Professional Documents
Culture Documents
Abstract: Strengthening of RC beams and slabs using fiber-reinforced polymers (FRPs) is known to control crack width. However, no
specific provisions are provided by most international design codes for predicting crack width in FRP-strengthened members. In this paper,
models for predicting crack width of conventional RC members in five international codes are extended to FRP-strengthened beams by
introducing appropriate modifications to account for the effect of FRP composites. Predictions of the crack width by the resulting models
are evaluated and compared with experimental results. Both analytical and experimental results showed significant reductions in crack widths
due to FRP strengthening. The efficiency of controlling crack width, however, decreased with increased amount of steel reinforcement and
increased with increasing the ratio of FRP plate to beam width. Comparison of the codes’ predictions of the maximum crack width against
experimental results revealed that some predictions are in good agreement with the measured values while others overestimate the maximum
crack width by a significant margin. DOI: 10.1061/(ASCE)CC.1943-5614.0000805. © 2017 American Society of Civil Engineers.
Author keywords: Concrete beams; Strengthening; Fiber-reinforced polymers; Cracking behavior; Crack width.
Introduction models and experimental data (ACI 2008; CSA 2012; Saliger 1936;
Thomas 1936; Base et al. 1966; Gergely and Lutz 1968; Frosch
Crack control is an important serviceability criterion in the design 1999), and many of the factors affecting the crack width have been
of RC members. Cracking of concrete is related to its limited tensile investigated (Ceroni et al. 2004; Ceroni and Pecce 2004; Ueda et al.
deformation capacity, and is usually expected to develop in RC 2002; Tan and Saha 2008). Makhlouf and Malhas (1996) studied
members under service loads during the lifetime of the structure. the effect of thick concrete cover on the maximum flexural crack
However, there should be some control over the width and distri- width under service load. Beeby (1971) and Nawy and Blair (1971)
bution of structural cracks in order to maintain the aesthetics of
showed that the transverse reinforcement had a strong influence on
the structure, protect the internal steel reinforcement especially
the crack spacing.
when subjected to aggressive environments, and reduce the risk
The application of externally bonded (EB) FRP reinforcements
of debonding of the fiber-reinforced polymer (FRP) external
for the flexural strengthening of RC members is becoming popular
reinforcement.
worldwide. The purpose of FRP strengthening is to increase the
Cracking pattern and width mainly depend on the available
flexural reinforcement. The most important factor controlling ultimate capacity or improve service conditions. Generally, RC
flexural crack width under a given load is the magnitude of the members strengthened with EB FRP laminates have smaller and
tensile strain in the reinforcing steel. Other important factors affect- closely spaced cracks compared with the unstrengthened beams,
ing crack widths are the thickness of concrete cover, quality of con- either considering the same service load or considering the same
crete, ratio of steel reinforcement, size of steel bars, and distribution tension level in the steel (Ceroni and Pecce 2009). This is due to
of the steel bars in the tension zone (Broms 1965; Avril et al. 2005; the additional tension stiffening of the EB reinforcement that re-
Zhang et al. 2015). duces the crack spacing.
The crack width of an RC flexural member is obtained by multi- The large number of experimental tests on RC elements
plying the crack spacing by the mean strain of flexural steel strengthened by externally bonded FRP laminates and sheets avail-
reinforcement. Many research works found in the literature pre- able in the literature focused on studying the flexural behavior at
dicted the crack width of RC flexural members based on theoretical ultimate conditions (e.g., Al-Negheimish et al. 2012; Al-Zaid et al.
2014). Only limited studies dealt with the cracking behavior of
FRP-strengthened beams (e.g., Ceroni and Pecce 2007, 2009;
1
Zhang et al. 2015). In this paper, models for predicting the crack
Ph.D. Candidate, Dept. of Civil Engineering, King Saud Univ.,
width in conventional RC members from leading international co-
P.O. Box 800, Riyadh, Kingdom of Saudi Arabia 11421. E-mail:
malsaawani@ksu.edu.sa des and design guidelines are presented. These prediction models
2
Associate Professor, Center of Excellence for Concrete Research and are then extended to strengthened beams by introducing appropri-
Testing, Dept. of Civil Engineering, King Saud Univ., P.O. Box 800, ate modifications to account for the effect of FRP composites. On
Riyadh, Kingdom of Saudi Arabia 11421 (corresponding author). E-mail: the other hand, an experimental investigation was conducted to
ahelsayed@ksu.edu.sa study the effect of several design variables on the crack width of
3
Professor, Dept. of Civil Engineering, King Saud Univ., P.O. Box 800, RC shallow beams strengthened with EB carbon FRP (CFRP) plates.
