Sei sulla pagina 1di 13

Construction and Building Materials 30 (2012) 347–359

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials


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

Structural behavior of double-lap joints of steel splice plates bolted/bonded


to pultruded hybrid CFRP/GFRP laminates
Nguyen Duc Hai ⇑, Hiroshi Mutsuyoshi
Department of Civil and Environmental Engineering, Saitama University, Saitama 338-8570, Japan

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

Article history: This paper presents results from experimental and numerical studies on the structural behavior of
Received 29 July 2010 double-lap joints of steel splice plates bolted/bonded to pultruded hybrid CFRP/GFRP laminates. A total
Received in revised form 30 November 2011 of 45 tensile coupon specimens with double-lap bolted/bonded joints and six large-scale flexural beam
Accepted 1 December 2011
specimens with/without butt joints in the mid-span section were conducted. Two types of double-lap
Available online 30 December 2011
joints were investigated including bolted joints and hybrid joints (bonded-and-bolted). The results
indicate that the double-lap hybrid joint which combined of stainless steel bolts, adhesive bonding,
Keywords:
and V-notched splice plates is an effective method of joining the hybrid FRP laminate. The rough surface
Hybrid CFRP/GFRP laminates
Double-lap joints
of the V-notched splice plates and the adhesive bonding contribute to improve the joint stiffness. Numer-
Bolted joints ical analysis was carried out to predict load–displacement curve of the hybrid joints and the results
Hybrid joints (bonded-and-bolted) showed a good correlation with the experiments. A design idea for use in the practical design of joints
V-notched splice plates was proposed.
Joint stiffness Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction failure [2]. Numerous experimental and analytical investigations


on joining of FRP laminates have been conducted, with the main
Fiber-reinforced polymers (FRPs) composites have been focus on aerospace applications [3–8]. However, limited published
increasingly used in civil infrastructure applications due to advan- works are available in the field of civil infrastructure applications.
tageous properties such as high specific strength, low weight, and Furthermore, those studies that have been carried out cover a wide
corrosion resistance. Recently, a hybrid FRP (HFRP) I-beam consist- range of FRP laminates with various fiber architecture and matrix
ing of carbon and glass fibers has been developed. The idea is to use types, fiber lay-up and stacking sequences, etc., which result in
the superior strength of carbon FRP (CFRP) in the flanges while varying joint behavior. Thus, additional investigations are required
keeping the material costs low by using glass FRP (GFRP) in the to fully understand the characteristics of FRP joints. This study
flanges and web. Various laboratory tests had been conducted to focuses on the experimentally and numerically determined behav-
investigate structural behavior of such HFRP beams and the results ior of double-lap joints of steel splice plates bolted/bonded to the
showed the potential of applying HFRP beams for bridge compo- hybrid CFRP/GFRP laminate. A total of 45 tensile coupon specimens
nents [1]. However, two major issues still need further investiga- with double-lap bolted/bonded joints and six large-scale flexural
tion before HFRP beams can be used in real bridge applications. beam specimens with/without butt joints in the mid-span section
One is the development of a joining method for HFRP laminates. were conducted. Two types of double-lap joints were investigated
The other is evaluation of environmental aspects such as the including bolted joints and bonded-and-bolted joints (hereafter
amount of carbon dioxide emissions (CO2) and the life-cycle cost called the ‘‘hybrid joints’’). The effects of bolt-end distance, bolt
of HFRP bridges. As the second step of an ongoing FRP research torque, V-notched splice plates, and adhesive bonding on joint
project supported by the Japanese government, this study focuses strength and failure mode were discussed.
on the first issue with emphasis on developing an effective joining
method for HFRP laminates. 2. Tensile joint tests
FRP laminates are commonly joined by adhesive bonding,
mechanical fastening, or a combination of the two methods. The 2.1. Joint specimens
combination is often employed as a safeguard against defects with-
in the adhesive layer, which may lead to premature or catastrophic A total of 45 HFRP coupon specimens with double-lap bolted/
bonded joints were tested under tensile loading as shown in
⇑ Corresponding author. Tel.: +81 90 9971 5028. Fig. 1. All specimens (HFRP laminates) were cut from the flanges
E-mail address: hai.nguyen.2004@gmail.com (N.D. Hai). of the large-scale HFRP I-beams. The HFRP beams were

0950-0618/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.conbuildmat.2011.12.001
348 N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359

P manufactured using the pultrusion process. Layup and stacking se-


w quence of the HFRP I-beam are schematically shown in Fig. 2. The
HFRP laminate consist of CFRP and GFRP. The angle of CFRP was
fixed at zero degree while the angles of the GFRP were 0°, 90°,
and ±45° or with Continuous Strand Mat (CSM). Fiber orientations
Metal Grip and volume contents of the FRP laminae stacked in the HFRP
I-beam are shown in Table 1. The mechanical properties of materi-
als provided by suppliers are listed in Table 2.
All specimens have the same width of w = 80 mm, nominal
s g s Gusset Plate thickness of t = 14 mm, bolt-side distance of s = 20 mm, and longi-
Splice Plate (Steel) tudinal and transverse hole spacing of p = g = 40 mm. They were
e
configured with the same ratios of laminate width to bolt diameter
p Stainless Steel Bolts (w/d = 8), bolt-side distance to bolt diameter (s/d = 2), longitudinal
spacing (pitch) to bolt diameter (p/d = 4), transverse spacing (gage)
p d b = 10 to bolt diameter (g/d = 4), bolt diameter to splice plate thickness (d/
e tsp = 1.1), washer diameter to bolt diameter (dw/d = 2.5), and hole
clearance (dh  d = 0.05d). Stainless steel bolts with a nominal
HFRP Laminate
diameter of 10 mm were used. Holes were machined in the speci-
mens using diamond tips. The specimens were laterally clamped
Aluminum Plate with bolts through a pair of 9 mm thick steel splice plates. A preset
torque wrench was applied equally to all the bolts in joint speci-
mens. The same joint geometry was used for the specimens with
bolted joints and hybrid joints (bonded-and-bolted). For the hybrid
joint specimens, the preset torque was applied to the bolts at the
same time as the adhesive was applied. The adhesive was then
cured in the experimental room for 24 h at a temperature of
20 °C before each test. Sikadur-30 was used for the adhesive. It
was a two-component and solvent-free adhesive having moderate
P viscosity and excellent mechanical strength. The mechanical prop-
Fig. 1. Configuration of tensile joint tests.
erties of the adhesive are shown in Table 3. Two types of double-
lap joints including bolted joints and hybrid joints were tested to
examine their strength and behavior. The effects of bolt-end dis-
tance, bolt torque, V-notched splice plates, and adhesive bonding
on joint strength and failure mode were investigated. Coupon joint
FLANGE
test variables are listed in Table 4.

