Professional Documents
Culture Documents
An experimental program was conducted to examine the bond characteris- This paper describes an experimental program, conducted
tics af carbon fiber reinforced plastic prestressing strands in pretensioned at the University of Manitoba, to examine the bond charac-
concrete beams. The bond characteristics are examined through measure- teristics of carbon fiber reinforced plastic prestressing
ments of the transfer and development lengths as well as their correspond-
strands in pretensioned concrete beams. The bond character-
ing bond stresses for 12.5 mm diameter and 15.2 mm diameter seven-wire
carbon fiber composite cable (CFCC) strands. Twenty prestressed concrete istics are examined through measurements of the transfer
beallls pretensioned bv CFCC strands were constructed and tested using and development lengths and their corresponding bond
different shear span mlues. This paper proposes equations to predict trans- stresses for 12.5 mm diameter and 15.2 mm diameter seven-
fer and flexural bOlld lengths for CFCC strands. The proposed equations wire CFCC strands.
are in good correlation with the measured values. This information is
important for the design of concrete structures prestressed by CFCC and
prol'ides data for the development of design considerations and codes for RESEARCH SIGNIFICANCE
concrete beams pretellsioned by CFCC strallds. This paper investigates the various bond characteristics
of carbon fiber reinforced plastic strands used to preten-
Keywords: bond; concrete; fiber reinforced plastics: development length; sion concrete beams. The information is valuable for de-
prestressed concrete: stress transfer. signers using CFCC for prestressing of concrete structures
and for the development of design guidelines currently un-
INTRODUCTION dertaken by ACI Committee 440 for concrete structures
In reinforced concrete structures, corrosion of reinforce- prestressed by FRP reinforcements. This paper examines
ment has been identified to be one of the most severe problems the results of bond tests and proposes equations that pre-
causing deterioration. This has been addressed quite often and dict the transfer and development lengths of CFCC
many solutions have been presented. Some of these solutions strands.
are the use of coatings for the reinforcement, such as epoxy
coatings, other plastic coatings. or cathodic protection. In re- MATERIAL PROPERTIES
cent years structures reinforced by coated or cathodic protect- The FRP reinforcement used in this experimental pro-
ed bars have been deteriorating, leading to the belief that these gram is carbon fiber composite cables (CFCC) produced by
solutions are not as effective as originally claimed. Therefore, a manufacturer in Japan. The individual wires of CFCC are
many engineers are turning to the use of advanced composite manufactured by a roving prepreg process using polyacry-
materials, or more specifically, fiber reinforced plastics (FRP) lonite (PAN) type carbon fibers. The prepreg is twisted to
to replace steel reinforcement. l-5 The materials are relatively create a fiber core and then wrapped by synthetic yarns.
new technology; therefore, much research efforts are needed The purpose of the yarn is to protect the fibers from ultra-
in order for designers to gain confidence in this material. violet radiation, mechanical abrasion, and also to improve
Performance of concrete members depends primarily on the bond properties of the wire to concrete. Cables are
the bond between the reinforcement and the surrounding made from one, seven, nineteen, or thirty-seven wires that
concrete. In pre tensioned concrete structures, the stress are twisted to allow better stress distribution through the
transfer is dependent on the bond between the concrete and cross-section. 6
the reinforcement. Bond strength is affected by many factors
such as configuration of the reinforcements, surface condi-
tions, size of the reinforcements, and variations in the load- ACI Structural Journal. V. 95, No.3. May-June 1998.
ing conditions. Many studies, both analytical and experi- Received May 7, 1996, and reviewed under Institute publication policies. Copyright
© 1998, American Concrete Institute. All rights reserved. including the making of
mental, have been done to examine the effect of these factors copies unless permission is obtained from the copyright proprietors. Pertinent discus-
sion will be published in the March-April 1999 ACI Structural Journal if received by
on the bond characteristics. November I. 1998.
281
Nolan Domenico is a structural engineer with Reid Crowther & Panners Limited,
Edmonton. Albena. Canada. He received his BS and MSfrom the University of Mani·
toba. Winnipeg. Manitoba. Canada. His research interests include the use of advanced
composite materials in strustural applications.
TypeC TypeD
mal coefficient of expansion in the direction of the fibers is
approximately 0.6x 1O.6/C (0.33x 106/F), which is about 1/20
Group 2
that of steel. The relaxation is about 3.5 percent after 30 Fig. i-Beam cross sections
years at 80 percent of the ultimate load, which is about
50 percent less than that of steel. 6 while the beams in the second group were cast on their sil
Concrete used in the beam specimens was provided by a to allow the demec points to be attached to the bottom of tl
local ready mix supplier. The mix proportions and the aver- beams. In Beams BT7 through BT10 in Group 2 the embe
age concrete strength at time of release and at time of testing ment length was varied by changing the debonded length
are given in Table 1. the mid-span of the beam as shown in Fig. 2. The avera~
concrete strength used in Group 2 was lower than that USt
EXPERIMENTAL PROGRAM for Group I as given in Table l.
