Construction and Building Materials xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Fatigue durability of cracked steel beams retrofitted with high-strength

materials
Qian-Qian Yu a,⇑,1, Yu-Fei Wu b,1
a
Department of Structural Engineering, Tongji University, Shanghai, China
b
School of Engineering, RMIT University, Melbourne, Australia

h i g h l i g h t s

Mechanically strengthened steel beams were tested under fatigue loading.

SIF at the crack front was theoretically and numerically analyzed.

Crack constraining effect on the SIF is non-negligible in analytical solution.

a r t i c l e i n f o a b s t r a c t

Article history: Steel structures under dynamic loading are prone to fatigue damage and a considerable number of metal-
Received 14 July 2017 lic infrastructure in the world are structurally deficient. The fatigue behavior of defected steel beams
Received in revised form 10 September mechanically retrofitted by different high-strength materials was experimentally studied in this work.
2017
The steel beam was I section with an artificial cut at the mid-span and tested under fatigue loading.
Accepted 12 September 2017
Available online xxxx
The specimens were strengthened on the tension flange by carbon fiber reinforced polymer (CFRP) lam-
inate, high strength steel (HSS) plate or SafStrip (SAF) plate. Three mechanical anchorage schemes were
designed to test their efficiencies. The fatigue crack propagation was recorded to explore the crack
Keywords:
High strength material
growth rate. Strain gauges were installed on the strengthening material to monitor the strain develop-
Fatigue ment. The stress intensity factor (SIF) at the crack front was calculated based on analytical solution
Steel beam and numerical simulation; then, the results were compared with experimental findings. It was demon-
Strengthening strated that the strengthening significantly increased the fatigue life of the beams. The attached strength-
Stress intensity factor ening materials not only globally shared a portion of the far-field load on the cracked steel section but
Crack constraint also provided a local constraining effect, both of which reduced the crack mouth opening displacement
(CMOD) and SIF. The different CMODs resulted from the different retrofitting configurations indicated
significant variation in the crack constraining effect caused by various retrofitting systems, which cannot
be neglected in analytical modeling.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Fiber reinforced polymer (FRP) material initially applied to air-

craft structures, has a high strength-to-weight ratio, excellent cor-
Metallic structures are prone to fatigue damage at the zone of rosion resistance and fatigue performance. It is now used to
stress concentration. Cracks may initiate and propagate, even lead- strengthen structurally deteriorated infrastructure and to build full
ing to catastrophic accidents. Traditional retrofitting techniques or hybrid FRP structures in civil engineering.
cut and replace or attach extra steel plates through welding or bolt Extensive experimental studies on cracked steel plates con-
joints, which may introduce new stress concentration or firmed that external attachment using carbon FRP (CFRP) could
imperfections. effectively retard crack propagation and extend fatigue life [1–8].
Research on CFRP strengthened steel beams was also reported,
which further investigated the feasibility of the retrofitting
⇑ Corresponding author. approach [9–18]. Special attention was paid to exam the effect of
E-mail address: [email protected] (Q.-Q. Yu). drilled holes which are normally found in riveted metallic girders
1
Formerly Department of Architecture and Civil Engineering, City University of on the strengthening scheme [19–21].
Hong Kong, China.

https://doi.org/10.1016/j.conbuildmat.2017.09.051
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Q.-Q. Yu, Y.-F. Wu, Fatigue durability of cracked steel beams retrofitted with high-strength materials, Constr. Build.
Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.051
2 Q.-Q. Yu, Y.-F. Wu / Construction and Building Materials xxx (2017) xxx–xxx