2.2. Test setup and measurements

CFRP 0° All tests were conducted using a universal testing machine


GFRP 0/90° (UTM) with a load capacity of 500 kN. A data logger was used to
GFRP CSM record load, displacement, and strain data. The both ends of the
GFRP ±45°
specimens were clamped in the grips of the testing machine and
the tensile force was applied at the top. Clip gages were attached
to both sides of the specimen to measure the relative displacement
between the HFRP laminate and the splice plates. Each specimen
was instrumented with back-to-back strain gages attached to the
WEB HFRP laminate and the splice plates. The tensile joint test setup
is schematically illustrated in Fig. 3.

2.3. Test results and discussion

Fig. 2. Layup and stacking sequence of the HFRP I-beam.


2.3.1. Load and displacement curve
The load versus relative displacement curves of specimens with
a 6-bolt double-lap joint configuration are compared in Fig. 4. For
identification purposes, the specimen with bolted joints is referred
Table 1 to as specimen A0 and those with hybrid joints (bonded-and-
Fiber orientation and volume content of FRP laminae stacked in the HFRP I-beam.
bolted) is referred to as specimen A1 (with flat splice plates) and
HFRP I-beam Type of lamina Fiber orientation (°) Volume content (%) specimen A2 (with V-notched splice plates), as shown in Table 4.
Flange CFRP 0 34 It is noted that the relative displacement in the figure indicates
GFRP 0/90 17 the difference in the averaged elongation of clip gages attached to
±45 41 the both sides of the specimen (where one end of a clip gage is at-
CSMa 8
tached to the splice plate and the other end to the specimen). As
Web GFRP 0/90 43 can be seen, the load–displacement curve for specimen A0 can be
±45 43
subdivided into four stages. In the first stage (OA), there was no
CSMa 14
slip between the splice plates and the HFRP laminate with the load
a
Continuous Strand Mat. ranging from 0 to 15 kN. This is due to the effect of the torque
N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359 349

Table 2 was the load reduction that occurred after exceeding the failure
Mechanical properties of materials. strength of the bolts.
Parameter Notation CFRP GFRP 0°/ GFRP GFRP Similarly, the load–displacement curves for specimen A1 can be
0° 90° ±45° CSM subdivided into four stages. The first stage (O0 A0 ) represented the
Volume fraction (%) Vf 55 53 53 25 slip resistance of joints up to the load of approximately 50 kN. This
Young’s modulus E11 128.1 25.9 11.1 11.1 value is three times higher than that of specimen A0 due to the
(GPa) E22 = E33 14.9 25.9 11.1 11.1 contribution of adhesive bonding. It should be noted that the
Shear modulus G12 = G13 5.5 4.4 10.9 4.2 bolt-hole clearances were filled with the adhesive during the prep-
(GPa) G23 3.9 4.4 10.9 4.2 aration of specimen A1. However, the holes were only partly filled
Poisson’s ratio (–) m12 = m13 0.32 0.12 0.58 0.31 due to the moderate viscosity of the adhesive. Hence, the second
m23 0.45 0.12 0.58 0.31 stage existed in the case of specimen A1, but the slip was signifi-
cantly reduced as compared with specimen A0. Indeed, the second
stage (A0 B0 ) indicated linear behavior of the load–displacement
curve with the load ranging from 50 to 130 kN, corresponding to
relative displacements of 0.05–0.2 mm. The third stage (B0 C0 D0 )
Table 3 exhibited a gradual development of bearing failure in the HFRP
Mechanical properties of adhesive (Sikadur-30).
laminate/splice plates combined with local debonding of the adhe-
Property Test method Curing Standard sive layers and shear/bending of the bolts when the load increases
condition value from 130 to 360 kN, corresponding to relative displacements of
Bending strength (N/mm2) JIS K 7203- 20 ± 2 °C 50a–62b 0.2–3.5 mm. The final stage (D0 E0 ) was the sudden failure of speci-
1995 7 days men A1 at about 345 kN.
Compressive strength JIS K 7208- 70a–76b
It is clear from Fig. 4 that the load–displacement curve of spec-
(N/mm2) 1995
Compressive elastic modulus JIS K 7208- 4000a– imen A1 is much stiffer than that of specimen A0. This suggests
(N/mm2) 1995 6400b that adhesive bonding can improve considerably the slip resistance
Tensile strength (N/mm2) JIS K 7113 20a–32b and stiffness of joints.
Tensile shear strength JIS K 6850 10a–18b
(N/mm2)
2.3.2. Effect of adhesive layer thickness
Impact strength (kJ/m2) JIS K 7111 1.5a–2.9b
The effect of adhesive layer thickness was investigated using
a
Minimum value. specimens with the hybrid joints of flat splice plates to the HFRP
b
Maximum value.
laminate (the same configuration as specimen A1). Three distinct
specimens TAL01, TAL05, and TAL15 were prepared with the thick-
ness of the adhesive layers (TALs) set at approximately 0.1–0.2 mm
for specimen TAL01 (no thickness control of the adhesive layers)
Table 4
and set at 0.5 mm and 1.5 mm respectively for specimens TAL05
Coupon joint test variables.
and TAL15 (thickness control of the adhesive layers) using the
Specimen Type of Type of splice No. of Torque TALc EDd adhesive thickness holders. The load–displacement curves of all
joint plate bolts (Nm) (mm)
specimens are shown in Fig. 5.
A0 Ba Flat 6 20 n/a 3d This figure shows that all specimens have the same failure load
A1 BBb Flat 0.1 and failure mode. The failure mode was debonding of the adhesive
A2 BBb V-notched 0.1
layers and shearing of the bolts. The figure reveals that the stiffness
TAL01 BBb Flat 6 20 0.1 3d of the load–displacement curve is proportional to the thickness of
TAL05 0.5
TAL15 1.5
the adhesive layers. Specimen TAL15 had the highest stiffness
while specimen TAL01 had the lowest. However, it is interesting
E2d-B Ba Flat 2 20 n/a 2d
E3d-B 3d
to note that specimen TAL15 has almost the same stiffness as spec-
E4d-B 4d imen TAL05 up to the point where the major debonding takes
E2d-BB BBb Flat 2 20 0.1 2d
place. This indicates that increasing the thickness of the adhesive
E3d-BB 3d layers from 0.5 to 1.5 mm results in an insignificant increase in
E4d-BB 4d joint stiffness. The load at which this major debonding occurred
T5 BBb Flat 2 5 0.1 3d was higher for specimen TAL05 than for specimen TAL15. These re-
T20 20 sults may suggest the use of an adhesive layer thickness of 0.5 mm
T30 30 for hybrid joints of the HFRP laminate. However, by looking at the
a
Bolted joint. initial loading stage of the load–displacement curve at a smaller
b
Bonded-and-bolted joint. scale, it can be seen that specimen TAL01 shows a stable behavior
c
Thickness of adhesive layer. while specimens TAL05 and TAL15 exhibit a zigzag behavior. The
d
Bolt-end distance.
zigzag behavior may be caused by stress concentration in the area
of the adhesive thickness holders leading to local debonding of the
adhesive layers at these positions. Therefore it is not a good idea to
applied to the bolts. In the second stage (AB), the load–displace- use the adhesive thickness holders to control thickness of the
ment curve was linear as the load increases from 15 to 60 kN, cor- adhesive layers in practical applications. In the hybrid joints using
responding to a relative displacement of 0.0–1.0 mm. This flat splice plates, it is recommended to use no thickness control of
indicates that the bolts gradually slip into a bearing region around the adhesive layers (resulting in an adhesive layer thickness of
the holes in the HFRP laminate. The third stage (BCD) was nonlin- approximately 0.1–0.2 mm).
ear behavior of the load–displacement curve, corresponding to rel-
ative displacements from 1.0 to 5.5 mm. This might be due to the 2.3.3. Effect of V-notched splice plates
development of bearing failure in the HFRP laminate/splice plates To examine the effect of splice plates on strength of the hybrid
combined with shear/bending of the bolts. The final stage (DE) joints, two specimens with the same configuration but with
350 N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359

