Professional Documents
Culture Documents
JQ Yn SKJ CS (2020) 247
JQ Yn SKJ CS (2020) 247
Composite Structures
journal homepage: www.elsevier.com/locate/compstruct
Bolted and bonded FRP beam-column joints with semi-rigid end conditions
Jawed Qureshi a,⇑, Yashida Nadir b, Shaise K John b
a
School of Architecture, Computing and Engineering (ACE), University of East London, 4-6 University Way, Beckton London E16 2RD, United Kingdom
b
Department of Civil Engineering, College of Engineering Trivandrum, Thiruvananthapuram 695016, Kerala, India
A R T I C L E I N F O A B S T R A C T
Keywords: Presented are test results from eight full‐scale pultruded FRP beam‐to‐column joint subassemblies. Moment‐
Semi‐rigid action rotation behaviour, failure modes, joint classification and load enhancement due to semi‐rigid end conditions
FRP profiles are discussed. Testing is divided in two series: first had FRP beam‐to‐steel column joints and second FRP beam‐
Beam‐to‐column joints to‐FRP column joints. The joints are either flange and web cleated or flange cleated only. The connection
FRP composite structures
method is bolting or ‘hybrid’ combining both bolting and bonding. Test parameters include effects of adhesive
Adhesive bonding
Hybrid joints
bonding, column flexibility, cleat material and joint configuration. Bolted and bonded joints not only increased
moment resistance but stiffness as well. Using steel cleats instead of FRP resulted in a 50% increase in the
moment resistance. Four failure modes, shear‐out failure, adhesive debonding with shear‐out failure, tensile
tearing of the column flanges from the web and delamination cracking of cleats were observed. All joints were
classified as semi‐rigid. This semi‐rigid action was used to increase the beam’s load carrying capacity. For a
span‐to‐depth ratio of 20, the increase in load was 50% for the bolted case and 70% for the bolted and bonded
joints. This increase provides a design flexibility to designers, who are restricted by limited off‐the‐shelf FRP
section sizes.
⇑ Corresponding author.
E-mail addresses: J.Qureshi@uel.ac.uk (J. Qureshi), yashidanadir@cet.ac.in (Y. Nadir).
http://dx.doi.org/10.1016/j.compstruct.2020.112500
Received 19 December 2019; Revised 12 May 2020; Accepted 15 May 2020
Available online 20 May 2020
0263-8223/© 2020 Elsevier Ltd. All rights reserved.
J. Qureshi et al. Composite Structures 247 (2020) 112500
Combining two joining methods may be unnecessary [14] as the load architecture that is prone to delamination cracking which makes them
is mainly taken by adhesive due to stiff load path. Yet, the hybrid joint unsuitable for practical use. Moreover, there is also lack of evidence to
is used as a safeguard against defects in the adhesive layer and sudden suggest that FRP cleats would not experience any delamination dam-
failure of the bonded joint. This provides a fail‐safe mechanism, just in age during service life [9]. This damage can be detrimental to the
case if one joining method fails other takes over [15]. There are sce- durability of FRP structures, especially in chemically ingressive envi-
narios where combined bolting and bonding can be very useful ronments. The main reason for use of steel with FRP members is lack
[15–17]: of understanding about durability performance of FRP cleats. Some
researchers [27,28] used multi‐bolted pultruded FRP plates instead
• Hybrid joints are very effective in resisting load from various direc- of leg‐angles for beam‐to‐column joints. Joints between tubular FRP
tions. Adhesive is good at resisting shear loads, while direct tension profiles were investigated in papers [29–31]. Recently, the effect of
and transverse loads can best be taken by bolts. adhesive bonding is studied by Razaqpur and his associates
• Bolts may enable the joint to resist fire exposure. [13,32,33]. Limited research exists on pultruded joints with steel
• Bolting can help in long‐term performance of bonded joints. cleats. Only a few research papers have been published on FRP joints
• Use of adhesive in bolted joints can increase fatigue life perfor- with steel cleats, including research by Mottram and Zheng [23],
mance of joints. Turvey [34], Qureshi and Mottram [9,10]. The current design guideli-
nes in pultruders’ manuals [4–6] and ASCE pre‐standard for pultruded
Moreover, fabrication cost is increased by combining bolting and FRP structures [35] make no distinction between behaviour of FRP
bonding. Any benefits of using combined joints must be carefully eval- and steel cleated joints. More research is needed to quantify the prop-
uated against extra cost of fabrication. Hybrid joints may not be appro- erties of FRP joints with steel cleats.
priate in following situations: Mottram and Zheng [23] were the first to conduct physical testing
on pultruded FRP joints with steel flange cleats. They used
• In high temperature service environments, the adhesive will soften 203 × 203 × 9.53 mm FRP member with top and seat flange cleats
leading to reduced joint stiffness. cut from 100 × 100 × 8 mm steel angles. To prevent outward flexural
• In situations where electrical continuity is required, such as process deformation of column, 20 mm dia. steel rods were used to connect
plants. opposite column flanges with steel cleats. The authors suggested using
either steel cleats or composite cleats with different shape, fibre archi-
Frames are classified into simple and moment frames. Joints in sim- tecture and manufacturing process. However, the connection details
ple frames are assumed to be nominally pinned and in moment frames were not adopted in practice. Instead of FRP column, Turvey’s [34]
fully rigid. Web‐cleated connections represent simple joints, and flange joint consisted of FRP beam connected to stiff steel support with steel
and web cleated connections produce moment joints. Stability in sim- angles. The steel support eliminated the flexibility of the column
ple frames is achieved via bracing between beams and columns. The flanges. The author used 102 × 102 × 6.4 mm FRP beam with
joints only take shear forces from connected members in simple 100 × 100 × 6 mm stainless steel cleats. Three joint details with
frames. Simple joints normally allow rotations at the ends. Due to this, web, flange and web, and flange cleats were tested. As much as 80%
no moment is transferred to the joints from the connected members. increase in the initial rotational stiffness was observed by using steel
Contrarily, moment frames transfer both shear and moment from the cleats in place of FRP cleats. The moment capacity of steel and FRP
member to the joint. Stability is derived from fixity by rigid or moment cleats was not discussed.
