Anda di halaman 1dari 12

Composites: Part B 33 (2002) 493–504

www.elsevier.com/locate/compositesb

Development length of FRP straight rebars


E. Cosenza, G. Manfredi, R. Realfonzo*
Department of Structural Analysis and Design, University of Naples Federico II, Via Claudio 21, 80125 Naples, Italy
Received 25 June 2001; revised 8 July 2002; accepted 13 July 2002

Abstract
In recent years, some attempts have been performed to extend general design rules reported in the codes for steel reinforced concrete to
Fiber Reinforced Polymer (FRP) materials; this is the case of relationships adopted in the evaluation of the development length clearly
derived by extension of the formulations used for steel rebars. However, such relationships seem to be inappropriate for FRP reinforcing bars:
in fact, experimental test results have shown that bond behaviour of FRP bars is different from that observed in case of deformed steel ones.
As a consequence, a new procedure for the evaluation of development length based on an analytical approach is needed in order to directly
account for the actual bond-slip constitutive law as obtained by experimental tests on different types of FRP reinforcing bars.
An analytical solution of the problem of a FRP rebar embedded in a concrete block and pulled-out by means of a tensile force applied on
the free end is presented herein. Such solution leads to an exact evaluation of the development length when splitting failure is prevented.
Finally, based on the analytical approach, a limit state design procedure is suggested to evaluate the development length. q 2002 Elsevier
Science Ltd. All rights reserved.
Keywords: FRP reinforcing bar

1. Introduction properties of bars and the compressive strength of concrete.


In particular, the interaction phenomena are governed by
From design point of view, the study of concrete shear strength and deformability of ribs, which are
structures reinforced using FRP reinforcing bars has been remarkably lower than those of steel bars; this leads to
initially developed by extending the wide body of increased slips between rebars and concrete and different
information gathered in a century of use of steel reinforced failure mechanisms [13,15,16]. Mechanical properties of
concrete. Studies have been often carried out by comparing resin, which the matrix is made of, have a remarkable
performances obtained by using steel or FRP reinforcing influence on the interaction behaviour since they strongly
bars; moreover, the manufacturing technologies have been affect strength and deformability of ribs and indentations
oriented to fabricate composite bars which are similar, in located on the outer surface.
shape and dimensions, to those made of deformed steel. Therefore, by comparing FRP and steel rebars, it has been
One of the critical aspects of structural behaviour is the noted that the differences in bond behaviour are due to some
development of an adequate bond behaviour; a number of properties of FRP reinforcing bars; this underlines the
tests have been performed by several authors on FRP inadequacy of extending the design rules for steel reinforced
reinforcing bars in order to study their bond performance concrete to FRP reinforcing bars. Therefore, a critical review
and to compare such bond properties with those evidenced of the design methodology is needed in order to introduce this
by deformed steel bars. On this topic, three state-of-art new type of bars as reinforcement for concrete structures.
reports have been recently published by Cosenza et al. [8], Despite the above remarks, some attempts have been
Tepfers [21] and fib Task Group 5.2 [12]. performed in order to extend to such new materials, with
From the experimental results, it was concluded that some minor modifications, the general design rules reported
bond between FRP reinforcement and concrete is controlled in the codes traditionally used for steel reinforced concrete;
by several factors such as the mechanical and geometrical this is the case of the relationships adopted in the evaluation
* Corresponding author. Tel.: þ 39-81-7683485; fax: þ 39-81-7683406.
of the embedment length clearly derived by extension of the
E-mail address: robrealf@unina.it (R. Realfonzo). formulation used for steel bars [3,14].

1359-8368/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 5 9 - 8 3 6 8 ( 0 2 ) 0 0 0 5 1 - 3
494 E. Cosenza et al. / Composites: Part B 33 (2002) 493–504

Nomenclature Ldb basic development length


s slip
Ab transverse section area of the bar sm slip at peak bond strength
cb bar perimeter wlim allowable crack width
f bar diameter a,p parameters of the Eligehausen et al., modified
E elastic modulus of the FRP rebar bond-slip law (Eqs. (16) and (17))
Ef elastic modulus of steel rebars 1 tensile strain
0
f ck characteristic compressive strength of concrete gE safety factor that affects the elastic modulus
fb0d design bond strength gg global safety factor
fc compressive concrete strength gm material safety factor
f c0 compressive concrete strength s tensile stress
fd design tensile strength of the FRP rebar su tensile strength of the FRP bar
ff tensile stress of the FRP rebar suk characteristic value of tensile strength of the
ft tensile concrete strength FRP rebar
fu tensile strength of the bar syk characteristic yielding strength of steel rebars
fyf yielding strength of steel rebar in tension t bond stress
Ld development length tm maximum bond strength

