Beruflich Dokumente
Kultur Dokumente
Advisor(s)
Su, KL
Author(s)
Siu, Wing-ho.; .
Citation
Issued Date
URL
Rights
2009
http://hdl.handle.net/10722/56994
FLEXURAL STRENGTHENING OF
REINFORCED CONCRETE BEAMS BY
BOLTED SIDE PLATES
by
SIU, WING HO
B.Eng., HKU, HKSAR, PRC
PHD THESIS
Other than the deterioration problem, a large number of old buildings are idled
due to economic restructuring in modern society. In order to revive these idled
buildings, the concept of adaptive reuse is often employed. This concept
involves adaption of existing buildings for the purpose other than those
originally planned so that demolition of buildings can be avoided. During this
adaption process, structural load paths may be changed and strengthening is
often required.
can be maintained. Due to the above advantages, the BSP technique was
investigated in this study.
This study aims to improve the understanding of and develop suitable analysis
for BSP beams in order to predict the response of BSP beams and facilitate the
development of suitable design guidelines. To begin with, full scale tests of BSP
beams were conducted. The overall response, in particular the partial interaction
behaviour was captured. It was observed that the bolt-plate arrangement and the
formation of plastic hinges in RC have dominant effects on the behaviour of
BSP beams.
In conclusion, this study provides an original and useful tool for predicting the
response of BSP beams.
(484 Words)
ii
DECLARATION
I declare that the thesis and the research work thereof represents my own work,
except where due acknowledgement is made, and that it has not been previously
included in a thesis, dissertation or report submitted to this University or to any
other institution for a degree, diploma or other qualifications.
Signed .
(SIU, Wing Ho)
iii
ACKNOWLEDGEMENTS
The author would like to express his sincere gratitude to his supervisor Dr. Ray
Su for his guidance in the research work and all his invaluable advices given
throughout the whole doctoral study.
The generous financial supports given by the Research Grant Council of Hong
Kong (Project no. HKU7166/08E) and technical supports by the HILTI
Corporation are gratefully acknowledged.
Finally, the author wishes to express his gratitude to his family and Ms. Silvia
Sze for their loving encouragement and supports.
iv
CONTENTS
ABSTRACT .
DECLARATION ..
ACKNOWLEDGMENTS
CONTENTS .
LIST OF FIGURES ..
LIST OF TABLES ...
i
iii
iv
v
x
xv
1
1
5
9
9
9
9
10
plates ...
2.3 Characteristics of RC Beams strengthened with bolted side plates
under flexure ..
2.3.1 Shear connection between RC beam and steel plates .
12
13
13
16
16
18
19
22
25
20
27
27
28
28
29
33
38
40
40
40
41
42
46
46
46
48
49
55
55
55
58
60
63
50
54
68
interaction .
4.8 Partial interaction response at critical section
4.8.1 Variation of strain factor of the critical section ..
4.8.2 Variation of curvature factor of the critical section
4.9 Longitudinal slip response .
73
75
77
81
83
vi
93
93
95
95
97
99
102
102
112
112
115
115
116
6.2.1 Assumptions
6.2.2 Numerical model for bolts under multi-directional shear ...
6.2.2.1 First stage yielding ...
6.2.2.2 Second stage yielding ...
6.2.3 Incremental load-deformation relationship of bolt groups ..
6.3 Implementation of non-linear bolt group model
116
118
119
120
123
125
106
5.4 Numerical verification 108
5.4.1 Non-linear finite element analysis .. 108
vii
135
BEAMS
7.1 Overview
7.2 Formulation
7.2.1 General
135
137
137
138
143
145
147
153
159
159
160
161
165
167
169
169
170
173
173
179
181
183
183
184
186
191
191
viii
196
198
202
207
210
CHAPTER 9 CONCLUSIONS
9.1 Summary
9.2 Conclusions
213
213
215
216
224
227
233
237
LIST OF PUBLICATIONS ..
243
ix
LIST OF FIGURES
Figure 1.1
Figure 1.2
Figure 1.3
Figure 2.1
Figure 2.2
Figure 3.1
Figure 3.2
Figure 3.3
Figure 3.4
Figure 3.5
Figure 3.6
Figure 3.7
Figure 3.8
Figure 3.9
Reinforcement cage 40
53
Figure 4.1
56
57
Figure 4.4
Figure 4.5
62
Figure 4.7
Figure 4.8
Figure 4.9
Figure 5.2
Figure 5.3
xi
Figure 5.4
Figure 5.5
Figure 5.6
Figure 5.7
Figure 5.8
Figure 5.9
Figure 6.1
Figure 6.2
Figure 6.3
Figure 6.4
Figure 6.5
Figure 6.6
Figure 6.7
Figure 6.8
Figure 6.9
Figure 7.1
Figure 7.2
Figure 7.3
Figure 7.4
Figure 7.5
Figure 7.6
Figure 7.7
xii
Figure 7.9
Figure 8.2
Figure 8.3
Figure 8.4
Figure 8.5
Figure 8.6
Figure 8.7
Figure 8.8
xiii
zone 185
Figure 8.9
Figure 8.10 Section analysis with strains at centroidal level of the steel
plates as pivotal points ... 195
Figure 8.11 Layered model adopted in modified section analysis 196
Figure 8.12 Load-slip model of bolts adopted in the current analysis ... 198
Figure 8.13 Program algorithm .. 201
Figure 8.14 Comparison of moment-curvature responses a) Specimen
SBSP, b) Specimen WBSP, c) Specimen WBWP, and d)
Specimen SBWP 205
Figure 8.15 Comparison of moment contributed by various actions a)
Specimen SBSP, b) Specimen WBSP, c) Specimen WBWP,
and d) Specimen SBWP . 206
Figure 8.16 Specimen to be strengthened in the example . 207
Figure 8.17 Bolt-plate arrangement of design example . 208
Figure 8.18 Simulated moment-curvature response of the mid-span
section . 210
xiv
LIST OF TABLES
Table 3.1
Table 3.2 Concrete mix adopted for producing a cubic meter of concrete . 34
Table 3.3 Cube compressive strength of concrete .. 34
Table 3.4 Material properties of T-16 reinforcements 34
Table 3.5 Material properties of T-10 reinforcements 35
Table 3.6 Material properties of strengthening steel plates 35
Table 3.7 Experimental scheme of bolt shear test .. 46
Table 3.8 Cube compressive strength of concrete in bolt shear tests . 47
Table 3.9 Values for bi-linear model of HAS-E bolts 52
Table 4.1 Failure modes and the increases in strength of BSP beams 63
Table 4.2 Displacement ductility factors and post-elastic strength
enhancement factors 63
Table 4.3 Strengthening performance of BSP beams . 75
Table 6.1 Specimen detail in the experimental study by Crawford and
Kulak ... 129
Table 6.2 Mechanical properties of bolts used in the simulations .. 129
Table 6.3 Comparison of the numerical results with the experimental
results .. 131
Table 7.1 Material properties adopted in simulations . 153
Table 7.2 Comparison of peak loads obtained from the present numerical 156
analysis and previous experiments ..
Table 8.1 Comparison of predicted strength with experimental results .. 203
xv
LIST OF ABBREVIATIONS
BSP
Bolted side-plated
DCBA
FRP
Fibre-reinforced plastic
LDT
LVDT
RC
Reinforced concrete
SBSP
SBWP
WBSP
WBWP
xvi
LIST OF NOTATIONS
A
: Area of section
Ac
: Area of concrete
Ap
Ast
Asc
: Strain-displacement matrix
Ba
Bb
: Breath of section
: cos
dc
cT
cT
cT
: Displacement vector of the RC beam ( d1 , d 2 K d n )
dic
dip
dp
d sc
d st
Ec
Ep
Es
ET
: Youngs modulus
EAc
EAp
xvii
EI c
EAp
Fi
Fp , Fp ( x)
: Plate force
Fp , fi
Fx , Fy , M
f ci
f co
f cu
f yp
f ys
gi
: Constant
hc
hna
hp
I ccy
xviii
I cs
I ssx
I zz
K bi
K ic
K ip
K cc
K cs
K ss
Kx
Ky
ke
km ( x)
kmi
kp
k p
k p
kr
kx
ky
: Length of element
xix
LLVDT
Lrc
Lrs
Ls
: Length of span
lab
lab
: Translation direction
M ( x)
M c ( x)
M p ( x)
Mu
: Ultimate moment
nc
np
np
: Bolt force
Pn
Psh
Psh , fsc
xx
Pt
Pu
Px
Py
Pyf
Py
Py
: Constant
Qx , Qy , Q
Rccy
Rcsx
Rcsy
Rssx
Rx
Ry
: sin
sl , sv , s
sl ( x)
sl ( x)
xxi
the beam
sl ,max
snorm ( x)
su
sv ( x)
sy
ua , ub
b
un , ut
ux , u y
: Volume
V ( x)
va , vb
b
vc ,mid
Wc
Wp
W ,W ( x)
: Applied load
w1 , w2 , w3
xxii
xi , yi
xo , yo
: Strain factor
: Curvature factor
lc, hp
lp,hp
cv , stz
vp, stz
: Strain
axial
bending
ci
co
xxiii
hc
hp
hslip
: Strain at level y
, (x)
: Curvature
c ( x)
p ( x)
it
: Angles
a , b
and node b
: Stress
: Stress of concrete
sc
st
xxiv
CHAPTER 1
INTRODUCTION
1.1 OVERVIEW
Hong Kong has undergone a rapid development in the 1960s and 70s.
Lots of buildings were constructed at that time. According to the Buildings
Department, the number of buildings aged over 30 years in Metro areas
dramatically raised from 8600 in 1998 to over 13000 in 2008 and will be further
increased by 30% to over 16000 in year 2016. Many of these buildings are
dilapidated and subjected to problems like concrete spalling, corrosion
reinforcement etc. Without proper way to upgrade these buildings, the health
and safety of the general public is jeopardized.