The stress intensity factor (SIF) is a dominant parameter in fati- critical for prediction of fatigue behavior. The contribution of the
gue analysis, which determines the crack propagation rate and mounted overlay was considered equivalent to an external com-
consequently the fatigue life of a structure. Evaluation of the SIF pressive force applied to the tension flange. By comparing with
at the crack front of retrofitted components could help to estimate the experimental findings, it was found that this equivalence
the strengthening efficiency and to give some optimal suggestions ignored the local effect of the crack restraint, which led to an over-
on the repair method. estimation of the SIF especially those with strong constraint from
With respect to cracked steel plates, classical theoretical solu- the anchorage configuration. It was concluded that the local effect
tions of SIF at the crack front are available in handbooks for the of crack restraint is non-negligible in the analytical solution.
cases without strengthening. Numerical simulation based on the
finite element method [22–24] and boundary element method 2. Brief summary of experimental program
[25–27] as well as theoretical analysis [28,29] was adopted to pre-
dict the SIF in CFRP bonded specimens. Parametric analysis was Six notched steel beams were tested under fatigue loading in
carried out to examine the effects of overlays’ mechanical proper- the Heavy Structures Testing Lab at City University of Hong Kong,
ties and bond configurations [30,31]. In regard to steel beams with China, among which, two were unstrengthened as control speci-
a mid-span crack, the research on SIF at the crack front is less mens, and the other four were strengthened with high-strength
reported. Kienzler and Herrmann [32] presented formulae for materials [13].
cracked steel beams with rectangular and circular cross sections
based on the elementary beam theory. Dunn et al. [33] later
2.1. Configuration of specimens
extended the solution to I-beams with a crack going through the
beam flange and into the web. A nondimensional parameter in
I-shaped steel beams were used in the experimental study as
the expression was derived from dimensional analysis and numer-
shown in Fig. 1. The cross section was 150 mm high and 75 mm
ical simulation. Recently, Ghafoori and Motavalli [34] forwarded an
wide; the thicknesses of the flange and the web were 7 mm and
analytical approach of SIF for cracked I-beams based on the
5 mm, respectively. The mid-span of the specimen was artificially
approach of crack surface widening energy release rate. The
wire-electrode cut to simulate the initial damage and to guide
method was verified by comparison with the aforementioned the-
the crack propagation. The slot was 15 mm long and 0.3 mm wide,
oretical analysis [32,33].
which went through the beam flange and a portion of the beam
In Hmidan et al. [35], a total of 1240 three-dimensional finite
web (Fig. 1b). All the specimens were manufactured in one batch
element models of wide-flange steel girders were developed to cal-
and therefore, it was assumed that they had the same initial state.
culate the SIF at the crack front. A correction factor composed of
For the strengthened scenarios, specimens were attached with
the notch-depth-to-height ratio, CFRP-to-steel area ratio, and
high strength materials 800 mm long and 50 mm wide on the ten-
CFRP-to-steel modular ratio was obtained for the CFRP-
sion flange.
strengthened beams.
Colombi and Fava [12] conducted a series of fatigue tests on I-
2.2. Strengthening schemes and material properties
beams repaired by CFRP strips and revealed a debonded area
around the crack front, which was detrimental to the strengthen-
The steel beams were Grade Q345b according to Chinese Stan-
ing efficiency. With the debonded region introduced into the
dard GB 50017-2003 [37]. Tensile coupon tests were performed to
numerical analysis, the strain on CFRP compared well with the
determine the mechanical properties (Australia Standard AS 1391
experimental findings. More recently, Colombi and Fava [36] pro-
[38]). The measured average yield strength, ultimate tensile
posed an analytical model to predict the strain on the bonded CFRP
strength, and Young’s modulus were 378.2 MPa, 519.0 MPa, and
material as well as the fatigue crack propagation. A good agree-
192.8 GPa, respectively.
ment was achieved between the theoretical results and the test
The CFRP laminates (BASFÒ Masterbrace Lam CF210/3000) used
data.
in this experiment were provided in rolls of 50 mm wide with a
The authors of the present paper conducted a preliminary study
nominal thickness of 1.4 mm. The Young’s modulus, ultimate ten-
on damaged steel beams strengthened by high strength materials
sile strength and ultimate strain at fracture were determined
[13]. Both adhesive bonding and mechanical anchorage were
according to ASTM D3039 [39], i.e. 200.4 GPa, 3022.4 MPa, and
investigated and the test results demonstrated that the strengthen-
15,081 le, respectively.
ing scheme was effective to reduce crack propagation rate and to
Since the CFRP strip is brittle, it could not be directly fastened to
prolong fatigue life. This paper performed theoretical and numeri-
the steel beam. Two anchorage schemes were designed, as illus-
cal analyses on the SIF at the crack front of the specimens, which is
trated in Figs. 2a and b. In configuration A, the anchor plate was

205 205
7

Notch

5
136
150

50 300 50 50 300 50
800
15
7

1200
1400 75

(a) Strengthened specimen (b) Notch details

Fig. 1. Specimen geometry (dimension in mm, not to scale).

Please cite this article in press as: Q.-Q. Yu, Y.-F. Wu, Fatigue durability of cracked steel beams retrofitted with high-strength materials, Constr. Build.
Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.051
 Q.-Q. Yu, Y.-F. Wu / Construction and Building Materials xxx (2017) xxx–xxx 3

CFRP CFRP HSS/SAF

r4 r4

15 20 15
30

50
8.5 58 8.5
HSS
75 21 83 21
125

CFRP CFRP
SAF

(a) (b) (c)

Fig. 2. Anchorage configurations (dimension in mm, not to scale): (a) anchorage scheme A for CFRP; (b) anchorage scheme B for CFRP; (c) anchorage scheme C for HSS and
SAF.