joints between beams and column. It means, in theory, no rotation More recently, Qureshi and Mottram [9] tested joints with FRP
should take place at the joint. However, in practice, all joints exhibit members and steel web cleats. Two‐ and three‐bolted joint details with
some form of moment and rotation resulting in a semi‐rigid behaviour. 254 × 254 × 9.53 mm FRP profiles and 100 × 100 × 10 mm steel
The aim of this paper is to utilise this semi‐rigidity to offer design flex- web angles were used. Specimen repetition and no clearance hole were
ibility to structural designers. the main contributions of this research. Compared with companion
Despite being identical in shape with steel sections, design of pul- joints with FRP cleats in Qureshi and Mottram [11], the joints with
truded FRP members is controlled by the deflections rather than the steel cleats in [9] showed rotational stiffness and moment capacity
strength. In modern structural design, ultimate and serviceability limit three and two times more, respectively. In another study, Qureshi
states should be satisfied for reliable performance of a structure. While and Mottram [10] found the joints with steel cleats to be stiffer and
martial strength of FRP is comparable with steel, it does have a draw- stronger than FRP cleated joints. Establishing the serviceability deflec-
back of low stiffness with elastic modulus (20–30 GPa) 10 times lower tion limits for pultruded FRP joints with either steel or FRP cleats was
than steel (210 GPa). This means design of FRP sections is often con- the primary objective of this research. FRP members consisted of
trolled by serviceability or deflection limit instead of ultimate strength 203 × 203 × 9.53 mm profiles with steel cleats as
limit. Normally, this is not a big issue in steel design as next larger sec- 75 × 75 × 10 mm and FRP as 75 × 75 × 9.53 mm leg‐angles. The
tion is chosen when deflection limit is not satisfied. It limits the design mid‐span vertical deflection limit with steel cleats was span/650 and
flexibility in FRP structures as the range of FRP beam sizes is currently for FRP cleats it was span/300; both limits were far less than the rec-
very small (100, 150, 200, 250, 300 mm deep beams [4–6]). As FRP ommendations by manufacturers [4–6]. The rotational stiffness and
beams are generally designed with simply supported end conditions, the moment capacity in steel cleated joints were double than the joints
this leads to higher midspan deflections requiring larger sections. with FRP cleats.
Because the range of FRP sections is limited, the alternative solution The primary aim of this research is to study the behaviour of pul-
will be to use semi‐rigid end connections. Semi‐rigid action has shown truded FRP beam‐to‐column joints with steel or FRP cleats. Eight
in the past by Turvey [18] to reduce mid‐span deflections and increase full‐scale physical tests are carried out to investigate the effect of adhe-
load carrying capacity. One main objective of this research is to exploit sive bonding, column and cleat material, additional stiffener and joint
the semi‐rigid end joint to minimise deflections and enhance loading configuration. Testing includes flange and web cleated, and only
capacity. flange cleated joints using either bolting or bolting and adhesive bond-
Previous research on pultruded beam‐to‐column joints [11,19–26] ing combined. One test also has additional angle stiffeners bolted to
has mainly focussed on joints with FRP cleats. However, in practice, the column. The important joint properties are characterised,
many fabricators use steel cleats instead of FRP cleats in bolted joints. moment‐rotation curves plotted, and failure patterns identified.
This is due to inability of FRP cleats to resist prying deformation gen- Joints are also classified according to their rotational stiffness as
erated from applied moments. Generally, FRP cleats have fibre pinned, semi‐rigid or rigid as per Eurocode 3 part 1–8 [36]. The
2
J. Qureshi et al. Composite Structures 247 (2020) 112500
2. Test configuration
cleats
No
No
The testing rig consists of a steel frame with two hydraulic jacks.
beam’s bolted region
First hydraulic jack had a capacity of 50 tonne with a stroke of
150 mm. It was used to supply moment through vertical load at the
free end of the beam in the stub column‐single cantilever beam test
set up. This arrangement was cost‐effective compared with stub
SFStc1A
No
—
only uses a single cantilever beam. Second hydraulic jack with a capac-
ity of 25 tonne and a travel of 150 mm was used to apply axial load to
the column. A constant compressive load of 20% of the column’s axial
Shear-out failure at beam’s bolted
region
No
—
shear‐out failure of the beam’s bolted region and tensile tearing of col-
Adhesive debonding and shear-out failure at
of the axial capacity of the column has shown very little effect on
Steel 50 × 50 × 6 mm
Steel 50 × 50 × 6 mm
steel socket for ensuring vertical alignment of the load. The load is
Yes
centre of rotation was assumed to be at the column centre (at the cen-
tre of column web between two flanges). Joint rotation is equal to ver-
tical displacement at the load point near the free end divided by a fixed
horizontal distance of 700 mm between the load point and the column
centre. The resulting joint rotation in radians is then multiplied by
region
SFSc1
No
bonding
stiffener
Adhesive
mode
Test Ref
Failure
Table 1
Table 2
Second series joint configuration and detailing using FRP column and FRP beam.
Joint detailing
5% error margin. Any slip rotation due to bolt clearance hole was Increase in load carrying capacity due to semi‐rigid end condition is
neglected. Its effect is minimal as bolts were tight‐fitting. The column quantified.
did not rotate due to compressive load applied at its top.