This paper analyses the problem of evaluating the embedded in a concrete block for a length Ld and subjected
development length of deformed FRP reinforcing bars and to a tensile force T is governed by the following equilibrium
presents a new approach based on an analytical formulation. equation:
Such a procedure is based on the analytical study of the T ¼ Ld ptf ð1Þ
problem of a rebar embedded in a concrete block and
pulled-out by means of a tensile force applied at one end. In If Ab is the rebar area and ff the tensile stress, the tensile
order to integrate the differential equation that governs such force T can be written as:
problem, the definition of a suitable bond-slip constitutive T ¼ Ab ff ð2Þ
relationship is needed.
Recently, Pecce et al. [18] proposed a numerical procedure From Eqs. (1) and (2), it follows:
for evaluating the constitutive bond-slip relationship. Based Ab ff
on the experimental bond test results, such procedure is able Ld ¼ ð3Þ
ptf
to identify values of the parameters which define the
modified version [8] of the Eligehausen, Popov and Bertero or, alternatively:
relationship [10]. In this way, a suitable constitutive t – s law fff
Ld ¼ ð4Þ
was proposed in the case of a recently introduced GFRP 4t
reinforcing bars (C-Bare by Marshall Inc.); however, the
For deformed steel bars, it has been found that bond strength
above-mentioned numerical procedure could allow to derive tm is a linear function of the square root of the compressive
a constitutive relationship for any type of bars. concrete strength f 0c [10]:
The presented analytical approach has been developed by pffiffiffi
considering the Eligehausen et al., modified law as bond- tm ¼ k f 0c ð5Þ
slip (t –s ) relationship; a numerical example is carried out
where k is a constant.
in case of C-Bare using the constitutive t –s law suggested
Therefore, from Eqs. (3) and (5):
by Pecce et al. [18].
Finally, based on the analytical approach, a design Ab ff
Ld ¼ pffiffiffi ð6Þ
procedure is proposed to evaluate the development length. pkf f 0c
The method, which represents a first proposal, takes into
and setting K ¼ pkf; it follows:
account the actual bond-slip constitutive laws of the different
FRP rebars and introduces some different safety factors. Ab ff
Ld ¼ pffiffiffi ð7Þ
K· f 0c
Eq. (7) represents the well known ‘basic development
2. Evaluation of the basic development length length’, generally indicated with Ldb; K depends on the
relationship between the bond strength and the compressive
Under the assumption of constant distribution of bond concrete strength and on the bar diameter.
stresses t, the problem of a diameter f reinforcing bar In case of #3 to #11 deformed steel reinforcing bars, the
E. Cosenza et al. / Composites: Part B 33 (2002) 493–504 495

ACI 318-89 [1] assumed a value of K equal to 25 thus reinforcing bars since they have been derived by adopting
leading to the following expression of Ldb: a linear relationship between bond strength and the square
root of the compressive concrete strength f 0c : Several
Ab fyf
Ldb ¼ 0:04 pffiffi0ffi ð8Þ investigators have shown that such a relationship does not
fc hold true for FRP reinforcing bars [6,13,15,16].
where fyf and f 0c are the yielding strength of steel bars and Therefore, the development of a new procedure to
compressive concrete strength, respectively (psi), and Ab, is evaluate the development length of FRP bars, based on an
the rebar area (in.2). analytical approach, is needed. It should take into account
According to ACI 318-89, the development length Ld is the actual bond-slip constitutive law, as obtained by
provided by: experimental tests.
Y 
Ld ¼ fi Ldb ð9Þ
Q 3. The problem of the rebar pull-out
where fi indicates the product of some modification
factors that take into account the influence on bond of some The problem of a rebar embedded in a concrete block and
key parameters (i.e. cover, spacing, transverse reinforce- pulled-out by means of an applied tensile force is analysed
ment). in the following.
Some modifications of Eq. (8) have been subsequently A closed form solution is obtained by adopting linear
reported by ACI 318-95 [2]. elastic constitutive laws for the materials and the Eligehau-
In order to extend Eq. (7) to FRP reinforcing bars, several sen et al., modified relationship [8,18] as bond-slip
investigators have attempted to evaluate experimentally constitutive law. Such an analytical solution allows to
values of K for different types of FRP bars [5,9,11,19]. obtain slip, normal stress and bond stress distributions along
Furthermore, based on experimental results, in case of the rebar (i.e. at a generic abscissa x ). Furthermore, such a
FRP reinforcing bars, some authors proposed simplified solution leads to an exact evaluation of the development
expressions of Ldb; these expressions, clearly design length.
oriented, are not suitable for all types of FRP reinforcing The studied case is schematically shown in Fig. 1, where
bars, because they are practically appropriate only for the the origin of the x-axis is in the free end of the bar.
selected bars. An example is given by [6,7]: The differential equation that governs the bond problem
Ld ¼ 20f ð10Þ [20] is obtained by considering:

Recently, formulations for evaluating the basic develop- † the equilibrium of rebar:
ment length have been proposed in new codes for design
of concrete structures reinforced with FRP bars. This is pf2
the case of the Japan Society of Civil Engineers (JSCE) ds ¼ pft dx ð13Þ
4
Design Code [14] and of the ACI Committee 440 Guide
[3]. † a linear elastic behaviour for the rebar that, if the
In the case of JSCE code, Ldb is clearly derived from contribution of concrete in tension is neglected, is given
Eq. (4): by:
ffd
Ldb ¼ a1 ð11Þ ds
4fb0d s ¼ E1 ø E ð14Þ
dx
where fd is the design tensile strength, fb0d, the design bond
strength and a1 is a coefficient less than 1. where E and f are the elastic modulus and the diameter
The Japanese code states that the basic development of the rebar, respectively.
length shall not be taken less than 20 times the bar diameter.
The ACI Committee 440, for failure controlled by From Eqs. (13) and (14), the following differential
pullout, proposes (in SI units): equation is obtained:
fffu
Ldb ¼ ð12Þ d2 s 4
18:5 2
2 tðxÞ ¼ 0 ð15Þ
dx Ef
where Ldb is in mm, ffu (ultimate design strength of FRP
reinforcing bars) in MPa and f is in mm. The Eligehausen et al. modified model [8,18]—shown in
In the ACI guide, two modification factors greater than 1 Fig. 2—is considered herein. Such a constitutive law is
are considered in order to prevent splitting of concrete and given by:
to evaluate the development length of top bars. (A) for s , sm ; an ascending branch which is formally
The above relationships seem to be unsuitable for FRP coincident with the first branch of the Eligehausen, Popov
496 E. Cosenza et al. / Composites: Part B 33 (2002) 493–504

tm, the maximum bond strength, sm, the slip at peak bond
strength and su is the ultimate slip.
By using such a law when integrating Eq. (15), two cases,
A (s # sm) and B (s . sm) have to be separately considered.

