-
ployed in ten of the specimens with different ratios, while
only two of the specimens were fabricated without hoop
winding. The fiber volume fraction was measured during the
manufacturing process by determining the weight of fiber
and resin used and transforming those weights into volume
fractions. The configurations as well as the fiber volume
fraction for the tested specimens are listed in Table 2.
The fixed support for the specimens consisted of a square
reinforced concrete base measuring 800 mm wide by
1000 mm high. It was made up of two segments with a ta-
pered circular hole in the middle with dimensions that
851
Table 1. Properties of E-glass and vinylester resin.
Vinylester resin
Properties E-glass (DERAKANE 470-300)
Tensile modulus (GPa) 72.4 3.58
Poisson’s ratio 0.2 0.3
Tensile strength (MPa) 2400 85
Shear modulus (GPa) 30 1.38
Density (gm/cm3) 2.54 1.08
matched the outer diameter of the specimens. The two seg-
ments of the base were clamped together using four dwyidag
bars, as shown in Fig. 1. To ensure the stability of the por-
tion of the specimen within the base, a GFRP tapered sleeve,
1000 mm long and 25 mm thick, was inserted in the speci-
men. The concrete base was clamped to a rigid structural
wall using four hollow square steel (HSS) sections (250 ×
250 × 6 mm), two on each side of the base, as shown sche-
matically in Fig. 1. Four high strength dwyidag bars were
used to clamp the HSS section and the concrete base to the
structural wall. To ensure full fixity of the specimen within
the concrete base, the gap between each specimen and the
concrete base was filled with plaster of paris.
The load was applied horizontally, 600 mm below the top,
according to the ANSI Standard 05.1 (1992). An electronic
load cell was used to monitor the applied load, as shown in
Fig. 1. Loading was applied at a rate of approximately
0.25 mm/s.
The lateral deflection of the poles at the loading position
was monitored through an electronic linear measurement
transducer (LMT). The LMT was mounted on a fixed steel
column 10 000 mm away from the specimen. The stroke
range for each LMT was 2500 mm.
Two LMTs attached on two opposite sides of the poles,
approximately 1000 mm above the base, were used to moni-
tor the change in the specimen diameter, as shown in Fig. 1.
These LMTs had a range of 350 mm. The difference in the
readings of the two LMTs represented the change in the
specimen diameter at that height. This method allowed con-
tinuous monitoring of the change in the diameter with load-
ing.
Strain along the specimen near the base was monitored
through 24 electrical resistance strain gauges. The strain
gauges had a 5 mm gauge length and a 120 W electrical re-
sistance. Three strain gauges were mounted at eight different
locations to measure strains in the longitudinal and cir-
cumferential directions as well as at 45° off-axis. Four strain
gauges were located on the tension side and the other four
on the compression side at heights 30, 400, 800, and
1200 mm above the concrete base.
The load cell, the LMTs, and the strain gauges were con-
nected to a computer-controlled 32-channel data acquisition
system to monitor and record all data. Visual observations
were also made during the test. The load was applied until
complete failure of the specimen occurred.
Test results
A summary of all test results is shown in Table 3. The
equivalent wooden pole class is also shown in this table. The
classification method of distribution and transmission poles
is used to compare the strength of poles made from different
© 2000 NRC Canada
852 Can. J. Civ. Eng. Vol. 27, 2000

Table 2. Configuration and fiber volume fraction of full-scale specimens.

Specimen Longitudinal Number of Total number Base thickness Fiber Total weight
No. fiber orientation hoop layers of layers (mm) volume (%) (kg)
1 (10/–10) 0 8 5.5 49 99.40
2 (10/–10) 2 8 5.0 60 80.25
3 (10/–10) 4 8 4.5 60 68.15
4 (10/–10) 2 6 4.0 54 60.85
5 (10/–10) 2 4 2.4 60 39.00
6 (20/–20) 0 8 5.5 53 94.30
7 (20/–20) 2 8 5.0 60 81.00
8 (20/–20) 4 8 4.9 58 72.05
9 (20/–20) 2 6 4.0 57 59.40
10 (20/–20) 2 4 2.75 51 38.20
11 (5/–5) 2 6 4.6 50 62.50
12 (5/–5) 2 4 2.8 52 38.00