No. of buildings
18000
16000
14000
12000
10000
8000
6000
4000
2000
0
1998
2000
2002
2004
2006
2008
2010
2012
2014
2016
Year
buildings
were
converted
into
residential,
office,
retail,
Column/Wall pier
Column/Wall pier
Bolts
Steel plates
b)
a)
a) Different external strength methods for RC beams are compared. After that,
previous studies on BSP beams are reviewed and a brief introduction to the
behaviour of BSP beams is given to provide background knowledge for
readers. Literature focused on partial interaction in standard composite
beams are also presented
b) Full scale BSP beam specimens are tested to reveal the behaviour of BSP
beams. Particular concerns are put on the effect of bolt-plate arrangement on
post-yield behaviour of BSP beams, the mechanical slip of the connecting
bolts and the strength reduction occurred due to partial interaction between
the RC beam and the steel plates.
c) A numerical model for the non-linear load-deformation for bolts and bolt
groups are developed to model the behaviour of the connecting media. With
the used of the numerical model for the connecting media, a finite element
model for BSP beams is developed. By carrying out simulation using the
developed model, the non-linear behaviour of BSP beam can be revealed.
d) Theories governing the partial interaction between the RC beam and the
steel plates are developed. The longitudinal slip along the beam is related to
the response of section with critical moments by two new parameters which
quantify the relationship between strain profiles of the RC beam and the
steel plates in BSP beams. The theory is validated by a simple spreadsheet
and illustrated through a design example
processes and instrumentation are included. The arrangement and results of the
accompanied shear test of anchor bolts are also included.
CHAPTER 2
LITERATURE REVIEW
2.1 OVERVIEW
In this chapter, a detailed literature review on bolted side-plated (BSP)
beams is presented. A brief introduction of different available external
strengthening methods for RC beams is first given. Then a detailed review on
previous researches on the behavioural characteristics of BSP beams under
flexure is presented. Previous numerical studies on composite beams which
behave similarly to BSP beams are then reported in detail. Finally analyses
currently available for BSP beams are also described.
The peeling failure problem, which the concrete cover adjacent to the plate
end crack and peel off from the RC core bounded by the stirrups, was discussed
by (Zhang et al., 1995). As the steel plate is attached to the cover of the RC
beam, the tensile stress developed in the steel plates has to be transferred by the
concrete cover to the structural core. This induced a high stress state in the
concrete cover and the concrete cover may detached from the structural core as
no transverse reinforcement is presence in the concrete cover, leading to this
type of premature failure.
10
As gluing plates on the surface of beams can lead to premature failure due
to the lack of structural integrity, the use of bolts to anchor steel plates on RC
beams appears to be the solution of the problem. As bolts are extended into the
interior of the structural core, the structural integrity is greatly improved and the
chance of premature failure, in terms of debonding or shear peeling failure, is
lowered.
Concerning both the strength and ductility concern, Ahmed et al. (2000)
studied the feasibility of bolting steel plates to the side faces of RC beams for
strengthening the beams. As the steel plates are arranged vertically, the moment
of inertia is much higher. This means that besides serving as additional
reinforcements and working with concrete in compression for additional
flexural resistance, the strengthened plates work independently in flexure to
resist part of the total moment. A number of BSP beams with different bolt-
11
Despite of the above advantages, the FRP plated beams are prone to
premature brittle failures like the debonding failure, which is either induced by
the high interfacial stress between concrete and the plates, leading to the
anchorage ends of plates debonding from the concrete face, or induced by the
shear/flexural shear cracks, results in separation of the concrete cover from the
concrete core. (Teng et al., 2002). In order to prevent such failure, additional
arrangements like the use of U-jacket and anchoring bolts at the plate ends are
needed, which in turn increases the difficulties in construction. Another major
problem concerning the use of FRP plates is the need of fire protection cover
which is usually provided by spraying concrete on the surface of the FRP plates.
12
This compensates the advantage of high strength to weight ratio and so the use
of FRP is not supreme to other strengthening techniques.
The strength of shear connection has a crucial effect on the failure mode
and the ultimate flexural strength of the composite beam. When the strength of
13
Psh
.
Psh , fsc
(2.1)
One of the early works on the effects of degree of shear connection on the
behaviour of composite beams was carried out by Chapman and Balakrishnan
(1964). In their works, 17 simply-supported composite beams with different
degrees of shear connection, ranging from 0.79 to 2.50 were tested under three
point bending. Concrete crushing failure was observed for all specimens with
full shear connection, while shear failures of shear connection dominated the
specimens with partial shear connection.
14
Johnson and May (1975) pointed out that the composite beam with full
shear connection may subject to shear failure since the flexural strength of the
composite beam can be too high. This is not desirable as the shear failure of
beam is relatively brittle. To ensure a desirable flexure mode of failure, it was
suggested that the partial shear connection design can be adopted so that the
flexural strength is reduced in order to ensure the ductile flexural failure of the
beam.
15
out and it was observed that addition of bolts in the partial interaction case can
reduce the maximum slip of shear connectors and hence prevent the premature
shear connector failure.
Analytical studies were also carried out in the same study (Newmark et al.
1951) and an elastic model for longitudinal slip was developed. It was derived
that the longitudinal slips and shear flow between the components was governed
16
17
18
19
In the research works by Smith et al. (1999a, 2000), the buckling problem
in BSP beams is simplified as a unilateral buckling problem of steel plates
restrained at discrete boundary points. Analytical study was carried out based on
the Rayleigh-Ritz method. Using approximate buckling deformations applicable
to different restraint and loading conditions, the expressions for the local
buckling coefficient was generated. The analytical results and the proposed
buckling coefficient was validated by a series of local buckling push tests
(Smith et al. 1999b, 2001), where steel plates bolted at boundaries were tested
under combined axial force and bending. With a good agreement between
experimental and analytical results, the proposed expression for local buckling
coefficient can be used in practical design and serve as a guideline for bolt
arrangement to prevent buckling of plates.
20
21
concrete slab, steel beam and the shear connectors. The method of finite
difference, while can simulate the beam responses in some simple cases, is
problem specific as the boundary conditions involved is different in each case.
From this point of view, the usefulness of the finite difference method is less
than the finite element approach.
It can be seen that the finite element technique is useful and effective in
handling the problem of partial interaction in composite beam systems.
However, the previous approaches assumed a uni-directional slip of the shear
connections, which is generally valid in standard composite structures but not in
bolted side-plated beams. In bolted side-plated beam, the situation is even more
complicated. As side-plates are used, the connecting bolts are subject to partial
interaction in both transverse and longitudinal directions simultaneously. Being
loaded with shear forces of unknown magnitudes in longitudinal and transverse
directions, the situations changes from an uni-directional load-deformation
problem to a multi-directional one. To model the response of BSP beams with
shear connectors subject to multi-directional shear, a suitable model has to be
first developed.
22
When the strength of the shear connection is less than the plate force
obtained in the full-shear-connection full-interaction analysis, partial shear
connection occurs and the flexural strength can then be estimated by the partialshear-connection partial-interaction analysis as illustrated in Figure 2.2. The
strain profiles of steel plates and the RC beams are moved independently with
the same curvature until each of the resultant force of RC beams and steel plates
as shown in Figure 2.2b are equal to the strength of the shear connection Psh.
Then the partial-shear-connection partial-interaction flexural strength can be
solved by taking moments about any levels.
23
Fc
Fp
Fs
Fc+ Fs+Fp=0
a) Strain profile
Fc
Fp=Psh
Fs
Fc+ Fs =Psh
Resultant stress and force block
Strain profile
The above analysis assumes that partial interaction occur only when partial
shear connection occurs. While such assumption is commonly adopted for
typical composite beams, this is not appropriate as partial interaction occurs
even in the presence of full shear connection, as evidenced by the study carried
out by Ahmed et al. (2000). It was observed experimentally that the effect of
partial interaction can affect the strength of the beam even in the case of full
shear connection and the strength can be over-estimated by 15% of the increase
of the flexural strength due to the addition of the strengthening scheme. Similar
comment was made for typical composite beams when the strength of the steel
girder is greater than the strength of the RC slab (Oehlers et al., 1997a). Such a
24
reduction is due to the finite stiffness of the shear connection, which cannot be
considered in a rigid plastic analysis since the components are assumed either
unloaded or fully-loaded and no stiffness is exist for the components. To
probably consider the flexural strength of BSP beams, Oehlers et al. (2000)
proposed a 15% reduction in increase in strength for consider the effect of
partial interaction. However, the suggestion is only based on limited
experimental results and in order to understand better the strength reduction due
to partial interaction, more experimental study and a better analysis like the
non-linear moment-curvature section analysis, with the incorporation of
appropriate stiffness of connection is required.
25
sectional strain profile between the RC beam and steel plates since these
behaviour affects fundamental design parameters like flexural strength and
stiffness of BSP beams.
26
CHAPTER 3
EXPERIMENTAL STUDY ON BOLTED SIDEPLATED BEAMS WITH PARTIAL INTERACTION
3.1 OVERVIEW
A number of studies, as described in Chapter 2, had been carried out to
investigate the flexural behaviour of bolted side-plated RC beams (BSP beams).
These studies showed that BSP beams are effective in strengthening existing
RC beams. However, the occurrence of partial interaction in BSP beams could
significantly affect the behaviour of the beams and experimental studies have
yet to be carried out to investigate in detail the partial interaction response of
BSP beams in either the structural element level, which is the slip between the
RC beam and steel plates along the whole beam, or the sectional level, which is
separation of the strain profile of steel plates from that of the RC beam. Also
other aspects being important to the application of BSP beams remain uncertain,
including the behavioural differences of BSP beams with different bolt-plate
combinations, the strength reduction resulted from the occurrence of partial
interaction and the ductility performance of BSP beams.
1)
the partial interaction responses in both structural element level, in which the
components being connected by bolts and slip between each other, and sectional
level, in which the strain profile of steel plates separates from that of the RC
beam, 2) the performance of BSP beams in terms of flexural strength and
ductility, 3) the strength reduction of BSP beams due to partial interaction and,
27
28
symmetrically in four-point bending with point loads 600mm from either side of
the mid-span location. Loading was applied by a 1000kN hydraulic jack
arranged vertically, and divided into two equal point loads through a transfer
beam hinged at three points, one to the hydraulic jack and two to the specimen.