First series used steel column and FRP beam and in second series 3.1. Joint properties
both column and beam were of FRP material. The reason for using
steel column in first series was to eliminate outward flexural deforma- Measured joint properties for all tests are presented in Table 3. Test
tion in column flange outstands, which happens when FRP column and reference is given in column (1). Initial moment (Mi), initial rotation
steel cleats are used. The joint moment produces this deformation in (ϕi) and initial rotational stiffness (Si = Mi/ϕi) are given in columns
column flanges, as the steel cleat material is stiffer than the FRP col- (2), (3) and (4), respectively. Initial joint properties correspond to lin-
umn. Companion tests in second series are performed with FRP col- ear elastic moment‐rotation response of a joint. Initial rotational stiff-
umn to study the effect of column flexural deformation. This local ness, Si, represents the slope of the linear elastic region of M‐ϕ curve.
deformation in column flange outstands can potentially have serious Secant joint properties are defined at damage onset. Damage onset or
consequences in practice, as many fabricators still use steel cleats first failure is defined as a point on moment‐rotation curve when either
instead of FRP cleats in bolted joints. First series consisted of five tests: audible acoustic emissions were first heard or delamination cracking
steel flange and web angles bolted, steel flange and web angles bolted in cleats was first seen. Delamination cracking only happened in the
and bonded, steel flange angles bolted, steel flange angles bolted and specimen, SFFc1, where cleats were of FRP material. At this point,
bonded, and FRP flange and web angles bolted. Second series con- the material damage was assumed to be enough to cause fibre expo-
tained columns and beams made up of FRP profiles. It had three tests: sure. The damage onset/first failure properties are given by Mj, ϕj
steel flange and web angles bolted, steel flange and web angles bolted and Sj (=Mj/ϕj) in columns (5), (6) and (7). Values of damage onset
and bonded, and steel flange and web angles bolted with additional are indicated by a solid circle symbol in moment‐rotation plots of
stiffener cleat bolted on the column side. First two tests in both series Figs. 7–15. Columns (8) and (9) show maximum moment Mmax and
were same except column material was steel in first series and FRP in corresponding rotation, ϕmax. The maximum rotation can also be ter-
the second. med as the maximum rotation capacity of the moment‐rotation curve.
In Tables 1 and 2 the test reference (e.g. SFSc1A) follows a four‐ Four main failure modes were observed. These include shear‐out
letter format, a number and a letter. First two letters indicate the col- failure (SFSc1 and SFStc1), adhesive debonding with shear‐out failure
umn and beam material; “S” for steel and “F” for FRP. Similarly, third (SFSc1A and SFStc1A), tensile tearing of the column flanges from the
letter shows material for cleat. Fourth small letter denotes joint config- web (FFSc1, FFSc1A and FFSc1S) and delamination cracking of cleats
uration; “c” for bolted joints with flange and web angles and “tc” for (SFFc1). Specimens labels in brackets are used to identify which fail-
bolted joints with flange angles only. The number indicates type of ure mode each test had. First failure mode was shear‐out failure of
loading; “1” for monotonic loading. The last letter “A”, used in some the FRP beam’s bolted region in specimens, SFSc1 and SFStc1. This
specimens, shows adhesive bonding together with bolting, whereas happened in joints, where the steel column was bolted to the FRP
‘S’ in one specimen denotes additional stiffener (additional double beam with steel leg angles. Second failure was due to adhesive
web angle bolted to the column’s web and flange). debonding and shear‐out failure of the beam’s bolted zone in speci-
mens SFSc1A and SFStc1A. This was accompanied by cracking at the
3. Results and discussion web‐flange junction of the FRP beam. This failure occurred in steel
cleated bonded and bolted joints with steel column and FRP beam.
Monotonic beam‐to‐column joint tests are conducted with FRP Third failure mode was tensile tearing of the column flange from its
beam‐to‐steel column and FRP beam‐to‐FRP column joints with either web in specimens FFSc1, FFSc1A and FFSc1S. It took place within
steel or FRP leg angles. Effects of cleat material, joint configuration, the web‐flange junction of the FRP column marked by outward flexu-
column flexibility, adhesive and additional stiffener cleat are studied. ral deformation of the flange at top flange cleats’ bolted zone. The fail-
The essential joint properties are established from moment‐rotation ure happened when both column and beam were of PFRP material,
curves. Failure modes are identified, and joints are classified. and steel cleats were used. Fourth failure was delamination cracking
J. Qureshi et al. Composite Structures 247 (2020) 112500
Table 3
Joint properties for beam-to-column joint tests.
Fig. 3. Shear-out failure mode in PFRP beam: (a) specimen SFSc1; (b) specimen SFStc1.
Fig. 4. Adhesive debonding and shear out failure mechanism in SFSc1A: (a) Cracking at the web-flange interface of the PFRP beam; (b) Adhesive debonding at the
flange cleat.
at root radius of FRP flange and web cleats. This happened in specimen stiffness of steel profiles than FRP counterparts. Fig. 3(a) and (b) show
SFFc1, where FRP cleats were used. the failure pattern for SFSc1 and SFStc1. The moment applied to the
The first failure mode in flange and web cleated joint, SFSc1 and joint generates out‐of‐plane forces in the connected members. These
flange cleated joint, SFStc1 was due to shear‐out failure of the bolted forces include prying forces at top and compressive forces at the bot-
region. This failure usually happens in plate‐to‐plate connections due tom of the joint. Thus, the bending moment causes top cleat to be
to bearing of bolt against FRP plate with short end distance. pulled away from the column flange and seat cleat to be compressed
Connection detailing and material strength of connected components into the column flange. These prying forces at the top portion of the
control the failure mode of a joint. The compression bearing force at joint result in shear‐out failure in beam flange and web above its neu-
the beam’s bolted region remains the same regardless of beam mate- tral axis, as shown in Fig. 3.