Fig. 1. Schematic drawing of the full-scale test setup.
materials under the effect of lateral loads and is employed
for preliminary design purpose only. It should be noted that
if poles made from different materials share the same class,
this does not mean that they would have the same strength
under different loading conditions such as combination of
axial and lateral loading. Specimen 2 failed prematurely in-
side the concrete base because of lack of internal stiffener in
this region. Two specimens (4 and 11) satisfied the require-
ment of class 1 poles, according to ANSI Standard 05.1
(1992), by carrying an ultimate load equal to or exceeding
20 kN (4500 lb). The top displacement of those two speci-
mens was less than 10% of the free height of the pole. Spec-
imens 4 and 11 consisted of only 6 layers and weighed about
60 kg. Specimen 9, which also consisted of 6 layers but had
a fiber orientation of ±20°, failed at an ultimate load of 18.9
kN. This is less by 5% than that required for class 1, so this
specimen was classified as class 2 pole. Five specimens (1,
3, 6, 7, and 8) exceeded the class 1 requirement and were
classified in the heavier categories (H2 and H3). The maxi-
mum lateral displacement of the poles tested was less than
12.7%.
Since each specimen had a different weight and a different
load carrying capacity, the load capacity-to-weight ratio
(kN/kg) was used to compare the performance of the various
specimens. Specimen 7, consisting of eight layers (six longi-
tudinal layers and two circumferential layers), had the high-
est performance ratio of approximately 0.47 kN/kg. Speci-
mens 3 and 8, both consisting of eight layers (four
longitudinal layers and four circumferential layers), had a
load-to-weight ratio of 0.4 kN/kg. Specimens 1 and 6, both
consisting of eight longitudinal layers without any cir-
cumferential layers, did not perform as well as the other
eight-layer pole group and had a load-to-weight ratio of 0.32
kN/kg.
A significant drop in the performance ratio was observed
when the number of layers was reduced to six or four layers.
This was due to the high reduction in the wall thickness,
which in turn reduced the buckling load significantly. As
shown in Table 3, the weight ratio for specimen 4 was ap-
proximately 0.34 kN/kg. This is almost the same value as
that of specimen 1, which had eight longitudinal layers with-
out circumferential layers. This result indicates the impor-
tance of including circumferential layers to improve the
performance of poles.
The load capacity-to-weight ratio of a class 1 6.1 m (20 ft)
long wooden pole is 0.095 kN/kg. The load-to-weight ratio
for an equivalent class 1 GFRP pole (specimen 4 or 11) is
0.3 kN/kg. This means that GFRP poles are three times more
efficient than wooden poles and are approximately only one
third the weight of equivalent wooden poles.
The number of circumferential layers has substantial in-
fluence on the local buckling failure load. Circumferential
layers provide confinement for the longitudinal fibers. For
GFRP poles with the same number of longitudinal layers,
higher local buckling failure load can be achieved by incor-
porating more number of circumferential layers. This was
observed on specimens 3 and 4. Both specimens have four
longitudinal layers; however, specimen 3 has four circumfer-
ential while specimen 4 has two circumferential layers. Con-
sequently, the local buckling failure load of specimen 3 is
about 45% higher than that of specimen 4. The same phe-
nomenon was observed for specimens 8 and 9.
The failure mechanism of specimen 6 is shown in Fig. 2.
A diagonal crack extending to a height between 200 and
1200 mm above the base developed on the compression side.
Specimen 6 was made from eight longitudinal layers at ±20°
with respect to the vertical axis. In this case, the longitudinal
bending stresses would have substantial components in the

© 2000 NRC Canada

Ibrahim et al. 853

Table 3. Performance of full-scale GFRP poles.

Specimen Weight Ultimate Top displacement Equivalent Displacement as Load/weight

No. (kg) capacity (kN) (mm) wooden pole class % of free height (kN/kg)
1 99.40 34.2 486 H3 9.7 0.34
2 80.25 * * * * *
3 68.15 30.1 608 H2 12 0.44
4 60.85 20.8 430 1 8.6 0.34
5 39.00 7.55 240 6 4.8 0.19
6 94.30 30.4 431 H2 8.6 0.32
7 81.00 36.4 634 H3 12.7 0.47
8 72.05 28.1 593 H2 11.9 0.39
9 59.40 18.9 416 2 8.3 0.32
10 38.20 6.9 252 6 5.0 0.18
11 62.50 20 403 1 8.1 0.32
12 38.00 8.8 278 5 5.6 0.23
*This specimen failed prematurely as a result of the lack of internal stiffener within the concrete base.

