Fiber-Reinforced Polymer Strengthening of Steel

Beams under Static and Fatigue Loadings

Bibek Regmi Bagale, S.M.ASCE 1; and Azadeh Parvin 2
Downloaded from ascelibrary.org by Auckland University Of Technology on 10/28/20. Copyright ASCE. For personal use only; all rights reserved.

Abstract: In this paper, nonlinear finite element analysis (FEA) was used to investigate the behavior of fiber-reinforced polymer (FRP)
retrofitted steel beams subjected to static and fatigue loadings. Under static loading, damaged and undamaged steel beams were retrofitted
with carbon FRP (CFRP) to examine the effect of bond length and transverse anchorages on flexural behavior and failure modes. Under
fatigue loading, damaged control, basalt FRP (BFRP), and aramid FRP (AFRP) retrofitted steel beams were analyzed to predict their fatigue
life. The CFRP laminates provided an increase in the flexural capacity for the undamaged and damaged beams by approximately 1.7 and 3
times, respectively, as compared to their counterpart control beams. The effective FRP laminate bond length under static loading condition
was found to be 2=5 of the simply supported beam length. Furthermore, with the addition of anchorages, the flexural capacity was increased
by 1.5 times under fatigue loading, a regression analysis was performed to generate new S-N curves, and formulas were proposed to predict
the fatigue life of BFRP and AFRP strengthened steel beams. DOI: 10.1061/(ASCE)SC.1943-5576.0000534. © 2020 American Society of
Civil Engineers.
Author keywords: Steel; Fatigue loading; Fiber-reinforced polymer; Bond; Finite element analysis; Flexural strengthening.

Introduction mode of failure is the tensile rupture of FRP (Linghoff et al. 2010),
which can be attained using longer laminate lengths (Lenwari et al.
Many older steel structure in-service steel bridges are structurally 2005) or providing anchorages. Transverse wrapping at the lami-
deficient (Colombi and Poggi 2006). These deficiencies can be nate ends prevented end debonding and increased the load capacity
attributed to harsh environmental exposure such as freeze-thaw of CFRP-strengthened reinforced concrete beams (Kalfat et al.
cycles, sea water in the coastal areas, as well as creep, fatigue, 2013).
aging, and lack of proper maintenance (Al-Saidy et al. 2004). A Additionally, traditional retrofit techniques such as welding and
recent report on the state of infrastructure by ASCE (2017) revealed bolting of external steel plates not only increase the self-weight but
that four in 10 bridges in the United States are more than 50 years also lead to newer stress concentration zones, creating more vulner-
old. Although the ratio of deficient to total bridges has decreased in ability to fatigue failure. Experimental studies performed on the
recent years, a large number (56,007) are structurally deficient. application of CFRP overlays on welded joints showed significant
To rehabilitate structurally deficient steel structures, several tra-
reduction in stress concentration and crack propagation (Alemdar
ditional methods have been used, including replacing the damaged
et al. 2012; Chen et al. 2012). More recently, Yu and Wu (2017)
section of the member by steel plates or adding steel plates to
reported that the use of CFRP slowed the crack propagation in
existing members by bolting or welding. These methods are incon-
damaged beams and increased fatigue life.
venient in terms of scale and workability. Thus, there is a need for a
Existing experimental studies in the literature on the steel beams
more efficient rehabilitation scheme (Kim and Brunell 2011; Zhao
or plates subjected to fatigue loading mainly involve the use of
and Zhang 2007). In recent years, fiber-reinforced polymer (FRP)
CFRP laminate (Chen et al. 2018; Jiao et al. 2012; Kim and
composites have been commonly regarded as one of the efficient
methods to retrofit steel structures due to its high strength-to-weight Harries 2011; Yu and Wu 2017). Additionally, most investigations
ratio and resistance to corrosion (Yu and Wu 2017). on FRP strengthening of steel beams under fatigue are experimen-
Experimental studies have revealed that strengthening of steel tal, and the use of basalt FRP (BFRP) as fatigue reinforcement of
beams using carbon FRP (CFRP) laminates increases the flexural steel beams (Wu et al. 2012) is extremely limited. BFRPs are rel-
capacity and stiffness of the member (Colombi and Poggi 2006; atively new, ecofriendly materials made out of natural fibers with a
Linghoff et al. 2010). The most commonly observed failure re- significantly lower cost than carbon and with superior properties
ported is FRP laminate debonding, which occurs either at the plate over the glass fibers. BFRP composites have shown similar benefits
ends due to high interfacial stress concentration or at the plastic as CFRP in flexural reinforcement of steel beams under static load-
hinge or defect locations (Teng et al. 2012). Another more desired ing (Bagale and Parvin 2019). In the present study, application of
various FRP types and configurations to steel beams under static
1
M.S. Student, Dept. of Civil and Environmental Engineering, Univ. of and fatigue loadings using three-dimensional nonlinear finite
Toledo, Toledo, OH 43606. element analysis (FEA) was investigated. As mentioned, debond-
2
Professor, Dept. of Civil and Environmental Engineering, Univ. of ing is a major issue in FRP-strengthened steel beams and can be
Toledo, Toledo, OH 43606 (corresponding author). Email: azadeh.parvin@ delayed or avoided through anchorage by transverse FRP wrap-
utoledo.edu
ping. Therefore, detailed parametric studies were carried out to
Note. This manuscript was submitted on November 3, 2019; approved
on July 22, 2020; published online on September 25, 2020. Discussion per- investigate the effects of anchorages on the bond of CFRP-
iod open until February 25, 2021; separate discussions must be submitted strengthened steel beams and to compare the performance of CFRP,
for individual papers. This paper is part of the Practice Periodical on Struc- glass FRP (GFRP), and BFRP wraps as anchorages. Another sig-
tural Design and Construction, © ASCE, ISSN 1084-0680. nificant objective of the FEA study was to generate new S-N curves