rial. The beam deformation, on the other hand, would be significantly The second failure mode, adhesive debonding with shear‐out, hap-
less if steel beam was used instead of FRP. This is due to higher pened in specimens with adhesive bonding and steel cleats, SFSc1A
J. Qureshi et al. Composite Structures 247 (2020) 112500
Fig. 5. Tension tearing failure mode – tearing of FRP column flange from web: (a) Specimen FFSc1A; (b) Specimen FFSc1S.
and SFStc1A. Fig. 4(a) shows cracking at top web‐flange junction of specimen SFSc1is bolted only and uses no bonding, while SFSc1A is
the FRP beam prior to adhesive failure. Prying forces exerted at the both bolted and bonded. M‐ϕ curves indicate that the specimen with
top joint, due to applied moment, caused shear failure of adhesive adhesive bonding and bolting is stiffer than the bolted only specimen.
bonding near top steel cleat, as shown in Fig. 4(b). Shear‐out failure Initial rotational stiffness, Si, is 125 kNm/rad and 250 kNm/rad for
of the PFRP beam was delayed by using adhesive bonding. SFSc1 and SFSc1A, respectively. This shows that a combination of bolt-
The third failure mode was tension tearing mode in specimens ing and bonding increases the joint’s initial stiffness. Rotational stiff-
FFSc1, FFSc1A and FFSc1S. The failure happened when PFRP beam ness at damage onset also increased in a similar proportion. Moment
and column, and steel leg angles were used to assemble the joint. at damage onset and maximum moment for bolted and bonded joint
Tensile tearing of the column flange from the web for FFSc1A is shown were about 1.5 and 1.6 times higher than the bolted case. The rotation
in Fig. 5(a). This type of failure has previously been reported by capacity (ϕmax) was not much different, 47 mrad and 43 mrad for
Turvey and Zhang [45]. Prying action caused significant outward flex- SFSc1 and SFSc1A.
ural deformation in FRP column near top flange cleats. At ultimate Second comparison includes specimens SFStc1 and SFStc1A with
moment, tensile stresses developed in the web‐flange junction of the top and seat flange cleats with and without bonding. Their moment‐
column resulting in tensile tearing of the column flange from its rotation curves are presented in Fig. 8. As shown in Table 3, the initial
web. Even the inclusion of additional stiffeners in specimen FFSc1S, stiffness for both specimens remained the same, 100 kNm/rad. Use of
in the form of double web cleats connected on the column side, could adhesive could not improve the initial stiffness. Damage failure hap-
not prevent the tension tearing failure of the web‐flange interface. The pened at a moment level of 2.8 kNm and 6.3 kNm, respectively for
FRP column flange’s outward flexural deformation is shown in Fig. 5 bolted, and bolted and bonded specimen. Rotational stiffness at dam-
(b) for FFSc1S. age onset for bonded and bolted test was about 30% higher than the
The fourth failure mode – delamination cracking happened when bolted only specimen. The rotation capacity (ϕmax) was about 64 mrad
cleat material was FRP and the connected members were steel column and it remained the same for both bolted, and bolted and bonded case.
and FRP beam in specimen SFFc1. The failure initiated at the root Similar to flange and web cleated joints, using adhesive in flange
radius of FRP flange angle, as indicated by delamination cracking in cleated joints increased the maximum moment by 60% compared with
Fig. 6(a). This cracking was a direct result of prying forces at top joint bolted only joint. However, the rotation capacity was much higher in
due to out‐of‐plane bending caused by the applied moment. The FRP flange cleated joint, about 64 mrad, compared with the maximum
web cleat also experienced delamination cracking at its root radius. rotation of 45 mrad in flange and web cleated joints.
Prying forces caused further delamination cracking in both top flange, Third configuration consists of specimens FFSc1 and FFSc1A, with
and web cleat. Eventually, the joint failed by excessive cracking in the flange and web cleats, shown in Fig. 9. The column profile was PFRP,
top flange, and web cleats, as seen in Fig. 6(b). contrary to first and second comparison, where column was of steel
material. The specimen without bonding, FFSc1, had quite extended
3.3. Effect of bonding linear portion of moment‐rotation graph. Initial rotational stiffness
for both specimens was not much different, with hybrid bolted and
Effect of adhesive bonding is assessed by comparing bolted joints bonded specimen exhibiting slightly higher stiffness. The moment at
with bolted and bonded joints. Three joint arrangements are presented damage onset coincided with each other at 1.75 kNm with rotations
for the comparison. These include flange and web cleated joints (SFSc1 of 21 mrad and 17 mrad for FFSc1 and FFSc1A. The M‐ϕ curves fol-
versus SFSc1A), flange cleated joints (SFStc1 versus SFStc1A) and lowed a comparable pattern up to damage onset. The maximum
flange and web cleated joint with FRP column (FFSc1 versus moment with bolted and bonded joint (4.2 kNm) was only 20% higher
FFSc1A). First two configurations had steel column and PFRP beam than bolted joint (3.5 kNm). The rotation capacity was about 50 mrad,
and third set up contained PFRP column and beam. not much different from the bolted case, 45 mrad.