75 mm wide, which was equal to the beam flange. This anchor distance between the loading points was 410 mm. Two compres-
plate was mounted to the specimen by two bolts through the sive cyclic loading spectrums applied to the spread beam were
bottom flange whereas in configuration B, a larger anchor plate 2–20 kN and 3.5–35 kN. The maximum loads were approximately
with more bolts was used, by which the tension flange was kept 36 and 63% of the calculated yield load of the control beam, respec-
intact. tively, based on the midspan notched cross section, which was
Besides CFRP laminates, two other materials were also selected, comparative to those reported in the references [9,11,12].
i.e. high strength steel (HSS) plate and SafStrip (SAF) plate, both of The load frequency and stress ratio R (minimum load/maximum
which were drillable. The HSS plate had a nominal thickness of load) were 6 Hz and 0.1, respectively.
1 mm. Coupon tests showed that it had a much higher yield Crack propagation with fatigue cycles of Beams 1 and 2 was
strength (978 MPa), ultimate tensile strength (1076 MPa) and ulti- recorded by the beach marking technique, while that of Beams 3
mate strain (25,247 le) than normal steel, and the Young’s modu- to 6 were detected by strain gauges attached to the potential crack
lus was 198 GPa [40]. The SAF plate is a combined fiber patch path on the beam web. Details of the beaching marking method
wherein one layer of unidirectional carbon is covered by E-glass and arrangement of strain gauges refer to Yu and Wu [13]. Static
fibers and E-glass strand mats, which has a high bearing strength loads from zero to the maximum load of the fatigue loading spec-
[41]. It had a nominal thickness of 3.2 mm and the tested Young’s trum were inserted to the fatigue load at the beginning of the test
modulus, ultimate tensile strength, and ultimate strain were and after a certain number of fatigue cycles. During these static
65.8 GPa, 1071 MPa, and 16,277 le, respectively [40]. Attributed loads, the crack mouth opening displacement (CMOD) was
to the drillable feature, the HSS plate and SAF plate were fastened recorded by two clip-shape displacement transducers installed at
to the steel beam by anchorage configuration C, as shown in Fig. 2c. each end of the initial cut on the tension soffit, and the strain dis-
Each strengthened beams were installed with four anchors, tribution along the high strength materials was monitored by
among which two were away from the initial crack for 50 mm strain gauges. A typical layout of strain gauges along the strength-
(the distance between the centerlines of the anchor and the crack) ening patch of Beam 3 is displayed in Fig. 3. Due to different retro-
and the other two were fixed at the ends of the strengthening fitting schemes, the arrangement of strain gauges on the overlays
material (Fig. 1). In the anchorage system, all the bolts had the of Beams 4 to 6 was slightly different.
same diameter of 8 mm. The screws were driven in by a torque
wrench with a torque moment of 56 kNm. One layer of extra 2.4. Test results
fine sand paper (400 Cw) was placed between the CFRP and steel
to avoid stress concentration and consequent splitting of this The initial slot propagated towards the compression flange as
brittle material. the repeated loading was applied. The fatigue life was identified
as the cycle number when the crack reached the tip of the web,
2.3. Fatigue loading and measurements and the specimen ruptured. The experimental program and results
are presented in Table 1. Note that the strengthening of Beam 4
All the beams were loaded under a four-point bending rig. As kept intact after 2 million cycles and the test was stopped. Beam
depicted in Fig. 1, the span of the beam was 1200 mm, and the 3 failed due to the fracture of one bolt and consequently became

Please cite this article in press as: Q.-Q. Yu, Y.-F. Wu, Fatigue durability of cracked steel beams retrofitted with high-strength materials, Constr. Build.
Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.051
4 Q.-Q. Yu, Y.-F. Wu / Construction and Building Materials xxx (2017) xxx–xxx

Anchor Anchor Anchor Anchor

Steel
CFRP
tension
flange

100 75 25 24 52 24 24 52 24 25 75 100

Fig. 3. Typical layout of strain gauges along the strengthening patch (Beam 3, dimensions in millimeters, not to scale).

Table 1
Experimental program and results.

Specimen High strength material Patch configuration DF (kN) Fatigue life Extension ratio
Beam 1 – – 2–20 71,259 –
Beam 2 – – 3.5–35 16,700 –
Beam 3 CFRP Anchorage A 3.5–35 263,259 15.8
Beam 4 CFRP Anchorage B 3.5–35 2,000,000 119.8
Beam 5 HSS Anchorage C 3.5–35 128,668 7.7
Beam 6 SAF Anchorage C 3.5–35 70,198 4.2

invalid in terms of fatigue life of the retrofitting system. In Beams 5 3. Calculation of stress intensity factor
and 6, the HSS plate and SAF plate fractured, respectively.
The extension ratio of fatigue life was obtained by the fatigue 3.1. Stress intensity factor of plain steel beams without strengthening
life of a strengthened specimen divided by that of a corresponding
unstrengthened one, as listed in Table 1. Clearly, the fatigue life Dunn et al. [33] proposed a closed-form expression for the SIF of
was significantly prolonged with the presence of the strengthening cracked I-beams with a crack propagated through the flange and
systems. The fatigue life improvement was comparable to that into the web when subjected to bending (Eq. (1))
with the adhesive bonding system [13], which confirmed the effi- sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
ciency of the mechanical anchorage schemes. bM Is
The crack length versus fatigue cycles of all the specimens is KM ¼M 1 ð1aÞ
I
Is  tw Icr
plotted in Fig. 4. The horizontal axis represents the fatigue cycle
numbers and the vertical axis denotes the crack propagation length
measured from the tension soffit to the crack front. The asterisk for bM ¼ 1:16n0:374
1 ð1bÞ
Beam 4 denotes that it did not fracture after 2 million cycles. It was
clearly observed that the strengthening significantly delayed the n1 ¼ a=h ð1cÞ
crack propagation and extended the fatigue life. The best perfor-
mance was achieved by the CFRP strengthening with anchorage where M is the bending moment; bM is a nondimensional parame-
configuration B. ter; Is and Icr are the moments of inertia of the uncracked I-beam
In addition to the crack propagation with fatigue cycles, the and the remaining cross section, respectively; tw is the thickness
CMOD against the crack propagation length was also monitored of the beam web; n1 is defined by the ratio of the crack length to
and is displayed in Fig. 5. In comparison with Beam 2 where the the beam height, indicating the degree of damage.
CMOD increased more and more quickly as the crack grew, The SIF results derived from the experimental findings were
the CMOD variations in Beams 3 and 6 occurred at a much slower obtained by using the ‘‘secant method” as recommended by ASTM
pace. E647-11e1 [42]. The average crack growth rate over the crack

150 3.5
Crack mouth opening displacment (mm)

Beam 1
Crack propagation length (mm)

Beam 2
3.0 Beam 3
120 Beam 4
Beam 5
2.5
Beam 6
Displacement
90 transducer
2.0

60 1.5
Beam 1
Beam 2 1.0
Beam 3
30 Beam 4*
Beam 5 0.5
Beam 6

0 0.0
0.0 0.1 0.2 0.3 0.4 0.5 0 30 60 90 120 150
Fatigue cycle numbers (million) Crack propagation length (mm)

Fig. 4. Crack propagation versus fatigue cycles. Fig. 5. CMOD versus crack propagation length.