3.1. Case A (s # sm)

Considering Eq. (16), it is possible to rewrite Eq. (15) as:


2
ds 4t m a
2 s ¼0 ð18Þ
dx2 Efsam
By integrating Eq. (18) with the following boundary
conditions:
 
ds
sð0Þ ¼ 0; ¼ 1ð0Þ ¼ 0
dx x¼0
i.e. considering a perfect anchorage of the bar, the following
solution is obtained:
" #1=ð12aÞ
2tm ð1 2 aÞ2
sðxÞ ¼ x2=ð12aÞ ð19Þ
Efsam ð1 þ aÞ

that provides the trend of the slip s along the bar.


Trends of the bond stress t and of the tensile stress s
are derived by considering Eqs. (13) and (14), respect-
ively:
pf ds ds
tðxÞ ¼ ; sðxÞ ¼ E ð20Þ
4 dx dx
Using the above-presented relationships, two characteristic
Fig. 1. The studied cases. limit values can be derived from Eqs. (19) and (20):

and Bertero law [10]: † the limit tensile stress in the bar s1;
 a † the limit development length lm.
s
tðsÞ ¼ tm ð16Þ
sm The first value represents the stress in the bar
corresponding to a slip equal to sm:
(B) for sm , s , su a softening branch given by:
  sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s 8E tm sm
tðsÞ ¼ tm 1 þ p 2 p ð17Þ s1 ¼ sðsm Þ ¼ ð21Þ
sm f 1þa
where a is a coefficient which describes the ascending The value lm represents an upper bound of the development
branch, p, a coefficient which defines the softening branch, length related to the ascending branch of the bond-slip law,
i.e. the development length that corresponds to a stress
applied to the rebar equal to s1. In fact, setting s ¼ sm in Eq.
(19), the corresponding value of x represents lm:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Ef sm ð1 þ aÞ
lm ¼ ð22Þ
2 tm ð1 2 aÞ2

Furthermore, considering Eqs. (21) and (22), lm can be also


written as:
s1 f 1 þ a 1þa
lm ¼ ¼ l0m ð23Þ
4t m 1 2 a 12a
where l0m is the development length evaluated for s ¼ s1
Fig. 2. Eligehausen et al., modified t –s law. and t ¼ constant ¼ tm.
E. Cosenza et al. / Composites: Part B 33 (2002) 493–504 497

A reduction of the rebar elastic modulus E results in conditions:


an increment of slips s (Eq. (19)) and in a reduction of  
ds s1
the embedment length (Eq. (22)). Eq. (23) confirms that sðlm Þ ¼ sm ; ¼ 1ðlm Þ ¼
dx x¼lm E
lm is greater than l0m, since (1 þ a )/(1 2 a ) is greater
than 1. the function sðxÞ is obtained:
Eqs. (19), (20) and (22) lead to the following useful (
relationships: sðxÞ 1
 ps ¼ 1 þ p 2 cos½vðx 2 lm Þ
sðxÞ x 2 sm p
¼ where ps ¼ ð24Þ sffiffiffiffiffiffiffiffiffi 9
sm lm 12a =
 ps 2p
sðxÞ x 1þa þ sin½vðx 2 lm Þ ð31Þ
¼ where ps ¼ ¼ ps 2 1 ð25Þ 1þa ;
s1 lm 12a
 pt
tðxÞ x 2a where:
¼ where pt ¼ ¼ ps 2 2 ð26Þ
tm lm 12a sffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffi
4ptm 1 2p 1 þ a
Eqs. (24) – (26) provide simple expressions of s, s and t as a v¼ ¼
E f sm lm 1 þ a 1 2 a
function of x=lm : In particular, it can be noticed that:
Finally, by substituting Eq. (31) into Eq. (17), the
† for a ¼ 0, the slip s is a quadratic function of the abscissa distribution law of bond stresses along the bar is given by:
x, while the normal stress s is linear and the bond stress t sffiffiffiffiffiffiffiffiffi
assumes the constant value tm; tðxÞ 2p
¼ cos½vðx 2 lm Þ 2 sin½vðx 2 lm Þ ð32Þ
† for a ¼ 1/3, sðxÞ is cubic, s(x) is parabolic and t(x) is tm 1þa
linear.
while, remembering Eq. (14), sðxÞ is provided by:
For s , s1, the development length l can be evaluated sffiffiffiffiffiffiffiffiffi
from Eq. (25) by setting x ¼ l; then, the following sðxÞ 1þa
¼ cos½vðx 2 lm Þ þ sin½vðx 2 lm Þ ð33Þ
expression of l is obtained: s1 2p
 ð12aÞ=ð1þaÞ
s It is possible to demonstrate by Eq. (30) that the
l ¼ lm ð27Þ development length l can be obtained by:
s1
8 sffiffiffiffiffiffiffiffiffiffiffiffiffiffi" sffiffiffiffiffiffiffiffi#
For 0 , s # s1, Eq. (27) provides values of the length l less < ð1 2 aÞ2 A t2
than the value lm derived from Eq. (23). l ¼ lm 1 þ arcsin
: 2pð1 þ aÞ Atmax
According to Eq. (23), another expression of l can be
easily derived from Eq. (27): sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi9
 2a=ð1þaÞ fs2 =
s 1þa 2 arcsin 1 2 ð34Þ
l ¼ l0 1 ð28Þ 8EAtmax ;
s 12a
where: where lm is given by Eq. (22) or Eq. (23), while:
t s t s t s
fs At 2 ¼ m m and Atmax ¼ m m þ m m
l 0 ¼ l 0 ð sÞ ¼ ð29Þ 2p 1þa 2p
4t m
Eq. (34) provides the development length for values of
is the development length evaluated in case of t ¼
s $ s1; obviously, for s ¼ s1 (i.e. s ¼ sm ) Eq. (34)
constant ¼ tm.
provides l ¼ lm :
It has to be underlined that Eq. (29) can be obtained from
Neglecting the frictional contribution to the bond
Eq. (28) by setting a ¼ 0, i.e. by assuming a rigid-plastic
mechanism (Fig. 2), a theoretical limit threshold of tensile
bond-slip constitutive law.
stress s2 can be evaluated. It is easy to demonstrate that such
Finally, from Eq. (28), it can be seen that the value of the
threshold, which corresponds to a maximum value of slip (at
development length l is greater than l0.
the loaded end) equal to the ultimate one s0, can be obtained
by the following expression:
3.2. Case B (s . sm)
sffiffiffiffiffiffiffiffiffiffiffiffi
8E
Considering Eq. (17), the differential Eq. (18) becomes: s2 ¼ sðs0 Þ ¼ A ð35Þ
f tmax
2
ds 4ptm 4ð1 þ pÞtm
2
þ s¼ ð30Þ It can be noticed that such a relation is very similar to that
dx Efsm Ef
obtained for the threshold s1 (Eq. (21)).
and integrating Eq. (30) with the following boundary This means that in the previously stated hypotheses, a
498 E. Cosenza et al. / Composites: Part B 33 (2002) 493–504