With this arrangement, a pure bending zone with constant moment was present
so that the flexural behaviour under pure bending could be studied.
1000kN hydraulic jack
Hinges
Loading beam
Roller
Strong ground
1200
1200
1200
29
2T10
317
350
30
225
3T16
30
Psh
,
Fp , fi
(3.1)
where Psh and Fp,fi are the total strength of bolts on a shear span and the plate
force at ultimate limit state in the full interaction case respectively.
31
relationship for steel plates and reinforcement. In each load step, incremental
curvatures were applied to the section. The neutral axis level was shifted until
horizontal equilibrium was reached. With the strain profile being fixed, the
moment of the section and the plate force can be solved by taking moment
about the neutral axis.
The plate force at the ultimate state and the degree of shear connection of
the strengthened specimens are listed in Table 3.1. The degree of shear
connection of Specimens SBSP and SBWP are 1.00 and 1.26, respectively,
which imply that sufficient bolt forces have been provided. Thus the bolt
arrangement is Strong. Conversely, the degree of shear connection of
Specimens WBSP and WBWP are 0.38 and 0.75, respectively, which signify
that these specimens have Weak bolt arrangements.
Unit SBSP and Unit WBSP
250
Unit SBSP
150
Unit WBSP
Unit WBWP
250
75
Unit SBWP
Anchor bolt
Steel plate
32
Specimen
Specimen
Specimen
Specimen
SBSP
WBSP
WBWP
SBWP
608
228
228
380
6150dp.
6150dp.
675dp.
675dp.
605
605
302
302
1.00
0.38
0.75
1.26
(kN)
Plate size# (mmmm)
Plate force at ultimate state in full
interaction analysis Fp,fi (kN)
Degree of shear connection (Pb/Fp,fi)
Specimens SBSP and WBSP were strengthened by the same size of plates
which are 6mm150mm deep on each side face of the beams, but with Strong
and Weak bolt arrangements respectively. Specimens SBWP and WBWP have
the same plate arrangement but with different degree of shear connections. By
comparing these two pairs of specimens, the effect of strong and weak bolt
arrangements on the behaviour of specimens could be studied. Meanwhile, the
bolt arrangements of Specimens WBSP and WBWP are identical but with
different plate arrangements. By comparing the results of these two specimens,
the effect of plate geometry on the performance of the beams could be
investigated.
33
The average concrete compressive strength of cubes for the control specimen is
35.2 MPa and for the BSP specimens are given in Table 3.3.
Table 3.2 Concrete mix adopted for producing a cubic meter of concrete
kg/m
water
cement
fine aggregate
10mm aggregate
199.6
278.9
1024.5
837.6
Specimen
Specimen
Specimen
Specimen
NBNP
SBSP
WBSP
WBWP
SBWP
35.5
34.5
34
34.3
35.5
36.2
35
37
34.1
35.5
34
34.2
31.9
37
35.0
Average
35.2
34.6
34.3
35.1
35.3
545
189
525
184
540
188
Average
537
187
34
550
185
540
194
Average
545
190
150mm plate
Youngs
modulus
(GPa)
Yield
strength
(MPa)
Youngs
modulus
(GPa)
Yield
strength
(MPa)
202
325
214
335
215
355
216
350
191
335
206
320
Average
203
338
212
335
HAS-E anchor rods of 12mm diameter (Hilti, 2005) were used as the
mechanical connectors of all specimens. The HAS-E rod was in Grade 5.8, with
a minimum 5m galvanized surface. Dynamic sets, instead of ordinary washers
and nuts were used in this study. A spherical washer, an injection washer and an
ordinary nut are included in the dynamic set, as illustrated in Figure 3.5. With
this system, epoxy grouts could be easily injected into the gaps between
connecting components and the bolt shaft, so that any undesirable slips due to
the presence of gaps between components could be avoided. Bolt shear tests
were conducted and described in detail in Section 3.3. The load-slip response of
35
600
Stress (MPa)
500
400
300
200
75mm - Specimen 1
75mm - Specimen 2
75mm - Specimen 3
100
0
0
0.05
0.1
0.15
0.2
0.25
Strain
600
Stress (MPa)
500
400
300
200
150mm - Specimen 1
150mm - Specimen 2
150mm - Specimen 3
100
0
0
0.05
0.1
0.15
0.2
0.25
Strain
36
Nut
Spherical
washer
Injection
washer
Back
Front
Injection washer
(a)
(b)
Figure 3.5 Dynamic set washer a) Diagrammatic illustration and b) Actual
arrangement
Bolt force P
P2 (Fracture point)
P1 (Turning point)
A bi-linear model
Deformation s
P1
P2
s (mm)
0.625
4.0
P (kN)
50
76
37
In the RC fabrication stage, the RC core of the specimens was and cast in
wooden formwork. The reinforcement cage was first fabricated and set at the
required position in formwork, as shown in Figure 3.9. Holes for future
installation of bolts were also reserved. In real construction, holes for the
installation of anchor bolts are drilled on the existing beams; and in some cases,
the existing shear reinforcement may disturb the installation. To prevent this,
through holes were reserved prior to pouring of concrete using aluminium tubes
with internal diameter of 17mm, as shown in Figures 3.7 and 3.8. The tubes
would be cast in the specimens during the load test. Considering the load path
of the bolt plate system, the in-plane shear force within the steel plates is
transferred through the bolts in shear and then directly to the concrete core
through the bearing layers between the bolt shaft and concrete, i.e. concrete and
epoxy grout in real situation and concrete, epoxy and the aluminium tube
embedded in this experimental study. It is believed that the change in bearing
stiffness due to the use of aluminium tubes is small comparing with the stiffness
of the bolts in shear. The addition of aluminium tubes in the anchorage system
would not significantly affect the experimental results. The specimens were
then cast and left for at least 14 days of curing before proceeding to the
strengthening stage.
38
In the strengthening stage, the steel plates for strengthening were fixed
onto the RC specimens. Bolt holes were first drilled at the required locations on
the steel plates. A 2mm hole clearance was adopted so as to facilitate bolt
installations. With holes drilled, the steel plates were mounted on the RC
components with clamps and set at the required positions. After that, RE-500
grout was injected into the holes previously reserved in the RC fabrication stage.
The dosage of injection was calculated to prevent overflow of grout. The anchor
rods on the two sides of the holes were then installed into the hole
simultaneously until reaching the appropriate depth. After the grout leaving for
curing of a minimum of 24 hours, as referring to the manual from Hilti (2005),
dynamic sets were installed and the same grout was injected again through the
holes on dynamic set to fill up the gap between anchor rods and bolt holes on
steel plates. The new grout filled was cured for another 24 hours. Finally, the
clamps mounting the steel plates were removed and the whole fabrication
process was completed.
Steel plates
Concrete
Concrete
Dynamic set washer
Aluminium tube
Adhesive RE-500
39
3.2.6 Instrumentation
3.2.6.1 Strain gauge arrangements
Strain gauges were attached on the steel plates, the longitudinal
reinforcements to measure the internal strains which were then used to
determine the internal stress distributions in specimens. These strain gauges
were installed along the direction of the beam span at five different sections.
Three sections were within the constant moment zone and two at the middle of
the left and right shear spans, as illustrated in Figure 3.10. Three sets of strain
gauges were installed in the constant moment zone to measure the average
response of the zone. The measured data may be affected by localized effects
40
800
400
400
800
600
a)
600
800
400
400
800
50
50
150
600
75
800
400
400
600
32.5 37.5
32.5
75
b)
800
41
600
600
D1
600
D3, D4
D2
600
D5, D6
600
D7, D8
600
D9
D10
B
D13
D12
D14
250
D11
B
LVDT
LDT
D9, D10
D3, D5, D7
D4, D6, D8
Section A-A
Section B-B
R2, R4
R1, R3
a
Specimens SBSP,
WBSP and WBWP
200
800
Specimen SBWP
175
850
42
planes AD and EF of the RC beam and the steel plate, respectively, was
invented to trace the slip responses (see Figure 3.12).
The reference points B and C of the target plane EF of the steel plate
were set on the left and right hand sides of the target plane. As the reference
points B and C were not on the target plane, a metal strip was used to extend the
target plane to the reference points. The metal strip was anchored on the target
plane with two screws so that it would move rigidly with the target plane EF
under both translational and rotational movements. To ensure accurate
measurements, the metal strip was offset from the steel plate so that the strip
and the plate were not in contact, except at the anchorage points. By measuring
the relative distances between reference points, the slip between the target
planes could be evaluated. The details can be found in Appendix 1.
43
44
To data logger
LVDT
Extendible rod
(a)
Concrete beam
A
E
B
Plate
C
F
Metal
strip
D
(b)
45
10
12
16
46
10
12
16
36.5
38.4
36.4
38.8
40.7
34.5
38.6
39.7
35.0
Average
38.0
39.6
35.3
The presence of small gaps between the steel plates and the bolts was
undesirable as it would affect the load-slip response of bolts. Adhesive RE-500
(Hilti Corporation, 2005) was used to fill up the gaps within the bolt holes and
between steel plates and bolt shafts. Dynamic set washers (Hilti Corporation,
2005) were also used so that adhesives can be injected to fill up the gap between
47
bolt shafts and plate holes. By filling up all gaps with adhesive, all structural
components were in contact from the beginning when loadings were applied.
Holes with 2mm clearance were drilled in concrete for the installation of anchor
rods. The glue was injected from the dead end of the hole so that the air lock
problem would not exist. The glue was cured for a minimum of 24 hours and
the steel plates were set to the required position and fastened afterwards.
In reality, there was a slight variation in the location of bolt as the bolt hole
on the concrete block was 2mm larger than the diameter of bolts. This results in
slight difference of the top levels of the two vertical steel plates, leading to
uneven distribution bolt forces. To solve this imperfection, mechanical
connections, instead of welding was adopted in the connections of setup so that
the top levels can be adjusted to compensate its effect.
