113

A new steel anchorage system for

post-tensioning applications using carbon fibre
reinforced plastic tendons
Ezzeldin Y. Sayed-Ahmed and Nigel G. Shrive

Abstract: During the past half century, the use of prestressing in different structures has increased tremendously. One of the
most important techniques of prestressing is post-tensioning. The main problem associated with post-tensioning in different
structures is the corrosion of the prestressing steel tendons even with well-protected steel. New materials, fibre reinforced
plastics or polymers (FRP), which are more durable than steel, can be used for these tendons/strands and thus overcome the
corrosion problem. However, different shortcomings appear when FRP tendons are introduced to post-tensioning prestressing
applications. For carbon fibre plastic tendons (CFRP), there is no suitable anchorage system for post-tensioning applications.
Some of the anchorages developed by others for use with FRPs are therefore described and assessed. A new anchorage system
developed by the authors, which can be used with bonded or unbonded CFRP tendons in post-tensioning applications, is
described. The results of direct tension and fatigue tests on CFRPs anchored with the new system are presented.
Key words: anchorage system, cyclic loading, fatigue, fibre reinforced plastics, finite element analysis, post-tension,
prestressed concrete, prestressed masonry, strands, tendons.

Résumé : Au cours des 50 dernières années, l’utilisation de la précontrainte dans différentes structures s’est accrue
remarquablement. Une des principales techniques de précontrainte est la post-tension. Le principal problème relatif à la
post-tension dans différents ouvrages est la corrosion des câbles en acier précontraints, même s’il s’agit d’acier bien protégé.
De nouveaux matériaux, tels que les plastiques ou polymères renforcés de fibres (PRF), qui s’avèrent plus durables que
l’acier, peuvent être utilisés pour ces câbles/torons, et enrayer les problèmes de corrosion. Toutefois, différents défauts
apparaissent lorsque les câbles en PRF sont soumis à des applications précontraintes en post-tension. Pour les câbles en
plastique renforcé de fibre de carbone (PRFC), il n’existe aucun système d’ancrage adéquat pour les applications de
post-tension. Quelques ancrages développés par d’autres auteurs utilisant des PRF sont décrits et évalués. Un nouveau
système d’ancrage que nous avons développé, lequel peut être employé avec des câbles en PRFC scellés ou non dans des
applications de post-tension est décrit. Les résultats de tests de traction directe et de fatigue sur des PRFC ancré avec le
nouveau système sont présentés.
Mots clés : système d’ancrage, charge cyclique, fatigue, plastiques renforcés de fibre, calculs d’éléments finis, post-tension,
béton précontraint, maçonnerie précontrainte, torons, câbles.
[Traduit par la rédaction]

Introduction times the bending resistance of a cavity wall; prestressing pro-

Prestressing of structural members was established early this
century, and the technique has been refined over time. In gen-
eral, masonry and concrete are strong in compression but weak
in tension; thus, prestressing improves their structural behav-
iour, particularly under flexural loading. As an example, for
masonry walls, Curtin et al. (1982) showed that for a 20%
increase in cross-sectional area, a diaphragm wall can have 15
Received February 11, 1997.
Revised manuscript accepted June 14, 1997.
E.Y. Sayed-Ahmed. Department of Structural Engineering,
Faculty of Engineering, Ain Shams University, Cairo, 11571,
Egypt.
N.G. Shrive. Department of Civil Engineering, The University
of Calgary, Calgary, AB T2N 1N4, Canada.
Written discussion of this article is welcomed and will be
received by the Editor until June 30, 1998 (address inside
front cover).
vides a further increase in resistance, perhaps up to 150 times
that of the original cavity wall. Shaw (1996) suggested that the
multiplying factor can be as high as 500. Prestressing of ma-
sonry has typically been applied in the form of post-tensioning
with unbonded steel tendons, as it is the most simple, yet most
effective and cheap technique of prestressing masonry. Post-
tensioning has been used both in new masonry structures and
to rehabilitate old ones (Ganz and Shaw 1997).
One of the most common problems associated with steel is
corrosion, even when the “right” protection technique is used.
Significant loss of prestressing may occur as a result of tendon
corrosion which may, in turn, lead to catastrophic failure. Even
if failure does not occur, the serviceability advantage of post-
tensioned concrete and (or) masonry may disappear if the
prestressing force in steel is lost or reduced as a result of cor-
rosion. Various measures have been used to provide protection
against corrosion in steel tendon post-tensioned masonry
(Foster 1970; Bradshaw et al. 1982). However, corrosion of
the steel tendons remains the major shortcoming of post-tensioning
applications (sometimes even with pre-tensioning applications).