Please cite this article in press as: Q.-Q. Yu, Y.-F. Wu, Fatigue durability of cracked steel beams retrofitted with high-strength materials, Constr. Build.
Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.051
 Q.-Q. Yu, Y.-F. Wu / Construction and Building Materials xxx (2017) xxx–xxx 5

length increment (ai+1  ai) is calculated by Eq. (2). Afterwards, the Good agreement was achieved among the results obtained by
SIF at the averaged crack length a ¼ ðaiþ1 þ ai Þ=2 could be obtained these three approaches for both Beams 1 and 2, i.e., analytical solu-
by the Paris Law [43] (Eq. (3)) tion, test data and numerical results.

ðda=dNÞa ¼ ðaiþ1  ai Þ=ðNiþ1  Ni Þ ð2Þ 3.2. Stress intensity factor of retrofitted steel beams

   1=n The analytical solution of SIF for a bare steel beam under bend-
1 da
ðDK eff Þa ¼ ð3Þ ing was extended to a scenario purely loaded by an axial force [34],
C dN a
as expressed by Eq. (6)
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
DK eff bN As
DK ¼ ð4Þ K NI ¼N 1 ð6Þ
U As  tw Acr

 ¼ ðaiþ1 þ ai Þ=2
a ð5Þ where N is the axial force applied to the cracked section; As and Acr
are the cross-sectional areas of the uncracked I-beam and the
where a is the crack length; N is the number of fatigue cycles; DKeff remaining cross section, respectively; bN is also a nondimensional
is the effective SIF range at the crack front; C and n are material con- parameter.
stants that are determined experimentally, which are taken as Colombi and Fava [36] further extended the solution to cracked
3.98  1013 and 2.88, respectively (da/dN in mm/cycle and DK in steel beams with CFRP bonded to the tension flange. The contribu-
MPamm1/2), according to the recommendation of British Standards tion of the retrofitting overlay was considered as a compressive
Institution (BS7910:2005) [44]; DK is the SIF range at the crack force transferred by the composite patch. Since the compressive
front; U is a factor considering the crack closure effect, which is force was applied to the bottom soffit of the steel beam, it was
taken as 0.9 as reported in Hosseini et al. [45]. equivalent to an axial compressive force and a bending moment
The SIF ranges of Beams 1 and 2 calculated based on Eqs. (2)–(5) acted at the centroid of the cracked section.
are given in Fig. 6. The results were compared with the test data as Therefore, the SIF at the crack front of a steel beam with retro-
well as the numerical results of finite element analysis carried out fitting on the tensile flange is given by Eq. (7) as
by the authors [46].
)
1/2

4000
10000
Y Stress intensity factor range (MPa· mm

3500
8000
3000
Strain (με)

6000 2500

2000
4000
1500
Beam 3
2000 Analytical solution Beam 4
1000 Beam 5
Test data
Numerical results Beam 6

0 500
0 30 60 90 120 150 0 30 60 90 120 150
Crack propagation length (mm) Crack propagation length (mm)
(a) Beam 1 (a) Strain versus crack propagation length
Y Stress intensity factor range (MPa· mm )
1/2

40
10000

8000
30
Axial force (kN)

6000
20

4000

10 Beam 3
2000 Analytical solution Beam 4
Test data Beam 5
Numerical results Beam 6
0 0
0 30 60 90 120 150 0 30 60 90 120 150
Crack propagation length (mm) Crack propagation length (mm)
(b) Beam 2 (b) Axial force versus crack propagation length

Fig. 6. Stress intensity factor ranges of Beams 1 and 2. Fig. 7. Strain and axial force on the retrofitting patch.

Please cite this article in press as: Q.-Q. Yu, Y.-F. Wu, Fatigue durability of cracked steel beams retrofitted with high-strength materials, Constr. Build.
Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.051
6 ) Q.-Q. Yu, Y.-F. Wu / Construction and Building Materials xxx (2017) xxx–xxx

)
1/2

6000

1/2
6000
Stress intensity factor range (MPa· mm

Stress intensity factor range (MPa· mm

Theoretical solution Theoretical solution
Test data Test data
5000 5000

4000 4000

3000 3000

2000 2000

1000 1000

0 0
0 30 60 90 120 150 0 30 60 90 120 150
Crack propagation length (mm) Crack propagation length (mm)
(a) Beam 3 (c) Beam 5
)

)
1/2

6000

1/2
6000
Stress intensity factor range (MPa· mm

Stress intensity factor range (MPa· mm

Theoretical solution Theoretical solution
Test data Test data
5000 5000

4000 4000

3000 3000

2000 2000

1000 1000

0 0
0 30 60 90 120 150 0 30 60 90 120 150
Crack propagation length (mm) Crack propagation length (mm)
(b) Beam 4 (d) Beam 6

Fig. 8. Stress intensity factor ranges of Beams 3 to 6.