A total of 20 pretensioned concrete beams were tested. The T-shaped beams used as Group 1 specimens we
The first group consisted of ten T -shaped pretensioned con- 3200 mm (126 in.) long with a 330 mm (13 in.) depth and 2f
crete beams. The second group included ten rectangular- mm (10.2 in.) effective depth. The beams are tested quasi-st,!
shaped beams as shown in Fig. 1. Detailed descriptions of ically as simply supported with a span of 2800 mm (110 ic
the 20 beams are given in Table 2. The embedment length of and 200 mm (8 in.) overhanging the supports. The prestressic
the first group and BTl through BT6 of the second group tendons were debonded over the 200 rum overhang at ear
was varied by changing the shear span. The demec points end in order to eliminate the bond length outside the span ar
were attached to the sides of the beams for the first group confinement caused by the support reaction forces as shown i
?R?
Group 1 Type A&B
2 br st 6 mm diam.@ 100 mm
/'
t CFCC Strand/ t
Embedment Length Embedment Length
·1·50·1
Group 2 Type D
t CFCC Strand/ t
Embedment Length +-
-+Debonded Length Embedment Length
283
~ 1111===;;;===11
=
=
ce
...c:
C,.) 1I1I1==~~=l1
gages. Strain of the concrete at the location of the strands passing through the average constant strain zone. It should
was measured after release by demec point gages. The mea- be noted that strain data in the transfer zone obtained from
mred transfer lengths, based on strain gages and demec point demec point gages are scattered due to the small magnitude
readings are summarized in Table 3. of the measured concrete strain. Concrete strength at release
The transfer lengths were determined using strain gage was determined using standard concrete cylinder.
lnd demec point readings as shown in Fig. 7 and 8, respec- The program provided a total of 36 measured values of
tively, for typical data. In these figures, the transfer length transfer length for the two CFCC sizes considered in this study
was determined as the length from the support to the point of using different concrete covers, concrete strengths and pre-
intersection of the varying strain zone and the horizontal line stressing levels. Literature review indicates that the transfer
<\CI Structural Journal/May-June 1998 285
_100
C
J
~
80
j 60
=
Strand Diameter 12.5 mm
S =
Prestress Level 50 % otfgu
... 40 =
Concrete Strength 51 MPa
~c: =
Concrete Cover 50 mm
Transfer Length = 125 mm
~
100 200 300 400 500 600 70C
Distance From Transfer Onset (mm)
Fig. 6-Testing setup Fig. 7-Transfer length measured by electric strain gages
length of a prestressing strand is greatly influenced by the the transfer length, and the development length. Test result~
Hoyer effect. 7-8 The Hoyer effect is caused by swelling of the and modes of failure for all beams are given in Table 4.
strand in the transfer zone after release as a result of Poisson's Using variable shear span length specimens, the develop-
ratio. Since the swelling is resisted by the surrounding con- ment length was determined as the length range at which the
crete, the induced confining stresses normal to the strand en- failure mode changed from bond slippage failure to rupture
hances the bond strength at the interface. Hence the bond of the cables. This is shown in Fig. 11 and 12 for Series A
enhancement caused by the Hoyer effect is directly related to and B specimens of Group 1. The test results are compared
the concrete strength and the coefficient of friction between to the ACI prediction for the development length of an
the two materials. Since the modules of elasticity of CFCC is equivalent beam prestressed by steel strands. 12 It should be
about 70 percent of that for steel strands, the longitudinal noted that the slope of the curves within the flexural bond
strain and consequent lateral strain is larger than that for steel zone represents the flexural bond stresses. The slope increas-
strands with the same prestressing stress and Poison's Ratio. es with decreasing the shear span and consequently the fail-
This behavior increases the induced stresses perpendicular to ure mode changes from rupture to bond slip. The limiting
the fibers and results in shorter transfer lengths for CFCC in value of the slope represents the flexural bond strength of the
comparison to steel strands. The wrap coating of the CFCC strand.
also enhances the frictional properties and therefore reduces
Specimens BT7, BT8, BT9, and BTlO in Group 2 were
the transfer length of CFCC.
tested using one concentrated load at mid-span. The devel-
Based on the measured data, the transfer length LT can be
opment length for these beams was measured by varying the
predicted in terms of the diameter of the strand d, the initial
debonded length of the strands at mid-span of the beam, as
prestressing levelfpi, and the concrete strength at transfer fe,.'
shown in Fig. 2. Test results are given in Table 4 according
as proposed in Eq. 0):
to the main parameters used in this program.