Can. J. Civ. Eng. 25: 113–127 (1998) © 1998 NRC Canada

114
Serious problems have been reported regarding the perform-
ance of unbonded steel tendons used for post-tensioned struc-
tures (Elliott and Morrison 1995). Thus, what starts as a dream
of having an economic and aesthetic structural concrete/ma-
sonry element may turn into a nightmare of expensive reha-
bilitation due to the corrosion of steel.
Instead of protecting the steel strands with a coating that
can be damaged and becomes ineffective, or curing the exist-
ing corrosion problems by adding more prestressing steel
strands, the idea of replacing steel with a new type of material
has emerged. This new type of material is fibre reinforced
plastics (FRP). The main advantages of FRP tendons over steel
are their higher durability and corrosion resistance, lighter
weight, and higher tensile strength. Due to these advantages
over steel, FRPs have been pioneered in recent years as an
alternative to prestressing steel tendons. Most FRPs used today
are reinforced with glass (GFRP), aramid (AFRP), and (or)
carbon (CFRP). Both CFRP and AFRP have been used for
both pre- and post-tensioning applications. The world’s first
highway bridge, with a span of 47 m, prestressed with GFRP
was built in Germany in 1986. The tendon, which is comprised
of nineteen 7.5 mm diameter pultruded GFRP rods, has a poly-
mer-mortar type anchorage system (Ballinger 1991). In 1994,
two masonry footbridges were lowered into place in the United
Kingdom. One of them incorporated Parafil rope prestressing
tendons; the other was prestressed with traditional steel ten-
dons (Shaw and Baldwin 1995; Shaw et al. 1995). In Canada,
a bridge with a number of girders prestressed with CFRP ten-
dons was built in Calgary (Grant et al. 1995).
Of the FRP considered, CFRP exhibits the highest tensile
strength (Hercules Aerospace Co. 1995), excellent fatigue
strength compared with steel tendons (Rostasy et al. 1993),
and very low relaxation (Rao 1992; Santoh et al. 1993). The
most attractive properties of CFRP are the high durability and
corrosion resistance compared with steel tendons. An exami-
nation of some of the properties of CFRP compared with other
FRPs reveals the advantage of CFRP. In Table 1, most of the
mechanical properties (collected from different references) of
the GFRP, AFRP, and CFRP tendons are shown and compared
with steel tendons.
As can be seen in Table 1, CFRP exhibits the highest short-
term static tensile strength and the highest modulus of elastic-
ity, compared with other FRPs. Poisson’s ratio in the fibre
direction is nearly the same as for other fibre types. CFRP also
has excellent dynamic (fatigue) strength when subject to cyclic
loading. When subjected to 2 × 106 cycles of loading, the
Leadline (CFRP tendons produced by Mitsubishi Chemical
Corporation) tendon results in an endurance limit of 1100 MPa
and CFCC (CFRP tendons produced by Tokyo Rope Manu-
facturing Co., Ltd.) 1300 MPa. Stress ratios of 0.1 and 0.54
were used for the Leadline and CFCC respectively (Tokyo
Rope 1990; Mitsubishi1).
The most significant advantage of FRP in general and
CFRP in particular is the high durability and corrosion resis-
tance compared with steel tendons. Sen et al. (1992) reported
complete deterioration in about 18 h for GFRP prestressed
piles. Nanni et al. (1992) tested AFRP rods and noticed that at
elevated temperature the rods preserved 97% of their capacity
1 2
Mitsubishi Chemical Corporation. 1996. Personal
communication with Mr. Sam Samejima.
Can. J. Civ. Eng. Vol. 25, 1998
after 2160 h in pH 13 alkali solutions at 80°C. Tests performed
by Mitsubishi on the Leadline (CFRP) showed a loss of about
0.5% of the strength after 528 h in alkali solution at 40°C, and
Tokyo Rope reported that CFCC (CFRP) strands in pH 13
alkali solution at 60°C retained 97.6% of their strength after
2500 h. The solutions used in all of the above tests were to
simulate direct contact between FRP and concrete with no
protection against alkaline reactions. GFRP coated with epoxy
was used to reinforce a concrete bridge in Calgary (Tadros2).
FRP tendons also have some disadvantages. Typically, they
do not exhibit an inelastic response like steel tendons; no duc-
tility is observed and failure strains are relatively low. This
shortcoming must be addressed in design procedures before
there will be any widespread practical usage of FRP in
prestressing applications. The second shortcoming of CFRP
tendons is the relatively high price compared even with well-
protected steel strands where CFRP costs can be 1.7–3.0 times
the cost of the steel strands. However, the cost of handling and
transportation is a plus for FRP, as they are much lighter than
steel. Furthermore, comparing CFRP with steel, the price and
performance of the tendon should be considered rather than
the price alone, which yields an advantage for CFRP tendons.
The price of FRP in general has been going down very quickly
over the past few years, so the cost disadvantage may disap-
pear in time.
Other than these disadvantages, the key problem facing the
application of CFRP (and FRP in general) tendons in post-ten-
sioning is how to anchor them. An anchorage system must be
designed which allows for the development of the full strength
of the prestressing cable with minimal creep and loss of load
at transfer, whilst also satisfying all the other technical require-
ments of a post-tensioning anchorage–tendon assembly.
The traditional anchorages for CFRP tendons involve ep-
oxy resin or expansive cement anchorages which were devel-
oped mainly for pre-tensioning applications or the traditional
wedge anchorages (e.g., the die-cast wedge used by Tokyo
Rope for CFCC tendons). Loss of load due to displacements
and expected large creep in the resin- or cement-based anchor-
age systems is likely to make them inadequate for “short” span
post-tensioned members. On the other hand, wedge systems
affect the load capacity of the FRP as they tend to dig into the
surface of the tendon causing premature failure. In the case of
using the die-cast wedge system (CFCC tendons), the main
disadvantage arises from having to predefine the length of the
tendon. Hence, the first stage for post-tensioning with CFRP
tendons is the development of an appropriate anchorage sys-
tem which can be used throughout the life time of the
prestressed member without affecting the strength of the FRP
tendons.
Objective and background
The objective of the work described here was to develop a new
anchorage system for post-tensioning applications which does
not have the disadvantages of the anchorages currently used
for CFRP tendons and simultaneously satisfies the require-
ments for post-tensioning anchorage systems.
Tadros, G. 1997. Personal communication regarding Crowchild
Bridge in Calgary.
© 1998 NRC Canada
Sayed-Ahmed and Shrive 115