Moment‐rotation curves for flange and web cleated joints with and Generally, bonded and bolted joints fail in a brittle manner, as
without bonding (SFSc1 versus SFSc1A) are plotted in Fig. 7. Test shown in Figs. 7–9. There is a sudden drop in moment‐rotation curves
J. Qureshi et al. Composite Structures 247 (2020) 112500
Fig. 6. Delamination cracking in specimen SFFc1: (a) Initiation of delamination cracking in FRP cleats; (b) Final failure of FRP top flange cleat and web cleat.
Fig. 7. Effect of bonding: comparison of flange and web cleated joints with Fig. 9. Effect of bonding: comparison of flange and web cleated joints (using
and without bonding. FRP column) with and without bonding.
J. Qureshi et al. Composite Structures 247 (2020) 112500
Fig. 10. Effect of column flexibility: comparison of flange and web cleated Fig. 13. Effect of additional stiffener: comparison of flange and web cleated
bolted joints with steel and FRP column. bolted joints with and without additional column web stiffener cleat.
Fig. 11. Effect of column flexibility: comparison of flange and web cleated Fig. 14. Effect of joint configuration: comparison of flange cleated joints with
bonded and bolted joints with steel and FRP column. flange and web cleated joints with bolting only.
Fig. 12. Effect of cleat material: comparison of flange and web cleated bolted Fig. 15. Effect of joint configuration: comparison of flange cleated joints with
joints with steel and FRP cleats. flange and web cleated joints with bolting and bonding.
J. Qureshi et al. Composite Structures 247 (2020) 112500
as a column, can potentially have serious consequences. Using a steel in moment capacity in the stiffened joint was at the expense of brittle
column instead of a FRP column helps prevent failure in the column. failure.
The influence of column flexibly is studied by comparing FRP beam‐
to‐steel column joints with FRP beam‐to‐FRP column joints. The bolted 3.6. Effect of joint configuration
case, with tests SFSc1 versus FFSc1, is presented in Fig. 10. The
bonded and bolted tests SFSc1A versus FFSc1A are compared in Two joint configurations with flange cleats only and flange and
Fig. 11. The joints in both comparisons had steel flange and web web cleats are compared here. These configurations include bolting
angles. only and combination of bolting and bonding. For bolted joints,
The rotational stiffness of the bolted joint with steel column at moment ‐rotation response of flange and web cleated joints (SFSc1)
damage onset is twice as much as the stiffness of the joint with FRP versus flange cleated joints (SFStc1) is plotted in Fig. 14. Table 3 indi-
column, as indicated in Fig. 10. The moment at damage onset and fail- cates that the initial rotational stiffness of the joint with flange and
ure for tests with steel column are 2.5 and 1.8 times the corresponding web cleats is 25% more than the joint with flange cleats only. There
values with FRP column. The shape of moment‐rotation curve for both is only 10% increase in maximum moment by using flange and web
steel and FRP column tests was almost the same. Fig. 11 shows the cleated joints instead of only flange cleated. Failure in flange cleated
comparison of tests using steel and FRP column with adhesive bonding joints happened in bolted region of tension flange of FRP beam.
in addition to bolting. The stiffness at damage onset in FRP beam‐steel Flange and web cleated joints failed by shear‐out failure in bolt holes
column joints was about 2.3 times higher than the joints with FRP col- above mid depth of the beam. M‐ϕ curve for flange cleated joint
umn. The maximum moment in steel column tests was about 2.5 times showed almost a linear trend until failure; and flange and web cleated
the moment in FRP column joints. Joints with steel column failed in a joint exhibited nonlinear pattern after initial linear part.
more brittle manner as compared to FRP column. The moment‐ Next comparison relates to the effect of adhesive bonding when all
rotation response of steel column joints was nearly linear up to the other parameters are same as described earlier. Tests with flange and
ultimate moment followed by a sudden drop in the moment carrying web cleats (SFSc1A) and flange cleats only (SFStc1A) are compared in
capacity. Overall, moment and rotational stiffness values were almost Fig. 15 by plotting their M‐ϕ curves. Adhesive bonding enhanced the
double in case of joints with steel column compared with the same maximum moment by 60% in both flange and web cleated, and flange
joints with FRP column. There was not much difference, though, in cleated joints. Due to larger surface area of adhesive bonding in flange
maximum rotation capacity of the joint. and web cleated joints, the moment rotation response of these joints
Cleat material affects moment and rotation capacity of a joint. was stiffer than bonded flange cleated joints as shown in Fig. 15.
Moment‐rotation behaviour of FRP beam‐to‐steel column joints with The maximum moment in flange and web cleated joint was only mar-
steel and FRP cleats is shown in Fig. 12. Both FRP and steel cleated ginally higher than flange cleated joint. Adhesive bonding in both
joints exhibit same moment‐rotation response up to a moment level cases caused sudden drop in moment carrying capacity leading to a
of 1.4 kN and rotation of 12.9 mrad. Beyond this point, the test with brittle failure. Failure mode was also similar in both joints with
steel cleat shows higher stiffness and moment resistance compared shear‐out failure of the beam’s bolted zone followed by adhesive
to the joint with FRP cleats. An average increase of 50% was observed debonding. Beneficial effect on moment capacity was insignificant
in damage onset and ultimate moment by using steel cleats in place of due to web cleats. Flange and web cleated joints only increased the
FRP cleats. Failure in FRP cleated joints was due to delamination rotational stiffness without proportional increase in the moment when
cracking of cleats. In steel cleated joints, failure is transferred from compared with flange cleated joints.
cleats to FRP beam’s bolted region.
3.7. Classification of joints
3.5. Effect of additional stiffener
Joints can be classified according to their strength and stiffness.