transverse to fiber direction as well as in-plane shear stress. Fig. 2. Diagonal fracture failure mode.
Since GFRP has very low strength in the transverse direc-
tion as well as low in-plane shear, failure in the transverse to
fiber direction would occur before local buckling load is
reached. On the other hand, specimen 1 was made from
eight longitudinal layers at ±10°. Because of that small fiber
angle, the component of the longitudinal bending stress in
the transverse to fiber direction was not critical and failure
due to local buckling was observed 200 mm above the base.
Moreover, local buckling failure was combined with exces-
sive splitting and delamination of the fibers because of the
lack of any confinement.
All specimens with circumferential layers (with the excep-
tion of specimen 2) failed by local buckling on the compres-
sion side of the specimens at a height that varied from 200
to 800 mm above the concrete base, as shown in Fig. 3. In
those specimens, circumferential layers resisted all the trans-
verse and in-plane shear stresses and prevented any failure in
the transverse direction.

Theoretical analysis
Material properties and strength
Glass fiber-reinforced plastic was used as the construction
material for the poles investigated. The lamina properties
were derived from the material properties of the E-glass fi-
ber and the vinylester resin given in Table 1.
These properties were used to derive effective moduli of the fact that the stresses in fibers and matrix are not equal
the composite material based on micromechanical models. under the corresponding loading conditions. The Tsai–Hahn
The rule of mixture, verified by Adams (1987), was used to equations can be expressed in terms of the fiber and matrix
evaluate the modulus of elasticity in the fiber direction (E1) properties as follows:
and the major Poisson’s ratio (m12) as follows:
(uf + hum )
[1] E1 = Ef uf + Em um [3] E2 =
é uf hum ù
[2] m12 = m f uf + m m um ê + ú
ë Ef Em û
where uf and um are the fiber and the matrix volume ratio, (uf + hum )
respectively; Ef and Em are the fiber and the matrix modulus [4] G12 =
of elasticity, respectively; mf and mm are the fiber and the ma- é uf hum ù
ê + ú
trix Poisson’s ratio, respectively. ëGf Gm û
For calculating the effective Young’s modulus in the trans-
verse direction (E2) and the shear modulus (G12), the Tsai– where h is the stress partitioning parameter. It was found that
Hahn approach (1980) was used. This approach is based on a value of h = 0.5 yields accurate predictions based on com-

© 2000 NRC Canada

854
Fig. 3. Local buckling failure mode.
parison of theoretical results and experimental data for glass
epoxy (Tsai and Hahn 1980). Gf and Gm are the fiber and
the matrix shear modulus, respectively.
Equations [1]–[4] were used to determine the effective
material properties for the composite lamina. The fiber vol-
ume ratio, uf , and the matrix volume ratio, um, were mea-
sured during the manufacturing process.
The strength of the composite can be determined in a sim-
ilar manner based on a micromechanical model and ad-
vanced elasticity methods. The strength of the composite
depends on the relative properties of fibers and matrix and
their corresponding volume fractions. Detailed calculations
of strength are given in Adams and Doner (1967) and
Agarwal and Broutman (1980).
Finite element model
The ANSYS program (1995) was used to develop a non-
linear finite element model of GFRP poles. The geometric
boundaries of the poles were defined as the first step in the
modeling process. Eight-node quadrilateral layered shell ele-
ments were used to model the poles as shown in Fig. 4. For
each layer within the element, orthotropic material proper-
ties in the fiber and transverse to the fiber direction were de-
fined. Fiber orientation for each layer was specified by
defining the fiber angle with respect to the element axes as
shown in Fig. 4. Detailed formulation of this element is
given in Yunus et al. (1989). The portion of the specimen
that was embedded in the concrete base was also included in
the model. In the current study, a geometric nonlinear analy-
sis was used, taking into account the cross-section distortion
Can. J. Civ. Eng. Vol. 27, 2000
as well as the large deflection at the top of the poles. The
specimens were discretized into 784 elements as shown in
Fig. 5. The circumference was divided into 16 elements. In
the longitudinal direction, the element length was taken
equal to 100 mm at the base and gradually increased to
185 mm near the top. This variation in element size was
used to provide a fine mesh near the area of maximum stress
and anticipated failure zone. The model developed was used
to predict the failure load, whether this was due to instability
(local buckling) or material failure. To determine the ulti-
mate load of the modeled GFRP poles, a load step was spec-
ified. This load step was automatically divided into unequal
substeps whose range depended on the behavior of the pole
in the previous substep. At the end of each substep, the
ANSYS program adjusted the stiffness matrix to reflect any
nonlinear changes in the pole stiffness. Ultimate loads were
determined on the basis of the Tsai and Wu (1971) failure
criterion unless divergence was detected due to structural in-
stability (i.e., local buckling) before reaching this failure cri-
terion.
A comparison between the experimental and the predicted
ultimate loads for the tested specimens is shown in Fig. 6.
The average ratio of the experimental-to-theoretical ultimate
load was approximately 0.99 with a standard deviation of
12%. As evident from this figure, there is a strong correla-
tion between the results obtained from the proposed model
and the experimental results.
The load–deflection curves obtained from both the experi-
mental and the finite element results are shown in Fig. 7 for
specimens with ±10° longitudinal fiber angle. It can be seen
© 2000 NRC Canada
Ibrahim et al. 855