Table 1. Properties of GFRP, AFRP, and CFRP compared with prestressing steel tendons.*
GFRP AFRP CFRP Prestressing
Property E-glass S-glass Arapree FiBRA Technora Parafil Leadline CFCC steel
Fibre/resin Glass/ Glass/ Twaron/ Kevlar 49/ Technora/ Carbon/ Carbon/
epoxy polyimide epoxy epoxy vinyl ester epoxy epoxy
Anchorage system Resin Wedge
Resin Resin Grouted potted or resin Barrel/ Wedge/soft Resin
anchor anchor wedge wedge potted spike metal anchor
Minimum fibre volume ratio 0.55 0.5 0.45 0.65 0.65 0.65 0.65 —
Density (g/cm3) 2.1 2.0 1.25 1.28 1.3 1.44 1.53 1.5 7.85
Longitudinal tensile strength (GPa) 1.08 1.28 1.2~1.5 1.25~1.4 1.7~2.1 1.21 2.25~2.55 1.8~2.1 1.86
Transverse tensile strength (MPa) 39 49 30 57 1860
Longitudinal modulus (GPa) 39 43 62~64 65~70 54 120 142~150 190
Transverse modulus (GPa) 8.6 8.9 5.5 10.3 190
In-plane shear strength (MPa) 89 69 49 71
In-plane shear modulus (GPa) 3.8 4.5 2.2 7.2 73.1
Major Poisson’s ratio 0.28 0.27 0.38 0.34~0.6 0.35 0.27 0.3
Minor Poisson’s ratio 0.06 0.06 0.02 0.02 0.3
Bond strength (MPa) 7.7 12.2 9.7~13 6.6~7.1
Maximum longitudinal strain (%) 2.8 2.9 2.4 2.0 3.7 1.5 1.3~1.5 1.57 4.0
Maximum transverse strain (%) 0.5 0.6 0.6 4.0
Longitudinal compressive strength
(MPa) 620 690 335 1440 1860
Transverse compressive strength
(MPa) 128 158 158 228 1860
Longitudinal thermal expansion
coefficient (/°C) 7 × 10–6 5 × 10–6 –2 × 10–6 –2 × 10–6 –3 × 10–6 –0.9 × 10–6† 0.5 × 10–6 11.7 × 10–6
Transverse thermal expansion
coefficient (/°C) 21 × 10–6 26 × 10–6 50 × 10–6 60 × 10–6 27 × 10–6 11.7 × 10–6
Relaxation ratio (%) at room
temperature 11–14% 6–9% 2–3% 1% 8%
*
Source: Daniel et al. 1994; Nanni et al. 1996; Benmokrane et al. 1996; Holte et al. 1993; Tokyo Rope 1990; Mitsubishi 1996; Mathys et al. 1995;
and Santoh 1993.
†
According to tests by the authors on Leadline.

The main requirements for a post-tensioning anchorage are

as follows:
1. The anchorage system must develop a minimum of 95%
of the nominal ultimate tensile strength of the tendon, which
is called the anchorage efficiency. This is a major requirement
for the anchorage. Here, we will consider that the nominal
ultimate tensile strength of the tendon to be the one defined by
the manufacturer and guaranteed to be the minimum strength
of the tendon. The tendon should thus fail at loads above 95%
of its maximum nominal tensile strength rather than slip out of
the anchorage, fail prematurely, or cause anchorage failure.
The anchorage efficiency can be defined by
(Ftu)meas (Ftu)meas
[1] η= = least 50 load cycles between 50% and 80% of the tendon
(Ftu)theo Asp ftu
where η is the anchorage efficiency; (Ftu)meas is the failure
force of the tendon–anchorage assembly obtained for a di-
rect tension test; (Ftu)theo is the nominal failure force of the
tendon based on its maximum nominal strength; Asp is the
cross-sectional area of one tendon; and ftu is the nominal ten-
sile strength of the tendon.
2. At the release of the jacking force, the anchorage must
undergo a very small, predictable displacement to reduce the
amount of prestressing loss in the tendons, especially in the
“short” lengths which may be expected in masonry walls or
some prestressed concrete beams.
3. Long-term static stress (prestress) should not reduce the
tensile strength of the tendon significantly. This means that the
tendon–anchorage assembly should have adequate creep rup-
ture strength over the service life of the structure.
4. The dynamic stress amplitude of the service load state
should not reduce the efficiency of the anchorage–tendon as-
sembly. In other words, fatigue failure of the anchorage com-
ponents should not occur. For prestressing steel strands, the
Post-tension Institute (PTI) recommended that 500 000 load
cycles between 60% and 66% of the specified tendon strength
should be withstood. Then the same sample should resist at
strength without failure. Following the fatigue test, the anchor-
age system should demonstrate at least 95% efficiency. We
have applied the same concept to the CFRP tendon–anchorage
assembly.
5. The anchorage must perform at the same level throughout
the service life of the structure. The stressing operation should
only have to be performed once. Thus, creep in the anchorage
must be minimal.
6. Environmental effects must not affect or reduce the
strength of the FRP tendon–anchorage assembly significantly.