The influence of additional stiffener is studied by comparing tests Based on their initial rotational stiffness, joints can be classified as
FFSc1 versus FFSc1S and plotting their M‐ϕ curves in Fig. 13. The con- rigid, nominally pinned or semi‐rigid. This classification can be
nected members in tests were of FRP material; and the joints had steel applied to any joint regardless of its material. Hence, Eurocode 3 part
web and flange angles. The test FFSc1S had additional steel double 1–8 [36] for steel joints is applicable to FRP joints. The initial rota-
web leg angle attached to the column side. One leg of this angle was tional stiffness Sj,ini limits in Eurocode 3 [36] are compared with exper-
bolted through column web to the angle on other side of the column, imental initial rotational stiffness Si (=Mi/ϕi). Fig. 16 presents the
and the other leg was bolted via column flange to the beam’s web moment‐rotation behaviour with the joints’ classification boundaries
cleat, as seen in Fig. 5(b). The moment‐rotation curve for both tests, for rigid joints indicted by a dotted line and nominally pinned by a
with or without extra stiffener, shows same linear elastic response solid line. The joint classification is divided into zone 1, 2 and 3 for
up to a moment of 1.4 kNm and rotation of 17 mrad. After this, M‐ϕ rigid, semi‐rigid and nominally pinned joints. Zone 1 (rigid) is
curve for the test with stiffener branches off to gain about 10% higher between vertical axis and the dotted line, zone 2 (semi‐rigid) between
stiffness than the control test. the dotted and solid lines and zone 3 (nominally pinned) between solid
When a moment level of 3.5 kNm is reached, the joint without stiff- and horizontal axis. The limit for zone 1 is Sj,ini ≥ kb EIb/Lb and for
ener starts to fail and the joint with stiffener takes more moment. At a zone 3 it is Sj,ini ≤ 0.5EIb/Lb, where kb is 8, based on the assumption
moment level of 6.3 kNm, there is a sudden drop in moment carrying that the frame is braced. Detailed calculations of initial rotational stiff-
capacity characterised by failure within PFRP column flanges. The ness, Sj,ini for the classification boundaries are given in Appendix A as
joint without stiffener failed in a similar fashion. The rotation at max- per Eurocode 3 part 1–8 [36] provisions. The limit for zone 1 and 3 are
imum moment in unstiffened joint was about 51 mrad compared with 895 and 56 kNm/rad using a span‐to‐depth of 20 with a span (L) of
64 mrad in the joint with the stiffener. Although the stiffened joint 3000 mm for 150 × 100 × 10 mm FRP beam.
resulted in higher rotation capacity than the unstiffened joint, it Bolted steel flange and web cleated joints with steel column and
showed a brittle failure marked by a sudden drop in the moment in FRP beam, SFSc1 (filled circle marker), are classified as semi‐rigid
Fig. 13. Employing an extra steel stiffener with double web cleats on joints as shown in Fig. 16. Addition of bonding to this configuration
the column side increased the moment capacity to almost double leads to 50% increase in initial stiffness, as indicated by SFSc1A (solid
(1.8 times the unstiffened joint moment). However, this improvement triangle marker), and classed as a semi‐rigid joint. Joints with only
J. Qureshi et al. Composite Structures 247 (2020) 112500
steel top and seat flange angles, SFStc1 and SFStc1A, and web and
flange cleated joints with FRP column, FFSc1 and FFSc1S are also clas-
sified as semi‐rigid. Similarly, the joints SFFc1 and FFSc1A are cate-
gorised as semi‐rigid too. Generally, use of steel cleats and adhesive
increases the initial rotational stiffness of the joint resulting in semi‐
rigid behaviour. In steel structures, joints with either web or flange
clips, partial depth end plate or fin plate allow enough rotation for
them to classified as nominally pinned joints. Use of steel flange cleats
with FRP beam transfers the failure to FRP beam resulting in lesser
rotations and semi‐rigid response. Adhesive bonding inhibits rotations
too and this often leads to higher initial rotational stiffness in the
joints. All joints in this paper are classified as semi‐rigid joints. This
semi‐rigid action should be exploited to decrease mid‐span deflection,
increase load carrying capacity and increase spanning capability of
beams.
• Adding steel double web angle stiffener on the column side, to steel Joint’s rotational stiffness that lies between the boundaries of zone
flange and web cleated joint, increased the moment capacity by 1 (rigid) and zone 3 (nominally pinned) are categorised to be in zone 2
80%. However, this increase was at the expense of brittle failure. (semi‐rigid). In this case, any initial rotational stiffness lying between
• Effect of joint configuration was studied by adding web cleats to 895 kNm/rad and 56 kNm/rad is classed as semi‐rigid joint.
the flange cleated joint. The resulting configuration only enhanced Zone 3: Nominally pinned joints, if Sj,ini ≤ 0.5EIb/Lb
rotational stiffness without corresponding increase in the moment. 0:5EI b
• Based on rotational stiffness at damage onset, all joints are classi- Sj;ini ≤ A:2
Lb
fied as semi‐rigid as per Eurocode 3 part 1–8 [36]. This semi‐
rigid behaviour can be utilised to increase load carrying capacity 0:5 28 11985980
or spanning capability and reduce mid‐span deflections of beams. Sj;ini ≤
3000 1000
• A significant increase in load carrying capacity was seen in steel
cleated FRP beam‐to‐steel column joints using Turvey’s [18] Sj;ini ≤ 56kNm=rad
expressions. This load enhancement is due to semi‐rigid ends of
the beam compared with otherwise identically loaded beam with
simply supported ends. Corresponding to a practical span‐to‐ Appendix B
depth ratio of 20, this increase was about 50% for the bolted case
and 70% for the bolted and bonded joint. For steel cleated FRP A beam with a uniformly distributed loading is considered to work
beam‐to‐FRP column joints, this gain is only 30% regardless of out load enhancement index (λQ) as per method in Turvey [18]. Only a
adhesive bonding. sample calculation for the test SFSc1 is provided and same procedure
is adopted for the remaining tests. The deflection reduction index (λδ)
CRediT authorship contribution statement must be calculated first from:
1 þ k2 β
λδ ¼ B:1
Jawed Qureshi: Data Curation, Formal analysis, Writing ‐ Original k4 þ k2 β
Draft, Writing ‐ Review & Editing. Yashida Nadir: Conceptualization,
where
Supervision, Funding acquisition. Shaise K John: Methodology,
k2 From Table 1 Turvey [18] is 10.