© ASCE 04020046-1 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2021, 26(1): 04020046

 and to propose equations for fatigue life of BFRP- and aramid FRP Finite Element Analysis Approach
(AFRP)-strengthened steel beams using regression analysis.
The nonlinear finite element analysis software program ANSYS
(Mechanical APDL v18.0) was used to analyze three-dimensional
Description of Experimental Programs for FEA FRP-strengthened steel beams. For simplicity, an idealized bilinear
Beam Models elastic-perfect plastic stress versus strain curve [Fig. 3(a)] was used
to input the steel material properties. The beam failure occurred
This section summarizes the experimental programs (Jiao et al. when the steel reached its maximum value of strain (εm ) or stress
2012; Lenwari et al. 2005) adopted in FEA modeling of steel beams (σm ). SOLID185 element, which has eight nodes with three degrees
under static and fatigue loadings. of freedom in each node and supports large deflection, isotropic
For static loading, FEA beam models were based on the exper- plasticity, hyperelasticity, and large strain, was selected to model
imental study conducted by Lenwari et al. (2005). A four-point the steel beams. A similar approach was applied by other studies
Downloaded from ascelibrary.org by Auckland University Of Technology on 10/28/20. Copyright ASCE. For personal use only; all rights reserved.

bending test was performed on an 1,800-mm simply supported on steel beams (Kodur and Naser 2014; Mahmoud 2016). Fiber-
W100 × 17.2 steel beam with an 8-mm flange and 6-mm web reinforced polymer (FRP) materials were assumed to exhibit linear
thickness. A concentrated load with half of the total intensity elastic brittle failure [Fig. 3(b)] and were modeled using SOLID65
was applied at 75 mm from the midpoint of the beam on each side. to allow the possibility of failure in three directions: along the
To avoid the local failure of the compression flange, a 200-mm length, width, and thickness of the laminate. Poisson’s ratio was
wide by 12.2-mm thick steel plate was welded throughout the total assumed as 0.3. To avoid localized stress concentration, steel pads
length of the beam. The CFRP laminates were glued to the tension were placed at supports and loading locations. They were modeled
using SOLID45 having a linear elastic behavior. The modulus of
flange with a total width of 100 mm, a thickness of 1.4 mm, and
elasticity value was 200 GPa.
various lengths along the span. A schematic diagram of the beam
The effect of adhesive that glues the FRP to the substrate is vital
test is illustrated in Fig. 1.
to accurately predict the failure modes (Wang et al. 2016). A con-
For fatigue loading, FEA beam models were based on the study
tact pair comprising two element types, TARGET170 (stiff
by Jiao et al. (2012). The setup was a four-point bending test on
substrate) and CONTA174 (flexible reinforcement capable of
150UB14 steel beams with the span length of 1,400 mm and sup-
debonding and delaminating), was used to model the effects of
ports located 100 mm from each end. A notch with a 2-mm width
epoxy. The contact pair is a zero-thickness element that operates
and 22-mm height was introduced at the midspan, depicting initial
based on Coulomb’s friction model. The approach to model the
damage. Various periodic loads, representing cyclic fatigue load, interface is referred to as cohesive zone modeling (CZM) and
were applied on the compression flange at 205 mm away from was found to realistically simulate the bond between the FRP
the midspan of the beam with half of the total load intensity. A and the substrate (Korta et al. 2014). The bond properties were
75-mm-wide and 1.2-mm-thick CFRP laminate was applied to based on the bilinear bond-slip model proposed by Xia and Teng
the bottom flange with a total bond length of 200 mm (100 mm (2005) (Fig. 4), where the bond properties between steel and FRP
each side of the notch). Fig. 2 illustrates the schematic diagram materials are estimated through Eqs. (1)–(3).
of the beam test. Mechanical properties of the materials used in
the present study for static load (Lenwari et al. 2005), fatigue ta τ f
load (Jiao et al. 2012), and further parametric studies [BFRP δ1 ¼ ð1Þ
Ga
(Papanicolaou et al. 2011), AFRP (Idris and Ozbakkaloglu
2014)] are shown in Table 1. where ta = thickness; Ga = shear modulus of the adhesive; and δ 1 =
the initial slip at local-bond strength (τ f ), which is calculated as