torque that would not lead to damage at the outermost surface


of the specimen. Furthermore, it is important to emphasize that
the V-notched splice plates can be used effectively to control thick-
ness of the adhesive layers without any adhesive thickness holders
(the thickness of the adhesive layers equals to the depth of V-
notches) resulting in good bonding strength of the hybrid joints.
Fig. 6 compares the load versus relative displacement relationship
for specimens A1 and A2. The figure shows that the stiffness of
these specimens is almost the same up to 130 kN (points B and
B0 in Fig. 6). However, specimen A2 shows higher stiffness than
specimen A1 when the load increases over 130 kN. This difference
in stiffness seems to be larger as the load increases. This may be
because the rough surface of the splice plates in specimen A2 con-
tributes to improve bonding between the splice plates and the
HFRP specimen. It is interesting to note that the load at which ma-
jor debonding of the adhesive layers happens in the case of speci-
men A2 (point C0 ) is slightly higher than its ultimate load (point D0 )
Fig. 3. Tensile joint test setup. due to the effect of V-notches. The major debonding load of spec-
imen A2 is approximately 14% higher than that of specimen A1
as shown in Fig. 6. These results reveal that the adhesive layers
may carry most of the load before the major debonding. When
400 the adhesive layers fail, the load is carried solely by the bolts. Final-
D'
E' ly, both specimens A1 and A2 failed due to shearing of the bolts.
350 D
E
300
Applied Load (kN)

C'
Specimen A0 (Bolted-only)
2.3.4. Effect of bolt-end distance
250
C Specimen A1 (Bonded-and-Bolted) To investigate the effect of bolt-end distance on failure strength
200 OA, O'A': Slip resistance
of the bolted/bonded joints, a total of 18 specimens with a 2-bolt
AB, A'B': Bolts slip into bearing region configuration varying end distances (e) from 2d to 4d (where d is
150 BCD, B'C'D': Bearing (plates), shear/bending (bolts)
B' DE, D'E': Load reduction the nominal bolt diameter) were conducted. Two series of speci-
100 C: Major shear/bending (bolts) mens with bolted joints and hybrid joints (bonded-and-bolted)
C': Major shear/bending (bolts), major debonding (adhesive)
A' B were investigated.
50 D, D': Ultimate load
E, E': Final failure load Fig. 7 shows the relationship between load and displacement of
A
0 the bolted joint specimens. It can be seen that the specimen E2d-B
0 O, O' 1 2 3 4 5 6 7
(with a bolt-end distance of 2d) has the lowest failure load and fails
Relative Displacement (mm)
due to shear failure of the HFRP laminate. Specimens E3d-B and
Fig. 4. Load versus relative displacement curve. E4d-B (with bolt-end distances of 3d and 4d, respectively) have al-
most the same failure load but different failure modes. The failure
mode of specimen E3d-B was shear failure of the HFRP laminate
while that of specimen E4d-B was bearing failure of the HFRP lam-
400 inate at the bolt-holes as presented in Fig. 8a. This indicates that a
bolt-end distance of 4d is the most appropriate for bolted joints of
350
the HFRP laminate.
300 Looking at the specimens with hybrid joints (E2d-BB, E3d-BB,
Applied Load (kN)

250 TAL01 (t = 0.1 mm) and E4d-BB), all failed with the same failure load and failure mode
TAL05 (t = 0.5 mm) (Fig. 8b), even though the bolt-end distance was varied in the same
200 way as for the bolted joint specimens. The load–displacement
TAL15 (t = 1.5 mm)
150
curves for the hybrid joint specimens are shown in Fig. 9. Since

100 P/2
P
P/2 400
50 C' D
Bonded-and-bolted joint (6-bolt configuration) E
350
0 14% D'
0 2 4 6 E'
300
Applied Load (kN)

C OA, O'A': Slip resistance


Relative Displacement (mm) AB, A'B': Bolts slip into bearing region
250 BCD, B'C'D': Bearing (plates), shear/bending (bolts)
DE, D'E': Load reduction
Fig. 5. Effect of adhesive layer thickness.
200 C, C': Major shear/bending (bolts), major debonding (adhesive)
D, D': Ultimate load
E, E': Final failure load
150
different types of splice plates were tested. Specimen A1 was con- B, B' V-Notch splice plates
100
figured with flat splice plates while specimen A2 used V-notched Specimen A1 (Flat plates)
splice plates which were processed with V-shaped notches on 50 Specimen A2 (V-notch plates)
Flat splice plates
A, A'
one side of their surfaces (where the depth and spacing of the
0
V-notches are 0.5 mm and 1.0 mm, respectively). The original idea O,0 O' 1 2 3 4 5 6 7
of using the V-notched splice plates was to provide more clamping
Relative Displacement (mm)
force between the splice plates and the HFRP specimen (the
V-notches bit into the specimen) with an appropriate amount of Fig. 6. Effect of splice plates.
N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359 351

120 120

100 100
Applied Load (kN)

Applied Load (kN)


80 E2d-B (e = 2d) 80
E2d-BB (e = 2d)
E3d-B (e = 3d)
60 60 E3d-BB (e = 3d)
E4d-B (e = 4d)
E4d-BB (e = 4d)
40 40
P/2 P/2
P P
20 P/2 20 P/2

Bolted joint (2-bolt configuration) Bonded-and-bolted joint (2-bolt configuration)


0 0
0 2 4 6 8 0 2 4 6
Relative Displacement (mm) Relative Displacement (mm)

Fig. 7. Effect of bolt-end distance (bolted joint). Fig. 9. Effect of bolt-end distance (hybrid joint).

the load was mostly carried by the adhesive during the initial load-
ing stage, the load reductions at around 40–70 kN could be attrib- 120
uted to local debonding of the adhesive. The failure mode of these
specimens was bearing failure of the HFRP laminate at the bolt- 100
holes. The difference in failure mode of between bolted joint and