Investigation, Writing ‐ Original Draft.
k4 From Table 1 Turvey [18] is 5.
β Dimensionless rotational flexibility of the beam end connection
Declaration of Competing Interest
given by (EI/KL), where
E Longitudinal modulus of elasticity for pultruded beam, (is 28 kN/
The authors declare that they have no known competing financial
mm2 from Fiberline Design Manual [5])
interests or personal relationships that could have appeared to influ-
I Major‐axis second moment of area of beam cross‐section,
ence the work reported in this paper.
11985980 mm4.
K Rotational stiffness of the beam end connection (is Sj = 170
Acknowledgements
kNm/rad from Table 3 for SFSc1)
L Beam span is 3000 mm (based on a span‐to‐depth ratio of 20 for
The authors wish to thank Centre for Engineering Research and
150 × 100 × 10 profile).
Development (CERD), APJ Kerala Technological University (KTU)
λQ is calculated from the inverse of Equation (B.1):
Trivandrum, Kerala, India (Research grant No: KTU/RESEARCH
3/1459/2017) for funding this research project as a part of Research 1
λQ ¼ ðB:2Þ
Seed Money. λδ
Assuming the pultruded FRP beam is shear‐rigid, β, λδ (B.1) and λQ
Appendix A (B.2) are determined as:
A sample calculation of the joint classification boundaries for a EI 28 1000 1:198 107
β¼ ¼ ¼ 0:66
beam span of 3000 mm corresponding to a practical span‐to‐depth KL 170 10 3000
6
[4] Strongwell. Strongwell design manual. Bristol, Virginia, USA: Strongwell [26] Turvey GJ, Cooper C. Characterisation of the short-term static moment– rotation of
Corporation; 2010. bolted connections between pultruded GRP beam and column WF connections. In:
[5] Fiberline Composites Fiberline design manual 2002 Fiberline Composites A/S El-Badry M, editor. Adv. Compos. Mater. Bridg. Struct., Canadian Society for Civil
Kolding, Denmark. Engineering; 1996, p. 927–934.
[6] Creative Pultrusions The new and improved Pultex pultrusion design manual 2004 [27] Russo S. On failure modes and design of multi-bolted FRP plate in structural joints.
Creative Pultrusions Inc. Alum Bank, PA, USA. Compos Struct 2019;218:27–38. http://dx.doi.org/10.1016/j.compstruct.2019.
[7] Bank LC. Progressive failure and ductility of FRP composites for construction: 03.048.
review. J Compos Constr 2013;17:406–19. http://dx.doi.org/10.1061/(ASCE) [28] Feroldi F, Russo S. Structural behavior of all-FRP beam-column plate-bolted joints.
CC.1943-5614.0000355. J Compos Constr 2016;20.
[8] Girão Coelho AM, Mottram JT. A review of the behaviour and analysis of bolted [29] Smith SJ, Parsons ID, Hjelmstad KD. An experimental study of the behavior of
connections and joints in pultruded fibre reinforced polymers. Mater Des connections for pultruded GFRP I-beams and rectangular tubes. Compos Struct
2015;74:86–107. http://dx.doi.org/10.1016/j.matdes.2015.02.011. 1998;42:281–90. http://dx.doi.org/10.1016/S0263-8223(98)00082-8.
[9] Qureshi J, Mottram JT. Behaviour of pultruded beam-to-column joints using steel [30] Smith SJ, Parsons ID, Hjelmstad KD. Experimental comparisons of connections for
web cleats. Thin-Walled Struct 2013;73:48–56. http://dx.doi.org/10.1016/j. GFRP pultruded frames. J Compos Constr 1999;3:20–6.
tws.2013.06.019. [31] Martins D, Proença M, Correia JR, Gonilha J, Arruda M, Silvestre N. Development
[10] Qureshi J, Mottram JT. Response of beam-to-column web cleated joints for FRP of a novel beam-to-column connection system for pultruded GFRP tubular profiles.
pultruded members. J Compos Constr 2014;18:04013039. http://dx.doi.org/ Compos Struct 2017;171:263–76. http://dx.doi.org/10.1016/j.compstruct.2017.
10.1061/(ASCE)CC.1943-5614.0000392. 03.049.
[11] Qureshi J, Mottram JT. Moment-rotation response of nominally pinned beam-to- [32] Ascione F, Lamberti M, Razaqpur AG, Spadea S, Malagic M. Pseudo-ductile failure
column joints for frames of pultruded fibre reinforced polymer. Constr Build Mater of adhesively joined GFRP beam-column connections: an experimental and
2015;77:396–403. http://dx.doi.org/10.1016/j.conbuildmat.2014.12.057. numerical investigation. Compos Struct 2018;200:864–73. http://dx.doi.org/
[12] Vallée T, Tannert T, Meena R, Hehl S. Dimensioning method for bolted, adhesively 10.1016/j.compstruct.2018.05.104.
bonded, and hybrid joints involving Fibre-Reinforced-Polymers. Compos Part B [33] Razaqpur AG, Ascione F, Lamberti M, Spadea S, Malagic M. GFRP hollow column
Eng 2013;46:179–87. http://dx.doi.org/10.1016/j.compositesb.2012.09.074. to built-up beam adhesive connection: mechanical behaviour under quasi-static,
[13] Ascione F, Lamberti M, Razaqpur AG, Spadea S. Strength and stiffness of cyclic and fatigue loading. Compos Struct 2019;224:. http://dx.doi.org/10.1016/
adhesively bonded GFRP beam-column moment resisting connections. Compos j.compstruct.2019.111069111069.