τ f ¼ 0.8σmax ð2Þ

where σmax = the tensile strength. The value of slip at complete

debonding (δ f ) is given by
   
σmax 0.56 t0.27
a
δ f ¼ 62 ð3Þ
Ga τf
Fig. 1. Schematic diagram of the beam test for static load (all dimen-
sions in mm). Fig. 4 depicts the schematic representation of these equations.
The adhesive type and properties used in the present study are
shown in Table 1 and were obtained from Lenwari et al. (2005) and
Jiao et al. (2012). Two other parameters needed in FEA modeling
of adhesives were shear modulus (Ga ) and tensile strength (σmax ).
Ga was calculated using the elastic modulus and Poisson’s ratio of
0.3, and σmax was obtained from the product data sheet. As men-
tioned, these parameters were used to simulate the interface.
The FEA beam models were meshed with a maximum element
size of 10 mm and with denser meshing at support and loading
locations to increase the accuracy. Similarly, denser meshing in
critical regions such as supports and loading locations of the beam
was adopted (Narmashiri and Jumaat 2011). By virtue of sym-
Fig. 2. Schematic diagram of the beam test for fatigue load (all dimen-
metry, only half of the beam was modeled for static loading to re-
sions in mm).
duce analysis time [Fig. 5(a)]. However, a full beam was modeled

© ASCE 04020046-2 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2021, 26(1): 04020046

 Table 1. Mechanical properties of materials
Study Material Elastic modulus (GPa) Ultimate strain (%) Density (g=cm3 )
Steel beam (W100 × 17.2) 200 — 7.8
Static load CFRP laminate (Sika CarboDur H514) 300 >0.46 1.6
Epoxy (Sikadur-30) 11.2 >0.29 —
Steel beam (150UB14) 207.4 — 7.8
Fatigue load CFRP laminate (Sika CarboDur M512) 210 >1.35 1.6
Epoxy (Araldite 420) 1.5 >4.6 —
BFRP laminate 89 >3.15 2.7
AFRP laminate 128.5 >1.86 —
Parameters CFRP wrap (SikaWrap Hex-230C) 230 >1.5 1.8
GFRP wrap (SikaWrap Hex-100G) 72.4 >4.0 2.5
Downloaded from ascelibrary.org by Auckland University Of Technology on 10/28/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Stress versus strain curves for (a) steel; and (b) different types of FRP materials.

in the case of fatigue loading to capture the stress concentration

effect at the notch [Fig. 5(b)].

Validation of FEA Beam Models

In the present study, the FEA beam models were developed and
validated against strengthened beam specimens with three different
CFRP laminate lengths under static loading (Lenwari et al. 2005).
Generally, FRP laminates in the tension flange of the steel beam
fail either by debonding or rupture. Debonding is attained when
Fig. 4. Schematic representation of bond-slip model. the strain in the epoxy exceeds its ultimate value (Narmashiri
and Jumaat 2011). Consequently, laminate rupture occurs when

Fig. 5. FEA beam model for (a) static; and (b) fatigue loadings.

© ASCE 04020046-3 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2021, 26(1): 04020046

 Table 2. Comparison of experiment and FEA beam models results under static loading
Difference between FEA
Experiment FEA and experiment
CFRP laminate
length (mm) Load (kN) Deflection (mm) Load (kN) Deflection (mm) Load (%) Deflection (%)
500 90.2[D] N/A 94.16 — 4.39 —
650a 102.2[D] 11.5b 105.53[D] 12.09 3.26 5.13
1,200 143[R] 19b 144.2[R] 18.06 1 4.95
Note: [D] = failure by debonding; and [R] = failure by rupture.
a
Average of values from the experiment.
b
Approximate value read off the graph.
Downloaded from ascelibrary.org by Auckland University Of Technology on 10/28/20. Copyright ASCE. For personal use only; all rights reserved.