value lu of the development length can be evaluated from † for fc , 30 MPa, pull-out failure was due to the breaking
Eq. (34) that corresponds to s ¼ s2; however, by increasing of the surrounding concrete, while the rebar was
such value of length, it is not possible to transfer a normal undamaged;
stress greater than s2. Obviously, this is true in the adopted † for fc . 55 – 60 MPa, failure was due to damage
hypothesis of neglecting the frictional contribution. concentrated on the ribs of the FRP reinforcing bar,
Therefore, it has to be pointed out that the evaluation of whereas damage of the surrounding concrete was
the development length requires a complete characterisation negligible;
of the bond-slip constitutive law and an estimate of the † for 30 , fc , 55 MPa damage both of the outer surface
elastic modulus of the bar. In particular, from Eqs. (22), (27) of the reinforcing bar and within the concrete was
and (34), it can be verified that the development length observed.
decreases if: the elastic modulus E of the bar decreases; the
rigidity of the bond-slip law increases (i.e. minor values of a The possibility that the interaction phenomena could be
and sm); the bond strength tm increases; the slope of the dependent on mechanical properties of the ribs results in a
softening branch of the bond-slip law decreases (i.e. minor substantial difference compared to the case of bond of steel
values of p ). reinforcement, where the weak element is always concrete.
Observing values shown in Table 1, the following bond-
slip relationship is suggested:
4. Analytical vs. experimental results
(a) for s # 0.25 mm:
The assessment of the t –s law by means of experimental
tb ðsÞ ¼ 20:5s1=4 ð36Þ
tests is not straightforward; generally, pull-out or beam tests
are performed to evaluate such a constitutive law and
tests are carried out on specimens characterized by short (b) for s . 0.25 mm:
embedment lengths (about five times the bar diameter or tb ðsÞ ¼ 16:5 2 7:4s ð37Þ
less). Furthermore, values of t are generally obtained by
assuming a constant distribution of bond stresses in where s is in mm and t is in MPa.
the embedded portion while slips are measured at the free
end. As above-mentioned, values of parameters shown in Eqs.
In the case of FRP reinforcing bars, values of slips (36) and (37) were identified for a #4 GFRP C-Bare from
measured at the two ends are very different and the bond tests without splitting failure. Anyway, an appropriate
hypothesis of a constant bond stress distribution appears set of coefficients can be also obtained in case of splitting
to be inadequate. In order to overcome this problem, some failure by applying the identification procedure proposed by
authors reduced the length of the embedded portion to very Pecce et al. [18] to bond tests characterised by this kind of
short values (2 – 3 times the bar diameter) [15,22]; however, failure; the same can be said in the case of change of the
it has to be observed that in this way the influence of local rebar type.
irregularities can be magnified and, consequently, test Therefore, the design procedure shown in the following,
results may have a large scatter. based on the Eligehausen et al., bond-slip law, can be seen
For this reason, in a recent paper, Pecce et al. [18] as a general method.
proposed an identification procedure aimed to estimate, However, from a design point of view, specific design
by experimental bond test results, values of parameters rules should be proposed in order to avoid splitting failure.
defining the Eligehausen et al., modified constitutive Furthermore, a mean value of the elastic modulus E equal
law. By using such a procedure, the authors calculated to about 42,000 MPa and an average value of tensile
the mean values and the coefficient of variation (c.o.v.) strength of 600 MPa were found by performing tensile tests
of the above parameters for a #4 GFRP C-Bare; such on #4 (bar diameter f equal to 12.7 mm) C-Bare [18].
values are shown in Table 1 (the c.o.v. are in It is worth noting that the value declared by Marshall
parenthesis). modulus E is equal to the obtained one, while the declared
It has to be underlined that test results assumed as tensile strength sud ¼ 770 MPaÞ is greater than the obtained
reference in Ref. [18] come from beam tests with a large one.
concrete cover, i.e. they are representative of cases where For the identified values, a threshold s1 equal to about
the splitting collapse is prevented. Furthermore, such tests
were conducted on specimens having a compressive Table 1
concrete strength fc ranging between 39 and 52 MPa; in Mean values and c.o.v. of the bond parameters
this case the variation of concrete strength should not
sm (mm) tm (Mpa) a p
largely influence the bond strength.
This last observation is confirmed by pull-out tests results
0.253 (0.136) 14.65 (0.083) 0.245 (0.465) 0.128 (0.449)
on C-Bare obtained by Karlsson [15]:
E. Cosenza et al. / Composites: Part B 33 (2002) 493–504 499