Using the same parameters used before for the transfer
LT -~
- (1) length, the flexural bond length Lfb is related to the increase
4.8fc;0.67 of stresses from the prestress level fpe to the ultimate tensile
strengthfpu of CFCC as follows:
The proposed equation has a standard deviation of 12 per-
cent in comparison to the measured values as shown in Fig. 9. = (fpu - fpe)d (2)
The proposed equation is also compared to the transfer length 2.8f;0.67
equations for steel strands proposed by Zia,9 Balazs,1O Mitch-
ell,1I and the ACI Code 12 in Fig. 10. It can be seen that the
proposed equation forCFCC strands lies well below all of the The proposed equation has a standard deviation of 12 per-
other equations for steel strands which predict a transfer cent in comparison to the measured values. In Fig. 13, the
length approximately twice the value for CFCC. flexural bond length for CFCC based on Eq. (2) is compared
to flexural bond length equations for steel strands proposed
Flexural bond length by Mitchell, I I Zia,9 and the ACI Code. 12 The figure shows
The flexural bond length is defined as the embedment that the proposed equation for CFCC strand is in the order of
length beyond the transfer length required to develop the full 25 to 40 percent of that of steel strands.
tensile strength of the prestressing strand. Summation of the
transfer length and the flexural bond length is known as the de- Development length
velopment length. Therefore, to determine the flexural bond As previously stated the development length of a prestressing
length of a prestressing strand it is required to determine both strand is the summation of the transfer length and the flexural
0.2,-----------------------, 80i.=====~--------------~~--1
!t'ci 30 MPa= I
70
o o 0
o o
c
~ 0.1
Strand Diameter 12.5 mm =
ia Prestress Level = 50 % of fgu
Concrete S1rength 41 MPa
Concrete Cover 50 mm =
=
Proposed for CFCC
u
Transfer Length 175 mm =
0~__-4--~-+--+--r~--+--+--~---4
100 200 300 400 500 1000 1200 1400 1600 1800
Distance From Transfer Onset (mm) 600 800
Prestressing streSS (MPa)
Fig. 8-Typical transfer length measured by demec gages Fig. lO-Transfer length comparison
proposed model c
iii
!.
2500
2000
BeamA3
Bond Failure
,
3000r----------------------------------,
BeamA2
Rupture of CFCC
~
~ 1500
'C
c
f! 1000
iii
12 % standard deviation 500 / ' Development Length =707 mm
o 250 500 750 1000 1250 1500 1750 2000 2250 o 500 1000 1500 2000 2500 300e
Distance From End of Beam (mm)
fpl d I f~io.67
Fig. 9-Transfer length correlation of CFCC Fig. II-Failure of Series A beams in Group I (15.2 mm
strand)
bond length. Therefore, based on the previous discussion, the up to a mid-span deflection of 15 mm (0.6 in.). This step-wise
proposed development length of a CFCC prestressing strand is: behavior was due mainly to the displacement control loading
system used for testing of this beam. After stabilization of the
crack pattern, the beam continued to carry load up to a peak
(3) value at which the load resistance dropped suddenly by about
10 to 15 kN. The beam continued to resist more load at a rela-
tively reduced stiffness. This behavior suggests that at this
Flexural behavior peak load level the strand became completely debonded from
All beams exhibited a bilinear moment-deflection rela- the concrete within the constant moment zone and acted main-
tionship. Beams with sufficient embedment length failed by ly as a tie causing a reduction of the stiffness shown by the
rupture of the CFCC prestressing strand within the constant
drop in load resistance under displacement control loading
moment zone or at the location of the load points as shown
condition. However, since the bond strength in the transfer
in Fig. 14. Beams with embedment length less than the de-
zone was sufficient, the beam was able to carry additional load
velopment length failed by bond slip of the CFCC strand at
and acted as an unbonded post-tensioned member up to failure
the location of the load points. Three beams failed prema-
with the transfer zones anchoring the strand. Beam A3 even-
turely by a combination of shear and bond slip between the
tually failed by bond slippage and Beams B6 and B7 failed by
load point and the support. This type of failure occurred
when a large diagonal shear crack suddenly formed immedi- strand rupture.