Applied Load (kN)


hybrid joint specimens was due to the presence of the adhesive 80
T5 (T = 5 Nm)
layers. As previously discussed, the adhesive may carry most of
the load before the major debonding. Thereafter, the load was 60 T20 (T = 20 Nm)
transferred to the bolts, but they did not suffer shear failure as T30 (T = 30 Nm)
40
occured with the bolted joint specimens. This suggests that bolt
shear failure does not develop suddenly but gradually. It is con- P/2
20 P
cluded that failure strength and failure mode of the bolted joints P/2
largely depend on the bolt-end distance while these of the hybrid Bonded-and-bolted joint (2-bolt configuration)
0
joints are governed by strength of the adhesive before the major 0 2 4 6
debonding of the adhesive layers happens. It is interesting to note Relative Displacement (mm)
that the bolt-end distance has minor effect on failure strength and
failure mode of the hybrid joints and the hybrid joint specimens Fig. 10. Effect of bolt torque.
fail with a desirable mode. A minimum bolt-end distance of 3d is
recommended for designing bolted/bonded joints of the HFRP
laminate. amount of torque needs to be applied to hold the joint parts and
ensure full spreading of the adhesive over the bonding areas with
a reasonable adhesive layer thickness (for instance, an adhesive
2.3.5. Effect of bolt torque layer thickness of 0.1–0.2 mm for specimens with the flat splice
The same configuration of the hybrid joint specimens as used plates and that of 0.5 mm for specimens with V-notched splice
for the investigation of the effect of bolt-end distance was utilized plates). Based on the result of this study, a bolt-torque of 20 Nm
to examine the effect of bolt torque. Applied torques with values of is recommended for hybrid joints of the HFRP laminate.
5 Nm, 20 Nm, and 30 Nm were investigated. Fig. 10 shows the
load–displacement relationships for all specimens. These results
indicated that the load–displacement curves for specimens T5 2.4. Finite element analysis (FEA)
and T20 had almost the same stiffness and failure load. However,
specimen T30 had a slightly lower stiffness than the others. This 2.4.1. Finite element (FE) geometry and mesh
implies that a high torque (30 Nm) may result in approximately The experimental results discussed in previous section indi-
zero-thickness of the adhesive layers, leading perhaps to reduce cated the importance of using hybrid bolted-and-bonded joint to
bonding strength of the hybrid joints. The results suggest that improve the joint stiffness. However, the behavior of hybrid joints
the effect of bolt torque is simply to fix the splice plates to the is always complicated and it is necessary to verify the experimen-
HFRP laminate during the adhesive curing process. An appropriate tal results of hybrid joints by the numerical analysis. A three

(a) Bolted joint specimens (b) Hybrid joint


(bonded-and-bolted) specimens

Fig. 8. Failure mode of joint specimens.


352 N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359

Bolts Adhesive-1 Preload section


Splice-1
HFRP Laminate
P/2
P
P/2

Splice-2 Adhesive-2 Preload

Fig. 13. Preload applied at mid-surface of bolts at the step 1 of the FEA.

20 Hole-1
Bolt-1
40 80 δom = effective displacement at the initiation of damage
Hole-2 δfm = effective displacement at complete failure
20 Bolt-2 tno ( t so , tto )
A

50 30 30 50

Traction
Fig. 11. Joint configuration and dimensions of FE model.

dimensional finite element model of a hybrid double-lap joint was GC


developed using the ABAQUS [9] finite element code. Two-bolt
configuration was used for all simulations as shown in Fig. 11.
The HFRP laminate, splice plates, bolts and nuts were modeled B
using an 8-node linear brick, reduced integration with hourglass O δ mo δ mf
Separation
control element (element C3D8R). The adhesive was modeled
using an 8-node three-dimensional cohesive element (element Fig. 14. Typical traction–separation response and linear damage evolution of
COH3D8). The washers were assumed to have minor effect to the cohesive layers.

strength of the hybrid joints and hence did not include in the mod-
el to reduce the post-processing time. Details of 3-D FE model for T ¼ F p ðKDÞ ð1Þ
the double-lap hybrid joints are presented in Fig. 12.
A neat fit was assumed between the bolts and the HFRP lami- where T is the bolt torque (T = 20 Nm), Fp the initial preload, D the
nate/splice plates in all simulations. Contact was defined between nominal diameter of bolts (D = 10 mm), and K is the nut factor
the bolts/nuts and the HFRP laminate/splice plates on the hole sur- (K = 0.2 as recommended by Bickford [10]).
faces and on the outer surface of splice plates. The contact pair ap- In the second step (step 2), tensile load was applied to one end
proach based on a master–slave algorithm was used with small of the joint which was free to move in the longitudinal direction
sliding allowed between surfaces in contact. A friction coefficient only (i.e. uy = uz = 0). The opposite end of the joint was fixed in
of 0.2 was assumed between bolts/nuts and HFRP laminate/splice all direction (i.e. ux = uy = uz = 0). The effect of nonlinear material
plates. The outer surface of the bolts was tied to the inner surface properties of splice plates were used in the analyses.
of the nuts. A bolt torque of 20 Nm was used in all cases.
The analyses were performed in two consecutive steps. In the 2.4.2. Material models for FEA
first step (step 1), initial preload (pre-tension load) was applied 2.4.2.1. Cohesive model. As mentioned in the previous section,
at the mid-surface of the bolts as shown in Fig. 13. The value of cohesive element was used to model the cohesive layers of the
initial preload (Fp = 10 kN) was calculated using the short-form hybrid joints. It was assumed that the thickness of the cohesive
torque-preload equation given by [10]: layers is negligibly small. The progressive damage and failure of

Bolt-2 Bolt-1

Splice-1

HFRP laminate
Adhesive-1

Adhesive-2

Splice-2 Nut-1
Nut-2
(a) FE mesh (b) Closer view of the model
Fig. 12. 3-D finite element model for the double-lap hybrid joint (bonded-and-bolted).
N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359 353