the stress value reaches the rupture stress of the FRP laminate letter A in the case of the anchorage presence, and the last letter
(Schnerch et al. 2007). Flexural load and failure mode were used indicates the type of anchorage (C:CFRP, G: GFRP, and B: BFRP).
to calibrate the FEA beam models under static loading. Each beam For example, BS500-U-C-AC represents an undamaged beam
model was considered to be validated if it failed at approximately under static loading with a 500-mm CFRP laminate bond length
the same load level and identical failure mode as those reported in anchored with CFRP at both laminate ends.
the experimental study (Lenwari et al. 2005). Table 2 shows com-
parison of the load and midspan deflection at failure along with the
failure mode of FEA beam models and test specimen results, which Control and FRP-Strengthened Beam Models under
are in good agreement with maximum discrepancy of 4.4%. Static Loading
In the fatigue load experiment performed by Jiao et al. (2012), Under static loading, the flexural behavior and failure modes of BS
the stress range (ΔS ¼ Smax − Smin ) results were obtained for a series beam models were investigated. The CFRP laminates with
maximum load for each beam specimen. A similar approach various bond lengths were applied to damaged and undamaged
was used to validate the FEA beam models. The stress range steel beams. Additionally, CFRP, GFRP, and BFRP wraps were
for each beam model under the same maximum load were com- used to anchor the CFRP laminate at both ends.
pared to their corresponding beam identified as AP in the experi-
ment (Jiao et al. 2012) and were in good agreement. The models
were considered validated since the same failure mode was ob- Comparison of Undamaged and Damaged Control and
served for each beam and the maximum discrepancy of the stress CFRP-Strengthened Steel Beams
range values between the FEA and experiment was only 4.2%. One undamaged and one damaged beam were strengthened with
Constant amplitude, fully reversed loads (Browell and Hancq CFRP laminate. The damaged beam was characterized by a notch
2006) were applied to the FEA beam models as periodic sine func- representing partial loss of section due to various conditions such as
tions with the minimum-to-maximum load ratio of 0.1. An example excessive loading and environmental exposure. The notch was cre-
of such loading is shown in Fig. 6. The cyclic load application also ated in the midspan at the bottom flange and was 2 mm wide by
has been used commonly in fatigue life testing (Chen et al. 2018; 8 mm high. The geometry and boundary conditions of the control
Hu et al. 2016; Yu and Wu 2017). Table 3 shows the comparison undamaged FEA beam model was identical to the beam reported in
between the FEA and the experimental results. Lenwari et al. (2005). Fig. 7 shows the response of control and
CFRP-reinforced beams. The 750-mm value was selected based
on a parametric study performed on bond length, which will be
Results and Discussions discussed in the following section.
Application of CFRP laminate increased the flexural load-
In this section, the results of FEA steel beam models under static
carrying capacity of the undamaged beam by almost 70%; however,
and fatigue loadings are discussed. Beam models are identified as
the elastic stiffness of control and CFRP-reinforced beams were
follows: the first letter indicates the member being a beam, the sec-
almost identical. In the case of damaged control beam, initially
ond letter identifies the load type (S for static and F for fatigue
the yielding occurred near the notch at the intersection of the flange
loading), the first number refers to the bond length, the third letter
and the web. Upon further loading, the ultimate stress and strain
signifies the presence of a notch in the beam to represent damage
were attained indicating crack formation and subsequent web fail-
(U for undamaged and D for damaged), the fourth letter is for the
ure. Fig. 8 shows the stress contour at the failure of the web. The
FRP laminate type (C: CFRP, B: BFRP, and A: AFRP), followed by
load-carrying capacity was increased threefold upon CFRP lami-
nate application to the damaged beam, which also delayed the

Table 3. Comparison of experiment and FEA beam model results under

fatigue loading
Stress range (MPa)
Maximum Difference between
load (kN) Experiment FEA FEA and experiment (%)
24 50 48.6 2.8
30 63 60.79 3.51
35 74 70.92 4.16
Fig. 6. Applied fatigue load (example case I; max load: 24 kN, 43 90 87.13 3.19
R ¼ 0.1, f ¼ 7 Hz). 56 118 116.74 1.1

© ASCE 04020046-4 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2021, 26(1): 04020046

 Load versus deflection curves
160 web failure significantly. Furthermore, the elastic stiffness was also
increased by almost 15%. The failure of the CFRP-strengthened
140 beam was the rupture of the laminate at the notch immediately fol-
120 lowed by web failure.

100
Load (KN)

Effect of CFRP Laminate Bond Length on Undamaged

80 Steel Beam
60 Another parameter considered in this study was the CFRP laminate
40 bond length. Similar to the experimental study by Lenwari et al.
(2005), the validated steel beam models with CFRP laminate bond
Downloaded from ascelibrary.org by Auckland University Of Technology on 10/28/20. Copyright ASCE. For personal use only; all rights reserved.

20 BS-D BS-U lengths of 500 and 650 mm failed by the laminate debonding from
0 BS750-D-C BS750-U-C the substrate [Fig. 9(a)], and the 1,200-mm bond length model
0 5 10 15 20 25 30 failed by the rupture of laminate [Fig. 9(b)].
Deflection (mm.) As observed, the change in failure mode was affected by the
CFRP bond length. Thus, two models with bond lengths of
Fig. 7. Load versus deflection curves of control and CFRP-reinforced
750 mm (Bagale and Parvin 2019) and 900 mm, equivalent to
beams.
2=5 and to 1=2 of the beam length, respectively, were numerically
investigated. Both beams failed by CFRP rupture. As shown in
Table 2, the CFRP-strengthened beam with a 1,200-mm bond
length had higher flexural capacity and failed by the CFRP lami-
nate rupture, while the beam models with 500-and 650-mm CFRP
laminate lengths failed by debonding at a lower flexural load capac-
ity. Therefore, Beam BS1200-U-C was selected for further compar-
ative analysis. By decreasing the laminate bond length from 1,200
to 750 and 900 mm, the reduction in flexural capacity of the beams
were insignificant (4% and 3% reduction, respectively). However,
their deflection capacities increased by 43% and 26%, respectively,
as compared to the beam strengthened with 1,200-mm CFRP
laminate. Consequently, further increases in length did not show
any significant change in the flexural capacity for laminate bond
lengths greater than 750 mm, but the ductility decreased. Therefore,
the bond length equivalent to 2=5 of simply supported beam span
length was found to be the optimum selection in this study. Fig. 10
shows the failure load versus vertical deflection of the beams with
Fig. 8. Ultimate stress in web.
various laminate bond lengths.