One of the main concerns of this test was to prevent buckling of steel
plates under compression. The distance between the bolt and the loading plate
was minimized, while still providing sufficient distance for the slip of bolts; the
48
Concrete block
Movable platform
connected to hydraulic jack
Movable platform
connected to hydraulic jack
Figure 3.16 Specimen with loading plates and LVDTs before testing
3.3.4 Loading History and Instrumentation
The load was applied monotonically until fracture of either bolts were
observed. The primarily objective of this test was to measure the loaddeformation relationship of bolts, which was defined as the total deformation in
a localized area near the bolts due to all deformations of the bolts, RC and steel
plates.
49
The slip of each bolt was measured by two LVDTs. Two L-shaped steel
plates were screwed on concrete for the LVDTs to sit on. Two metal plates were
welded on the steel plates at the horizontal level where the bolt was located at
the capture the vertical displacement of the plate relative to the concrete. The
plates were positioned near the edge of the steel plates so that the local effects
near the bolt on the readings were minimized.
From the above observations, a bi-linear model for simplifying the nonlinear responses of bolts subjected to in-plane shear is proposed. The bi-linear
model was defined by bolt forces and deformations at the turning point (P1) and
the failure point (P2), as illustrated in Figure 3.18. The proposed values for
modelling the behaviour of 10, 12 and 16 HAS-E bolts were summarized in
Table 3.9 and compared with the experimental results in Figure 3.19.
50
(a)
(b)
(c)
Figure 3.17 Load-slip responses of HAS-E bolts with diameter a)
10mm, b) 12mm, and c) 16mm
51
Bolt force P
P2 (Failure point)
P1 (Turning point)
Deformation s
10
12
16
s (mm)
P (kN)
s (mm)
P (kN)
s (mm)
P (kN)
P1
0.46
40
0.625
50
1.58
126
P2
2.4
53
4.0
76
6.8
158
52
(a)
(b)
(c)
Figure 3.19 Comparison of bi-linear models with load-slip responses of
HAS-E bolts with diameter a) 10mm, b) 12mm, and c)16mm
53
54
CHAPTER 4
RESULT AND ANALYSIS OF EXPERIMENTAL
STUDY ON BOLTED SIDE-PLATED BEAMS
4.1 OVERVIEW
In this Chapter, the test results and the associated data analysis of the
experimental study described in Chapter 3 are presented. The general responses
of BSP beams are first presented and then a detailed examination of the ductility
performance of BSP beams is given. The various controlling factors on BSP
beams are discussed. The latter part of this Chapter focuses on the partial
interaction responses and the related aspects like the strength reduction due to
partial interaction and longitudinal slip responses. The effect of moment
contribution by various components due to the existence of partial interaction
are reported and discussed in detail.
55
(a)
(b)
(c)
(d)
56
Outward deflections of plates were also observed at the plate ends as the
beam approached to failure as shown in Figure 4.3. The deflection was due to
an uneven distribution of plate forces along the thickness of the plate, which
occurred simultaneously with the tilting of the bolts. The outward deflections of
plates were restrained by the anchoring bolts located nearest to the plate ends.
Bolt rod
Steel plates
Bolt shaft
Reinforced
concrete
Tilting
(a)
(b)
57
For Specimen WBSP, which has the same plate arrangement with that of
Specimen SBSP, but with lesser bolts installed, the behaviour of the beam was
governed by bolts. The elastic stage terminated when the applied moment
reached 131kNm (0.88Mu) due to yielding of the bottom reinforcement. At that
instant, the tangential stiffness was abruptly dropped to 26% of the stiffness just
prior to yielding. The stiffness dropped gradually when the beam was deformed
further and reached zero when the moment reached 147kNm (0.99Mu). The
gradual drop in stiffness was due to the stiffness drop in bolts as bolt slips
increased. Two bolts on the same shear span and at the same side fractured
58
consecutively began from the one closer to support, at failure when the midspan deflection was 51mm and 54mm respectively.
For Specimen WBWP, which was strengthened by the same Weak bolt
arrangement with that of Specimen WBSP, but the Weak plate arrangement,
the bottom reinforcements yielded at the end of the elastic stage when the
moment reached 124kNm (0.93Mu). The remaining stiffness was about 10% of
the stiffness prior to yielding. The remaining stiffness sustained up to the failure
point when the moment and deflection were 133kNm and 50mm respectively.
Failure of the beam was due to consecutive bolt fractures, as similar to
Specimen WBSP.
For Specimen SBWP, which had the same Weak plate arrangement as
that of Specimen WBWP but 60% more bolts to anchor the plates, the cracked
elastic stiffness of Specimen SBWP was 14% larger than that of Specimen
WBWP. The bottom reinforcement yielded when the moment was equal to
124kNm (0.86Mu). After yielding, a constant tangential stiffness, equal to 17%
of its stiffness just prior to yielding, remained to resist further loading. When
the applied load reached 145kN, the load-deflection curve changed gradually
from an ascending trend to a descending trend and crushing of concrete began.
The descending branch of Specimen SBWP was milder than Specimen SBSP
and the residual strength in beam reached 85% of Mu at the point when the
beam was further deformed by 22mm.
59
Moment (kNm)
180
120
60
Unit SBSP
Unit WBSP
Unit WBWP
Unit SBWP
Unit NBNP
0
0
20
40
60
80
60
61
Actual behaviour
Bi-linear approx.
62
Table 4.1 Failure modes and the increases in strength of BSP beams
Failure mode
Mu (kNm)
Specimen NBNP
Concrete Crushing
101.4
Specimen SBSP
Concrete Crushing
161.5
59
Specimen WBSP
Bolt fracture
149.2
47
Specimen WBWP
Bolt fracture
133.4
32
Specimen SBWP
Concrete Crushing
144.6
43
y (mm) u (mm)
Py (kN) Pu (kN)
Specimen NBNP
17.4
63.0
3.62
101.4
101.4
1.00
Specimen SBSP
20.8
35.2
1.70
154.4
161.5
1.05
Specimen WBSP
21.0
48.2
2.30
140.2
149.2
1.06
Specimen WBWP
19.4
49.9
2.57
124.0
133.4
1.08
Specimen SBWP
18.0
44.0
2.45
123.6
144.6
1.17
63
steel plates was estimated by the difference of strain gauge readings at different
levels of the steel plate divided by the vertical separation between them (see
Figure 3.9 for the arrangements of strain gauges). The moment-curvature
relationships of all four BSP specimens are plotted in Figures 4.6a to 4.6d.
For Specimen SBSP, the curvature of the RC beam and the steel plate were
the same up to the point when the applied moment reached 135kNm. Then the
curvature of the steel plate increased much less per unit applied moment than in
the initial elastic stage. Plastic hinges began to form in the RC beam when the
moment reached 154kNm as the moment-curvature relationship of the RC beam
became flat. Finally the beam failed when the curvatures of the beam and steel
plate were 3910-3 m-1 and 1510-3 m-1 respectively.
64
linearly up to the point when the moment and curvatures were 124kNm and
1610-3 m-1 respectively. Plastic hinges formed in the RC beam and the
corresponding curvature reached 3110-3 m-1 when failure occurred. The
curvature in the steel plate stopped to increase with that of the RC beam when it
reached a maximum of 2310-3 m-1.
The moment resistance of BSP beams composes the moment resistances of the
RC section and steel section as well as the axial force couple generated from the
axial force interaction between the steel plate and RC section. One of the
possible reasons for the curvature reduction between the elastic limit is that
once the bottom reinforcement yields, the moment resistance provided by the
RC section can no longer be increased. The moment demands on the force
couple are increased with the increase in the applied load. Such an increasing
demand in the force couple continues up to the point when the concrete core is
65
not able to provide additional compression i.e. the ultimate point. As only one
row of bolts was provided in Specimen SBWP, the rotational restrain of the
constant moment zone is relatively small. Hence the additional tension force
stretch and straighten the steel plates up to the ultimate point. The curvature in
the steel plates is therefore reduced accordingly. As the reduction of curvature
implies a reduction in the moment contribution by the steel plates in flexure, the
additional moments are resisted by the force couple which is more effective in
providing moment resistance than the flexural resistance of steel section in the
ultimate loading stage. Detailed analysis of the contributions to the total
moment resistance by various components will be presented in Section 4.11.
66
180
160
Moment (kNm)
140
120
100
80
60
RC
40
Steel plate
20
0
0
(a)
10
15
20
25
30
35
40
45
50
Curvature (0.001m-1)
160
140
Moment (kNm)
120
100
80
60
40
RC
20
Steel plate
(b)
10
15
20
25
30
35
40
35
40
Curvature (0.001m-1)
140
120
Moment (kNm)
100
80
60
40
RC
20
Steel plate
0
0
(c)
10
15
20
25
30
Curvature (0.001m-1)
180
160
Moment (kNm)
140
120
100
80
60
RC
40
Steel plate
20
0
(d)
10
20
30
40
50
60
70
Curvature (0.001m-1)
Figure 4.6 Moment-curvature responses of the RC beam and the steel plates
at the constant moment zone a) Specimen SBSP, b) Specimen WBSP, c)
Specimen WBWP, and (d) Specimen SBWP
67
68
69
70
(a)
P
C
A
(b)
P
D
C
A
E1
E2
(c)
P
B
71
When more bolts are added, the point corresponding to this strengthening
scheme (e.g. Specimen WBSP) in Figure 4.8 would be shifted upward; likewise,
when larger sectional area or higher strength of the steel plates is used, the point
(e.g. Specimen SBWP) in Figure 4.8 would be shifted along the horizontal
direction. For practical design, it is recommended that the amount of steel plates
provided should not exceed the value of the balanced failure as shown in
Figure 4.8. This is because in that case no matter how the bolts are arranged, the
strengthened beam will either be controlled by bolts or by concrete and will be
failure in a relatively brittle manner.