cohesive layers was modeled using traction–separation law. A lin- quired to cause failure in the normal, the first and the second shear
ear elastic traction–separation law was assumed prior to damage. directions, respectively. g is the material parameter and GC is the
However, once a damage initiation criterion is met, the failure of critical mixed-mode fracture energy.
cohesive layers is characterized by progressive degradation of the Input data of the cohesive model including elastic properties (E,
stiffness of cohesive layers according to a damage evolution law. G1, G2), quadratic nominal stresses (t on ; t os , and tot ) and critical frac-
2.4.2.1.1. Damage initiation criterion. The quadratic nominal stress ture energies (GCn ; GCs ; GCt ) are listed in Table 5. Material parameter
criterion was used in this study in which damage was assumed to (g) of 1.45 was used for all cases.
initiate when a quadratic interaction function involving the nomi-
nal stress ratios (as defined in the expression below) reaches a va- 2.4.2.2. Steel and HFRP model. The stainless steel bolts and steel
lue of one. This criterion can be represented by [9]: splice plates were modeled with elastic and plastic isotropic mate-
 2  2  2 rial properties as listed in Table 6. On the other hand, the HFRP
htn i ts tt laminate was modeled with equivalent material properties (homo-
þ o þ o ¼1 ð2Þ
ton ts tt geneous and orthotropic) to reduce complexity of the joint model
(Table 6). The in-plane tensile and compressive strength of the
where tn, ts and tt represent the nominal stress when the deforma- HFRP laminate were obtained from the material tests.
tion is either purely normal to the interface or purely in the first or
the second shear direction, respectively. 2.4.3. FE results and comparisons
t on ; t os , and t ot represent the peak values of the nominal stress 2.4.3.1. Load versus relative displacement. Fig. 15 shows compari-
when the deformation is either purely normal to the interface or sons of load versus relative displacement curves between experi-
purely in the first or the second shear direction, respectively. ment and FEA. A FE parametric study has been carried out on
It is noted that the Macaulay bracket indicated in Eq. (2) is used five cases to estimate the critical fracture energies of the adhesive
to signify that a pure compressive deformation or stress state does as presented in Table 5. However, only results of cases 1 and 2 are
not initiate damage. plotted in Fig. 15 to avoid multiple overlapping series in compari-
2.4.2.1.2. Damage evolution law. The damage evolution law de- sons with the experimental result. Results of other cases are sum-
scribes the rate at which the material stiffness is degraded once marized in Table 7.
the corresponding initiation criterion is reached. In the model, The load–displacement curve in case of the experiment was al-
damage evolution was defined based on the energy that was dissi- most linear before the local debonding of the adhesive layers at
pated as a result of the damage process (so-called fracture energy). point A. The local debonding in the experiment may be due to
The fracture energy is equal to the area OAB under the traction– the imperfect bonding between the splice plates and the HFRP lam-
separation curve (see Fig. 14). The linear softening behavior was inate. Stiffness of load–displacement curve was degraded after the
assumed. The mixed mode of deformation fields in the cohesive local debonding but the linear behavior of load–displacement
zone, which quantify the relative proportions of normal and shear curve continued up to point B where the load–displacement curve
deformation, was utilized. The dependence of the fracture energy became nonlinearly. Point B was defined as the major debonding of
on the mixed mode was specified by using Benzeggagh–Kenane the adhesive layers. Since the adhesive has much stiffer load path
(BK) fracture criterion [9] which the critical fracture energies in than the bolts, the experimental results verify that the load is car-
the first and the second shear directions were the same; i.e., ried mostly by the adhesive before the major debonding and the
GCs ¼ GCt . It is given by load is then transferred to the bolts up to the final failure at point
  G g C.
S
GCn þ GCs  GCn ¼ GC ð3Þ FEA results indicated that local debonding of the adhesive lay-
GT ers did not happen and the first debondings in all cases of FEA were
where GS = Gs + Gt is the total work done by shear tractions, assumed to be the major debondings. As shown in Table 7, the
GT = Gn + GS the total work done by normal and shear tractions, adhesives with higher critical fracture energies give larger major
and Gn, Gs, Gt is the work done by the traction and its conjugate rel- debonding loads. Similar to the experimental result, the load–dis-
ative displacement in the normal, the first, and the second shear placement curves in cases 1 and 2 of FEA behaved linearly before
directions, respectively. GCn ; GCs ; GCt is the critical fracture energies re- the major debondings at points D and G. After the major debon-
dings, the loads were significantly degraded up to the points E
and H respectively. Although the loads at which the major debon-
Table 5 dings happened were different in cases 1 and 2, the stiffness of
FE input data for the cohesive model. load–displacement curve after the total debondings at points E
Elastic property and H were almost the same. The reasons are the load have trans-
Young’s modulus: E First shear modulus: G1 Second shear modulus: G2
ferred to the bolts after the total debonding. The ultimate loads for
(N/mm2) (N/mm2) (N/mm2) cases 1 and 2 of FEA are coincident at points F and I. The ultimate
7500 7500 7500
loads for other cases have almost the equal values as cases 1 and 2
as listed in Table 7.
Quadratic nominal stress (damage initiation)
In comparisons with the experimental result, the stiffness of
Normal direction: ton First shear direction: t os Second shear direction: t ot load–displacement curve before the local debonding of the exper-
(N/mm2) (N/mm2) (N/mm2)
iment is almost the same with that of FEA. However, after the local
20 10 10 debonding of the experiment (point A), FEA result shows much
Critical fracture energy (damage evolution) stiffer behavior than that of the experiment up to the major debon-
Case Normal direction: First shear direction: Second shear direction: dings of FEA (points D and G). Significant load reductions after the
GCn (N/mm) GCs (N/mm) GCt (N/mm) major debondings happen in all cases of FEA while those in the
0 0.061 0.446 0.446
experiment do not happen. The reason for this difference is that
1 0.089 0.646 0.646 even after the major debonding of the experiment (point B), the
2 0.143 1.046 1.046 adhesive layers may not completely fail and the load is assumed
3 0.692 5.046 5.046 to carry mostly by bolts and trivially by the adhesive. Conversely,
4 2.749 20.046 20.046
the adhesive layers in FEA rapidly fail after the major debonding
354 N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359

Table 6
FE input data for the stainless steel bolts, steel splice plates and the HFRP laminate.

Name Type of steel Young’s modulus: E (N/mm2) Poisson’s ratio: m (–) Yield stress: ry (N/mm2) Plastic strain: ep (–)
Material properties for the stainless steel bolts and steel splice plates
Bolt and Nut SUS304 197,000 0.3 205 0
520 0.4
Splice plate SS400 206,000 0.3 240 0
450 0.21
E11 (kN/mm2) E22 = E33 (kN/mm2) G12 = G13 (kN/mm2) G23 (kN/mm2) m12 = m13 (–) m23 (–) rc11c (N/mm2) rc22d (N/mm2) rt11e (N/mm2) rt22g (N/mm2)
Equivalent material properties for the HFRP laminate
49.4a 18.1a 4.38b 3.07 0.23a 0.32 384a 196a 885a 170a
a
Obtained from material tests.
b
Obtained from classical lamination theory.
c
Longitudinal compressive strength.
d
Transverse compressive strength.
e
Longitudinal tensile strength.
g
Transverse tensile strength.