Fig. 9. Failure by (a) debonding; and (b) laminate rupture.

© ASCE 04020046-5 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2021, 26(1): 04020046

 Effect of Various FRP Anchorages on Flexural laminate length and was most susceptible to debonding. Commer-
Strengthening of Beams cially available SikaWrap Hex-230C and SikaWrap Hex-100G,
BS500-U-C beam model with 500-mm CFRP laminate was se- which are CFRP and GFRP wraps, respectively, were used to an-
lected to study the anchorage effect, because it had the shortest chor the beam’s laminate. A schematic representation of a beam
with anchorage is shown in Fig. 11. Sikadur-300 epoxy, recom-
mended for the SikaWrap, was used to bond the wraps to the
laminate at the bottom and to the cross-sectional sides of the
Load versus deflection steel beam. The anchorages were modeled with a ply thickness of
160
0.13 mm.
140 Fig. 12 shows the failure modes of beams without (BS500-U-C)
120 and with anchorages (BS500-U-C-AC). The presence of the
Downloaded from ascelibrary.org by Auckland University Of Technology on 10/28/20. Copyright ASCE. For personal use only; all rights reserved.

100
anchorage changed the failure mode of the beam from the laminate
Load (KN)

debonding at the ends to the rupture of the laminate at midspan.

80
Although the rupture of the anchorage initiated, the beam did
60
not completely fail until the laminate ruptured.
40 BS500-U-C BS650-U-C BS750-U-C Next, the effectiveness of the anchorage system was examined
20
BS900-U-C BS1200-U-C
by comparing the load versus deflection curves of the beams with a
0 shorter laminate length with and without anchorage, and a beam
0 5 10 15 20 25 30 with a longer laminate length without anchorage (Fig. 13). The load
Deflection (mm)
capacity of the beam with a 500-mm laminate length was enhanced
Fig. 10. Load versus deflection curves for steel beams with varying by 43% as a result of anchorage presence. Comparing the beam
CFRP laminate length. with 500-mm laminate and anchorage (BS500-U-C-AC) and the
beam with optimum laminate length of 750 mm and no anchorage
(BS750-U-C), similar flexural capacity was achieved with an
increased elastic stiffness of about 20%. Additionally, different
composite types such as CFRP, GFRP, and BFRP were applied
to BS500-U-C as anchorages. The beam with an optimum laminate
length of 750 mm (BS750-U) was also strengthened with CFRP
and BFRP laminates. The BFRP-strengthened beam showed sim-
ilar flexural capacity with slightly higher deflection capacity than
the beam strengthened with CFRP laminate. More details can be
found in Bagale and Parvin (2019). Use of BFRP wraps as anchor-
age showed similar performance as CFRP and GFRP wraps.
All types of composite materials used as wraps eliminated the
laminate debonding failure in the beam, and the failure mode
changed to laminate rupture. At the CFRP anchor rupture initiation
(115.2 kN), BFRP anchors had almost 14% higher and 5% lower
slip values than CFRP and GFRP wraps, respectively. However, the
wrap rupture of BFRP initiated much later than CFRP and was
Fig. 11. Transverse wrapping anchorage through plate ends.
almost similar to GFRP (Table 4).

Fig. 12. Failure mode of BS500-U-C and BS500-U-C-AC beams.

© ASCE 04020046-6 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2021, 26(1): 04020046

 Load Deflection 180
BS500-U-C-AC BS750-U-C BS500-U BF200-D-B BF200-D-A
160 160

140
Initial anchorage rupture 140

Stress range (MPa)

120 Laminate rupture
120
100
Load (KN)

80 100

60 R2 = 0.9553
80
40
60 R2 = 0.9564
20
Downloaded from ascelibrary.org by Auckland University Of Technology on 10/28/20. Copyright ASCE. For personal use only; all rights reserved.