Riyadh, Kingdom of Saudi Arabia 11421. E-mail: negaimsh@ksu.edu.sa
The cracking behavior of the tested beams as well as the effect of
Note. This manuscript was submitted on September 25, 2016; approved
on December 22, 2016; published online on March 10, 2017. Discussion
test variables on the measured crack width are presented and dis-
period open until August 10, 2017; separate discussions must be submitted cussed in this paper. In addition, the resulting prediction models
for individual papers. This paper is part of the Journal of Composites for are evaluated by comparison with the results from the conducted
Construction, © ASCE, ISSN 1090-0268. experimental program.
the loading duration (kt taken as 0.6 for short-term loading and 0.4
for long-term loading); the effective tensile strength of concrete, The value of A in Eq. (10) is then calculated as A ¼ ð2btr =nb Þ,
f ctm ; the effective reinforcement ratio, ρp;eff ; and the modular where b is the beam width and nb is the number of tensile steel bars.
ratio (ns ¼ Es =Ec ). The lower limit is suggested to the difference
ðεsm − εcm Þ, which is set greater than or equal to 0.6fs =Es .
When external fiber reinforcement is applied, Eqs. (7) and (8) Japanese Standard Specifications
can be modified (Ceroni and Pecce 2009) to introduce the contri- For unstrengthened RC members, the crack width can be calculated
bution of the external reinforcement in ρp;eff as follows: according to JSCE-07 using the following equation:
As þ Af ðEf =Es Þ
ρp;eff ¼ ð9Þ f 0
Ac;eff wmax ¼ 1.1α1 α2 α3 ½4cc þ 0.7ðcs − db Þ se þ εcsd ð12Þ
Es
where Af and Ef = area and the modulus of elasticity of the external in which cc = concrete cover (mm); cs and db = center-to-center
FRP reinforcement, respectively. The external FRP reinforcement distance and diameter of the steel reinforcement (mm), respec-
is also accounted for in Eq. (8) by the reduced stress in steel, which tively; α1 = coefficient that characterizes the effect of surface geom-
is obtained from considering the RC cracked section with the ex- etry of reinforcement (α1 ¼ 1.0 for deformed bars and α1 ¼ 1.3 for
ternal FRP laminate. plain bars and prestressing steel); and α2 = constant to take into
account the effect of concrete quality, calculated as
American Concrete Institute Approach 15
α2 ¼ þ 0.7 ð13Þ
The approach of the American Concrete Institute (ACI), through fc0 þ 20
ACI 224.1R-07 (ACI 2007), is based on the statistical analysis
of Gergely and Lutz (1968) of crack width data from a number in which f c0 = compressive strength of concrete (N=mm2 ). Here, α3
of sources. Based on the statistical analysis, the equation that pre- is a constant that takes into account the effect of multiple layers of
dicts the maximum crack width in bending is given in a simplified tensile reinforcement; it may be calculated as
form by
pffiffiffiffiffiffiffiffi 5ðnl þ 2Þ
α3 ¼ ð14Þ
wmax ¼ 0.011β ACI f s 3 dc A × 10−3 ð10Þ 7nl þ 8
0
where wmax = most probable maximum crack width (mm). The steel in which nl = number of layers of tensile reinforcement. Also, εcsd
stress is the most important variable affecting crack width and is is the compressive strain for evaluation of the increment of crack
taken into account in the ACI equation with a factor f s. Also, the width due to shrinkage and creep of concrete, and f se is the incre-
maximum crack width is directly affected by the thickness of con- ment of stress of reinforcement from the state in which concrete
crete cover, dc , considered from the tension concrete fiber to the stress at the portion of reinforcement is 0 (N=mm2 ).
center of the closest steel bars in millimeters. In addition, the ACI For FRP-strengthened members, the crack width according to
equation takes into account the effect of the area of concrete sub- the Recommendations for Upgrading of Concrete Structures with
jected to tensile stresses by the factor, A, which is calculated as the Use of Continuous Fiber Sheets (JSCE 2001) is approximately 0.3
area of concrete symmetric with reinforcing steel divided by num- to 0.7 times the width of cracks in members not bonded with FRP.