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
bM Is bN As length. In Beam 5, the theoretical solution compares quite well
K I ¼ ðM  N  ðy þ t=2ÞÞ 1 N 1 with the test data, while in Beam 6, the general variation trend is
Is  t w Icr As  tw Acr
similar between the two curves.
ð7Þ The deviation was mainly due to the local effect of crack con-
straint, which was not taken into consideration in Eq. (7). For the
where y is the distance between the top of attachment and centroid defected steel beam strengthened on the tension flange, the
of the cracked cross section; t is the thickness of the high strength strengthening patch helps to share the far-field load and also
material; bN here is assumed to be equal to bM. serves to constrain the crack opening, both of which reduce the
For the strengthened specimens, strain gauges were mounted to SIF at the crack front, leading to an improved fatigue behavior. Tak-
the strengthening materials to record the development of tensile ing Beams 5 and 6 as an example: At the same crack length, the
strain distribution with fatigue cycles (Fig. 3). Based on the strain axial forces of the HSS plate and SAF plate at a certain crack length
reading at the section beneath the crack on the strengthening were approximately the same (Fig. 7), leading to a similar SIF
patch under the maximum load (35 kN) as shown in Fig. 7a, the according to the analytical solution (Fig. 9a). However, the SIF
corresponding tensile force is presented in Fig. 7b. Consequently, results of Beams 5 and 6 derived from the experimental results
the SIF ranges according to Eq. (7) are calculated and compared graphed by Fig. 9b clearly show that the two are different from
with the experimental findings in Fig. 8. It should be pointed out each other and the difference increased with the crack length.
the horizontal coordinates of the data points are not exactly the The variation tendency of the deviation was proportional to that
same which is because the average crack length was adopted for of the CMOD values of Beams 5 and 6 (Fig. 5).
the theoretical analysis as indicated in Eq. (5). Numerical analysis was also conducted to further validate the
The deviation between the analytical solution and test results is influence of crack opening on the SIF. Three-dimensional finite ele-
much more significant in the cases with strengthening. Especially ment models were developed by using the commercial software
for Beams 3 and 4, which were strengthened by CFRP laminates, ABAQUS. Due to the symmetry of the specimen geometry and
the SIF is dramatically overestimated by the theoretical approach the load condition, only one-quarter of the model was considered,
and the deviation considerably increases with the increase of crack as displayed in Fig. 10a. The specimen configuration and dimen-

Please cite this article in press as: Q.-Q. Yu, Y.-F. Wu, Fatigue durability of cracked steel beams retrofitted with high-strength materials, Constr. Build.
Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.051
 Q.-Q. Yu, Y.-F. Wu / Construction and Building Materials xxx (2017) xxx–xxx 7

front element size l was set as one-fifth of the crack length a to cre-
)
1/2

6000
ate a relatively fine mesh [46]. The detailed mesh in the vicinity of
Stress intensity factor range (MPa· mm

the crack is depicted in Fig. 10b.

5000
The external load transferred from the actuator of the test
machine was applied to the corresponding position of the beam
4000 model. The force, N, attributed to the strengthening system, was
applied as an external compression load to the centroid of the
3000 cracked cross section. The value of N was calculated based on the
strain at the section beneath the crack on the strengthening patch
2000 under the maximum load (35 kN), as shown in Fig. 7. A bending
moment expressed as N(y + t/2) due to the shift of the point of
1000 application was also applied as an external load.
Beam 5 Symmetric boundary conditions were set on the corresponding
Beam 6
surfaces and the crack surface was left free. To investigate the local
0
0 30 60 90 120 150 effect of crack constraint on the SIF, two extreme cases were stud-
Crack propagation length (mm) ied as shown in Fig. 11: (I) the crack opening was set free, (II) the
CMOD was set as zero at the tension soffit, which implied that the
(a) Theoretical results
attached strengthening material is fully attached to the bottom of
the beam without any interfacial slip.
)

6000
1/2

The SIF results obtained from the theoretical solution, experi-

Stress intensity factor range (MPa· mm

mental results, and numerical simulation with boundary condi-

5000 tions (I) and (II) of Beams 3 to 6 are plotted in Fig. 12. The
variation trends of the curves with respect to the crack propagation
4000 length in scenarios with different boundary conditions were appar-
ently different, which confirmed the local effect of the crack open-
3000 ing on the stress field around the crack front.
It was found that the theoretical results by Eq. (7) and numeri-
cal analysis with boundary condition (I) were approximately the
2000
same, and the maximum difference was 25.6%. The test data is
located between the numerical results with the two boundary con-
1000 ditions. The SIF ranges of different specimens obtained from the
Beam 5
Beam 6 experiments showed different variation tendencies. In Beams 3
0 and 4, it decreased with the crack length while the value kept fairly
0 30 60 90 120 150
constant in Beam 5, and increased in Beam 6 (Fig. 12). It was, there-
Crack propagation length (mm) fore, demonstrated that in Beams 3 to 6, the local effect of crack
(b) Experimental results restraint had various degrees of contribution to the reduction of
SIF, which was consistent with the different CMODs observed in
Fig. 9. Comparison of stress intensity factor ranges between Beams 5 and 6.
Fig. 5. Again taking Beams 5 and 6 as an example, for the crack
length developed from 15 mm to 75 mm, the finite element models
had the same geometry and dimensions due to the same crack
sions were all according to the experimental program. In the length, and the applied load were also approximate (Fig. 7b). Con-
numerical simulation, only bare steel beams were adopted since sequently, similar COMD and SIF of Beams 5 and 6 are expected.
the contribution of the strengthening patch was regarded as an However, Figs. 5 and 9b clearly indicate the difference of CMOD
external compressive load. The steel was considered below the and SIF between these two specimens, respectively.
yield strength under the fatigue loading; therefore, a linear elastic The difference between the analytical solution and the test
constitutive relationship was adopted with the Young’s modulus results is caused by the local restraining effect of the retrofitting
and Poisson’s ratio of 192.8 GPa and 0.3, respectively. Three- material, or the local stress field due to CMOD. The analytical solu-
dimensional quadratic brick elements (C3D20R) were selected. tion is equivalent to the case with an external force N applied at the
Around the crack front, 15-node wedge elements were adopted bottom soffit of the beam. If we move the same force along the bot-
to simulate the stress and strain singularity; besides, the crack tom soffit to the mid-span, it produces a local stress field around

Vertical load
l

Crack
front

Bending moment

Horizontal load Support

(a) One-quarter model (b) Crack front mesh
Fig. 10. Finite element modeling.