Fig. 3. Analytical vs. experimental results.

280 MPa has been evaluated from Eq. (21) and a along the rebar can increase too. This effect is strictly
corresponding development length lm equal to about dependent on the bond-slip constitutive law.
100 mm has been calculated from Eq. (23). Because of the hypothesis of a slip equal to zero at the
The stress s vs. the dimensionless development length free end, the above-presented analytical formulation does
L/f curve has been obtained by adopting values of Table 1 not allow to evaluate the maximum stress s in pull-out
and the above-mentioned elastic modulus. Such a curve, conditions; in order to calculate such stress, for a given
shown in Fig. 3, has been derived by computing embedment length, a numerical integration of Eq. (15) is
the development length for each value of the normal stress, needed. Fig. 3 also depicts the stress s vs. L/f curve
using Eq. (28) if s # s1, and Eq. (34) if s . s1 (closed obtained by means of the numerical integration; this
form solutions). In Fig. 3, results of bond tests performed by curve shows a better agreement to the experimental results
Karlsson [15], Benmokrane and Masmoudi [4] and Pecce and confirms the more conservative trend provided by the
et al [18] are also shown. close form solution (i.e. when slip is equal to zero at the free
Karlsson carried out an experimental program on #4 and end). Finally, examining Fig. 3, it can be observed that, in
#5 C-Barse by adopting three values of the embedment some cases, results from Pecce et al. show a bar tensile
length (3f, 5f and 7f ): a pull-out failure was observed in failure with values of the normal stress less than 600 MPa:
all the cases. this was probably due to the unavoidable coupling between
The same failure was observed by Benmokrane and tension and bending of the rebar in case of beam tests.
Masmoudi testing #5 C-Barse with an embedment length In Fig. 4, trends of normal stresses, bond stresses and
equal to five times the bar diameter. slips are reported in case of a #4 GFRP C-Bare in tension
Pecce et al. performed beam tests on #4 C-Barse with embedded in a concrete block for a length equal to 200 mm.
embedment length ranging from 5 to 30 times the bar Such trends have been obtained by using Eqs. (24) –(26) in
diameter: both pull-out and tensile failure were observed the embedded portion where s is less than the threshold
depending on the embedment length value. s1, and by Eqs. (31) – (33) where s is greater than the
Fig. 3 shows a good agreement between the threshold s1.
experimental results and the analytical law; however, Two levels of normal stresses are considered (610 and
results from Karlsson tests are underestimated by the 280 MPa); they are respectively associated to a develop-
numerical curve, probably due to the well known ment length of 200 and 100 mm. Because both cases are
‘Achillides effect’ [12]. In fact, Karlsson tests give a characterised by s $ s1, the above-mentioned value of l has
higher bond strength because of an end anchor contribution been evaluated from Eq. (34). The origin of the axes is
due to the protruding free bar end pulled into the bond placed in the point where:
region.  
ds
However, the analytical curve is conservative in s ¼ 0; ¼1¼0
dx
comparison with the actual behaviour; in fact, for a
given development length L, the evaluated corresponding and it can be observed that for x equal to lm:
s— s(L )—is less than the experimental value.
Immediately before the pull-out failure occurs, i.e. when s ¼ s1 ¼ 280 MPa; t ¼ tm ¼ 14:6 MPa;
the rebar starts to slip at the free end, the normal stress s
can yet increase because the integral of bond stresses t s ¼ sm ¼ 0:25 mm
500 E. Cosenza et al. / Composites: Part B 33 (2002) 493–504

Fig. 4. Trends of normal stress, bond stress and slip along a GFRP C-Bare.