© 1998 NRC Canada

116
7. Galvanic or other corrosion reactions between the FRP
tendon and the anchorage or between the anchorage and the
surrounding media (if embedded) must be avoided.
Many anchorage systems have been developed for post-
tensioning with FRPs which satisfy some, but not all, of the
above requirements. The most common failure modes of FRP
anchorage systems can be summarized as follows:
1. Rupture of the cable/rod within its free length. This is the
ideal mode which indicates that the anchorage is working well
because the tensile capacity of the FRP cable/rod is developed
totally.
2. Shear failure in the anchorage zone. The cable/rod may
be pinched due to the large shear stress concentration that oc-
curs with certain anchorage geometries. The shear stress
causes premature failure of the tendon. This kind of failure
occurs if the anchorage systems currently used for steel ten-
dons are used for CFRP tendons.
3. Bond failure between the epoxy and the tendon. In many
epoxy resin anchorage systems developed for FRPs, the bond
between the resin and the tendon fails. There is no subsequent
load transfer between the anchorage and the tendon.
4. Excessive deflection and (or) long-term creep. The low
elastic modulus epoxy resin is very sensitive to high tempera-
ture and also exhibits long-term creep deformation. As a result,
undesired longitudinal deformation resulting from these two
factors may lead to significant loss of prestress force. This
serviceability type of failure occurs with epoxy resin anchor-
age systems.
5. Slip failure between the tendon and the grip. This type
of failure is catastrophic and leads to complete loss of the
prestressing force due to the tendon pulling out from the an-
chorage.
CFRP tendons, and FRP in general, have very poor lateral
and shear strengths (Table 1). Thus, they are much more sen-
sitive to loads in the transverse direction compared with steel
tendons. However, conventional anchorage systems cause
stress concentrations in the transverse direction around the
wedge teeth, which generally lead to premature failure of the
tendon (failure mode type 2). Thus, the conventional wedge
type anchorage systems currently used to post-tension steel
strands typically cannot be used with CFRP strands. New an-
chorage concepts are required for CFRPs. Currently there are
some anchorage systems, mostly used for pre-tensioning ap-
plications, where the FRP tendon is bonded to the concrete.
The prestressing force is transmitted to the concrete after it
hardens through bond rather than through anchorage. Some of
the more common types of FRP anchorage systems used to
date are shown in Fig. 1 and summarized below (Holte 1993a,
1993b):
1. Split wedge anchorage system where a metal wedge in a
conical housing is used to grip the tendon (Tokyo Rope 1990).
The main anchorage concept is that the wedges compress the
perimeter of the tendon and teeth in the wedges grip it. Since
FRPs have very low lateral and shear strengths, the tendons
fracture easily due to the biting action of the wedge. Enka
(1986) used a plastic wedge system, but use of this system is
limited to pre-tensioning prestressing.
2. Plug-in-cone anchorage system where a bundle of ten-
dons is placed in a conical housing socket. A solid cone (spike)
is then drawn into the bundle centre to splay out the tendons,
gripping them individually between the spike itself and the
Can. J. Civ. Eng. Vol. 25, 1998
socket. The system is reported to perform well under a static
load (Burgoyne 1990). The system has the advantage of not
using resins around the tendons and thus suffers no creep de-
formation and is not sensitive to elevated temperature. The
main disadvantage of this system with respect to FRP tendons
is that the tendons are not straight at the front of the anchorage,
which may shatter the tendons. A similar system composed of
a barrel and spike (not shown in Fig. 1) is used with Parafil
(Burgoyne 1993). With this system, to ensure that the rope is
the correct length, allowance has to be made for the bed down
of the spike.
3. Resin–sleeve anchorage system where an epoxy resin is
injected between a cylindrical steel shell (sleeve) and the ten-
dons. The inside surface of the sleeve is usually deformed or
threaded to improve the load transfer (Wolff and Miesser
1989; Tokyo Rope 1990). In addition to suffering excessive
creep deformation and being sensitive to moisture and thermal
loads, rod bond failures have also been reported for this an-
chorage system (Holte et al. 1993a).
4. Resin-potted anchorage system where the resin–sleeve
anchorage system is modified to this geometry to achieve bet-
ter anchorage. The resin-potted anchorage system is actually
a combination of the split wedge system and the resin–sleeve
system where the compressive action of the split wedge is
developed while the continuous bond of the resin relieves the
biting action of the teeth. However, creep deformation and
sensitivity to thermal loading and moisture are still major prob-
lems for this type of anchorage (Dolan 1991). Numerical finite
element analysis (Holte et al. 1993a) showed that a very high
concentration of shear stress occurs at the anchorage front end.
To reduce the stress concentration, Holte et al. (1993a) sug-
gested using a parabolic change in the angle of slope of the
outer sleeve. Meier (1995) suggested changing the stiffness of
the resin through the length of the anchorage and has used this
type of anchorage for cable stays.
5. Soft-metal overlay anchorage system, used by Mitsubishi
for the Leadline (CFRP) tendons. The gripping pressure is
transferred to the FRP rods through a soft-metal tube (sleeve).
With this configuration the metal sleeve is permanently at-
tached to the cable and gripping is achieved using a conven-
tional strand chuck. Typically, the soft metal is aluminum or
an aluminum alloy. These materials corrode in concrete and
are thus unsuitable with post-tensioning or unbonded tendons.
The use of this system is again limited to pre-tensioning appli-
cation with bonded tendons.
6. Swaged anchor where the rod/cable is embedded in a resin
and transverse stress is generated along a steel shell using bolts
and nuts. Increased friction along the surface of the tendons is
generated and provides the required gripping (Sippel 1992).
Besides the above systems, another anchorage has been de-
veloped using highly expansive materials (Harada et al. 1995;
Khin et al. 1996). This anchorage system has been used in
Japan with ordinary steel and AFRP and CFRP tendons. Khin
et al. (1996) showed that after few cycles of repeated loading,
the stress concentration is less with this anchorage than with
the epoxy resin one. This anchorage is still in the research stage
and requires a special technique for use in the field.
Consideration of the systems described above and their
various limitations indicates that a new system is required: one
that meets all the required technical criteria, is simple to use,
and can be utilized in post-tensioning applications. The latter
© 1998 NRC Canada
Sayed-Ahmed and Shrive
Fig. 1. Schematic of different anchorage systems currently available for CFRP tendons.