120 120
G C Holes on Holes on
F, I splice plates HFRP laminate
100 100
Major
Hole-1 (HFRP Laminate)
debonding
B Hole-2 (HFRP Laminate)
Point O: Zero coordinate 80 Hole-1 (Splice-1)
Applied Load (kN)

80 D Point A: Local debonding (experiment)


Applied Load (kN) Hole-2 (Splice-1)
Point B: Major debonding (experiment) Hole-1 (Splice-2)
Point C: Ultimate load (experiment) 60 Hole-2 (Splice-2)
60 A Point D: Major debonding (case 1 - FEA) 1.0
H Point E: Total debonding (case 1 - FEA)

Applied Load (kN)


0.8
Point G: Major debonding (case 2 - FEA) 40 Total
Point H: Total debonding (case 2 - FEA) debonding 0.6
40 Point F, I: Ultimate load (cases 1&2 - FEA)
E 0.4
20 Scale up of load versus 0.2
Experiment bearing stress curve in step 1 0.0
20 C C
Case 1: Gs =Gt = 0.646 N/mm (FEA) of the finite element analysis 0 10 20 30 40 50
0 Average Bearing Stress (N/mm )
2
C C
Case 2: Gs =Gt = 5.046 N/mm (FEA)
0 Step 1
O -20 Step 2 (tensile load)
(bolt load)
0 1 2 3 4 5 6 7 8 9 0 200 400 600 800 1000 1200
Relative Displacement (mm) 2
Average Bearing Stress on Bolt Holes (N/mm )
Fig. 15. Load versus relative displacement.
Fig. 16. Load versus average bearing stress on bolt holes
(case 1 : GCs ¼ GCt ¼ 0:646 N=mm).

and totally debonded at points E and H and the loads are therefore
transferred solely by bolts.
Fig. 15 shows that the major debonding load in case 1 of FEA of the analysis. However when the tensile load applied to the joints
(point D) is very close to that of the experiment (point B). The stiff- in the step 2, bearing stresses increased insignificantly before the
ness of load–displacement curve after the total debonding in case 1 major debonding. This is supported by the experimental results
of FEA is only 6% higher than that of the experiment after the local discussed in Section 2.3.3 that the adhesive layers carry almost
debonding and the ultimate load in case 1 of FEA is just 2% differ- the load before the major debonding. The bearing stresses sud-
ence compared with that of experiment (Table 7). This suggests denly increased after the major debonding which indicated that
that the critical fracture energies of the adhesive used in case 1 bolts start bearing against the holes. After the total debonding of
of FEA shows good correlation with the experimental result. the adhesive layers, the bearing stresses increased linearly with
the increase of the tensile load from 40 to 90 kN. The bearing stres-
2.4.3.2. Bearing stress. Fig. 16 shows the relationship between the ses became nonlinearly when the tensile load increased from
average bearing stress on bolt holes of HFRP laminate/splice plates 90 kN to the failure at 111.6 kN. Failure due to bearing in the HFRP
and applied load in case 1 of FEA. It can be seen that the bearing laminate started at a tensile load of 67 kN when the bearing stres-
stresses on bolt holes increase with the increase of the bolt-torque ses on the bolt holes reached the compressive strength of the HFRP
from 0 to 20 Nm (with an equivalent preload of 10 kN) in the step 1 laminate (Table 6). The bearing failure was then developed until

Table 7
Comparisons of debonding and ultimate load between the experiment and FEA.

Experiment Finite element analysis


Local debonding load (kN) Major debonding load (kN) Ultimate load (kN) Case Major debonding load (kN) Total debonding load (kN) Ultimate load (kN)
0 71.8 40 111.6
61.1 80.8 109.1 1 78.3 42 111.6
2 90.7 45.7 111.6
3 116.6 61.7 111.7
4 131.9 100.4 112.3
N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359 355

Deformation of
Splice
adhesive layer
plates

Shear failure
of bolt

HFRP laminate

Fig. 17. Deformation of the hybrid double-lap joint at failure (cut through the
center of bolt-2).

Fig. 20. Design idea for hybrid joints (bonded-and-bolted).

Bolt-1
were shearing of bolts, bearing of the HFRP laminate and total deb-
onding of the adhesive layers. Shearing of bolts happened at the
Shear failure shear planes of the hybrid joints as shown in Fig. 18. All bolts failed
of bolts at at a yield stress of 520 N/mm2 corresponding to a plastic strain of
shear planes 400,000 l.
Fig. 19 shows the damage dissipation energy of the adhesive
layers at major debonding. It is found that high dissipation
energies are concentrated at the free edges of the adhesive layers
Unit: N/mm2 (lines AA and BB in Fig. 19) and the debonding may initiate at these
edges and develop towards to the left side of these edges. There-
fore the bolt-end distance from center of the bolt-holes to the free
Bolt-2 edges of the splice plates may affect to the debonding load of the
hybrid joints. The use of large bolt-end distance may decrease
Fig. 18. Stress distributions in bolts at failure (case 1 : GCs ¼ GCt ¼ 0:646 N=mm).
the debonding load because of lower contact pressure induced by
the final failure of the hybrid joints corresponding to the maximum bolt-torque between splice plates and the HFRP laminate at the
average bearing stress of 1185 N/mm2 at hole-1. This value of bear- free edges. The use of small bolt-end distance may result in shear
ing stress is 3.1 times larger than the longitudinal compressive failure of the HFRP laminate. Based on results of this study, a
strength of the HFRP laminate. However, the experimental data re- bolt-end distance of 3d is recommended for hybrid joints of the
vealed that the shear failure of the HFRP laminate did not happen HFRP laminate.
in the hybrid joints for all specimens with the end distance of 3d
(as shown in Fig. 8b). It is anticipated that the adhesive filled in
2.5. Design idea for hybrid joints
the gaps of bolt holes may allow the HFRP laminate attains stron-
ger bearing resistance.
Fig. 20 shows a design idea for hybrid joints of the HFRP
laminate based on results of this study. As noted previously, the
2.4.3.3. Failure mode. Fig. 17 shows the deformation of the double- adhesive carries most of the full load before the major debonding.
lap hybrid joints at failure. The failure modes of the hybrid joints The load is then transferred to the bolts up to the failure. This

Adhesive-1

HFRP laminate
A
B

Unit: N/mm
or kJ/m

A
Adhesive-2
B
Fig. 19. Damage dissipation energy of the adhesive layers at major debonding (case 1 : GCs ¼ GCt ¼ 0:646 N=mm).
356 N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359

20 55 20

30 V-notch splice plate


(t = 9 mm)

14
40

HFRP laminate Adhesive

30
250

40

40
V-notch splice plate
(t = 9 mm)

250
180
40

40
Stainless steel bolts M10
30

40
30
14
(a) Cross section of beams B-1 and B-2 at mid-span
30 40 30 30 40 30
P/2 P/2 5
Section A
(d) Section A (beam B-1)

14
250 1000 500 500 1000 250

30
1750 (HFRP-1) 1750 (HFRP-2)

40
3500

250
180
40
(b) Elevation of beam B-1

40
P/2 Section B P/2

30
14

30 40 40 30 30 40 40 30
250 1000 500 500 1000 250 5

1750 (HFRP-1) 1750 (HFRP-2) (e) Section B (beam B-2)


3500

(c) Elevation of beam B-2


Fig. 21. Details of joint specimens for large-scale flexural tests.

Table 8
Flexural beam test variables.