Struct 2017;160:1248–57. http://dx.doi.org/10.1016/j.compstruct.2016. [34] Turvey GJ. Moment-rotation tests on bolted end connections in pultruded GRP
11.021. beams-tests with stainless steel cleats and an assessment of their performance
[14] Kelly G. Load transfer in hybrid (bonded/bolted) composite single-lap joints. relative to GRP cleats. ECCM9, Compos. – from Fundam. to Exploit., Brighton:
Compos Struct 2005;69:35–43. http://dx.doi.org/10.1016/j.compstruct.2004. 2000.
04.016. [35] ASCE Pre-standard. Pre-standard for Load and Resistance Factor Design (LRFD) of
[15] Weitzenböck JR, McGeorge D. Science and technology of bolt-adhesive joints. Adv pultruded Fiber Reinforced Polymer (FRP) structures (Final). American
Struct Mater 2011;6:177–99. http://dx.doi.org/10.1007/8611_2011_54. Composites Manufacturers Association, American Society of Civil Engineers; 2010.
[16] Kelly G. Quasi-static strength and fatigue life of hybrid (bonded/bolted) composite [36] BS EN 1993-1-8:2005. Eurocode 3: Design of steel structures - Part 1-8: Design of
single-lap joints. Compos Struct 2006;72:119–29. http://dx.doi.org/10.1016/ joints. London, UK: British Standards Institution; 2005.
j.compstruct.2004.11.002. [37] IS 800. General Construction in Steel - Code of Practice. Bureau of Indian
[17] Anonymus. Guide to the structural use of adhesives. London, UK: Institution of Standards; 2007.
Structural Engineers; 1999. [38] Clarke JL. Structural Design of Polymer Composites- EUROCOMP Design Code and
[18] Turvey GJ. Analysis of pultruded glass reinforced plastic beams with semi-rigid Handbook. 1996.
end connections. Compos Struct 1997;38:3–16. http://dx.doi.org/10.1016/S0263- [39] Smith PA, Ashby MF, Pascoe KJ. Modelling clamp-up effects in composite bolted
8223(97)00036-6. joints. J Compos Mater 1987;21:878–97. http://dx.doi.org/10.1177/
[19] Bank LC, Mosallam AS, Gonsior HE. Beam-to-column connections for pultruded 002199838702101001.
FRP structures. Proc. 1st Mater. Eng. Congr. ASCE, Denver, USA: 1990, p. 804–13. [40] Gorenc BE, Tinyou R, Syam AA. Steel designers’ handbook. Sydney, NSW 2052,
[20] Bank LC, Mosallam AS, McCoy GT. Design and performance of connections for Australia: University of New South Wales Press Ltd.; 2005.
pultruded frame structures. J Reinf Plast Compos 1994;13:199–212. http://dx.doi. [41] Li B, Leong CL. Experimental and numerical investigations of the seismic behavior
org/10.1177/073168449401300302. of high-strength concrete beam-column joints with column axial load. J Struct Eng
[21] Bank LC, Yin J, Moore L, Evans DJ, Allison RW. Experimental and numerical 2015;141:04014220. http://dx.doi.org/10.1061/(ASCE)ST.1943-541X.0001191.
evaluation of beam-to-column connections for pultruded structures. J Reinf Plast [42] Allam SM, Elbakry HMF, Arab ISE. Exterior reinforced concrete beam column joint
Compos 1996;15:1052–67. http://dx.doi.org/10.1177/073168449601501005. subjected to monotonic loading. Alexandria Eng J 2018;57:4133–44. http://dx.
[22] Bass AJ, Mottram JT. Behaviour of connections in frames of fibre- reinforced- doi.org/10.1016/j.aej.2018.10.015.
polmer section. Struct Eng 1994;72:280–5. [43] Li B, Chua HYG. Seismic performance of strengthened reinforced concrete beam-
[23] Mottram JT, Zheng Y. Further tests on beam-to-column connections for pultruded column joints using FRP composites. J Struct Eng 2009;135:1177–90. http://dx.
frames: Flange-cleated. J Compos Constr 1999;3:3–11. doi.org/10.1061/(ASCE)0733-9445(2009)135:10(1177).
[24] Mottram JT, Zheng Y. State-of-the-art review on the design of beam-to-column [44] Mashaly E, El-Heweity M, Abou-Elfath H, Osman M. Finite element analysis of
connections for pultruded frames. Compos Struct 1996;35:387–401. http://dx.doi. beam-to-column joints in steel frames under cyclic loading. Alexandria Eng J
org/10.1016/S0263-8223(96)00052-9. 2011;50:91–104. http://dx.doi.org/10.1016/j.aej.2011.01.012.
[25] Turvey GJ, Cooper C. Semi-rigid pultruded GRP frame connections: tests to [45] Turvey GJ, Zhang Y. Tearing failure of web-flange junctions in pultruded GRP
determine static moment–rotation characteristics. 7th Eur. Conf. Compos. Mater. profiles. Compos Part A Appl Sci Manuf 2005;36:309–17. http://dx.doi.org/
ECCM-7, Cambridge: Woodhead Publishing Ltd.; 1996, p. 295–300. 10.1016/j.compositesa.2004.06.009.