0 40
0 5 10 15 20 25 30 0 100000 200000 300000 400000 500000 600000
Deflection (mm) Number of cycles (N)

Fig. 13. Load versus deflection curves of BS500-U, BS500-U-C-AC, Fig. 15. Regression analysis for S-N curves of BFRP- and AFRP-
and BS750-U-C beams. strengthened beams.

namely BFRP and AFRP, were applied to the steel beams as lam-
Table 4. Different types of FRP materials as anchorage inates. The material properties are listed in Table 1. To develop the
Anchorage Load at rupture initiation Slip at rupture S-N curves for BFRP- and AFRP-strengthened steel beams, the
type of anchorage (kN) initiation (μ mm) stress range and number of cycles to reach failure were obtained
from FEA results and the strain versus life curve, respectively.
CFRP 115.2 67.51
Regression analysis was performed to obtain the equations
GFRP 121.6 81.28
BFRP 119.4 77.34 that best fit the S-N curves of the BFRP- and AFRP-strengthened
steel beams with a coefficient of determination (R2 ) approximately
equal to 95% (Fig. 15). Fig. 16 shows the comparison between
S-N curves for CFRP- (Jiao et al. 2012), BFRP-, and AFRP-
Control and FRP-Strengthened Beam Models under strengthened FEA beam models in this study. As compared to
Fatigue Loading CFRP, the application of BFRP and AFRP to strengthen the steel
beam extended the fatigue life of the damaged beam further. How-
The strain life approach was used to predict the fatigue life of con-
ever, the BFRP- and AFRP-strengthened beams showed similar
trol and FRP-strengthened beams, named BF series. This approach
fatigue life prediction.
is suitable to predict the elastic and plastic material behaviors under
The fatigue life for a simply supported steel beam strengthened
cyclic loading. The total strain in such cases is the sum of both
by BFRP or AFRP laminates along with Araldite 420 epoxy can be
elastic and plastic strains. Further details on the strain life method
approximated using the following formulas:
can be found in Bannantine et al. (1990). The strain versus life
curve for control steel beam was generated using a similar method  
3,371 3.401
as described by Kim and Brunell (2011). As a conservative ap- N BFRP ¼ ð4Þ
ΔS
proach, the fatigue life of each material for the whole member
can be predicted based on localized points with a high stress con-  
2,827 3.52
centration such as notch roots (Wallbrink and Weiping 2010). The N AFRP ¼ ð5Þ
strain life of the control beam was generated using values of total ΔS
strain at the notch tip. Figs. 14(a and b) show the notch tip before
and after the load was applied, respectively. The crack geometry in
the present study was selected to be equivalent to Category E of CFRP AFRP BFRP

fatigue curves in AISC (2010). Two different types of fibers,

Stress Range (MPa)

100

10
1 10 100 1000 10000
Number of cycles x 103

Fig. 14. (a) Notch before loading; and (b) strain distribution at the Fig. 16. Comparison of S-N curves for CFRP-, BFRP-, and AFRP-
notch after loading. strengthened beams.

© ASCE 04020046-7 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2021, 26(1): 04020046

 where N BFRP and N AFRP = the fatigue life of BFRP- and AFRP- could be replaced by BFRP laminate in steel beam strengthen-
strengthened beams, respectively; and ΔS = the stress range for a ing for fatigue loading.
given loading. Both equations are also applicable to the linear elas-
tic fracture method (LEFM) in which the maximum stress level
cannot exceed 50% of the material yield stress (Bannantine Data Availability Statement
et al. 1990). Thus, the fatigue life curves were generated between
31 and 200 MPa stress level range, corresponding to fatigue All data, models, and code generated or used during the study
Category E threshold value and 50% steel beam’s yield stress, appear in the published article.
respectively.

Acknowledgments
Downloaded from ascelibrary.org by Auckland University Of Technology on 10/28/20. Copyright ASCE. For personal use only; all rights reserved.