ber of bars in square millimeters. For the specific case of large dead load, and the absence of concrete
The ACI recommended equation for crack control predicts the cracking prior to the application of FRP, or for structures governed
maximum crack width at the level of tension steel reinforcement by live loads, the flexural crack width may be calculated using
and does not require the evaluation of crack spacing. The width Eq. (15) [Eq. C6.5.1 in JSCE (2001)], in which the crack width
of a flexural crack increases with the distance from the neutral axis. is taken as 70% of the width calculated from Eq. (12) [Eq. 7.4.4
This is taken into account in the ACI prediction equation through in JSCE-07 (JSCE 2007)]
the use of an amplification factor, β ACI , to account for the change of
the strain over the beam section. This factor ðβ ACI ¼ h − x=ds − xÞ is f 0
wmax ¼ 0.7α½4cc þ 0.7ðcs − db Þ se þ εcsd ð15Þ
taken as the ratio of distance between neutral axis and concrete ten- Es
sion face to distance between neutral axis and the reinforcing steel.
It could be seen that the ACI recommended equation for crack In addition, the external FRP reinforcement is accounted for in
control predicts the most probable maximum crack width, which the Japanese Society of Civil Engineers (JSCE) prediction equation
was derived from regression analyses on data from several crack of the crack width by the reduced stress in steel reinforcement,
width studies. This usually means that expected crack widths in the which is calculated for the FRP-strengthened section based on
member have high probability to be below the calculated value. elastic cracked-section analysis.
Fig. 1. Test setup and instrumentation: (a) elevation; (b) beam cross sections; (c) locations of strain gauges along the CFRP plates
other two beams (C-0.5-35-0 and S-0.5-35-240) from Group I widths are obtained at both service and ultimate limits due to FRP
(ratios of bf =b are 0.6 and 0.9). Fig. 1 shows the beam dimensions strengthening. Considering the full loading history, the tension-
and reinforcements. Detailed information about the fabrication and stiffening effect due to FRP strengthening is evident even after
testing can be found in Al-Saawani et al. (2015). yielding of tension steel for all strengthened beams due to the elas-
tic behavior of FRPs (Ceroni et al. 2004; Ceroni and Pecce 2004).
The external FRP reinforcement produces the highest tension stiff-
Experimental Results and Discussion ening effect for beams with the lowest steel reinforcement percent-
age (Beam S-0.5-35-240) as indicated by Fig. 3(a). The effectiveness
The cracking patterns were observed and presented for different of EB FRP laminates in controlling the width of cracks, however,
load levels until failure. The average crack spacing and width for decreased with increased ρs of the tested beams.
the tested beams were measured in the central part (between the two Comparing the values of the measured crack width for the un-
concentrated loads) of each beam, where the bending moment is strengthened beams in Fig. 3(b) shows insignificant change in the
constant. Fig. 2 shows the crack patterns for the tested beams at crack widths with increasing the concrete strength, f c0 . A similar
yielding of tension steel. Generally, no significant difference in crack finding was noticed among their respective strengthened beams,
spacing was observed for the unstrengthened and the strengthened which are cast with varying fc0 (plotted in the same figure), where
beams. The steel stirrups spaced at 100 mm appear to have influ- no major change in the cracking behavior and crack widths was
enced the development of cracks during the loading history. During observed among these strengthened beams. On the other hand,
testing, the initial crack widths in the constant moment region were Fig. 3(c) shows considerable reductions in the average crack width
manually measured at the level of tension steel reinforcement using a with increasing the width of the CFRP plates (and consequently
handheld 50× microscope. With further loading, the widths of these increasing the FRP reinforcement ratio, calculated as ρf ¼ Af =bds ),
cracks were also measured using high-accuracy LVDTs up to failure. especially after yielding of tension steel. The greater the amount of
Fig. 3 shows curves of the load versus average crack width plot- EB FRP used (Beam S-0.5-35-360), the greater is the tension stiff-
ted for the tested beams in the different groups. Fig. 3(a) shows that ening effect and the contribution to crack width control.