Fig. 4. Eight-node layered shell element.

that the load–deflection relationship obtained from the finite Fig. 5. Finite element mesh for analysis of test specimen 4 with
element is stiffer than the experimental results. In addition, cantilever height of 5 m and fiber volume of 54%.
the finite element results are linear because progressive fail-
ure was not accounted for in the analysis.
When a GFRP pole is subjected to bending moment, the
tensile and compressive stresses resulting from the applied
moment tend to flatten or ovalize the cross section. This re-
duces the flexural stiffness of the GFRP pole as the curva-
ture increases. Moreover, the deformation of the cross
section increases the axial bending stresses and lowers the
local buckling load (Ibrahim and Polyzois 1999). The load
versus the ratio of the change in cross-section radius
(ovalization), at 1 m above the concrete base for the ±10°
specimens, obtained from the finite element model as well as
the experimental investigation is shown in Fig. 8. It is evi-
dent in this figure that the load–ovalization curve obtained
from the finite element model has the same nonlinear trend
as the experimental results. This figure indicates the ability
of the finite element model to account for the nonlinear dis-
tortion of the cross section. The ovalization predictions of
the finite element model are less than the experimental mea-
surements because progressive failure was not accounted for
in the analysis.

Optimization study
An extensive theoretical study involving 6.1 m (20 ft) and
18.3 m (60 ft) GFRP poles was performed to determine the
optimum cross-section diameter of these poles that would
meet the same strength criteria as class 1 wooden poles. Ac-
cording to ANSI Standard 05.1 (1992), class 1 wooden poles
are required to have a load carrying capacity of 20 kN
(4500 lb) applied horizontally at 600 mm (2 ft) from the top
of the poles. In the optimization procedure, the fiber angles
were assumed to be ±10° with the two external layers placed
in the circumferential directions, one inner and one outer
layer. The fiber volume ratio of 0.6 was used in calculating
the material properties. The taper ratio of the pole consid-
ered was 0.83% (1 in./10 ft). The embedded portion of the tively. For the 6.1 m (20 ft) specimen, the bottom diameter
GFRP poles under the ground line was 1 m (3.28 ft) and was varied from 152.4 mm (6 in.) to 508 mm (20 in.), while
2.44 m (8 ft) for the 6.1 m and 18.3 m specimens, respec- for the 18.3 m (60 ft) poles, it was varied from 457 mm
© 2000 NRC Canada
856 Can. J. Civ. Eng. Vol. 27, 2000

Fig. 6. Comparison between theoretical and experimental ultimate load.

Fig. 7. Comparison between experimental and finite element load–deflection relationship for 10° full-scale specimens.

(18 in.) to 1016 mm (40 in.). In each case, the wall thickness
of the specimen was determined by trying different number
of layers and the least number of layers that met class 1
strength requirements was considered. The thickness of each
layer was assumed to be 0.44 mm (0.0173 in.). Using the de-
signed wall thickness, the total weight of the poles was de-
termined. The optimum bottom diameter can be obtained by
plotting the total weight versus the bottom diameter. The re-
lationship between the bottom diameter and the weight for
the 6.1 m (20 ft) GFRP poles is shown in Fig. 9. As shown
in the figure, the optimum bottom diameter for the case ex-
amined in this program was 305 mm (12 in.). A similar opti-
mization study was conducted for the 18.3 m (60 ft) GFRP
poles, shown in Fig. 10. The optimum bottom diameter in
that case was 559 mm (22 in.).
Conclusions
An experimental study was used to investigate the ulti-
mate capacity and performance of tapered filament wound
GFRP poles subjected to cantilever bending. The study
showed that the ultimate load capacity of GFRP poles is