Conclusions
The authors would like to thank the graduate student, Mr. Aman
In the present study, nonlinear finite element analysis was per- Dhakal, for his help in the finite element modeling of the interface
formed by validating steel beam models with experimental studies between the FRP plate and the steel beam.
under static and fatigue loads. The validated beam model under
static load was further considered to investigate the effectiveness
and optimum bond length of CFRP laminates used in strengthening References
undamaged and damaged beams. Subsequently, various types of
AISC. 2010. Specification for structural steel buildings. Chicago: AISC.
transverse anchorages (CFRP, GFRP, and BFRP) were provided
Alemdar, F., A. Matamoros, C. Bennett, R. Barrett-Gonzalez, and S. T.
at CFRP laminate ends to examine the feasibility of using BFRP Rolfe. 2012. “Use of CFRP overlays to strengthen welded connections
versus CFRP and GFRP wraps. Additionally, the validated beam under fatigue loading.” J. Bridge Eng. 17 (3): 420–431. https://doi.org
model under fatigue loading was evaluated for fatigue strengthen- /10.1061/(ASCE)BE.1943-5592.0000230.
ing of steel beams using BFRP and AFRP sheets as compared to Al-Saidy, A. H., F. W. Klaiber, and T. J. Wipf. 2004. “Repair of steel
CFRP. Regression analysis was then performed to generate S-N composite beams with carbon fiber-reinforced polymer plates.” J. Com-
curves. Consequently, formulas for predicting fatigue life in BFRP- pos. Constr. 8 (2): 163–172. https://doi.org/10.1061/(ASCE)1090-0268
and AFRP-strengthened steel beams were developed. Following (2004)8:2(163).
are the major conclusions drawn from the study: ASCE. 2017. “Infrastructure report card.” Accessed April 5, 2019. https://
www.infrastructurereportcard.org/cat-item/bridges/.
1. Results of FEA beam models for static loading (BS500-
Bagale, B. R., and A. Parvin. 2019. “FRP—Flexural strengthening of steel
U-C, BS650-U-C, and BS1200-U-C) and for fatigue loading beams: A finite element analysis study.” In Proc., ISEC 2019—10th
(BF410-D-C) were in good agreement with less than 5% dis- Int. Structural Engineering and Construction Conf. Fargo, ND:
crepancy as compared to their counterpart beam test specimens International Structural Engineering and Construction Press.
with similar observed failure modes. Bannantine, J. A., J. J. Comer, and J. L. Handrock. 1990. Fundamentals of
2. CFRP strengthening of undamaged and damaged steel beams metal fatigue analysis. Upper Saddle River, NJ: Prentice Hall.
showed increases in flexural load capacity by 1.7 and 3 times, Browell, B. R., and A. Hancq. 2006. Calculating and displaying fatigue
respectively. Further, the elastic stiffness of undamaged beam results. Canonsburg, PA: ANSYS.
remained fairly unchanged; however, there was a 15% increase Chen, T., X. L. Gu, M. Qi, and Q. Q. Yu. 2018. “Experimental study on
in the damaged beam. fatigue behavior of cracked rectangular hollow-section steel beams re-
paired with prestressed CFRP plates.” J. Compos. Constr. 22 (5):
3. In the undamaged steel beam under static loading strengthened
04018034. https://doi.org/10.1061/(ASCE)CC.1943-5614.0000872.
with varying CFRP laminate lengths, the optimum bond length Chen, T., Q. Q. Yu, X. L. Gu, and X. L. Zhao. 2012. “Study on fatigue
was found to be approximately 2=5 of the simply supported behavior of strengthened non-load-carrying cruciform welded joints us-
beam length. ing carbon fiber sheets.” Int. J. Struct. Stab. Dyn. 12 (1): 179–194.
4. Transverse wrap anchorages at laminate ends of Beam BS500- https://doi.org/10.1142/S0219455412004586.
U-C (500-mm laminate length with no anchorage) improved the Colombi, P., and C. Poggi. 2006. “An experimental, analytical and numeri-
flexural capacity by 43%. Furthermore, the failure mode was cal study of the static behavior of steel beams reinforced by pultruded
changed from laminate end debonding to FRP rupture. CFRP strips.” Composites, Part B 37 (1): 64–73. https://doi.org/10
5. Beams BS500-U-C-AC (500-mm laminate length with .1016/j.compositesb.2005.03.002.
anchorage) and BS750-U-C (750-mm laminate length with no Hu, L. L., X. L. Zhao, and P. Feng. 2016. “Fatigue behavior of cracked
high-strength steel plates strengthened by CFRP sheets.” J. Compos.
anchorage) had almost the same failure load values. However,
Constr. 20 (6): 04016043. https://doi.org/10.1061/(ASCE)CC.1943
BS500-U-C-AC had a slightly higher elastic stiffness. This -5614.0000698.
means that if the optimum bond length is not attainable Idris, Y., and T. Ozbakkaloglu. 2014. “Flexural behavior of FRP-HSC-steel
due to limited accessibility or site constraints, shorter bond composite beams.” Thin Walled Struct. 80 (Jul): 207–216. https://doi
length with anchorages can provide similar benefits of beam’s .org/10.1016/j.tws.2014.03.011.
strengthening. Jiao, H., F. Mashiri, and X. L. Zhao. 2012. “A comparative study on fatigue
6. BFRP is comparable to commercially available versions of behaviour of steel beams retrofitted with welding, pultruded CFRP
CFRP and GFRP wraps. plates and wet layup CFRP sheets.” Thin Walled Struct. 59 (Oct):
7. Equations for fatigue life of the BFRP- and AFRP-strengthened 144–152. https://doi.org/10.1016/j.tws.2012.06.002.
Kalfat, R., R. Al-Mahaidi, and S. T. Smith. 2013. “Anchorage devices
steel beams were developed using regression analysis. The
used to improve the performance of reinforced concrete beams
fatigue life could be predicted accurately using these equations retrofitted with FRP composites: State-of-the-art review.” J. Compos.
as long as the same combination of FRP and adhesive types Constr. 17 (1): 14–33. https://doi.org/10.1061/(ASCE)CC.1943-5614
is used. .0000276.
8. BFRP-strengthened beam had higher fatigue life as compared to Kim, Y. J., and G. Brunell. 2011. “Interaction between CFRP-repair and
the one strengthened with CFRP laminate. Therefore, the CFRP initial damage of wide-flange steel beams subjected to three-point