increasing the tension steel reinforcement ratio, ρs , had a consid- Table 3 lists the average and maximum experimental crack
erable effect in reducing the crack width for the unstrengthened width measured at service load level (wave;serv and wmax;serv ) and at
beams when compared at the same load level. Applying external ultimate load level (wave;u and wmax;u ). It also shows the efficiency
FRP reinforcement modifies the cracking behavior due to the addi- of FRP strengthening in controlling the crack width of the strength-
tional stresses transferred to the concrete in tension by the external ened beams as compared with their respective unstrengthened (con-
FRP reinforcement. Generally, significant reductions in the crack trol) beams, which is indicated by the ratios of the crack widths of
Fig. 2. Crack patterns of the tested beams at yielding: (a) Group I beams; (b) Group II beams; (c) Group III beams
Fig. 3. Load-average crack width curves for the tested beams: (a) Group I beams; (b) Group II beams; (c) Group III beams
the strengthened beams to that of their respective control beams It is observed from Table 3 that applying EB FRP laminates
(wstr =wcrl ). The service limit considered in this table is taken cor- caused significant reductions in the measured crack width for most
responding to 60% of the ultimate load capacity for the respective strengthened beams. This reduction is quantified numerically in
control beams in the different test groups. Table 3 by the ratios of the average crack width of the strengthened
111
99
101
99
101
97
102
92
93
102
110
Pu , as listed in the table. Varying the steel reinforcement ratio, the
FRP strengthening of Beam S-0.5-35-240 (which is reinforced by
the smallest ρs of 0.5%) was very effective in controlling crack
width because the average measured crack width was only 33%
Number of
cracks, nc
10 of that of the corresponding unstrengthened beam. Increasing ρs
11
11
11
11
11
11
12
11
10
10
to 0.9% in the strengthened beam S-0.9-35-240 caused a reduc-
tion in the efficiency of FRP strengthening in controlling the crack
width because the average measured crack width was increased
to be 51% of its respective control beam. Further increase in ρs
wstr =wcrl
0.20
0.36
0.43
0.19
0.13
0.09
—
—
to 1.25% in the strengthened beam S-1.3-35-240 significantly re-
Downloaded from ascelibrary.org by University of California, San Diego on 04/16/17. Copyright ASCE. For personal use only; all rights reserved.
At Pu
width because the average measured crack width was 83% of its
Service limit considered in comparing the change in crack width due to CFRP strengthening is taken at 60% of ultimate load of the respective control beams.
wmax;u
0.26
0.61
0.79
0.57
0.42
0.21
ing the ratio of CFRP plate to beam width from bf =b ¼ 0.6 (for
—
C-0.5-35-0.
Fig. 4 shows the relationship between the average experimental
crack width and the corresponding strain in the steel reinforcement.
wstr =wcrl
0.30
0.33
0.17
0.12
0.11
—
the other hand, Fig. 4(b) shows the same plots for beams reinforced
Average crack width (mm)
while Fig. 4(c) shows plots of the relationship for beams rein-
forced with equal ρs and strengthened with different amounts of
CFRP. These figures show that the values of the crack width are
wstr =wcrl
0.51
0.83
0.53
0.49
0.26
—
are almost equal when measured at the same steel strain, irrespec-
tive of the amount of steel or CFRP reinforcements and loading
wave;serv
0.26
0.09
0.20
0.10
0.13
0.10
0.21
0.11
0.19
0.09
0.07
level considered.
62.7
101.1
119.9
97.4
92.8
62.7
Pserv a
63
54
28
30
16
132
S-0.9-35-240
S-1.3-35-240
S-0.9-24-240
S-0.9-17-240
S-0.5-35-360
C-0.5-35-0
C-0.9-35-0
C-1.3-35-0
C-0.9-24-0
C-0.9-17-0
Figs. 5–9 show curves for the load versus maximum crack width
III
of the tested beams in the different groups. The figures also show
II
I
Fig. 4. Measured crack width versus corresponding strain in steel for the strengthened beams: (a) control and strengthened beams with ρs ¼ 0.9%;
(b) beams with different ρs and strengthened with equal bf ; (c) beams with equal ρs and strengthened with different bf
Fig. 5. Comparisons of the measured crack widths with fib Bulletin 14 predictions: (a) comparisons of strengthened beams with varied ρs ; (b) com-
parisons of strengthened beams with varied fc0 ; (c) comparisons of strengthened beams with varied bf
comparisons of these experimental curves with the predicted values predictions are generally in good agreement with the experimental
given by the different codes, for the unstrengthened as well as the values for most of the strengthened beams regardless of the varying
FRP-strengthened beams. These figures were plotted to show com- values of the steel reinforcement [Fig. 5(a)], the concrete strength
parisons of load-crack width starting from cracking of concrete [Fig. 5(b)], or the width of FRP external reinforcement used
stage until yielding of tension steel for both unstrengthened and [Fig. 5(c)].