© 2000 NRC Canada

Ibrahim et al.
Fig. 8. Comparison between experimental and finite element load–ovalization relationship for 10° full-scale specimens.
Fig. 9. Optimization of 6.1 m (20 ft) GFRP poles.
equivalent to that of wooden poles. However, a significant
saving in the weight is achieved by utilizing GFRP. The
load capacity-to-weight ratio of GFRP poles is almost three
times higher than that of wooden poles. Maximum load ca-
pacity-to-weight ratio was achieved by incorporating cir-
cumferential winding along with longitudinal fibers. Of the
eight-layer GFRP poles tested, the maximum load capacity-
to-weight ratio was attained when two circumferential layers
were combined with six longitudinal layers. Local buckling
was the most dominant mode of failure in most of the speci-
857
mens tested because of the high radius-to-thickness ratio of
the specimens. Only one specimen suffered material failure
on the compression side because of the absence of cir-
cumferential layers in that specimen. The finite element
method used in this investigation provided an excellent pre-
diction of the critical buckling and material failure loads, as
well as the corresponding modes of failure for thin-walled
GFRP poles. Optimum cross-section dimensions for both
6.1 m (20 ft) and 18.3 m (60 ft) class 1 GFRP poles were
obtained theoretically using the finite element model.
© 2000 NRC Canada
858
Fig. 10. Optimization of 18.3 m (60 ft) GFRP poles.
Acknowledgements
The research project in this study was sponsored by ISIS-
Canada, Faroex Ltd. of Manitoba, and the Province of Mani-
toba. The specimens were fabricated at the ISIS–Faroex Fil-
ament Winding Research Facility, and the testing was
conducted at the Structural Engineering and Construction
R&D Facility at The University of Manitoba.
References
Adams, R.D. 1987. Damping properties analysis of composites. In
Engineering materials handbook. Edited by T.J. Reinhart et al.
ASM International, Materials Park, Ohio, Vol. 1, Composites,
pp. 206–217.
Adams, F.D., and Doner, D.R. 1967. Transverse normal loading of a
unidirectional composite. Journal of Composite Material, 1: 152.
Agarwal, B., and Broutman, L. 1980. Analysis and performance of
fiber composites. Wiley Interscience publication, New York.
ANSI. 1992. American national standard for wood poles 05.1,
specifications and dimensions. American National Standards In-
stitute, New York, N.Y.
ANSYS, Inc. 1995. ANSYS user’s manual for revision 5.2 —
Volumes I to IV. Houston, Pa.
Derrick, G.L. 1996. Fiberglass composite distribution and trans-
mission poles. Manufactured Distribution and Transmission Pole
Can. J. Civ. Eng. Vol. 27, 2000
Structures Workshop Proceedings, Electric Power Research In-
stitute, July 25–26, Denver, Colo., pp. 55–61.
Dow Plastics Company. 1997. Technical product information. Mid-
land, Mich.
Escher, G.A. 1982. Design and testing of an optimum fiberglass re-
inforced pole and market study on the demand for utility poles
in Canada. Report prepared for the Canadian Electrical Associa-
tion, Montreal, Que.
Ibrahim, S., and Polyzois, D. 1999. Ovalization analysis of fiber-
reinforced plastic poles. Composite Structures, 45: 7–12.
McClure, G., Boire, L., and Carriere, J.C. 1992. Applications of
advanced composite materials in overhead power lines and tele-
communications structures. In Advanced composite in bridges
and structures. Edited by K.W. Neale and P. Labossière. Cana-
dian Society for Civil Engineering, Montreal, Que., pp. 543–549.
Tsai, S.W., and Hahn, H.T. 1980. Introduction to composite materi-
als. Technomic Publishing Co., Lancaster, Pa.
Tsai, S.W., and Wu, E.M. 1971. A general theory of strength for
anisotropic materials. Journal of Composites Materials, 5: 58–80.
Vanderbuilt, M.D., and Criswell, M.E. 1988. Reliability analysis of
pole-type transmission structures. Computer and Structures,
6(2): 335–343.
Yunus, S., Kohnke, P.C., and Saigal, S. 1989. An efficient through-
thickness integration scheme in unlimited layered doubly curved
isoparametric composite shell element. International Journal for
Numerical Methods in Engineering, 28: 2777–2793.
© 2000 NRC Canada