ately followed by a sudden slippage of the strand. All beams
displayed extensive flexural cracking extending up to the top Transfer bond stresses
flange prior to failure with some beams exhibiting some It is known that the bond stresses within the transfer zone
flexural-shear cracking as shown in Fig. 14. between the prestressing strand and the surrounding concrete
Beams A3, B6, and B7 exhibited an interesting moment- vary along the transfer length. 7 However, for the purpose of
deflection relationship as shown in Fig. 15. The figure shows this analysis the bond stress along the CFCC strand within
that the beam behaved linearly up to the cracking load then be- the transfer zone is expressed as an average value. The aver-
haved in a step-wise fashion due to formation of new cracks age bond stress was calculated for all the tested beams based
')Q7
3500.----------------------------------;
Beam B8 Beam B7
\
3000 Bond Failure Rupture of CFCC
'l25oo
./
~
I
I
: Development Length = 520 mm
:/
500 1000 1500 2000 2500
Distance From End of Beam (mm)
250
... If'c = 40 MPa I Peak due to bond slip within
~ constant moment zone "
E 200
01
i5 Zia (19
~150
c
!l
~ 100
0
CD
Proposed for
a=700mm
=0.25
all
• t
E Mer =22.1 kNm
~
:> 50
Mu =50.2 kNm
)C
III CFCC _ _- - - - - -
u::
0
0 200 400 600 800 1000 1200 1400 o 10 20 30 40 50
Stress Change, (fpo -~) (MPa) Mid-span Deflection (mm)
Fig. 13-Flexural bond length comparison Fig. is-Moment-deflection relationship for Beam B6
on the measured transfer lengths, the initial prestressing average flexural bond stresses were found to be in the range
force, and the size of the CFCC strands. The average transfer of 3.5 to 6.3 MPa (507 to 914 psi) with an average value of
bond strength could be expressed in terms of a bond stress in- 5.0 MPa (725 psi). Using the expression suggested by the
dex Vf the compressive strength of the concrete at transfer, ACI Code l2 for the flexural bond length of steel strands, the
fe;' as follows: flexural bond strength for steel was calculated to be in the
range of 1.34 MPa (194 psi). Therefore, it can be seen that
V T -- U' 4' ,0.67 the average bond stress in the flexural bond zone for CFeC
T Jei (4)
is about four to five times greater than that of steel strands.
The flexural bond stress Vfo can also be expressed as a
For CFCC the average value of the bond strength index for flexural bond stress index Vfo' as given in Eq. (5) to account
the transfer zone was found to be 0.75 with a standard devi- for the variability of the concrete strengthfc'.
ation of 12.7 percent. Using the expression suggested by the
ACI Code 12 for the transfer length of steel strands, the aver-
age bond strength for steel was calculated and was found to vII> =
V'
Ib
1,0.67
c (5)
be in the range of 4.0 MPa (580 psi). According to Eq. (4),
the bond strength for CFCC given the same parameters and For CFCC the flexural bond stress index, Vfo' ranged from
concrete strength at transfer of 30 MPa is 7.3 MPa (l 058 psi) 0.40 to 0.45.
which is about twice that for steel strands.
Parametric analysis
Flexural bond stresses Proposed Eq. (1) and (2) show that both the transfer length
For the purpose of this analysis the bond stress along the and flexural bond length are proportional to the diameter of
CFCC strand in the flexural bond zone is expressed as an av- the prestressing reinforcement. The transfer bond index and
erage value. The average bond stress was calculated for all flexural bond index, given in Fig. 16 in terms of the nominal
beams based on the flexural bond lengths from testing, the diameter of CFCC, suggests that both the average transfer
effective prestressing force, the ultimate strength of the bond index and the average flexural bond index were not af-
strand, the concrete strength, and the size of the strand. The fected by a change in the strand diameter.
2. The average transfer bond strength and flexural bond
strength is not affected by changes is the strand diameter.
Q 3. The transfer length is proportional to the prestress level while
0.8
~
'CI
C
I
Transfer Bond
the transfer bond strength was not affected by the prestress level.
4. The amount of concrete cover had no bearing on the
;; 0.6 Flexural Bond transfer length nor on transfer bond strength. However, the
c
~ I range of concrete cover examined in the experimental pro-
t
+
fO. 4
a m
gram was 3.3 to 6 strand diameters, therefore, these findings
are only applicable for this range of concrete covers or greater.
~ 5. Flexural behavior of the beams was bilinear up to fail-
0.2
ure. Some beams exhibited a spike in the moment-deflection
relationship followed by a reduced stiffness up to failure.
11 12 13 14 15 16 This phenomenon could be attributed to debonding of the
Strand Diameter (mm) strand in the constant moment zone.