FRP-strengthened beams. For EC2-04, good predictions of the maximum crack width
Predictions of the maximum crack width for the RC beams were provided for the unstrengthened RC beams, as shown in
externally strengthened with FRP laminates using the simplified Fig. 6(a). These predictions, however, were slightly underestimated
method of the fib Bulletin 14 (fib 2001) presented previously in for the beams with larger ρs at small levels of steel stress. The com-
Eq. (5) are shown in Fig. 5. The figure shows that the crack width parisons shown in the figure indicate that the EC2-04 prediction
Fig. 6. Comparisons of the measured crack widths with EC2-04 predictions: (a) comparisons of unstrengthened beams with varied ρs ; (b) compar-
isons of strengthened beams with varied ρs ; (c) comparisons of unstrengthened beams with varied fc0 ; (d) comparisons of strengthened beams with
varied f c0 ; (e) comparisons of strengthened beams with varied bf
equation could capture the effect of varying ρs on the maximum reinforcing steel, fs , and also the thickness of cover from tension
crack width because it accounts for the stress in the reinforcing concrete fiber to the center of the closest bar as well as the effective
steel, computed on the basis of a cracked section. The EC2-04 area of concrete symmetric with reinforcing steel bars. Therefore,
equation also limited the value of the average strain εm to not be the effect of varying ρs on the maximum crack width was reflected
less than 0.6fs =Es . This is clearly important for avoiding underes- in the prediction equation, as shown in the comparison. On the
timated predictions of the crack width at low levels of steel stress other hand, predictions of the maximum crack width for the FRP-
for RC beams reinforced with low ρs. On the other hand, Fig. 6(b) strengthened beams were calculated using the modified ACI equa-
shows comparisons of the maximum crack width for the FRP- tion taking into account the external FRP reinforcement, according
strengthened beams reinforced with different ρs , calculated using to Eq. (11). Fig. 7(b) shows good correlation of the predicted crack
the modified equation of EC2-04 taking into account the presence width values with the experimental results at different levels of steel
of external FRP laminates. The figure indicates good correlation stress for the strengthened beams reinforced with different ρs . Sim-
with the experimental values at all levels of steel stress. Compari- ilarly, Figs. 7(c and d) show good predictions of the crack width for
son curves of the maximum crack width for tested beams with the unstrengthened and FRP-strengthened beams with different fc0 .
different compressive strength of concrete are shown in Fig. 6(c) On the other hand, Fig. 7(e) shows the effect of varying the width of
for the unstrengthened beams and in Fig. 6(d) for the FRP- the CFRP laminate on the predicted values of the maximum crack
strengthened beams. Predictions obtained by the EC2-04 equation width obtained by the modified ACI equation. The predictions were
were in good agreement with the experimental crack width values. in good agreement with the experimental values for both of the
Fig. 6(e) shows the effect of varying the CFRP plate width on FRP-strengthened beams.