requirement adds further criteria regarding durability and
long-term behaviour.
The new anchorage system
Description of the new anchorage system concepts
We have designed a new anchorage system by merging con-
cepts from the split-wedge, the plug-in, and the soft-metal
117
overlay anchorage systems (Shrive et al. 1996). The new an-
chorage is resin-free and is very easy to put together in the field;
it requires no new or advanced technology either to manufac-
ture or to use. The single strand anchor is made of three parts:
1. An outer steel cylinder (outer sleeve) with a conical hole
as shown in Fig. 2. The inside surface of the hole is very
smooth and grease is added to facilitate movement between
this outer sleeve and the inner wedge.
© 1998 NRC Canada
118
Fig. 2. Schematic details of the new designed anchorage system for CFRP tendons.
2. A four-piece wedge (spike) which has a smooth outer
surface. The wedge has a central hole and the surface of this
hole is sand-blasted. The edges of the four pieces at this hole
are rounded to reduce the stress concentration on the tendon
when the spike is seated inside the outer sleeve. The angle of
inclination of the outer surface of the spike is slightly larger
than that of the inner surface of the outer sleeve. The inner
hole of the spike is drilled to the outer diameter of the inner
sleeve.
3. The inner sleeve is made of steel or copper and has a
Can. J. Civ. Eng. Vol. 25, 1998
small wall thickness. The inner diameter of the sleeve is drilled
to the diameter of the tendon. The outer surface of the sleeve
is sand-blasted.
Early prototypes were made of mild steel to the dimensions
shown in Fig. 2, on the basis that a long gripping length would
be required to transfer load without a high stress concentration.
Test results and numerical analysis using the finite element
method allowed us to reduce the dimensions to make the an-
chorage more practical. These new dimensions are shown in
Fig. 3. This latter anchorage is made of high-strength stainless
© 1998 NRC Canada
Sayed-Ahmed and Shrive
Fig. 3. Modified dimensions of the new designed anchorage system for CFRP tendons.
steel (0.2% proof stress of 862 MPa and tensile strength of
1000 MPa) with a copper soft-metal overlay (Figs. 4 and 5).
The three main concepts of the system are as follows:
1. The small differential slope between the outer sleeve and
the spike. This distributes the pressure on the CFRP prestress-
ing tendon more evenly over the length of the anchorage, com-
pared to a normal (steel tendon) anchorage where the conical
hole and the wedge spike have the same angle. With such a
“normal” anchorage, high stress concentration occurs near the
119
leading edge of the tendon. A steel tendon yields and accepts
this biting action, but FRP tendons break.
2. The rounded edges of the inner surface of the 4-piece
wedge. As the wedge tightens on the soft-metal sleeve, the
metal in the sleeve yields and flows a small way into the gaps
between the four wedge pieces. The sharp corner of the wedge
“digs” into the soft-metal overlay and the stress concentration
is passed on to the tendon, causing premature failure in fatigue
tests. Rounding the corner reduces the stress concentration.
© 1998 NRC Canada
120
Fig. 4. The new anchorage system.
Fig. 5. The new anchorage system; one anchorage is assembled with the Leadline.
3. The inner thin deformable sleeve which uniformly distributes
lateral pressure from the wedge around the CFRP cross section.
It is possible that materials other than steel and copper could
be used for the anchorage, and variations on the theme here
could be used to make multi-strand anchorages.
Testing the new anchorage system
Two types of test were performed on the CFRP tendons with
Can. J. Civ. Eng. Vol. 25, 1998
the new anchorage system. The first was a direct tensile test to
obtain the efficiency of the anchorage and the static short-term
tensile strength of the tendon–anchorage assembly. The second
was a fatigue test to check one aspect of the long-term behaviour
and the dynamic axial strength of the tendon–anchorage assem-
bly when subject to dynamic cyclic loading. Tests were per-
formed on both the prototype anchorage shown in Fig. 2 and
the practical one shown in Figs. 3–5.
© 1998 NRC Canada
Sayed-Ahmed and Shrive
Fig. 6. Test arrangement for direct tension and fatigue testing.
Short-term axial strength (direct tension test)
The direct tension test with the anchorage was performed with
8 mm diameter indented spiral Leadline (CFRP) tendon on a
standard tensile testing machine. All the CFRP tendon speci-
mens used in the tests were about 1 m in length (Fig. 6). The
maximum tensile strength of the 8 mm Leadline as defined by
the manufacturer is 104 kN, which was taken to be the nominal
failure load of the tendon.
Our first aim was to achieve 95% efficiency in the an-
chorage, which implies a failure load of 99.0 kN. Modifica-
tions after initial failure loads of 104 kN gave an average
failure load of 114 kN (range: 105–124 kN), exceeding the
specified nominal strength of the tendon, guaranteed by the
manufacturer, by a minimum of 10%. Some of the fibres
which were left after failure of the Leadline are shown in
Fig. 7. A total of 24 tests were performed: 10 using the
prototype anchorage and 14 using the smaller, refined an-
chorage.
With the exception of one test, final failure occurred in the
free length of the Leadline away from the anchorage zone
despite the fact that some cracks tend to extend to (and may
be initiated at) the anchorage zone. Failure typically started
with the splintering of a small amount of fibres at more than
99% of the failure load. Catastrophic failure followed very
quickly with the remainder of the test specimen shattering
121
apart. In fact, failure of the CFRP tendon was very sudden.
When we filmed failure using 60 frames/second video-cam-
era, in one frame a dark cloud appeared around the free
length of the CFRP tendon and in the very next frame shat-
tering of the tendon took place leaving cracked tendon at
the anchorage zone and nothing but shattered fibres in the
free length. The front edge of the anchorage after failure is
shown in Fig. 8.
At failure, with the preliminary tests using the prototype
model, the maximum seating of the anchorage was about
5 mm from our lightly placed initial assembly position. When
the anchorage system was stressed up to 70 kN, seating the
order of 3 mm was observed, which may be a more appropriate
value for practical post-tensioning applications. This 3 mm
displacement was eliminated in the subsequent design by seat-
ing the anchorage with an oil pump and jacking frame before
use.
Dynamic strength (fatigue testing)
The relation between maximum stress and fatigue life (S–N
curve) is usually used to define fatigue behaviour and the dy-
namic strength of the material. Fatigue life of any material
usually depends on the maximum stress, stress range, stress
ratio, rate of loading/unloading, specimen geometry, grips,
temperature, humidity, etc. Thus, the evaluation of the fatigue
© 1998 NRC Canada
122 Can. J. Civ. Eng. Vol. 25, 1998