72
Bolt strength
Platecontrolled
Region
Specimen SBSP
Specimen
SBWP
Concrete-controlled
Region
Balanced
failure
Specimen WBWP
Specimen WBSP
Bolt-controlled Region
Plate strength
73
74
N.A
1.00
0.38
0.75
1.26
From the
present tests
101.4
161.5
149.2
133.4
144.6
60.1
47.8
32.0
43.2
95.9
193.5
193.5
153.3
153.3
97.6
97.6
57.4
57.4
Mu/Mfsc,fi
0.62
0.49
0.56
0.75
38
51
44
25
Ultimate moment
Mu (kNm)
Additional moment
Mu (kNm)
From fullshearconnection
full-interaction
Ultimate moment
Mu,fsc,fi (kNm)
Additional moment
Mu,fsc,fi (kNm)
analysis
which
are
fundamentally controlled
by the
displacement
compatibility the RC beam, the steel plate and the anchoring bolts. The
longitudinal partial interaction is quantified by the strain factor of the critical
section, which is defined mathematically as
p,h
c ,h
(4.1)
75
where p ,hp and c ,hp are the longitudinal strains of the steel plates and the RC
beam at the centroidal level of the steel plate of the corresponding section
respectively. When the strain factor of the critical section is equal to 1, full
longitudinal partial interaction exists and the longitudinal strain at the plates
centroidal level is the same as the adjacent concrete. In this case, the maximum
plate force is induced in the steel plates. When partial interaction exists, the
strain factor is smaller than 1 and the longitudinal strain is reduced to c ,hp . In
the extreme case, the strain factor is zero and no net axial force is induced in the
steel plates.
Considering the compatibility between the RC beam, the steel plates and
slip at the critical section, as shown in Figure 4.9, the strain in concrete, and
hence the strain factor can be expressed in terms of the strain in the steel plates
and the slip strains at the centroidal level of the steel plates as
c , h = p ,h + slip ,h ,
p
p ,h
p , h + slip , h
p
(4.2)
(4.3)
On the other hand, the curvature factor of the critical section is used to
quantify the transverse interaction between the RC beam and the steel plate and
is defined as
p
,
c
(4.4)
where p and c are the curvatures of the steel plates and the RC beam of the
corresponding section respectively. Full transverse interaction is present when
76
the curvature is equal to 1 and the curvatures of the RC beam and the steel
plates are the same. When partial interaction exists, the curvature of the steel
plates is smaller than that of the RC beam and the curvature factor is between 0
and 1. When the curvature factor is 0, the beam exhibits no transverse
interaction.
hp
Strain factor =
hp
hc
Curvature factor =
Strain profile of RC
hpp
p
c
hcp
Figure 4.9 Definitions of strain factor and curvature factor (at the section
subjected to the critical moment)
(4.5)
c
where top
gauge is the average of the strain gauge readings from the top
77
of the centroidal level of the steel plate, and d c ,top gauge is the depth of the top
reinforcement where the strain gauge is attached.
Using Equations (4.1) and (4.5), the strain factor can be solved. The
variation of the strain factor against curvature of the RC beam in the constant
moment zone is plotted in Figure 4.10. The points corresponding to the yielding
of the bottom reinforcement, as evidenced by an abrupt decrease in the
tangential stiffness of the load-deformation response of the beams, are also
indicated in Figure 4.10.
78
hinge. This results in a rapid increase of concrete strain at the centroidal level of
the steel plate. As the concrete strain is the denominator of the strain factor (see
Equation (4.1)), such increase would reduce the strain factor and hence the
longitudinal partial shear interaction.
A significant drop in strain factor implies that the strain in the steel plate is
much lower than the strain in the adjacent concrete, at the centroidal level of the
steel plate. The strain reduction in the steel plates reduces the chance of plate
yielding, and can greatly reduce the moment capacity of the strengthened beam
as described in Section 4.7. The low chance of plate yielding at the peak
moment implies that assuming the steel plates to be plastic at the ultimate limit
state, as adopted in rigid plastic analysis, is not always true in BSP beams, and
the strength of the BSP beams can be significantly over-estimated if the steel
plate is assumed to be plastic. This suggests that in the ultimate load analysis of
BSP beams the strain reduction in the steel plate should be considered.
79
0.8
Unit SBSP
Unit WBSP
Strain factor
Unit WBWP
Unit SBWP
Yielding of the bottom
reinforcement
0.4
0
0
10
20
30
40
50
60
80
81
The decrease of the curvature factor shows that the difference in curvature
is more significant as a plastic hinge forms. As one of the governing behaviour
affecting the moment resistance of the BSP beam, the effects of plastic hinge on
difference in curvature should be considered in the ultimate limit state analysis
of BSP beams.
Curvature factor
1.5
Unit SBSP
Unit WBSP
0.5
Unit WBWP
Unit SBWP
Yielding of the bottom
reinforcement
0
0
10
20
30
40
50
60
82
Both Weak bolt specimens (Specimens WBSP and WBWP) failed in bolt
fracture. The initial responses of these two specimens are similar to Strong
bolt arranged specimens, as friction existed in all Specimens. After the bolts
were deformed and mobilized, slips gradually developed between the RC beam
and the steel plates. As the beam approached failure, the rate of longitudinal slip
also increased gradually. It should be noted that the rate herein is relative to the
deflection of the mid-span. A higher rate in increase of slip suggests that the
ductility demand for bolts may be greater than the ductility demand for the
beam, and should be remembered when choosing bolts in practical design. To
prevent premature failure of bolts, ductility of bolts should be considered
carefully if a partial-shear-connection design is carried out.
83
200 (a)
200 (b)
1000 (a)
1000 (b)
0
0
10
20
(a)
30
40
50
60
70
50
60
70
200 (a)
200 (b)
1000 (a)
1000 (b)
0
0
(b)
10
20
30
40
200 (a)
200 (b)
1000 (a)
1000 (b)
(c)
10
20
30
40
50
60
70
50
60
70
175 (a)
(d)
175 (b)
1025 (a)
1025 (b)
1
0
0
10
20
30
40
84
(4.6)
where c ,hp and p ,hp are the strains of concrete and plate, respectively, at the
centroidal level of the plate, Ls and x are half of the clear span and the distance
from support, respectively. The strain of the RC beam at the centroidal level of
the plate was estimated by the strain gauges attached on the top and bottom
reinforcement. Assuming that plane sections remain plane after bending, the
longitudinal strains of the RC beam at the centroidal level of plate in the
constant moment zone were estimated by linear interpolation. On the other hand,
the strains of steel plates at the centroidal level of the plate could be read
directly from the strain gauge attached at the corresponding level.
The longitudinal slip responses along the beam span of all strengthened
specimens at different load levels are plotted in Figure 12. When M=0.5Mu, all
specimens were elastic and the longitudinal slip profiles were smooth, as
illustrated in Figure 12 by the dotted line which connected the discrete
85
measuring points. The longitudinal slip was maximal at the supported ends and
decreased gradually until reaching zero at mid-span.
As a plastic hinge gradually formed below the loading point, the beam
response became non-linear. The plastic hinge zones of specimens are
approximated as a segment 200 mm long below the loading point, as shown in
Figure 4.13. The longitudinal slip profile at the instant when M=Mu is illustrated
by the dashed line in Figure 4.13. According to Figure 4.13, there is an abrupt
change in longitudinal slip within the plastic hinge zone, equivalent to 42% to
90% of the total change in longitudinal slip for the whole span. The slip strain
(the first derivative of the longitudinal slip with respect to the coordinate along
the beam) is accumulated in the plastic hinges but not for the remaining parts of
the beam. This means that the strain separation between the RC beam and the
steel plates is greatest in the plastic hinge, and such effects become greater as
the beam gradually approaches its ultimate state. Such an increase in strain
separation in the plastic hinge reduces the tensile strain in the steel plates and
eventually lowers the flexural strength of BSP beams.
86
M = 0.5Mu
Longitudinal slip (mm)
M = 0.75Mu
M = 0.9Mu
M = 0.95Mu
1
Plastic Hinge
M = Mu
0
0
600
1200
1800
(a)
5
M = 0.5Mu
4
Longitudinal slip (mm)
M = 0.75Mu
M = 0.9Mu
3
M = 0.95Mu
Plastic Hinge
M = Mu
0
0
(b)
600
1200
1800
M = 0.5Mu
4
Longitudinal slip (mm)
M = 0.75Mu
M = 0.9Mu
3
M = 0.95Mu
Plastic Hinge
M = Mu
0
0
600
1200
1800
(c)
2
M = 0.5Mu
Longitudinal slip (mm)
M = 0.75Mu
(d)
M = 0.9Mu
M = 0.95Mu
1
Plastic Hinge
M = Mu
0
0
600
1200
1800
Figure 4.13 Longitudinal slip responses along beam spans a) Specimen SBSP,
b) Specimen WBSP, c) Specimen WBWP, and d) Specimen SBWP
87
(4.7)
The internal moment of the RC core and the steel plates were back
calculated from the corresponding strain profiles captured by strain gauges
attached on different levels and estimated by linear interpolation along the depth
of sections. As the strain profiles of the RC beam and the steel plates are
different due to partial interaction, independent sets of strain gauges, as shown
in Figure 3.9 were used. According to Figure 3.9, the strain gauges used to
capture the strain profile of the RC beam were attached on the top and bottom
reinforcements, while those attached directly on top and bottom layers of steel
plates were employed for measuring the strain profile in the steel plate.
88
After solving the independent strain profiles of the RC beam and the steel
plates, the corresponding stress strain relationships were used to estimate the
longitudinal stress in the section considered. The stress-strain relationship
proposed by Attard and Stewart (1998), and modified by Su and Cheng (2008)
was used to model the response of concrete in compression. The concrete cube
strength required in the model was obtained through material test and the results
are summarized in Table 3.3. On the other hand, the tensile stress provided by
concrete was neglected in this study since the magnitude is relatively small and
becomes zero when the section is cracked.