the predicted values of the maximum crack width as compared with Fig. 8(a) shows that the JSCE-07 (JSCE 2007) predictions over-
the experimental results. The predicted values correlated well estimated the maximum crack width for the unstrengthened beams
with the experimental values for the strengthened beams as shown reinforced with varying ρs by a significant margin. Similarly, the
in the figure. JSCE prediction equation accounting for the presence of the exter-
Predictions of the maximum crack widths by ACI 224.1R-07 nal FRP reinforcement gave an overestimated maximum crack
(ACI 2007) for the unstrengthened RC beams were in good agree- width except for the beam reinforced with the largest ρs , as shown
ment with measured values, as shown in Fig. 7(a). The ACI equa- in Fig. 8(b). Comparison curves of the maximum crack width given
tion accounts for the effect of varying ρs by including the stress in by the JSCE for the beams cast with different fc0 also indicated
Fig. 7. Comparisons of the measured crack widths with ACI predictions: (a) comparisons of unstrengthened beams with varied ρs ; (b) comparisons of
strengthened beams with varied ρs ; (c) comparisons of unstrengthened beams with varied f c0 ; (d) comparisons of strengthened beams with varied f c0 ;
(e) comparisons of strengthened beams with varied bf
overestimated values [as shown in Fig. 8(c) for the unstrengthened considered in these comparisons was taken as 60% of the yielding
beams, and shown in Fig. 8(d) for the FRP-strengthened beams]. load of the unstrengthened beams. The comparisons show that
On the other hand, Fig. 8(e) shows a comparison of the maximum EC2-04 and ACI 224.1R-07 provided predictions that are generally
crack width for the strengthened beams with different CFRP plate in good agreement with the experimental crack width values with
width given by the modified JSCE equation accounting for the ex- the least scatter because the average ratios of the predicted to mea-
ternal FRP strengthening. The highly overestimated predictions of sured maximum crack width, wmax;calc =wmax;exp , are 0.93 and 1.11,
the maximum crack width given by JSCE are due to the large crack with coefficients of variation of 5 and 6%, respectively. On the
spacing values calculated by the code equation, as will be discussed other hand, the JSCE-07 and CAN/CSA S6-10 predictions over-
subsequently. estimated the crack width and showed more scatter, with the aver-
Fig. 9(a) shows overestimated values of the maximum crack age ratio of wmax;calc =wmax;exp being 1.43 and 1.36, and having
width given by the CAN/CSA S6-10 code (CSA 2010) for the un- coefficients of variation of 11 and 14%, respectively.
strengthened beams. Predictions of the crack width for the FRP- Comparisons of the maximum experimental crack widths
strengthened beams reinforced with various ρs were noticed to be (measured at the level of tension steel reinforcement) for the FRP-
slightly overestimated but in better agreement with the experimen- strengthened beams with predicted values by the modified equa-
tal values after implementing modifications for the prediction tions from the various design guidelines are shown in Table 5
equation to account for the external FRP strengthening [Fig. 9(b)]. and Fig. 10(b). The service load limit considered in these compar-
Also, predictions of the crack width for the unstrengthened beams isons was taken at 80% of the yielding load of the strengthened
[Fig. 9(c)] and the strengthened beams [Fig. 9(d)] with different fc0 beams, as prescribed by ACI 440.2R (ACI 2008). The comparisons
were shown to be overestimated. Fig. 9(e) shows highly overesti- indicated that each of the fib Bulletin 14, EC2-04, and ACI 224.
mated values of the maximum crack width provided by the modified 1R-07 provided predictions that are in good agreement with the ex-
equation of the CAN/CSA S6-10 code for the beams reinforced by perimental values of the crack width with the least scatter. This is
the smallest ρs and strengthened with different CFRP plate widths. evidenced because the average ratios of the predicted to measured
Table 4 and Fig. 10(a) show comparisons of the maximum ex- maximum crack width, wmax;calc =wmax;exp , are 1.01, 1.06, and 1.02,
perimental crack widths (measured at the level of tension steel rein- with coefficients of variation of 10, 11, and 9%, respectively.
forcement) for the unstrengthened beams with the predicted values On the other hand, the JSCE-07 and CAN/CSA S6-10 predictions
calculated by the different design guidelines. The service load limit overestimated the crack width and showed more scatter, with the
Fig. 8. Comparisons of the measured crack widths with JSCE predictions: (a) comparisons of unstrengthened beams with varied ρs ; (b) comparisons
of strengthened beams with varied ρs ; (c) comparisons of unstrengthened beams with varied f c0 ; (d) comparisons of strengthened beams with varied
f c0 ; (e) comparisons of strengthened beams with varied bf
average ratio of wmax;calc =wmax;exp being 1.17 and 1.24, and having spacing and also of the average strain in steel, as calculated by the
coefficients of variation of 22 and 11%, respectively. different codes and guidelines, for both the unstrengthened and the
These comparisons indicate that the modifications implemented FRP-strengthened beams at the service load limit (taken as 60% of
into the crack width prediction equations for the unstrengthened the yielding load of the unstrengthened beams and 80% of the
beams that are given by both EC2-04 and ACI 224.1R-07 resulted yielding load of the strengthened beams).