Fig. 7. Leadline after failure in a direct tension test.

Fig. 8. The prototype anchorage after a direct tension test showing failure of the Leadline.

behaviour usually requires a large number of tests. We have
begun to evaluate the behaviour of the tendon–anchorage as-
sembly subject to cyclic loading.
The new anchorage has been tested using a standard testing
machine with a load range of ±100 kN and stroke range of
±50.0 mm. The specimens used were 8 mm diameter Leadline
(CFRP) tendons with an approximate length of 1 m. Three
tests were performed on the prototype anchorages and two on
the refined anchorages as follows:
1. A total of 1.7 × 106 cycles (at a rate of 5 cycles/s) with the
following sequential load ranges and number of cycles:
— 30–50 kN which represents 29–48% of the nominal
strength with up to 800 000 cycles.
— 40–60 kN which represents 38–58% of the nominal
strength with up to 500 000 cycles.
— 50–70 kN which represents 48–67% of the nominal
strength with up to 330 000 cycles.
— 60–80 kN which represents 57–77% of the nominal
strength with 72 700 cycles before failure took place at
this load level.
2. A total of 2.0 × 106 cycles with the following sequentia-
load ranges and number of cycles:

© 1998 NRC Canada

Sayed-Ahmed and Shrive 123

— 40–60 kN which represents 38–57% of the nominal Fig. 9. Load against strain and deformation of cyclic loading test 5:
strength with up to 1 995 000 cycles. The rate used with (a) loading the specimen up to 40 kN and then unloading before
this load level was 5 cycles/s. cycling; (b) loading the specimen up to 90 kN and then unloading
— 6–60 kN which represents 5–57% of the nominal before cycling; (c) loading the specimen up 95 kN and then
strength with up to 5200 cycles, before failure at this unloading after performing the PTI cyclic load.
load level. The rate used with this load level was 2 cy-
cles/s.
The third test followed the PTI recommended test for steel
tendon–anchorage assemblies.
3. A total of 0.5 × 106 cycles with the following sequential
load ranges and number of cycles:
— 500 000 cycles between 62 and 68 kN which represents
60–66% of the nominal strength at 5 cycles/s.
— 50 cycles between 52 and 83 kN which represents
50–80% of the nominal strength at 1 cycle/s.
— A static tension test up to 99 kN (corresponding to
95.2% efficiency) with no failure.
— At this stage the PTI test had been completed success-
fully so the specimen was loaded between 5 and 55 kN
at 10 cycles/s. When the first crack appeared at about
8000 cycles, the dynamic loading was stopped, and the
specimen was loaded statically. The tendon carried up
to 70 kN before failure occurred.
The following two tests were performed on the refined an-
chorage systems which again start by satisfying the PTI re-
quirements then continue with different cyclic loadings:
4. A total of 2.25 × 106 cycles with the following sequential
load ranges and number of cycles:
— A static load up to 95 kN then unloading the specimen.
— 500 000 cycles between 62 and 68 kN which represents
60–66% of the nominal strength at 5 cycles/s.
— 50 cycles between 52 and 83 kN which represents
50–80% of the nominal strength at 1 cycle/s.
— A static tension test up to 95 kN with no failure then un-
loading the specimen.
— 1 750 000 cycles between 45 and 55 kN at 5 cycles/s.
— Failure occurs after cycling between 40 and 60 kN at
5 cycles/s. When the first crack appeared at about 5500
cycles, the dynamic loading was stopped and the speci-
men was loaded statically. The tendon carried up to
75 kN before failure.
5. A total of 2.42 × 106 cycles with the following sequential
load ranges and number of cycles:
— Three static loading and unloading cycles up to 40, 60,
and 90 kN at a rate of 1 kN/s.
— 721 000 cycles between 62 and 68 kN which represents
60–66% of the nominal strength at 5 cycles/s.
— 50 cycles between 52 and 83 kN which represents
50–80% of the nominal strength at 1 cycle/s.
— A static tension test up to 95 kN at a rate of 1 kN/s with
no failure then unloading the specimen.
— Failure occurred after 1 700 000 cycles between 45 and
55 kN at 10 cycles/s.
The previous preliminary cyclic loading tests performed on
the CFRP tendon–anchorage assembly showed that rate of
loading is less significant than the stress range. Four rates of
loading (1, 2, 5, and 10 cycles/s) were used and none of them
had a significant effect of the fatigue strength of the tendon.
On the other hand, the stress range appeared to have more
significant effect on the fatigue strength of the CFRP tendons.
With a narrow stress range, the maximum strength could have