© ASCE 04020046-8 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2021, 26(1): 04020046

 bending.” Compos. Struct. 93 (8): 1986–1996. https://doi.org/10.1016/j Schnerch, D., M. Dawood, S. Rizkalla, and E. Sumner. 2007. “Proposed
.compstruct.2011.02.024. design guidelines for strengthening of steel bridges with FRP materi-
Kim, Y. J., and K. A. Harries. 2011. “Fatigue behavior of damaged steel als.” Constr. Build. Mater. 21 (5): 1001–1010. https://doi.org/10.1016/j
beams repaired with CFRP strips.” Eng. Struct. 33 (5): 1491–1502. .conbuildmat.2006.03.003.
https://doi.org/10.1016/j.engstruct.2011.01.019. Teng, J. G., T. Yu, and D. Fernando. 2012. “Strengthening of steel struc-
Kodur, V. K. R., and M. Z. Naser. 2014. “Effect of shear on fire response of tures with fiber-reinforced polymer composites.” J. Constr. Steel Res.
steel beams.” J. Constr. Steel Res. 97 (Jun): 48–58. https://doi.org/10 78 (Nov): 131–143. https://doi.org/10.1016/j.jcsr.2012.06.011.
.1016/j.jcsr.2014.01.018. Wallbrink, C., and H. Weiping. 2010. “A strain-life module for CGAP:
Korta, J., A. Młyniec, P. Zdziebko, and T. Uhl. 2014. “Finite element analy- Theory, user guide and examples.” DSTO-TR-2392. Canberra,
sis of adhesive bonds using the cohesive zone modeling method.” Mech. Australia: Defence Science and Technology Organisation.
Control 33 (2): 51. https://doi.org/10.7494/mech.2014.33.2.51. Wang, H. T., G. Wu, Y. T. Dai, and X. Y. He. 2016. “Determination of the
Lenwari, A., T. Thepchatri, and P. Albrecht. 2005. “Flexural response of bond-slip behavior of CFRP-to-steel bonded interfaces using digital im-
steel beams strengthened with partial-length CFRP plates.” J. Compos. age correlation.” J. Reinf. Plast. Compos. 35 (18): 1353–1367. https://
Downloaded from ascelibrary.org by Auckland University Of Technology on 10/28/20. Copyright ASCE. For personal use only; all rights reserved.

Constr. 9 (4): 296–303. https://doi.org/10.1061/(ASCE)1090-0268 doi.org/10.1177/0731684416651342.

(2005)9:4(296). Wu, G., H.-T. Wang, Z.-S. Wu, H.-Y. Liu, and Y. Ren. 2012. “Experimental
Linghoff, D., M. Al-Emrani, and R. Kliger. 2010. “Performance of steel study on the fatigue behavior of steel beams strengthened with different
beams strengthened with CFRP laminate—Part 1: Laboratory tests.” fiber-reinforced composite plates.” J. Compos. Constr. 16 (2): 127–137.
Composites, Part B 41 (7): 509–515. https://doi.org/10.1016/j https://doi.org/10.1061/(ASCE)CC.1943-5614.0000243.
.compositesb.2009.05.008. Xia, S. H., and J. G. Teng. 2005. “Behaviour of FRP-to-steel bonded
Mahmoud, A. M. 2016. “Finite element modeling of steel concrete beam joints.” In Proc., Int. Symp. on Bond Behaviour of FRP in Structures,
considering double composite action.” Ain Shams Eng. J. 7 (1): 73–88. BBFS 2005, edited by J. F. Chen and J. G. Teng, 411–418. Kingston,
https://doi.org/10.1016/j.asej.2015.03.012. ON, Canada: International Institute for FRP in Construction.
Narmashiri, K., and M. Z. Jumaat. 2011. “Reinforced steel I-beams: A Yu, Q. Q., and Y. F. Wu. 2017. “Fatigue strengthening of cracked steel
comparison between 2D and 3D simulation.” Simul. Modell. Pract. beams with different configurations and materials.” J. Compos. Constr.
Theory 19 (1): 564–585. https://doi.org/10.1016/j.simpat.2010.08.012. 21 (2): 04016093. https://doi.org/10.1061/(ASCE)CC.1943-5614
Papanicolaou, C., T. Triantafillou, and M. Lekka. 2011. “Externally bonded .0000750.
grids as strengthening and seismic retrofitting materials of masonry Zhao, X. L., and L. Zhang. 2007. “State-of-the-art review on FRP strength-
panels.” Constr. Build. Mater. 25 (2): 504–514. https://doi.org/10 ened steel structures.” Eng. Struct. 29 (8): 1808–1823. https://doi.org/10
.1016/j.conbuildmat.2010.07.018. .1016/j.engstruct.2006.10.006.

© ASCE 04020046-9 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2021, 26(1): 04020046