5. Basic development length The first circumstance (Case A), i.e. s # s1, involves
Previously, two cases have been studied; they differ for only the ascending branch of the t – s law; for s $ s1 (Case
what follows: B), instead, the ‘softening’ branch of the constitutive law
has to be considered. Otherwise, the bound s1 for s limits
Case A—s # s1; s # sm ; l # lm . l0m the possibility to anchor the rebar.
Case B—s $ s1; s $ sm ; l $ lm However, the softening branch of the bond-slip curve
E. Cosenza et al. / Composites: Part B 33 (2002) 493–504 501

seems to be rather unstable and characterised by large c.o.v. Therefore, remembering Eq. (23), Eq. (42) becomes:
of parameter p [18], which such branch is dependent on ; for  ð12aÞ=ð1þaÞ
this reason, only the Case A will be examined in the design s
Ldb ¼ gt gc lm d
procedure presented in the following. s1
Furthermore, it has to be noticed that since the softening  
branch surely provides a further contribute in the anchorage fs1 1 þ a sd ð12aÞ=ð1þaÞ
¼ gt gc ð43Þ
of the rebar, in case of rebars characterized by different 4tm 1 2 a s1
softening behaviour, such design procedure provides
A further safety factor gg has to be applied as a global
different safety levels.
coefficient to obtain the development length value; such a
Neglecting the contribution of the softening branch, the
global coefficient accounts for all uncertainties arising in
above-mentioned value s1 can be seen as an upper threshold
the evaluation of the embedment length (in particular for the
for the normal stress s; in particular, s1 represents a limit
uncertainty regarding the bond-slip law). Therefore, the
beyond which-even though it is possible to anchor the rebar-
final value of the basic development length can be obtained
it is necessary to rely on the descending branch of the bond-
by:
slip law.
Therefore, it is possible to define a limit condition given Ld ¼ gg Ldb ð44Þ
by the following expression:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi Obviously, this approach can also cause large variations in
8E tm sm safety for different FRP rebars.
s # s1 ¼ ð38Þ
f 1þa It is well-known that in case of FRP reinforced concrete
structures, checks at the serviceability limit state may
or: become the main design criteria. In particular checks of the
8 E t m sm member deflections, of the amplitude of cracks width and of
$1 ð39Þ stress intensity must be performed; results from checks at
1 þ a s2 f
serviceability may indirectly lead to a further reduction of
and the computation of the development length can be the design stress level (sd).
performed by using Eq. (27). As an example, assuming—as first approximation—that
If Eq. (27) is used, a preliminary definition of the t – slip slip s is equal to the amplitude of crack w, a maximum value
constitutive relationship is needed; this law will depend on of slip slim can be fixed based on the allowable crack width
the type of rebar (i.e. on the type of fibres and matrix and on wlim, i.e. the width compatible with aesthetic and psycho-
the shape of outer surface and possible treatments), on the logical problems [14]. Then, a limit for normal stresses slim
concrete compressive strength and on some other par- can be evaluated as a function of slim by using the following
ameters (i.e. temperature, etc.) [8]. Furthermore, the equation:
definition of the elastic modulus E is also necessary. sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a a
The above-mentioned design procedure is briefly 8E tm s1þ lim 8E tm w1þ lim
slim ¼ a ¼ a ð45Þ
reported in the following. fsm 1 þ a fsm 1 þ a
Assigned the characteristic value of the ultimate strength
of the rebar suk, the design value sd has to be determined by A value of slim can be also obtained considering a threshold
considering a material safety factor gm: for s at the serviceability state level: for example, an
allowable stress limit slim ¼ 0.25ffu could be suggested in
suk
sd ¼ ð40Þ case of glass FRP (GFRP) bars.
gm This value of slim can indirectly provide also a limit to
This value sd must be compared to the threshold s1; if the avoid a creep failure of the FRP reinforcing bar; it is worth
design value sd is greater than s1, a new design value of the to notice that, in order to avoid failure due to creep rupture
normal stress should be used in the calculation: of FRP, in the ACI Committee 440 code a stress limit equal
to 0.20ffu under all sustained loads (i.e. dead load plus the
sd ¼ s1 ð41Þ sustained portion of the live loads) has been assumed in the
case of GFPR reinforcing bars [3].
According to Eq. (27), the basic development length will be
In order to verify that serviceability checks are satisfied,
evaluated by:
the ratio sd/slim has to be rather low; in particular, such ratio
 ð12aÞ=ð1þaÞ must be less than the ratio of the design load at the ultimate
s
Ldb ¼ lm d ð42Þ state level to the design load at the serviceability level
s1
(about 1.50). Conversely, if sd/slim is greater than the above
According to the ACI Guide [3], two modification factors ratio, the design will be conditioned by the checks at
should be considered: gt, bar location modification factor, serviceability and the stress level under ultimate load
and gc, concrete cover modification factor. conditions will be less than the design value sd.
502 E. Cosenza et al. / Composites: Part B 33 (2002) 493–504