Please cite this article in press as: Q.-Q. Yu, Y.-F. Wu, Fatigue durability of cracked steel beams retrofitted with high-strength materials, Constr. Build.
Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.051
8 Q.-Q. Yu, Y.-F. Wu / Construction and Building Materials xxx (2017) xxx–xxx

Vertical load

Horizontal load Horizontal load

Free-restrained
Bending moment Bending moment

(a) Boundary condition (I)

Vertical load

Horizontal load Horizontal load

Fully-restrained
Bending moment Bending moment
(b) Boundary condition (II)
Fig. 11. Schematic diagram of the load conditions.

)
)

1/2
1/2

7000 7000

Stress intensity factor range (MPa· mm

Stress intensity factor range (MPa· mm

Theoretical soultion Theoretical soultion

6000 Test data 6000 Test data
Numerical results (Boundary I) Numerical results (Boundary I)
Numerical results (Boundary II) Numerical results (Boundary II)
5000 5000

4000 4000

3000 3000

2000 2000

1000 1000

0 0

-1000 -1000
0 30 60 90 120 150 0 30 60 90 120 150
Crack propagation length (mm) Crack propagation length (mm)
(a) Beam 3 (c) Beam 5
)
)

1/2
1/2

7000 7000
Stress intensity factor range (MPa· mm
Stress intensity factor range (MPa· mm

Theoretical soultion Theoretical soultion

6000 Test data 6000 Test data
Numerical results (Boundary I) Numerical results (Boundary I)
Numerical results (Boundary II) Numerical results (Boundary II)
5000 5000

4000 4000

3000 3000

2000 2000

1000 1000

0 0

-1000 -1000
0 30 60 90 120 150 0 30 60 90 120 150
Crack propagation length (mm) Crack propagation length (mm)
(b) Beam 4 (d) Beam 6

Fig. 12. Stress intensity factor range of numerical results.

the tip of the crack that changes the SIF. This local stress field can- local stress that causes a closing effect on the crack hence a smaller
not be modeled by the analytical solution as it gives the same SIF for the same force in the strengthening material.
result for the two cases, which have the same far-field stress but In the experimental program in this study, two anchors were
different local field stresses. The magnitude of this local stress field fixed very close to the crack, especially in Beams 3 and 4. Conse-
depends on the CMOD. As CMOD is equal to the interfacial slip quently, a strong local effect restrained the value of CMOD which
between the strengthening material and the bottom face of the produced a relatively large local field stress. Whereas, in the sce-
steel beam, this local stress field is related to the attachment sys- nario of beams adhesively bonded with CFRP, significant debond-
tem of the strengthening material. In other words, a better anchor- ing near the crack would occur, which shifts the restraining force
age with smaller interfacial slip gives a smaller CMOD and a larger away from the crack tip, and consequently, reduces the local field

Please cite this article in press as: Q.-Q. Yu, Y.-F. Wu, Fatigue durability of cracked steel beams retrofitted with high-strength materials, Constr. Build.
Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.051
 Q.-Q. Yu, Y.-F. Wu / Construction and Building Materials xxx (2017) xxx–xxx 9