in good predictions of the maximum crack width for the FRP- Fig. 11(a) includes crack spacing for the strengthened beams
strengthened beams considering the different steel and FRP rein- predicted by fib Bulletin 14, EC2-04, JSCE-07, and CAN/CSA
forcement ratios and concrete compressive strength. On the other S6-10 codes, while no prediction was given by the ACI 224.1R-07
hand, modified equations of the JSCE-07 and CAN/CSA S6-10 for the crack spacing because the equation calculates the crack
codes to predict the maximum crack width for the FRP-strength- width directly without predicting the crack spacing. On the other
ened beams provided poor predictions with larger scatter. hand, Fig. 11(b) includes a comparison of the average strain in the
It is also observed from Table 5 that, for the same code, the pre- steel reinforcement predicted by the different codes, which repre-
dicted crack width values are almost equal when calculated at the sents the average tensile strain difference between the steel and the
same steel strain, irrespective of the amount of tension reinforce- concrete taking into account factors such as bond stress, tension
ment or loading level considered. This is more pronounced for fib stiffening, and shrinkage effects. The cause for the inaccurate pre-
Bulletin 14, EC2-04, and ACI 224.1R-07. This observation was dictions given by the JSCE-07 and the CAN/CSA S6-10 codes
evidenced experimentally as discussed previously and shown in is highlighted by the large variation of crack spacing shown in
Fig. 4, emphasizing the key role played by the strain in steel rein-
Fig. 11(a). The crack spacing predicted by fib Bulletin 14 was ap-
forcement in determining the crack width in RC beams.
proximately similar to the average experimental crack spacing of
100 mm, while both the JSCE-07 and CAN/CSA S6-10 predictions
Discussion of the Main Parameters Related were always higher than experimental values by significant mar-
to Code Predictions gins. No such variations were observed from comparison of the
average steel strain in Fig. 11(b), which shows only marginal differ-
Formulation of crack width in design is usually based on the aver- ences in the predicted values by various codes.
age crack spacing, srm , and the average reinforcement strain, εsm . In the JSCE-07 code, the only modification that was applied to
Therefore, it is convenient to show a comparison of the crack the prediction equation by JSCE-07 is represented by multiplying
Fig. 9. Comparisons of the measured crack widths with CAN/CSA S6-10 predictions: (a) comparisons for unstrengthened beams with varied ρs ;
(b) comparisons for strengthened beams with varied ρs ; (c) comparisons for unstrengthened beams with varied fc0 ; (d) comparisons for strengthened
beams with varied fc0 ; (e) comparisons of strengthened beams with varied bf
Table 4. Comparisons of the Maximum Experimental Crack Width for Unstrengthened Beams with Codes’ Predictions at Service Load
Eurocode 2 ACI 224.1R-07 JSCE-07 CAN/CSA S6-10
Experimental (CEN 2004) (ACI 2007) (JSCE 2007) (CSA 2010)
Pserv a wmax;exp wmax;calc wmax;calc = wmax;calc wmax;calc = wmax;calc wmax;calc = wmax;calc wmax;calc =
Beam (kN) (mm) (mm) wmax;exp (mm) wmax;exp (mm) wmax;exp (mm) wmax;exp
C-0.5-35-0 38.8 0.16 0.15 0.95 0.19 1.20 0.27 1.74 0.27 1.72
C-0.9-35-0 71.5 0.17 0.14 0.85 0.18 1.08 0.22 1.31 0.22 1.30
C-1.3-35-0 99.3 0.15 0.14 0.98 0.17 1.16 0.20 1.35 0.19 1.28
C-0.9-24-0 75.4 0.18 0.17 0.95 0.20 1.09 0.25 1.39 0.23 1.28
C-0.9-17-0 76.2 0.20 0.19 0.94 0.21 1.03 0.27 1.38 0.24 1.19
Mean — 0.93 — 1.11 — 1.43 — 1.36
Standard deviation — 0.04 — 0.06 — 0.16 — 0.19
Coefficient of variation — 5% — 6% — 11% — 14%
a
Service load limit considered in these comparisons is taken at 60% of the yielding load of the unstrengthened beams.
the prediction equation of crack width, which is calculated for the and the center-to-center distance of the outer-layer steel reinforce-
unstrengthened members, by a maximum crack width ratio of 0.7. ment, cs , are suggested to be modified in order to accommodate the
In addition to that, the expression of the strain in steel is modified effect of the external FRP reinforcement.