5.1. Evaluation of the design tensile strength sd Depending on the state level in the JSCE Recommen-
dation [14], two different proposals for the elastic modulus
In defining the tensile strength of FRP reinforcing are reported, thus obtaining two ‘tensile force – strain’
bars some different conversion factors should be used in design curve (see Figs. 3.4.1 and 3.4.2 in the JSCE
order to: Recommendation).
As underlined above, from Eq. (27), it can be noticed that
† transform the strength measured during experimental when the elastic modulus increases, the development length
tests to that of FRP reinforcement in the structure; increases. Therefore, in the evaluation of the development
† take into account the decrease of the strength due to long length, a safety factor gE less than 1 should be used in order
time load; to increase the value of E obtained from the experimental
† consider the deterioration of strength due to concrete test results:
environment.
Ed ¼ E=gE ð47Þ
All the factors should be less than 1 and depend on the
FRP rebar type. For what concerns the conversion factor, for Such an effect could be directly considered by assuming the
‘in situ strength’ a value ranging from 0.8 and 0.95 can be global coefficient gg (Eq. (44)).
adopted.
In the code proposed by ACI 440 [3], the design ultimate 5.3. Multiplication factors (gt and gc)
tensile strength is given by:
According to the ACI Guide [3] a bar location
ffu ¼ CE ffup ð46Þ modification factor (gt) equal to 1.3 should be considered
where CE is the ‘environmental reduction factor’ that when evaluating the basic development length of top FRP
depends on the type of fibre and on the exposure condition, bars.
while ffup is the ultimate tensile strength of the FRP bar, as A concrete cover modification factor (gc) of 1.5 should
reported by the manufacturer, defined as the mean strength be used when the concrete cover or the reinforcement
of a sample of test specimens minus three times the spacing is equal to the bar diameter (f ); for concrete cover
standard deviation. For GFRP reinforcing bars, CE is equal larger than 2f, a factor gc equal to 1 should be applied;
to 0.7 or 0.8 in case of exterior or interior exposure, finally, gc is provided by a linear interpolation of the two
respectively. previous values in the case of concrete cover ranging
The material safety factor gm should reflect the actual between f and 2f.
variability of properties of the reinforcement and the
calculation model for the resistance; moreover, compared 5.4. A numerical example
to steel reinforcement, the more severe consequences due to
the elastic-brittle behaviour of FRP reinforcing bars have to The procedure presented in the previous section is applied
be considered in defining gm. below to a #4 GFRP C-Bare, having an elastic modulus
In case of continuous fibre reinforcing materials (CFRP), equal to 42,000 MPa. A value of gE equal to 1 will be adopted
Okamura et al. [17] suggested to divide the characteristic in the following.
strength by a value of gm ranging from 1.0 to 1.3 since the In this case, values of tm, sm and a which characterised
following factors have to be considered: damage to CFRP the ascending branch of the bond-slip constitutive law are
occurring in the construction and transportation phases; those reported in Table 1 and the threshold stress s1—
differences in material properties between the experimental obtained by Eq. (21)—results equal to 280 MPa.
specimens and the structure; temperature and environmental According to the JSCE Recommendation, a value of gm
condition, etc. equal to 1.30 is considered; then assuming as characteristic
In the recent code proposed by JSCE [14], the following strength a value of s equal to 600 MPa the following design
values of gm are reported: strength is obtained: sd ¼ 600/1.3 < 430 MPa.
The application of the above described procedure for a #4
gm equal to 1.15 in the case of CFRP with carbon or GFRP C-Bare, leads to verify that:
aramid fibres;
gm equal to 1.30 in the case of CFRP with glass fibres. (a) the design value of normal stresses is greater than s1;
therefore, s1 is adopted as new design value (reduced
5.2. Evaluation of the elastic modulus E value):

sd ¼ s1 ¼ 280 MPa
The s – 1 curves obtained by tensile tests in case of FRP
reinforcing bars are not perfectly linear elastic. Therefore, (b) in the case of reinforcement located in the bottom of
some uncertainties exist in evaluating the elastic modulus the concrete element (gt ¼ 1) and for concrete cover
that characterised the linear elastic design s – 1 diagram. larger than 2f (gc ¼ 1), the development length Ldb
E. Cosenza et al. / Composites: Part B 33 (2002) 493–504 503

evaluated for s ¼ sd ¼ s1 by Eq. (43) is equal to: The application of the proposed procedure is very easy
 ð12aÞ=ð1þaÞ and the presented numerical example allows to draw the
s fs 1 þ a
Ldb ¼ gt gc lm d ¼ lm ¼ following conclusions:
s1 4t m 1 2 a
12:7 £ 280 1 þ 0:245 1. the use in practice of GFRP C-Barse seems to be limited
¼ ø 100 mm by their bond behaviour since, neglecting the softening
4 £ 14:65 1 2 0:245
branch of the bond-slip law, it is needed to reduce the
i.e. about eight times the bar diameter; value of design stresses in order to anchor these bars in
(c) assuming a global coefficient gg equal to 2.5, the the concrete matrix;
following development length is obtained (Eq. (44)): 2. the value of the design strength can be increased if the
Ld ¼ gg Ldb ¼ 2:5 £ 8f ¼ 20f unsafe softening branch is considered; in this case, an
adequate value of the basic development length must be
evaluated according to the analytical formulation (Case
The proposed procedure points out that the use in practice of
B);
such a deformed GFRP reinforcing bars appears to be quite
3. a value of the global safety coefficient has to be
limited by their bond behaviour; in fact, it is necessary to
established in order to evaluate the development
reduce the design stresses to the value s1 in order to anchor
length.
these bars in the concrete matrix.
However, in order to check the allowable stress limit at Furthermore, it has been observed that checks at the
the serviceability state limit, assuming as slim the value serviceability state level may become dominant design
0.25ffu it can be obtained: criteria thus leading to a reduction of the stress intensity
slim ¼ 0:25 £ 600 ¼ 150 MPa at the ultimate load conditions in comparison to the
design value. Therefore, the calculated development
and: length results in a conservative value and, actually, the
sd =slim ¼ 280=150 < 1:87 aforementioned limitations, due to the bond behaviour of
GFRP reinforcing bars, do not influence its evaluation.
Therefore, as expected, check at the serviceability limit state
may become a dominant design criteria and a stress level
less than sd will be reached at the ultimate load conditions; References
however, in order to be conservative, a development length
corresponding to sd should be evaluated. [1] ACI, Building code requirements for reinforced concrete. ACI
318-89, Detroit, MI: American Concrete Institute; 1989. revised
1992.
[2] ACI, Building code requirements for structural concrete. ACI 318-95,
Detroit, MI: American Concrete Institute; 1995.
6. Concluding remarks [3] ACI, Guide for the design and construction of concrete reinforced
with FRP bars. ACI Committee 440, Draft, Detroit, MI: American
The problem of the evaluation of the development Concrete Institute; 2000.
[4] Benmokrane B, Masmoudi R. FRP C-Bar as reinforcing rod for
length in case of FRP reinforcing bars has been studied.
concrete structures. In: El-Badry M, editor. Proceedings of the Second
Despite of an extensive state-of-the-art research, which International Conference on Advanced Composite Materials in Bridge
extended the code design formulas for steel reinforced Structures, Montreal, Quebec, Canada; 1996.
concrete structures, a new design procedure has been [5] Brown VL, Bartholomew CL. FRP reinforcing bars in reinforced
proposed based on the analytical formulation. concrete members. ACI Mater J 1993;90(1).
[6] Chaallal O, Benmokrane B. Pullout and bond of glass-fibre rods
It is considered that, since these bars are made of new
embedded in concrete and cement grout. Mater Struct 1993;26.
materials, the design procedure has to be derived in a [7] Chaallal O, Benmokrane B, Masmoudi R. An innovative glass-fibre
rational manner, i.e. neglecting approximate approaches composite rebar for concrete structures. Proceedings of the First
or formulations derived from classical theories (used for International Conference on Advanced Composite Materials in Bridge
steel reinforced concrete structures) and accounting for Structures, Sherbrooke, Quebec, Canada; 1992.
[8] Cosenza E, Manfredi G, Realfonzo R. Behaviour and modeling of
the actual bond-slip constitutive laws.
bond of FRP rebars to concrete. J Compos Construct, ASCE 1997;
Therefore, a suitable assessment of the actual bond 1(2).
behaviour is a main step of the proposed procedure. The [9] Ehsani MR, Saadatmanesh H, Tao S. Design recommendations for
constitutive bond-slip relationship has to be identified for the bond of GFRP rebars to concrete. J Struct Engng, ASCE 1996;
considered FRP rebar from experimental tests. The highest 122(3).
[10] Eligehausen R, Popov,EP, Bertero VV. Local bond stress –slip
bond strength will be obtained for specimens characterised
relationships of deformed bars under generalized excitations. Report
by a well-confined situation, while tests on elements in a not no. 83/23, EERC, University of California, Berkeley; 1983.
so well confined situation will allow the evaluation of the [11] Faza SS, Gangarao HVS. Theoretical and experimental correlation of
influence of the splitting on bond resistance. behavior of concrete beams reinforced with fiber reinforced plastic
504 E. Cosenza et al. / Composites: Part B 33 (2002) 493–504