Beam Type of joint Length of joint part Total length Number of bolts
L1 (HFRP-1) (mm) L2 (HFRP-2) (mm) L (mm) Flange Web
B0 (control beam) Without joint (prototype) – – 3500 – –
B1 Hybrid joint (bonded-and-bolted) 1750 1750 3500 8 8
B2 1750 1750 3500 12 16

means that the ultimate load of the hybrid joints is governed by (Padhesive) is higher than the ultimate load of the hybrid joints (Pulti-
the number and strength of bolts. However, the design load of mate) in some cases depending on various factors such as thickness
joints in practical applications must be in serviceability limit state of the adhesive layers (specimen TAL05 in Fig. 5) or types of the
and this is assumed to be dependent on the tensile shear strength splice plates (specimen A2 in Fig. 6). Therefore the safety factor
of the adhesive and the bonded area, as shown in the following (c) of the hybrid joints is defined as the ratio of maximum load
equation: to design load, as presented in the following equation:

Pdesign ¼ k  Padhesiv e ¼ k  2sta Aa ð4Þ c ¼ Pmax =Pdesign ¼ maxðPadhesiv e ; P ultimate Þ=Pdesign ð5Þ

where k is the design coefficient, Padhesive the load at which major where
debonding occurs, sta the tensile shear strength of the adhesive, Pultimate ¼ 2sub Ab N ð6Þ
and Aa is the adhesive area or bonded area.
The design coefficient (k) in Eq. (4) should be obtained from the sub is the ultimate shear strength of one bolt, Ab the cross-sectional
experiment based on the linear portion of the load–displacement area of one bolt, and N is the total numbers of bolts in the hybrid
curve. It is noted that the load at which major debonding occurs joints.
N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359 357

200
180
Load cell 160 Bolt failure

Applied Load (kN)


Major
140 debonding
120 (adhesive)

Spreader
100
High speed camera
beam 80
Major shear/bending(bolts)
60
Beam B1 (Top flange)
Major debonding (adhesive) 40 Beam B1 (Bottom flange)
Major shear/bending(bolts) Beam B2 (Top flange)
20
Beam B2 (Bottom flange)
Transducers 0
-1500 -1000 -500 0 500 1000 1500 2000
Strain ( )

Fig. 25. Load versus strain curve for splice plates.

Fig. 22. Test setup and measurement.


bolts, and approximately 0.5 mm thick adhesive layers. A bolt-
end distance of 3d was used. A torque of 20 Nm was applied
200 equally to all the bolts. The adhesive was cured in the experimental
room for 24 h at a temperature of 20 °C before each test. The fiber
180 Computed ultimate load (Beam B2)
lay-up and stacking sequence and mechanical properties of the
160 Control beam HFRP I-beam are the same as those of tensile joint specimens as
Applied Load (kN)

(Beam B0)
140 Computed ultimate load (Beam B1) presented in Tables 1 and 2 and Fig. 2. Details of the joint geometry
120 and beam specimens used for the large-scale flexural joint tests are
Beam B1
100 given in Fig. 21. Flexural beam test variables are listed in Table 8.
Beam B2
80
3.2. Test setup and measurement
60
40 All beams were simply supported and tested in four-point
20 bending with a total span of 3000 mm and a loading span of
0 1000 mm. The test setup and measurement are shown schemati-
0 10 20 30 40 50 60 cally in Fig. 22. Linear voltage displacement transducers (LVDTs)
Deflection (mm) and laser transducers were used to measure beam deflection in
the mid-span section and under the loading points. A number of
Fig. 23. Load versus deflection curve.
strain gages were attached in the flexural span, shear span, and
3. Large-scale flexural test of HFRP I-beam with butt joints in near the loading points to measure the strain distribution. A
the mid-span section high-speed camera was placed in front of the beams to record sud-
den failure.
3.1. Test specimens
3.3. Test result and discussion
The results of the tensile joint tests were used to design large-
scale flexural tests of HFRP beams with butt joints in the mid-span 3.3.1. Load–deflection curve
section. The double-lap hybrid joints with V-notched splice plates The load–deflection curves in the mid-span section of beams B0,
were selected as the joining method for all beams. Laboratory tests B1, and B2 tested under four-point bending are shown in Fig. 23.
were conducted on six large-scale HFRP I-beams with 95 mm wide, The results indicated that beams B0 and B2 behaved linearly up
250 mm high, 14 mm (flange) and 9 mm (web) thick. These to failure. Beam B2 failed at a somewhat lower load than beam
I-beams were fabricated by pultrusion process with a length of B0, but beam B2 exhibited higher stiffness due to the presence of
3500 mm. All HFRP beams were cut into two equal half lengths the splice plates and bolts in the flanges and web. The ultimate
of 1750 mm, which were then jointed in the flexural span. The load of beams B1 and B2 was predicted, with results that showed
hybrid butt joints were formed at the flanges and web using good correlation with the experiments as plotted in Fig. 23. Details
9 mm thick V-notched splice plates, 10 mm diameter stainless of the computation are given in Appendix A. It is noted that the

Beam B1 Beam B2

Bolt Web
failure splice Delamination
rotation

Web
crushing
Fig. 24. Failure mode of beams B1 and B2.
358 N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359