stress. In this case, the SIF is mainly caused by the far-field stress, [5] Q.Q. Yu, T. Chen, X.L. Gu, X.L. Zhao, Z.G. Xiao, Fatigue behaviour of CFRP
strengthened steel plates with different degrees of damage, Thin-Walled
and consequently, Eq. (7) is applicable. In fact, Eq. (7) was validated
Struct. 69 (2013) 10–17.
in the original work for cases with significant FRP debonding near [6] P. Colombi, G. Fava, L. Sonzogni, Effect of initial damage level and patch
the crack [36]. configuration on the fatigue behaviour of reinforced steel plates, Fatigue Fract.
Therefore, it is concluded that the effect of local crack constraint Eng. Mater. Struct. 38 (2014) 368–378.
[7] L.L. Hu, X.L. Zhao, P. Feng, Fatigue behavior of cracked high strength steel plates
or interfacial slip on the calculation of SIF cannot be ignored in a strengthened by CFRP sheets, J. Compos. Constr. 20 (6) (2016) 04016043.
strengthening scheme with mechanical anchorage. More research [8] B. Täljsten, C.S. Hansen, J.W. Schmidt, Strengthening of old metallic structures
on the effect of interfacial slip on SIF is planned in the future. in fatigue with prestressed and non-prestressed CFRP laminates, Constr. Build.
Mater. 23 (4) (2009) 1665–1677.
[9] Y.J. Kim, K.A. Harries, Fatigue behavior of damaged steel beams repaired with
4. Conclusions CFRP strips, Eng. Struct. 33 (5) (2011) 1491–1502.
[10] H. Jiao, F. Mashiri, X.L. Zhao, A comparative study on fatigue behaviour of steel
beams retrofitted with welding, pultruded CFRP plates and wet layup CFRP
In this paper, six cracked steel beams were tested under fatigue sheets, Thin-Walled Struct. 59 (2012) 144–152.
loading to investigate the strengthening efficiency of mechanically [11] G. Wu, H.T. Wang, Z.S. Wu, H.Y. Liu, Y. Ren, Experimental study on the fatigue
behaviour of steel beams strengthened with different fiber-reinforced
anchored high-strength materials on the fatigue life of steel beams.
composite plates, J. Compos. Constr. 16 (2012) 127–137.
Normal modulus CFRP laminate, HSS plate and SAF plate were [12] P. Colombi, G. Fava, Experimental study on the fatigue behaviour of cracked
adopted to retrofit I-beams with an initial crack at the mid-span. steel beams repaired with CFRP plates, Eng. Fract. Mech. 145 (2015) 128–142.
Three mechanical anchorage systems were investigated. Four- [13] Q.Q. Yu, Y.F. Wu, Fatigue strengthening of cracked steel beams with different
configurations and materials, J. Compos. Constr. 21 (2) (2017) 04016093.
point cyclic bending tests showed that the strengthening effec- [14] E. Ghafoori, M. Motavalli, Lateral-torsional buckling of steel I-beams
tively retarded the crack propagation and extended fatigue life. retrofitted by bonded and un-bonded CFRP laminates with different pre-
Retrofitting with CFRP showed a better performance while the stress levels: experimental and numerical study, Constr. Build. Mater. 76
(2015) 194–206.
strengthening schemes with HSS plate and SAF plate were also [15] E. Ghafoori, M. Motavalli, J. Botsis, A. Herwig, M. Galli, Fatigue strengthening of
effective and were recommended for repair of steel beams, consid- damaged metallic beams using prestressed unbonded and bonded CFRP plates,
ering the ease of installation. Int. J. Fatigue 44 (2012) 303–315.
[16] E. Ghafoori, M. Motavalli, A. Nussbaumer, A. Herwig, G.S. Prinz, M. Fontana,
Analytical solution was also adopted to study the SIF at the Design criterion for fatigue strengthening of riveted beams in a 120-year-old
crack front. The effect of the strengthening by the attached mate- railway metallic bridge using pre-stressed CFRP plates, Composites Part B 68
rials was considered as a compressive force applied to the tension (2015) 1–13.
[17] E. Ghafoori, M. Motavalli, Normal, high and ultra-high modulus carbon fiber-
soffit of the beam. By comparing the experimental results with the
reinforced polymer laminates for bonded and un-bonded strengthening of
analytical solution, a significant difference was found. This differ- steel beams, Mater. Des. 67 (2015) 232–243.
ence is attributed to the local stress field caused by the strengthen- [18] E. Ghafoori, M. Motavalli, Innovative CFRP-prestressing system for
strengthening metallic structures, J. Compos. Constr. 19 (6) (2015) 04015006.
ing material which is related to the CMOD.
[19] E. Ghafoori, M. Motavalli, X.L. Zhao, A. Nussbaumer, M. Fontana, Fatigue design
Three-dimensional finite element analysis was used to further criteria for strengthening metallic beams with bonded CFRP plates, Eng. Struct.
exam the SIF of the test beams. Two scenarios of free-restrained 101 (2015) 542–557.
and fully-restrained crack openings were evaluated. The analytical [20] E. Ghafoori, M. Motavalli, A retrofit theory to prevent fatigue crack initiation in
aging riveted bridges using carbon fiber-reinforced polymer materials,
solution was approximately the same to the case with free con- Polymers 8 (2016) 308.
straint, while the test results were between these two extreme [21] E. Ghafoori, M. Motavalli, A. Nussbaumer, A. Herwig, G.S. Prinz, M. Fontana,
cases. Determination of minimum CFRP pre-stress levels for fatigue crack prevention
in retrofitted metallic beams, Eng. Struct. 84 (2015) 29–41.
It was concluded that, besides the effect of a reduction in the far [22] P. Colombi, A. Bassetti, A. Nussbaumer, Analysis of cracked steel members
(remote) stress field caused by the strengthening material, the reinforced by pre-stress composite patch, Fatigue Fract. Eng. Mater. Struct. 26
local stress field related to CMOD should be taken into considera- (1) (2003) 59–66.
[23] K. Kaddouri, D. Ouinas, B. Bachir Bouiadjra, FE analysis of the behaviour of
tion when calculating the SIF of strengthened steel beams. The octagonal bonded composite repair in aircraft structures, Comput. Mater. Sci.
CMOD is related to the anchorage at the interface; therefore, differ- 43 (4) (2008) 1109–1111.
ent strengthening systems will lead to different SIF and fatigue life [24] C. Wu, X.L. Zhao, R. Al-Mahaidi, W.H. Duan, Mode I stress intensity factor of
centre-cracked tensile steel plates with CFRP reinforcement, Int. J. Struct. Stab.
of the beams.
Dyn. 13 (1) (2013) 1350005.
[25] H.B. Liu, X.L. Zhao, R. Al-Mahaidi, Boundary element analysis of CFRP
Acknowledgements reinforced steel plates, Compos. Struct. 91 (1) (2009) 74–83.
[26] Q.Q. Yu, X.L. Zhao, T. Chen, X.L. Gu, Z.G. Xiao, Crack propagation prediction of
CFRP retrofitted steel plates with different degrees of damage using BEM,
The work described in this paper was supported by the National Thin-Walled Struct. 82 (2014) 145–158.
Natural Science Foundation of China (Project No. 51508406) and [27] Q.Q. Yu, T. Chen, X.L. Gu, X.L. Zhao, Boundary element analysis of edge cracked
the Research Grant Council, Hong Kong Special Administrative steel plates strengthened by CFRP laminates, Thin-Walled Struct. 100 (2016)
147–157.
Zone, China (Project No. CityU124113). The authors wish to [28] P. Colombi, Plasticity induced fatigue crack growth retardation model for steel
express their heartfelt appreciation to Prof. Xiang-Lin Gu from elements reinforced by composite patch, Theor. Appl. Fract. Mech. 43 (2005)
Tongji University and Prof. Xiao-Ling Zhao from Monash University 63–76.
[29] Q.Q. Yu, X.L. Zhao, Z.G. Xiao, T. Chen, X.L. Gu, Evaluation of stress intensity
for supervising the present work. Thanks are due to all the techni- factor for CFRP bonded steel plates, Adv. Struct. Eng. 17 (12) (2014) 1729–
cal staff in the Heavy Structures Testing Lab at City University of 1746.
Hong Kong, especially Mr. Chan Wan Tong Allan and Mr. Ernest [30] R. Brighenti, Patch repair design optimization for fracture and fatigue
improvements of cracked plates, Solids Struct. 44 (3–4) (2007) 1115–1131.
Cheung, for their assistance in carrying out the tests. [31] H.T. Wang, G. Wu, Z.S Wu, Effect of FRP configurations on the fatigue repair
effectiveness of cracked steel plates, J. Compos. Constr. 18 (2014) 04013023.
References [32] R. Kienzler, G. Hermann, An elementary theory of defective beams, Acta Mech.
62 (1986) 37–46.
[33] M.L. Dunn, W. Suwito, B. Hunter, Stress intensity factors for cracked I-beams,
[1] X.L. Zhao, L. Zhang, State-of-the-art review on FRP strengthened steel
Eng. Fract. Mech. 57 (6) (1997) 609–615.
structures, Eng. Struct. 29 (8) (2007) 1808–1823.
[34] E. Ghafoori, M. Motavalli, Analytical calculation of stress intensity factor of
[2] J.G. Teng, T. Yu, D. Fernando, Strengthening of steel structures with fiber-
cracked steel I-beams with experimental analysis and 3D digital image
reinforced polymer composites, J. Constr. Steel Res. 78 (2012) 131–143.
correlation measurements, Eng. Fract. Mech. 78 (2011) 3226–3242.
[3] S.C. Jones, S.A. Civjan, Application of fibre reinforced polymer overlays to
[35] A. Hmidan, Y.J. Kim, S, Yazdani, Stress intensity factors for cracked steel girders
extend steel fatigue life, J. Compos. Constr. 7 (4) (2003) 331–338.
strengthened with CFRP sheets, J. Compos. Constr. 19 (5) (2015) 04014085.
[4] H.B. Liu, R. Al-Mahaidi, X.L. Zhao, Experimental study of fatigue crack growth
[36] P. Colombi, G. Fava, Fatigue crack growth in steel beams strengthened by CFRP
behaviour in adhesively reinforced steel structures, Compos. Struct. 90 (1)
strips, Theor. Appl. Fract. Mech. 85 (2016) 173–182.
(2009) 12–20.