With the longitudinal stress in the beam being solved, the moment in the
RC beam (Mc) can be estimated by considering the RC beams only and taking
moment about the elastic centroidal axis, while the moment in the steel plates
(Mp) can be estimated in a similar way by setting the bending axis at the middepth of the plates. The moment contribution by the axial force couple (Fph) is
solved by integrating the forces in plates and multiplying the force with the
separation of centroidal axis.
89
to represent the full range behaviour of the four specimens in Figure 4.14. In
general, the contributions of the three components were constant throughout the
whole loading history with deviation less than 10% of the applied moment.
90
100%
Concrete flexure
Plate flexure
80%
Axial couple
60%
40%
20%
0%
0
40
80
120
160
200
Moment (kNm)
(a)
100%
80%
60%
Concrete flexure
Plate flexure
40%
Axial couple
20%
0%
(b)
40
80
120
160
200
Moment (kNm)
100%
80%
Concrete flexure
60%
Plate flexure
Axial couple
40%
20%
0%
0
(c)
40
80
120
160
Moment (kNm)
100%
80%
Concrete flexure
60%
Plate flexure
Axial couple
40%
20%
(d)
0%
0
40
80
120
160
Moment (kNm)
91
92
CHAPTER 5
NON-LINEAR MODEL FOR BOLT GROUPS WITH
LINEAR-ELASTIC-PLASTIC BOLTS UNDER
COMBINED IN-PLANE MOMENT AND SHEAR
5.1 OVERVIEW
In BSP beams, the shear flow between the RC beam and the steel plates is
resisted by the connecting bolt groups. These bolt groups, when resist the inplane shear force, will slip in the corresponding directions, leading to
incomplete, or partial, interaction between the RC beam and the steel plates. As
illustrated in the experimental study, the behaviour of BSP beams is greatly
affected by the slip response of the connecting bolt groups. Thus an appropriate
model which is able to simulate the load-slip response of these connecting bolt
groups is necessary for an accurate prediction of response of BSP beams.
Previous research on anchor bolts and bolt groups was focused on their
response under uni-directional shear only (Ueda et al., 1991; Fuchs et al., 1995;
ACI Committee 355, 1995). The response of bolt groups in BSP beams is more
complicated since they are loaded by combined in-plane moment and shear. A
number of research on bolt groups in steelwork subjected to combined in-plane
shear and moment were carried out. These studies aimed at determining the
strength (Harrison, 1980; Crawford and Kulak, 1971; Abolitz, 1966; Surtees et
al., 1981; Nowak and Hartmann, 1993) and stiffness (Lim and Nethercot, 2004)
of connections. For the strength analyses, the deformations of individual bolts
were related to the rigid body movement of the bolt group, as defined by the
centre of rotation. Then, by assuming the load deformation relationship of bolts
as elastic (Harrison, 1980), plastic (Abolitz, 1966), elasto-plastic (Surtees et al.,
93
1981) or non-linear (Crawford and Kulak, 1971; Nowak and Hartmann, 1993),
the coordinates of the centre of rotation were solved by considering equilibrium
in translational and rotational directions. With the centre of rotation solved, the
ultimate strength could be found subsequently by summing up responses of
individual bolts. On the other hand, studies are also carried out to describe the
rotational stiffness of shear-type bolted moment-connections commonly used in
cold form steel portal frames (Lim and Nethercot, 2004). Formulas relating the
geometry and the stiffness of bolt groups were developed and the importance of
considering the semi-rigidity of bolt groups was pointed out. While the strength
and stiffness are the major behavioural characteristics of bolt groups, the full
range load-slip response is also important when full non-linear analysis of BSP
beams is to be carried out. However, no analysis or mathematical model has
been developed.
94
The bolts are perfectly fitted and connected by a plate, which acts as if
rigid.
2)
3)
The bolts are separated widely such that the interference between them can
be neglected.
4)
The first assumption idealizes the bolts as being perfectly fitted in the
connection media. In practice, clearance hole is often used to facilitate the
installation of bolts, making the bolts not perfectly fitted and affecting the load
deformation responses of bolts (Kulak et al., 1987; Owens and Cheal, 1989).
However, the influence is usually small and gradually ceases to exist as more
bolts are getting into close contacts with the connection media through slips. So
it is acceptable to ignore the above effect and assume the bolts as perfectly fitted.
Based on the first assumption, only rigid body motion is admissible for the bolt
group. As a result, the relative positions of all the bolts remain unchanged after
deformation. It implies that the movements of the bolts are rotated about a point
which is known as the instantaneous centre of rotation.
The second assumption limits the usage of the theory developed for bolt
groups comprising ordinary bolts where the pre-compressive force, and hence
the friction between media, is negligible. For high strength friction grip bolt
95
connections, the theory developed herein is not applicable as the applied shear
is primarily resisted by the friction between the connecting media. The slip
would be insignificant due to the high stiffness provided by the frictional
resistance. In this case, the connections can be considered as rigid.
Furthermore, according to the first and the third assumptions, the force
resisted by each bolt can be readily obtained from the deformation of the bolt.
After the polar moment of inertia and the stiffness of the bolt group have been
calculated, the relationship between the rigid body movement of the bolt group
and the external in-plane force can be established.
96
u xi = sl yi s and u yi = sv + xi s
(5.1)
Fy, sv
O
Fx, sl
M, s
Pxi k xi
P = 0
yi
0
k yi
s
k xi yi l
s
k yi xi v
s
(5.2)
97
Considering the global force equilibrium of the applied loads and the bolt
forces, one can have,
Fx = Pxi , Fy = Pyi and M = ( Pyi xi Pxi yi )
i
(5.3)
Substituting Equation (5.2) into Equation (5.3) and simplify, the applied
forces can be related to the rigid body movements as
Fx K x
F = 0
y
M Rx
0
Ky
Ry
Rx sl
Ry sv ,
I zz s
(5.4)
R y = k yi xi .
i
Solving Equation (5.4), the rigid body movements of the bolt group can be
determined as
sl
1
s =
v R2 K R2 K + I K K
x
y
y
x
zz x
y
s
I zz K y Ry2
Rx Ry
Rx K y
Rx Ry
2
x
I zz K x R
Ry K x
Rx K y Fx
Ry K x Fy .
K x K y M
(5.5)
The centre of rotation of the bolt group (as shown in Figure 5.2) can be
determined by the following equations,
xo =
sv
s
and yo = l .
s
s
(5.6)
It can be seen that the centre of rotation depends on the applied loads as
well as the bolt group arrangement. By substituting Equation (5.5) into Equation
(5.2), the forces at the ith bolt can be calculated from the applied forces.
98
( R K
2
x
Pxi
Ry2 K x + I zz K x K y )
=
Pyi
F
k xi ( I zz K y Ry2 yi Rx K y )
k xi ( yi Ry K x Rx Ry )
k xi ( Rx K y yi K x K y ) x
Fy
2
k yi ( Rx Ry + xi Rx K y )
k yi ( I zz K x Rx xi Ry K x ) k yi ( Ry K x + xi K x K y )
(5.7)
-yo
sv
-sl O
-xo
Centre of rotation (xo, yo)
99
Assuming that all the incremental applied loads (Fx, Fy, M) are equal,
the total external load applied to the bolt group after the j-1th loading step (Fx, Fy,
M)j-1 is simply equal to the sum of all the incremental loads from the 1st to the
j-1th loading steps, or (j-1)(Fx, Fy, M). Further assuming that some of the
bolts have been yielded and by considering the global force equilibrium of the
bolt group, the incremental load shared by both the yielded and un-yielded bolt
at the jth loading step is,
j
Fx Fx
Fx
+ Fy
.
Fy = Fy
M M
M
elastic
inelastic
(5.8)
As mentioned in the classic bolt group theory, the force vector of the
elastic bolts can be found according to Equation (5.4),
j
Kx
Fx
= 0
Fy
M elastic Rx
j
Rx sl
Ry sv .
I zz s
0
Ky
Ry
(5.9)
P = Pyfi
j
xi
P = Pyfi
j
yi
j 2
o
j
sgn(s ) ,
(5.10)
j 2
o
j
sgn(s ) ,
(5.11)
( yij yoj )
(x
j 2
o
) +(y
( xoj xij )
(x
j 2
o
) +(y
(5.12)
100
where Pyfi is the yielded force of the ith bolt. Hence the change in bolt forces of
the yielded bolt from the j-1th to the jth load step (as illustrated in Figure 5.3)
and the sum of the bolt forces contributing to balance the external load (Fx,
Fy, M) are,
j
Pxij = Pxij Pxij 1 , Fxinelastic
= Pxij ,
(5.13)
i
j
Pyij = Pyij Pyij 1 , Fyinelastic
= Pxij ,
(5.14)
i
j
P ij = P ij P ij 1 , M inelastic
= P ij ,
(5.15)
For the un-yielded bolts, making use of Equations (5.6) and (5.9),
j
Kx
Fx
= 0
Fy
M
Rx
elastic
0
Ky
Ry
Rx yoj
Ry xoj sj .
I zz 1
(5.16)
Equations (5.10) to (5.16) show that the bolt forces of both yielded and unyielded bolts are functions of the instantaneous centre of rotation ( xoj , yoj ) and
the incremental rotation sj of the bolt group. Hence, after the instantaneous
centre of rotation and the incremental rotation at each loading step have been
determined, the non-linear response of the bolt group could be solved readily.
101
102
Step 3: Compare the calculated displacement with the yield displacement of the
bolts to determine if bolts are yielded. In case none of the bolts are
yielded, the bolt group displacements ( slj , svj , sj ) in the jth loading
step is calculated. Then the instantaneous centre of rotation ( xoj , yoj ) can
be solved, the incremental displacements ( u xij , u yij ), and the
incremental bolt forces ( Pxij , Pyij ) can be evaluated accordingly. It can
then go back to Step 2. However, if any of the bolts are yielded during
the jth loading step, the process has to go to Step 4.