flexural strength of beams with butt joints in the mid-span section failure strength and failure mode of the hybrid joints and
is underestimated by about 10% when the effect of the web splice strength of the hybrid joints is governed by strength of the
plates is not taken into account (as assumed in this study in predic- adhesive before the major debonding of the adhesive layers. A
tion of the ultimate load of the hybrid joints). Hence, the good minimum bolt-end distance of 3d is recommended for design-
agreement between the experiments and analysis may be due to ing bolted/bonded joints of the HFRP laminate.
the effect of the V-notched splice plates. 3. The effect of bolt-torque and adhesive layer thickness were
The failure mode of beams B0 and B2 was crushing of the fiber investigated for hybrid joints of the HFRP laminate. The results
near the loading point (due to the stress concentration) followed indicate that the effect of bolt-torque is simply to fix the splice
by the delamination of the top flange of the HFRP I-beam. In addi- plates to the HFRP laminate during the adhesive curing process.
tion, web buckling in the flexural span was observed in the case of Bolt-torque of 20 Nm is recommended to apply equally to the
beam B2 due to the joint constraints in the flanges and web, as bolts to hold the joint parts and to ensure full spreading of
shown in Fig. 24. Visual monitoring of beam B2 during the test the adhesive over the bonding areas with an adhesive layer
indicated that failure occurred away from the joint as predicted. thickness of approximately 0.1–0.2 mm (no thickness control
In contrast to beam B2, beam B1 was expected to fail in the area of the adhesive layers) for specimens with the flat splice plates
of the joint. Indeed, beam B1 behaved linearly up to 130 kN. When and that of 0.5 mm for specimens with the V-notched splice
the load increases over 130 kN, the behavior of beam B1 becomes plates. The results reveal that thickness of the adhesive layers
nonlinear because of major debonding of the adhesive layers and in the hybrid joints of the HFRP laminate can be effectively used
shear/bending of the bolts. The final failure mode of beam B1 at 0.5 mm. The use of the adhesive layers thicker than 0.5 mm
was shearing of bolts in the bottom flange leading to rotation of may result in insignificant increase in stiffness of the hybrid
the splice plates in the web (Fig. 24). joints. Therefore, it is strongly recommended to use the V-
notched splice plates with the V-notches’ depth of 0.5 mm cor-
3.3.2. Load–strain curve of splice plates responding to a controlled adhesive layer thickness of 0.5 mm
Fig. 25 shows the relationship between the load and strain of for the hybrid joints of the HFRP laminate in practical
the splice plates at the top and bottom flanges in the mid-span sec- applications.
tion of beams B1 and B2. In the case of beam B1, both compressive 4. Experimental results and finite element analysis (FEA) verify an
and tensile strains behaved linearly to the point at which major important load transfer mechanism of the hybrid joints that the
debonding of the adhesive layers and major shear/bending of bolts adhesive carries most of the load before the major debonding.
occurred. Then the strain behavior became nonlinear as the bolts The load is then transferred to the bolts and the ultimate
experienced bending/shearing. On the other hand, the compressive strength of the hybrid joints depends on shear strength of the
strain in the case of beam B2 was linear up to failure while the ten- bolts. Additionally, the FEA results can well predict the failure
sile strain was nonlinear once major debonding of the adhesive mechanism of the hybrid joints which the failure initiates at
layers took place. the free edges of the adhesive layers due to the concentration
It is interesting to note that the load at which major debonding of high dissipation in these edges. The failure is then developed
occurs in the tension flange is lower than that in the compression towards to the left side of these edges until the major debond-
flange. Indeed, the load is mostly carried by the adhesive prior to ing of the adhesive layers takes place.
the major debonding of the adhesive layers and results from the 5. The FEA tensile test results of the hybrid joints with 2-bolt con-
tensile joint tests and FEA indicate that high peel stresses at the free figuration showed a good correlation with the experiment in
edge of the tension flange splice plates cause the premature deb- term of load–displacement curve. A design idea for use in prac-
onding. Conversely, peel stresses in the compression flange are typ- tical design of the hybrid joints was proposed.
ically not of concern with regard to the separation of joints and 6. HFRP beams with hybrid butt joints (bonded-and-bolted) in the
debonding of the adhesive layers in the compression flange may mid-span section subjected to the flexural loading exhibited
therefore happen at higher loads than that in the tension flange. almost the same strength and stiffness as the control beam
without joints. The stiffness of the hybrid joints most likely
depends on the bonding strength of the adhesive alone, while
4. Conclusions the ultimate load and failure mode of the joints are governed
by the number of bolts.
Experimental and numerical studies on the structural behavior
of double-lap joints of steel splice plates bolted/bonded to pultrud-
ed hybrid CFRP/GFRP laminates were investigated. Two types of Acknowledgment
double-lap joints were examined: bolted joints and hybrid joints
(bonded-and-bolted). The following main conclusions can be This research was funded by Ministry of Land, Infrastructure
addressed: and Transport of Japan (MLIT) Grant-in-aid for scientific research
in construction technology, which is gratefully acknowledged.
1. The combined use of stainless steel bolts, adhesive bonding, and
V-notched splice plates in double-lap hybrid joints is found to Appendix A. Prediction of ultimate load of HFRP I-beam with
be an effective method of joining the HFRP laminate. The rough butt joint in the mid-span section subjected to flexural loading
surface of the V-notched splice plates and the adhesive bonding
contribute to improve the joint stiffness. Furthermore, it is It is assumed that the flanges carry the full bending moment
important to emphasize that the V-notched splice plates can and the shear force is resisted by the web alone. In this study, there
be used effectively to control thickness of the adhesive layers is no shear force applied over the flexural span and the strength of
without any adhesive thickness holders (the thickness of the the joint is therefore governed by the magnitude of the bending
adhesive layers equals to the depth of V-notches) resulting in moment. The forces in the tension flange (T) and in the compres-
good bonding strength of the hybrid joints. sion flange (C), shown in Fig. A1, are given by:
2. Failure strength and failure mode of the bolted joints largely
depend on the bolt-end distance. On the other hand, it is inter- M PL
T¼C¼ ¼ ðA1Þ
esting to note that the bolt-end distance has minor effect on H  t 6ðH  tÞ
N.D. Hai, H. Mutsuyoshi / Construction and Building Materials 30 (2012) 347–359 359

t Note that the bearing load of the HFRP laminate and splice
C C plates must satisfy in the following equation:

Ppl u
br ¼ Ndb t pl rbr P P u ðA4Þ
H h=H-t where db is the nominal diameter of bolt, tpl the thickness of HFRP
M M laminate/splice plates, and rubr is the ultimate bearing strength of
T T HFRP laminate/splice plates.

References
Fig. A1. Calculation of flange forces.
[1] Hai Nguyen Duc, Mutsuyoshi Hiroshi, Asamoto Shingo, Matsui Takahiro.
Structural behavior of hybrid FRP composite I-beam. J Constr Build Mater
where M is the bending moment, H the height of HFRP beam cross 2010;24(June):956–69.
section, t the thickness of the flanges, P the applied load, and L is the [2] Kelly G. Quasi-static strength and fatigue life of hybrid (bonded/bolted)
composite single-lap joints. Compos Struct 2006;72:119–29.
span of beam. [3] Camanho PP, Matthews FL. Stress analysis and strength prediction of
The ultimate joint load is obtained when the tension and com- mechanically fastened joints in FRP: a review. Composites Part A
pression forces reach the ultimate shear capacity of the bolts: 1997;28A:529–47.
[4] Hart-Smith LJ. Adhesive bonding of composite structures: progress to date and
T ¼ C ¼ Pubolt N ¼ mAb sub N ðA2Þ some remaining challenges. J Compos Technol Res 2002;24(3):133–53.
[5] Tate MB, Rosenfeld SJ. Preliminary investigation of the loads carried by
where P ubolt is the ultimate shear load of one bolt, N the total number individual bolts in bolted joints. NACA TN 1051; May, 1946.
[6] Nelson WD, Bunin BL, Hart-Smith LJ. Critical joints in large composite aircraft
of bolts in one flange, m the number of shear planes (m = 2 for dou- structure. NASA CR-3710; August, 1983.
ble-lap joints), Ab the cross-sectional area of one bolt, and sub is the [7] Hart-Smith LJ. Design methodology for bonded-bolted composite joints.
ultimate shear strength of one bolt. Technical report AFWAL-TR-81-3154, Douglas Aircraft Company; 1982.
[8] Hart-Smith LJ. Bonded-bolted composite joints. J Aircraft
From Eqs. (A1) and (A2), the ultimate joint load can be com- 1985;22(11):993–1000.
puted using the following equation: [9] Abaqus User’s Manual. Dassault Systèmmes; 2008.
[10] Bickford John H. Introduction to the design and behavior of bolted
6mAb sub NðH  tÞ joints. Taylor and Francis Group, LLC; 2008.
Pu ¼ ðA3Þ
L

Potrebbero piacerti anche