considering the RC cracked section with external FRP reinforce- Similarly, in the crack spacing equation given by CAN/CSA
ment. No other modification was applied to the prediction equation S6-10, overestimated values could be avoided by reducing the ini-
because it did not include parameters that could be directly modi- tial crack spacing of 50 mm given by the crack spacing equation. In
fied to introduce the contribution of the external FRP reinforce- addition, the effective area of concrete in tension, Ac;eff , needs to be
ment. For this reason, the empirical factors available in the crack modified to accommodate strengthened beams with different steel
spacing equation, such as the multiplier of the concrete cover, cc , and FRP reinforcements.
control beams; (b) predicted to measured crack width for strengthened beams
Table 5. Comparisons of Maximum Experimental Crack Width of Strengthened Beams with Codes’ Predictions at Service Load
fib Bulletin 14 Eurocode 2 ACI 224.1R-07 JSCE-07 CAN/CSA S6-10
Experimental (fib 2001) (CEN 2004) (ACI 2007) (JSCE 2007) (CSA 2010)
Pserv a wmax;exp wmax;calc wmax;calc = wmax;calc wmax;calc = wmax;calc wmax;calc = wmax;calc wmax;calc = wmax;calc wmax;calc =
Beam (KN) (mm) (mm) wmax;exp (mm) wmax;exp (mm) wmax;exp (mm) wmax;exp (mm) wmax;exp
S-0.5-35-240 123.2 0.23 0.27 1.19 0.25 1.10 0.24 1.06 0.32 1.39 0.32 1.40
S-0.9-35-240 152.5 0.23 0.23 1.00 0.22 0.95 0.22 0.94 0.21 0.94 0.26 1.15
S-1.3-35-240 201.4 0.24 0.22 0.91 0.22 0.90 0.22 0.89 0.20 0.81 0.25 1.02
S-0.9-24-240 172.6 0.25 0.25 0.98 0.26 1.04 0.25 0.99 0.26 1.04 0.30 1.18
S-0.9-17-240 172.7 0.22 0.23 1.09 0.27 1.24 0.26 1.19 0.28 1.30 0.30 1.37
S-0.5-35-360 156.6 0.21 0.19 0.91 0.24 1.12 0.22 1.04 0.33 1.52 0.28 1.31
Mean — 1.01 — 1.06 — 1.02 — 1.17 — 1.24
Standard deviation — 0.10 — 0.11 — 0.10 — 0.26 — 0.13
Coefficient of variation — 10% — 11% — 9% — 22% — 11%
a
Service load limit considered in these comparisons is taken at 80% of the yielding load of the strengthened beams.
Fig. 11. Comparisons of crack spacing and steel tensile strain at service load: (a) comparison of crack spacing for the strengthened beams;
(b) comparison of steel strain for the strengthened beams
Conclusions • Steel strain was shown to be the key parameter affecting crack
width in both strengthened and unstrengthened RC members be-
The crack width prediction equations of conventional RC members cause the crack width values were found to be almost equal when
in several international codes and design guidelines were extended measured at the same steel strain, irrespective of the amount of
to FRP-strengthened beams by introducing appropriate modifica- steel, CFRP reinforcements, and loading level considered.
tions to account for the presence of external FRP reinforcement. • For unstrengthened RC beams, comparison of the crack width
The resulting modified equations were used to predict the crack predictions by the various codes with the experimental results
widths of FRP-strengthened beams and compare them with exper- revealed that the EC2-04 and ACI 224.1R-07 codes provided
imental results. Based on the results of this investigation, the fol- predictions that are generally in good agreement with the experi-
lowing conclusions can be made: mental crack width values with the least scatter. On the other
• Significant reductions in crack width were achieved at both hand, JSCE-07 and CAN/CSA S6-10 predictions overestimated
service and ultimate load limits due to strengthening with EB the measured crack width and showed more scatter.
CFRP laminates. However, the efficiency of FRP strengthening • For FRP-strengthened beams, the predictions of the crack
in controlling the crack width decreased with increased tension width given by the fib Bulletin 14 were in good agreement
steel reinforcement. On the other hand, varying the compres- with the experimental results. Similar good predictions were
sive strengths of concrete had little effect on the efficiency of also obtained using the modified equations of both the ACI
strengthening in controlling crack width. 224.1R-07 and EC2-04 codes, which provide confidence that