Please cite this article in press as: Q.-Q. Yu, Y.-F. Wu, Fatigue durability of cracked steel beams retrofitted with high-strength materials, Constr. Build.
Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.051
10 Q.-Q. Yu, Y.-F. Wu / Construction and Building Materials xxx (2017) xxx–xxx

[37] GB 50017-2003 Code for design of steel structures, 2003. [43] Japan Society of Steel Construction (JSSC), Fatigue design recommendations for
[38] AS 1391-1991, Methods of tensile testing of metals, 1991. steel structures, 1995.
[39] ASTM D3039/D3039M-08, Standard test method for tensile properties of [44] BS code British Standards Institution (BSI), Guide to methods for assessing the
polymer matrix composite materials, 2008. acceptability of flaws in metallic structures, BS7910:2005. London, 2005.
[40] Y.F. Wu, L. He, L.C. Bank, Bond-test protocol for plate-to-concrete interface [45] A. Hosseini, E. Ghafoori, M. Motavalli, A. Nussbaumer, X.L. Zhao, Mode I fatigue
involving all mechanisms, J. Compos. Constr. 20 (1) (2016) 04015022. crack arrest in tensile steel members using prestressed CFRP plates, Compos.
[41] L.C. Bank, Mechanically-fastened FRP (MF-FRP)—A viable alternative for Struct. 178 (2017) 119–134.
strengthening RC members, in: 2nd Int. Conf. on FRP Composites in Civil [46] Q.Q. Yu, Y.F. Wu, Fatigue behaviour of cracked steel beams retrofitted with
Engineering-CICE 2004, Taylor & Francis, London, 3, 2004. CFRP laminates, Adv. Struct. Eng. http://dx.doi.org/10.1177/
[42] ASTM E647 – 13e1 Standard Test Method for Measurement of Fatigue Crack 1369433217729518.
Growth Rates, American Society for Testing and Materials (ASTM), West
Conshohocken, USA 2013.

Please cite this article in press as: Q.-Q. Yu, Y.-F. Wu, Fatigue durability of cracked steel beams retrofitted with high-strength materials, Constr. Build.
Mater. (2017), https://doi.org/10.1016/j.conbuildmat.2017.09.051