rebars. In: Nanni A, Dolan CW, editors. Proceedings of [17] Okamura H, Kakuta Y, Uomoto T, Mutsuyoshi H. Design concept for
the International Symposium on Fiber Reinforced Plastic Reinforce- concrete members using continuous fiber reinforcing materials. In:
ment for Concrete Structures, ACI SP-138, Vancouver, Canada; 1993. Nanni A, Dolan CW, editors. Proceedings of the International
[12] fib, Bond of reinforcement in concrete. Bulletin 10. State of the art Symposium on Fiber Reinforced Plastic Reinforcement for Concrete
prepared by Task Group Bond Models, former CEB, Task Group 5.2., Structures, ACI SP-138, Vancouver, Canada; 1993.
CH-1015 Lausanne; August 2000. [18] Pecce M, Manfredi G, Realfonzo R, Cosenza E. Experimental and
[13] Hattori A, Inoue S, Miyagawa T, Fujii MA. Study on bond creep analytical evaluation of bond properties of GFRP bars. J Mater Civil
behavior of FRP rebars embedded in concrete. In: Taerwe L, editor Engng, ASCE 2001;13(4).
Proceedings of the Second International RILEM Symposium [19] Pleimann LG. Strength, modulus of elasticity, and bond of deformed
(FRPRCS-2), Ghent, Belgium; 1995. FRP rods. In: Iyer SL, editor. Advanced Composite Materials in Civil
[14] JSCE, In: Machida A, editor. Recommendations for design and Engineering Structures, Proceedings of the Specialty Conference, Las
construction of concrete structures using continuous fiber reinforcing Vegas, NV, USA; 1991.
materials. Concrete Engineering Series, 23, Tokyo, Japan: Research [20] Rehm G. Über die Grundlagen des Verbundes Zwischen Stahl und
Committee on Continuous Fiber Reinforcing Materials; 1997. Beton. Deutscher Ausschuss für Stahlbeton, Helft 1961;138.
[15] Karlsson M. Bond between C-Bar FRP reinforcement and concrete, [21] Tepfers R, Work No. 15. Bond of FRP reinforcement in concrete—a
Chalmers University of Technology, Pub. No.98:3, Work No. 22. state of the art in preparation, Pub. 97. Goteborg, Sweden: Chalmers
Graduation Thesis: E-91:1, Goteborg, Sweden; 1997. University of Technology; 1997.
[16] Nanni A, Al-Zaharani MM, Al-Dulaijan SU, Bakis CE, Boothby TE. [22] Tepfers R, Karlsson M. Pull-out and tensile reinforcement splice
Bond of FRP reinforcement to concrete—experimental results. In: tests using FRP C-Bars. Proceedings of the Third International
Taerwe L, editor. Proceedings of the Second International RILEM Symposium on Non-Metallic (FRP) Reinforcement for Concrete
Symposium (FRPRCS-2), Ghent, Belgium; 1995. Structures, Japan: Japan Concrete Institute; 1997.

Anda mungkin juga menyukai