Step 4: Iteration has to start to determine the instantaneous centre of rotation and
the incremental rotation so that force equilibrium can be satisfied in this
loading step. A trial instantaneous centre of rotation, denoted as
( xotj , yotj ), at the jth loading step is determined based on the previous
loading or iteration step. The subscript t stands for trial or temporary
value of which force equilibrium has not been satisfied completely.
Step 5: With the trial centre of rotation, together with the sign of incremental
rotation sgn( sjt ) (based on the previous loading or iteration step), the
force vector of any yielded bolt can be found by Equations (5.10) to
(5.12). Then Minelastic can be computed by adding up all the moment
contributed from individual yielded bolts. The incremental rotation sjt
can be solved by considering moment equilibrium according to
Equations (5.8) and (5.16). The translational movement ( sltj , svtj ) of
the bolt group can then be determined by Equation (5.6). Hence the
incremental displacements ( uitj , vitj ) of each bolt can be calculated by
Equation (5.1).
Step 6: The trial values of Pxitj , Pyitj , Fxtj and Fytj , where equilibrium has
not been fully satisfied, are calculated.
103
Step 7: Fxtj and Fytj is checked against the value Fx and Fy . If the error is
less than the preset tolerance, the trial instantaneous centre of rotation
can represent the true value. The iteration can be stopped and the process
can go to Step 8. Otherwise, a new trial instantaneous centre of rotation
is determined by the steepest descent method (Hageman and Young,
1981), which would be described in details in Section 5.3.2. Then the
iteration goes back to Step 4.
Step 8: The
incremental
displacements
are
added
to
the
cumulative
104
Start
Define
1. Load slip relation
2. Applied loading
3. Bolt group geometry
no
Check any bolt
yielded
yes
Assume a centre of rotation
Calculate
Rigid movement
of bolt group
x, y and
by elastic theory
Calculate Minelastic
Calculate xt, yt and t
Calculate Fxt, Fyt for the
assumed centre of rotation
Error within
tolerance
no
yes
Update ui, vi, Pxi, Pyi
yes
no
yes
Stop
105
xo ( k +1) xok
f (xok , yok ) / x
= k
.
y
f ( xok , yok ) / y
o ( k +1) yok
(5.17)
In the above equation, xo(k+1) and yo(k+1) refers to the instantaneous centre of
rotation to be determined in the kth iteration step. Define the error term of bolt
group forces in x- and y- directions as,
h( x, y ) = Fxt ( x, y ) Fx and g ( x, y ) = Fyt ( x, y ) Fy .
(5.18)
(5.19)
The above function combines the errors in bolt group forces in x- and ydirections. By minimizing f(x, y), both functions h(x, y) and g(x, y) converge to
zero. The direction f ( xok , yok ) is the one where function f has the greatest
decrement and is found by substituting the points at a fixed distance which is
preset by the program around the trial centre of rotation ( xok , yok ) . The value k
106
is defined as the distance between the trial point ( xok , yok ) and the point (x, y)
on the direction f ( xok , yok ) with a minimum value of f. It is determined by
performing a line search along the direction f ( xok , yok ) . Mathematically, it
can be represented by a simple formula,
f ( x , y )
f ( xok , yok )
, yok s y ok ok .
x
y
(5.20)
The iterative procedure is completed when the error terms in bolt group forces
in x- and y- directions are within the preset tolerance . That is,
h( xok , yok ) , g ( xok , yok ) .
(5.21)
Once the iterative procedure is completed, the solution point is returned to the
main program where the subsequent rigid body movements and bolt slips are
computed as previously described.
f(x,y) = C1
C1 > C2 > Cn 0
f(x,y) = C2
(xo1,yo1)
(xo3,yo3)
f(x,y) = Cn
- f(xo1,yo1)
(xo2,yo2)
Line Search on - f(xo1,yo1)
Solution centre of rotation
107
of
the
material
was
108
adjusted
so
that
the
300
200
100
0
0
Slip (mm)
(a)
100
100
B1
90
B2
y B3
90
B4
B5
B6
(b)
M=131 kNm
Fy=350 kNm
Fx=380 kNm
(c)
109
uni-directional response of the ring would match with the load-slip behaviour of
the bolt. The finite element mesh of the numerical example is shown in Figure
5.8. The bolts are connected by steel plate and behave as a group. The external
in-plane loads are applied onto the steel plate and distributed to the bolts. The
deformation of the bolt group is taken to be the displacement of the point on the
steel plate where the centroid of the bolt group is located. The load-slip
relationship in x-, y- and rotational directions obtained from the analysis are
plotted in Figure 5.9.
110
Bilinear material
simulating load-slip
behaviour of bolt
Steel plate
111
400
Force X (kN)
300
200
ATENA
100
BOGA
0
0
a)
0.4
0.8
1.2
Displacement X (mm)
400
Force Y (kN)
300
200
ATENA
100
BOGA
0
0
b)
0.4
0.8
1.2
Displacement Y (mm)
150
Moment (kNm)
100
50
ATENA
BOGA
c)
0.01
0.02
0.03
Rotation (rad)
Figure 5.9 Comparison of finite element analysis with the proposed analysis
in a) Force-displacement in x-direction, b) Force-displacement in y-direction
and c) Moment-rotation.
112
113
A computer program BOGA has been implemented to evaluate the nonlinear behaviours of bolt groups. By comparing the results of a numerical
example simulated by BOGA and the non-linear finite element analysis, it is
concluded that BOGA is reliable in modelling the response of bolt groups
consisting of linear-elastic-plastic bolts.
114
CHAPTER 6
NON-LINEAR MODEL FOR BOLT GROUPS WITH
NON-LINEAR KINEMATIC-HARDENED BOLTS
UNDER COMBINED IN-PLANE MOMENT AND
SHEAR
6.1 OVERVIEW
115
6.2.1 Assumptions
The following assumptions are made within the framework of the present
formulation.
1)
2)
3)
The bolts are separated widely such that the interference between them can
be neglected.
4)
5)
116
The first three assumptions are the same as those adopted for the elasticplastic model in Chapter 5. The meanings and implications of these assumptions
on the physical behaviour of the bolt groups can be found in Section 5.2. Other
than the first three assumptions, two more assumptions are adopted in this
model. The fourth statement assumes a tri-linear and kinematically-hardened
load-slip for bolts as illustrated in Figure 6.1. A non-linear load-slip relationship
with gradual drop in bolt stiffness is common in bolts (Crawford and Kulak,
1971; Yam and Chapman, 1968), and the kinematic hardening behaviour is
widely accepted by engineers to describe the non-linear cyclic behaviour of
metals. As bolts are, in general, made of metals, the assumption is likely to be
valid in the present study.
The fifth assumption is different from the normal assumption that bolt
failure is governed by the slip capacity. A limiting strength is chosen as the
failure criterion in the present study, because the bolts actually slip in curved
paths when subjected to eccentric shear, and the deformation capacity of bolts
within a bolt group will be reduced (Crawford and Kulak, 1971). Hence, it is
difficult to define a specific slip capacity for bolts, and the use of strength as the
failure criterion appears to be more appropriate.
P
Pu
kp
Py
kp
Py
ke
su
117
When initial loading is applied to the bolt, the force point P =(Px, Py)T lies
within the yield surface. In this case, the load-slip response is elastic and can be
expressed as
P = k e u ,
(6.1)
118
where ke is the elastic stiffness of the bolt, and P =(Px, Py)T and u =(ux,
uy)T are the incremental load and deformation, respectively, of the bolt.
As further load is applied, the force point gradually shifts away from
origin and will touch the yield surface, and the bolt yields. In general, the
stiffness of the bolt in the directions normal and tangential to the yield surface at
the force point can be written as
Pn k p
P = 0
t
0 un
,
ke ut
(6.2)
where the subscripts n and t denote the normal and tangential directions,
respectively, and kp is the corresponding plastic stiffness in the idealized loadslip curve, depending on the current yielding stage. The post-yield behaviour of
bolts can be divided into two stages as described as follows.
First stage yielding refers to the state when the force point of a bolt
(denoted by P) is on the yield surface (active surface), as illustrated in Figure
6.2b. The corresponding plastic stiffness equals kp, as shown in Figure 6.1. The
unitary normal vector on the yield surface at point P is denoted by n , where
n =
P
,
P
(6.3)
119
m = PP,
(6.4)
,
= = m
(6.5)
=
m
m
.
m
(6.6)
As the final force P+P lies on the new yield surface, we have
Py = P + P .
(6.7)
From Eqautions (6.4) to (6.7), the updated location of the centre of the yield
surface is readily solved with the given P, P and .
P = P.
(6.8)
n =
P
P
=
,
P
P
(6.9)
120
(6.10)
When the final force P+P is on the new bounding surface, we have
Py = P + P .
(6.11)
From Eqautions (6.9) to (6.11), the updated location of the centre of the
bounding surface is readily solved. The yield surface translates accordingly
so that the two surfaces remain touching at the force point, P+P. Using
Eqaution (6.9), we have
P + P
P + P
=
.
P + P
P + P
(6.12)
Substituting Eqautions (6.7) and (6.11) into (6.12) and simplifying, we have
= + ( Py Py ) ( P + P ) .
(6.13)
From Eqaution (6.13), the new location of the yield surface can be solved once
the new location of the bounding surface and the force increment are known.
121
P
,
,
Yield surface
Bounding surface
a) Elastic stage
n
P
m
P
Bounding surface
122
sv
s
sl
u =
sv .
0
1
x
i
y i
s
(6.14)
The load-slip response of the ith bolt can be represented by Eqaution (6.2) as,
Pn k p
P = 0
t i
0 u n
.
k e i u t i
(6.15)
123
s
0 c s 1 0 yi l
s .
ke i s c i 0 1 xi v
s
(6.16)
Here, c and s corresponds to cos and sin respectively, with the value as
indicated in Figure 6.4 and equal to arg( n ).
Py
u
P
un
ut
Pn
Pt
Yield Surface
Px
=
Py i cskr
cskr
ke s 2 k r
sl
(ke c 2 kr ) yi (cskr ) xi
sv .