© 1998 NRC Canada

124
Fig. 10. Finite element mesh used to model the anchorage system.
been increased up to 80 kN. As the stress range was increased
the maximum strength which corresponds to the fatigue
strength decreased. Despite the fact that more tests are required
to prove this conclusion, these preliminary results are in con-
trast with the specifications prepared by Tokyo Rope and Santoh
(1993) for the CFCC tendons, where the mean stress is con-
sidered as the governing (effective) factor against fatigue.
Analysis of test 5 results
In the cyclic loading test number 5 above, two sets of strain
gauges were mounted along the free length of the specimen to
collect data during the test: one on the smooth surface of the
Leadline specimen (strain gauge 1) and the other on the in-
dented surface (strain gauge 2).
During static loading, readings were taken at the rate of 1
reading/s, which corresponds to 1 reading/kN. During the cy-
clic loading the rate was 30 readings/s, which corresponds to
6 and 3 readings/cycle for the 5 and 10 cycles/s loading rates
respectively. Data were collected about every 200 000 cycles.
These data are shown in Figs. 9a, 9b, and 9c.
In Figs. 9a, 9b, and 9c, the deformation (stroke in millime-
tres) and the strain gauge readings are plotted against the load
corresponding to statically loading and unloading the speci-
men up to 40, 90, and 95 kN respectively. The last case
(95 kN) is the loading/unloading after performing the PTI cy-
cles. Readings from strain gauge 1, mounted on the smooth
surface of the Leadline, result in elastic moduli of 140, 148,
and 150 GPa. The first two were obtained before cyclic load-
ing with an average of 144 GPa and the third one was obtained
after 500 000 cycles. There may be a marginal effect of cyclic
loading on the stiffness of the tendon–anchorage assembly.
The corresponding results from strain gauge 2 (mounted on the
indented surface) are 158, 166, and 167 GPa which is 11–18%
higher than the value of elastic modulus of 142 GPa given by
the manufacturer (Mitsubishi 1996; Daniel and Ishai 1994).
Referring again to Fig. 9b, a permanent deformation of
2.5 mm remained after unloading the specimen; this is due to
seating of the wedges of each anchorage by about 1.25 mm
each. It is interesting to note that the seating started at a load
level of about 70 kN. The load used to seat the wedges before
testing was about 65–70 kN, which is why a higher load may
be required to induce further seating during testing. However,
Can. J. Civ. Eng. Vol. 25, 1998
in practice, the tendon will not usually be prestressed beyond
about 70% of its ultimate capacity (73 kN for the 8 mm Lead-
line used here). Thus, the amount of seating due to loading will
be negligible. Figures 9b and 9c reveal that no further seating
took place after the first loading beyond 70 kN.
Numerical analysis of the anchorage system using finite
element method
An axisymmetric finite element model of the anchorage was
used to assess the stress distribution along the CFRP cable
inside the anchorage zone. Eight-node isoparametric elements
were used for both the anchorage and the Leadline (Fig. 10).
Slide-line interface elements were placed between the Lead-
line and the wedges and between the outer sleeve and the
wedges with coefficients of friction of 0.5 and 0.05 respec-
tively as shown in Fig. 10. The model involved a total of 1124
elements. The CFRP cable is modelled as an orthotropic mate-
rial with the properties shown in Table 1 for Leadline. Two
types of analyses were performed. In the first, the steel anchor-
age was modelled as a linear elastic material with Young
modulus, E, and Poisson’s ratio, ν, of 200 GPa and 0.3 respec-
tively. The second model included the plastic behaviour of the
stainless steel anchorage. In this analysis, initial yield of the
stainless steel occurred at 862 MPa and the yield plateau at
1000 MPa. In both cases, geometric nonlinearity was included
in the analyses. When considering elastic properties for the
anchorage material, no failure criterion was specified for the
CFRP. However, in the second analysis, a failure stress of
2250 MPa was defined as the failure stress of the Leadline. The
analyses were performed in three load steps. The first was
seating of the anchorage by pushing the wedge in using applied
displacement along line a–a (Fig. 10) with all nodes on line
b–b restrained. In the second step, nodes on line a–a were
released; and in the third step, tensile force was applied to the
Leadline.
Distributions of radial, shear, and longitudinal stresses in
the CFRP cable along its length are shown in Figs. 11a, 11b,
and 11c respectively. Linear behaviour for the anchorage ma-
terial indicates that the radial stress in the Leadline (Fig. 11a)
at the back end of the anchorage is nearly 1.7 times its value
at the front end. However, using the more realistic constitutive
model shows that at the end of load step 3 (Fig. 11a), the radial
© 1998 NRC Canada
Sayed-Ahmed and Shrive
Fig. 11. Stress distribution in the Leadline inside the anchorage
resulting from the finite element analyses: (a) radial stress;
(b) shear stress; and (c) longitudinal stress. (NL, nonlinear analysis;
L, linear analysis.)
stress is nearly uniform along the Leadline throughout the an-
chorage length. It is interesting to note that during seating (load
125
Fig. 12. Radial stress distribution in the 4-piece wedge of the
anchorage resulting from the finite element analyses.
steps 1 and 2) the highest radial stress developed at the front
edge of the anchorage, with subsequent application of tension
to the Leadline (load step 3) relieving this stress concentration
without affecting the gripping action of the anchorage.
A relatively uniform distribution of shear stress along the
length of the Leadline in the anchorage zone is evident from
Fig. 11b. The figure shows that there is a concentration of
shear stress at the front end of the anchorage during seating
(load steps 1 and 2), which is again relieved at the end of the
loading step (step 3). The shear stress distribution resulting
from the analysis including plasticity of the anchorage steel is
more uniform than that from the purely elastic analysis. At the
end of the loading step (Fig. 11b), the shear stress ranges be-
tween –75 and +40 MPa.
The distribution of the third component of stress (longitu-
dinal stress) in the Leadline through the anchorage length is
shown in Fig. 11c. The analyses reveal that the force in the
Leadline is transferred almost linearly through the 80 mm
length of the anchorage. Finally, the radial stress at the end of
load step 3 along the length of the 4-piece wedge is shown in
Fig. 12. It can be seen from the figure that the compressive
stress is also nearly uniform along the length of the wedges.
Despite the fact that the model omits the effect of the soft-
metal overlay between the wedges and the tendon, the analyses
suggest that the anchorage design avoids the introduction of
stress concentration to the tendon. This was an underlying ob-
jective in the design of the anchorage in order to develop the
maximum strength of the tendon rather than induce failure
within or close to the anchorage.
Conclusions
A new anchorage system for CFRPs has been developed,
patented, and tested to yield the required efficiency for
post-tensioning applications, and to provide fatigue resis-
tance meeting the PTI requirements (adopted from steel
strands). Numerical analyses using the finite element
method provide an assessment of the stress distribution
within the anchorage system.
© 1998 NRC Canada
126
Acknowledgments
This work is supported by the ISIS (Intelligent Sensing in In-
novative Structures) Network Centre of Excellence. This sup-
port is gratefully acknowledged. The authors acknowledge the
help and support during the design and testing stages of the
anchorage system offered by Dr. Tadros, Mr. Damson, and
Mr. Tilleman, who share with the authors the anchorage patent
application. The great help of Mr. Blakeney and Mr. McCul-
lough during testing of the anchorage is also acknowledged.
References
Ballinger, C.A. 1991. Development of composites for civil engineer-
ing. In Advanced composite materials in civil engineering struc-
tures. Edited by S.L. Iyer and R. Sen. American Society of Civil
Engineers, New York. pp. 288–301.
Bradshaw, R.E., Drinkwater, J.P., and Bell, S.E. 1982. A multipur-
pose farm building incorporating diaphragm walling. Proceedings
of British Ceramic Society, Vol. 30, pp. 308–315.
Benmokrane, B., Xu, H., and Nishizaki, I. 1997. Aramid and carbon
fibre-reinforced plastic prestressed ground anchors and their field
applications. Canadian Journal of Civil Engineering,
24(6): 968–985.
Burgoyne, C.J. 1990. Properties of polyaramid ropes and implications
for their use as external prestressing tendons. In External
prestressing in bridges. Edited by A. Naaman and J. Breen. Ameri-
can Concrete Institute, Detroit, Mich. pp. 107–124.
Burgoyne, C.J. 1993. Parafil ropes for prestressing applications. In
Fibre-reinforced plastic (FRP) reinforcement for concrete struc-
tures: properties and applications. Edited by A. Nanna. Elsevier
Science Publisher, New York. pp. 333–351.
Curtin, W.G., Shaw, G., Beck, J.K., and Bray, W.A. 1982. Modern
philosophy of structural brickwork design and a change of outlook
for the brick industry. Proceedings of the 6th International Brick
Masonry Conference, Associazione Nazionale degli Industriali
dei Laterizi (ANDIL), Rome, Italy, pp. 939–948.
Daniel, I.M., and Ishai, O. 1994. Engineering mechanics of composite
materials. Oxford University Press Inc., Oxford, United Kingdom.
Dolan, C.W. 1991. Kevlar reinforced prestressing for bridge decks.
Transportation Research Record 1290, Transportation Research
Board, Washington, D.C.
Elliott, J., and Morrison, D.W. 1995. Acoustic system monitors and
predicts tendon failure. Concrete Repair Bulletin, ICRI, pp. 8–10.
Enka, B. 1986. Arapree — the Twaron resin prestressing element for
concrete. Corporate Report, Arnhem, The Netherlands.
Foster, D. 1970. Design and construction of a prestressed brickwork
and water tank. Proceedings of the 2nd International Brick/Block
Masonry Conference, Stroke-on-trent., United Kingdom,
pp. 287–294.
Ganz, H.R., and Shaw, G. 1997. Stressing masonry’s future. Civil
Engineering, ASCE, 67(1): 42–45.
Grant, L., Tadros, G., and Rizkalla, S. 1995. Towards development of
bridges in the next century. In Non-metallic (FRP) reinforcement
for concrete structures. Edited by L. Taerwe. E & FN Spon, Lon-
don, United Kingdom. pp. 645–662.
Harada, T., Matsuda, H., Khin, M., Tokumitsu, S., Enomoto, T., and
Idemitsu, T. 1995. Development of non-metallic anchorage de-
vices for FRP tendons. In Non-metallic (FRP) reinforcement for
concrete structures. Edited by L. Taerwe. E & FN Spon, London,
United Kingdom. pp. 41–48.
Hercules Aerospace Co. 1995. Task 6: cable and anchorage technol-
ogy, state-of-the-art report. Task Report, Technical meeting at
University of California, San Diego, Calif.
Holte, L.E., Dolan, C.W., and Schmidt, R.J. 1993a. Anchorage of
Can. J. Civ. Eng. Vol. 25, 1998
non-metallic prestressing tendons. University of Wyoming,
Laramie, Wyo., Technical Report UWYO-CE93.1.
Holte, L.E., Dolan, C.W., and Schmidt, R.J. 1993b. Epoxy socked
actors for non-metallic prestressed Tendons. Fibre-reinforced-
plastic Reinforcement for Concrete Structures, International Sym-
posium, SP-138, American Concrete Institute, Detroit, Mich.,
pp. 381–400.
Khin, M., Harada, T., Tokumitsu, S., and Idemitsu, T. 1996. The
anchorage mechanism for FRP tendons using highly expansive
material for anchoring. In Advanced composite materials in
bridges and structures. Edited by M. El-Badry. Canadian Society
for Civil Engineering, Montreal, Que. pp. 959–964.
Mathys, S., Blaere, B., and Taerwe, L. 1995. Relaxation of Aramid
fibre reinforced plastic AFRP. Internal Laboratory Report, De-
partment of Structural Engineering, Ghent University, Ghent, Bel-
gium.
Meier, U. 1995 Extending the life of cables by the use of carbon
fibres. Proceedings of the IABSE Symposium, San Francisco,
Calif., pp. 1235–1240.
Mitsubishi Chemical Corporation. 1996. CFRP Leadline. Corporate
Report, Mitsubishi Chemical Corporation, Tokyo, Japan.
Nanni, A., Matsubara, S., and Hasuo, K. 1992. Durability of braided
epoxy impregnated Aramid FRP rods. In Advanced composite
materials in bridges and structures. Edited by K.W. Neale and
Labossiere. Canadian Society for Civil Engineering, Montreal,
Que. pp. 101–109.
Nanni, A., Bakis, C.E., O’Neil, E.F., and Dixon, T.O. 1996. Perform-
ance of FRP tendon–anchor systems for prestressed concrete
structures. PCI Journal, 41(1): 34–43.
Rao, S.S. 1992. Reliability-based design. McGraw-Hill, New York.
Rostasy, F.S., and Budelmann, H. 1993. Principles of design of FRP
tendons and anchorages for post-tensioned concrete. Fibre-rein-
forced-plastic Reinforcement for Concrete Structures, Interna-
tional Symposium, SP-138, American Concrete Institute, Detroit,
Mich., pp. 633–649.
Post-tension Institute. 1985. Post-tensioning manual. 4th ed. Phoenix,
Ariz.
Santoh, N. 1993. CFCC (carbon fibre composite cables). In Fibre
reinforced plastic (FRP) reinforcement for concrete structures:
properties and applications. Edited by A. Nanni. Elsevier Science
publisher, New York. pp. 223–247.
Santoh, N., Kimura, H., Enomoto, T., Kiuchi, T., and Kuzuba, Y.
1993. Report on the use of CFCC in prestressed concrete bridge in
Japan. Fibre-reinforced-plastic Reinforcement for Concrete Struc-
tures, International Symposium, SP-138, American Concrete In-
stitute, Detroit, Mich., pp. 895–911.
Sen, R., Issa, M., and Mariscale, D. 1992. Feasibility of fibreglass
pretensioned piles in marine environment. University of South
Florida, Fla., Research Report CEM/ST/92/1,
Shaw, G. 1996. Innovative development of prestressed masonry. In
Worldwide advances in structural concrete and masonry. Proceed-
ings of the Committee on Concrete and Masonry Structures Sym-
posium held in conjunction with Structures Congress XIV. Edited
by A.E. Schultz and S.L. McCabe. American Society of Civil
Engineers, New York. pp. 13–24.
Shaw, G., and Baldwin, R.J. 1995. The construction of end built
prestressed masonry flat arch box girder footbridges. Masonry
International, 9(1): 1–5.
Shaw, G., Southcombe, C., and Easterbrook, D. 1995. Design, con-
struction and instrumentation of prestressed masonry box girder
footbridges. In Non-metallic FRP reinforcement for concrete
structures. Edited by L. Taerwe. E & FN Spon, London, United
Kingdom. pp. 673–679.
Shrive, N.G., Sayed-Ahmed, E.Y., Damson, E., Tilleman, D., and
Tadros, G. 1996. Prestressing anchorage system for fibre rein-
forced plastic prestressing tendons. U.S. Patent Application
08/754186.
© 1998 NRC Canada
Sayed-Ahmed and Shrive 127

Sippel, T.M. 1992. Design, testing and modelling of an anchorage List of symbols
system for resin bonded fibreglass rods used as prestressing ten-
dons. In Advanced composite materials in bridges and structures. Asp cross-sectional area of one tendon
Edited by K.W. Neale and Labossiere. Canadian Society for Civil ftu ultimate tensile strength of the tendon material
Engineering, Montreal, Que. pp. 363–372.
(Ftu)meas failure force of the tendon–anchorage assembly
Tokyo Rope Manufacturing Company, Ltd. 1990. Carbon fibre com-
posite cable. Corporate Report, Tokyo Rope Manufacturing Com- obtained for a direct tension test
pany, Ltd., Tokyo, Japan. (Ftu)theo granted failure force of the tendon based on its
Wolff, R., and Miesser, H.J. 1989. New material for prestressing and maximum strength
monitoring heavy structures. Concrete International, η anchorage efficiency
11(9): 86–89.

© 1998 NRC Canada