2
(ke s kr ) xi + (cskr ) yi i
s
(6.17)
Considering the global force equilibrium of the applied loads and the bolt forces,
we have
Fx = Pxi , Fy = Pyi and M = (Pyi xi Pxi y i ) .
i
124
(6.18)
K cc
sl
s = K
cs
v
s Rccy Rcsx
K cs
K ss
Rssx + Rcsy
K cc = ke ci2 kri
where
Rccy Rcsx
Rssx + Rcsy
I ccy + I ssx + 2 I cs
Fx
F ,
y
M
K ss = ke si2 kri
(6.19)
K cs = ci si kri
i
2
i ri
Rcsy = ci si yi kri ,
Rcsx = ci si xi kri
and
I cs = ci si xi yi kri .
i
S = ( sl , sv , s )n .
(6.20)
n =1
125
Step 1: The parameters for the analysis are defined. These include:
1) The properties of the bolts (e.g. the stiffnesses of the bolts and the
two yield points of the bolts);
2) The geometries of the bolt group (i.e. the number and coordinates of
bolts with respect to an origin defined by the user);
3) Details of the simulation (e.g. the total applied load and the number
of load steps).
Step 2: To begin with, an incremental load is applied to the bolt group.
Step 3: The yield conditions of the bolts are checked and the corresponding
value kpi is set in the stiffness matrix in Eqaution (6.2). Using Eqaution
(6.19) and the value ij, the global stiffness matrix is assembled.
Step 4: The rigid body movement (sl , sv , s ) j of the jth loading step is
solved using Eqaution (6.19). Then the incremental deformations and
forces on individual bolts are calculated by Eqautions (6.14) and (6.16).
Step 5: The yield conditions of the bolts are checked by considering the updated
bolt force vector. For the ith yielded bolts at the 1st stage yielding, the
ij is computed using
unitary translational direction of the yield surface m
ij , the updated locations of the
Eqaution (6.4). Using the computed m
centre of yield surface ij are solved using Eqautions (6.5) and (6.7).
For the ith yielded bolts at the 2nd stage yielding and using the value n ij
computed at the previous load step, the centres of the yield and bounding
surfaces are solved using Eqautions (6.10), (6.11) and (6.13).
Step 6: The centres of the yield surface and bounding surface i ( j +1) and i ( j +1)
are set as ij and ij , respectively. The normal to the yield surface at the
updated force point of the bolts n i ( j +1) is solved using Eqaution (6.3) and
126
the values of i(j+1) are also computed. These values are used to assemble
the stiffness matrices in the j+1th loading step.
Step 7: The cumulative bolt forces ( Px ,Py )i and deformations (u x ,u y )i are
computed and updated. The cumulative deformations are then used to
update the instantaneous locations of the bolts.
Step 8: The bolt forces are checked against the ultimate resistance of the bolts. If
any of the bolt forces exceed the limiting resistance, the program will be
terminated. Otherwise, the program proceeds to the j+1th loading step
(Step 2) and further loads are applied to the bolt group.
Start
Ultimate load
reached
No
Yes
Stop
127
128
Table 6.1 Specimen detail in the experimental study by Crawford and Kulak
Specimen
Number
Bolt group
arrangement
(row column)
B1
B2
B3
B4
B5
B6
B7
B8
51
51
51
61
61
42
42
52
Row Separation
Column
Eccentricity
(mm)
Separation (mm)
(mm)
63.5
76.2
76.2
76.2
76.2
76.2
76.2
63.5
63.5
63.5
63.5
203.2
254.0
304.8
330.2
381.0
304.8
381.0
381.0
P (kN)
ke
kp
kp
Py
Py
Pu
526.7
66.3
5.7
133.8
260.0
297.0
129
increases; a clear yield point and bounding point are difficult to define on the
load-slip curve. Hence, a notional yield point and bounding point that could fit
the actual load-slip curve were used for computational convenience. A
proportional load defined by the corresponding eccentricities was applied to the
bolt groups until failure, and the geometries of bolt groups and the mechanical
properties of bolts are summarized in Tables 6.1 and 6.2 respectively. The loadcarrying capacity, deformability, and complete load-slip relationship of bolts are
compared with the experimental results in the following section.
400
Idealized
Measured
Load (kN)
300
200
100
0
0
4
6
Deformation (mm)
10
130
Numerical
Exp a
Exp b
Numerical
B1
504.0
530.9
1408.5
2304.9
937.9
B2
515.2
508.5
1791.5
1743.3
1076.2
B3
425.6
430.1
1476.5
1476.5
921.5
B4
562.2
575.7
1923.5
2169.8
1477.2
B5
495.0
501.8
3594.2
3594.2
1263.8
B6
591.4
566.7
1294.5
1503.1
1112.1
B7
474.9
497.3
1168.6
1284.5
898.1
B8
595.8
633.9
2248.0
2236.8
1238.6
When the external loading applied on the bolt group increases and
approaches to its ultimate state, the numerical results agree closely with the
experimental results, as illustrated in Figures 6.8a to 6.8c and listed in Table 6.3.
It may be due to the fact that at high shear load level, relatively larger shear
deformations occur in bolts, releasing the axial elongation, and hence the bolt
preload (Foreman and Rumpf, 1961; Bendigo et al., 1963; Fisher et al., 1963).
131
So the friction between the faying surfaces gradually decreases and has no
significant effect on the shear strength of bolts.
600
500
Load (kN)
400
300
B1(a)
200
B1(b)
BOGAN
100
0
0
a)
3
4
Rotation (0.01 rad)
700
600
Load (kN)
500
400
300
B4(a)
200
B4(b)
BOGAN
100
b)
2
3
Rotation (0.01 rad)
500
400
Load (kN)
300
200
B7(a)
B7(b)
BOGAN
100
c)
0
0
132
600
500
Load (kN)
400
300
increment=4.448kN
200
increment=8.896kN
increment=17.792kN
100
increment=35.584kN
0
0
133
In this chapter, a new solution algorithm to calculate the non-linear loadslip response of bolt groups subjected to combined in-plane moment and shear
were presented. The bolts are assumed to be kinematically hardened when
yielded. By postulating the post-yield behaviour of bolts using Mrozs
translation rules for the yield surface, the stiffness of bolts in the x and y
directions were derived. Further applying classical theory of bolt groups, the
complete non-linear response of bolt groups, and the behaviour of individual
bolts were evaluated accordingly.
134
CHAPTER 7
NUMERICAL ANALYSIS OF BOLTED SIDEPLATED BEAMS WITH PARTIAL INTERACTION
7.1 OVERVIEW
Partial interaction in composite beam structures is a complicated global
compatibility problem. Newmark et al. (1951) showed that the equation
governing the compatibility of the steel, concrete and the connecting media
along the length of beam is a 2nd order differential equation with longitudinal
slip as the dependent variable and is a function of parameters including the
mechanical properties and arrangement of bolts, the geometric properties of RC
beam and steel plates and mechanical properties of material used. Alternations
in any part of the beam affect the response of the whole beam.
135
136
7.2 FORMULATION
7.2.1 General
Consider a BSP beam as shown in Figure 7.1(a) with a steel plate anchored
by n+1 bolt groups. The stiffness centroids of the bolt groups are located at the
geometric centroids of the steel plates. The corresponding idealized structural
model of the strengthened beam is shown in Figure 7.1(b). Nodes, defined at the
centroids of the bolt groups, are numbered from left to right. The whole beam is
discretized into n pairs of RC and steel plate non-linear macro-finite-elements.
The length of the ith element is equal to the separation between the ith and the
i+1th bolt groups. The formulation of the stiffness matrices of the elements is
provided in the next section. Since the centroidal axis of the RC is located
above the centroid of the steel plates, rigid arms are added at the ends of the RC
elements so that the concrete elements can interact with the steel plate elements
through the dimensionless bolt groups. The element models for RC and steel
plates are shown in Figure 7.1(c). The nodes on the left and right sides of the
RC and steel plate elements are named Node a and Node b, respectively. This
means that the ith bolt group is connected to Node b of the i-1th RC and steel
plate elements together with Node a of the ith RC and steel plate elements. To
account for the effects of partial interactions in the analysis, we adopted nonlinear bolt groups with finite stiffness as developed in Chapter 6, the details of
which will be described in Section 7.2.3.
137
Reinforced concrete
st
nd
ith
i+1th
n+1th
Steel plate
(a)
The ith RC element
1st
2nd
ith
i+1th
Flexural elements
n+1th
Rigid arms
(b)
L
a
e
a
(c)
Rigid Arms
b
(7.1)
v( x) = 0, 0, (1 3 2 + 2 3 ), (3 2 2 3 ), L( 2 2 + 3 ), L( 2 + 3 ) ,
(7.2)
138
x
b
ua
va
ub ua
vb
(a)
va
ub
vb
(b)
Figure 7.2 Sign conventions for a) Flexural elements, and b) RC beam elements
The strain at any point within the member can be divided into an axial
component and a bending component:
( x, y ) = axial ( x, y ) + bending ( x, y ) .
(7.3)
Assuming the axial strain is evenly distributed along the element and across the
section, we have
axial ( x, y ) =
ub u a 1 1
T
= , [ua , ub ] .
L
L L
(7.4)
Applying the hypothesis that plane sections remain plane after bending, the
bending strain bending ( x, y ) can be expressed in terms of the curvature (x) of
the section:
bending ( x, y ) = ( x) y .
(7.5)
y (1 2 ), 2 y (1 + 2 ), y (2 3 ), y (1 3 ) .
2
L
L
L
L
bending ( x, y ) =
(7.6)
139
6
2
2
1 1 6
B = , , 2 y (1 + 2 ), 2 y (1 2 ), y (2 3 ), y (1 3 ) .
L
L
L
L L L
(7.7)
(7.8)
6
2
2
1 1
6
where B a = , , Bb = 2 (1 + 2 ), 2 (1 2 ), (2 3 ), (1 3 ) .
L
L
L
L L
L
L B T B E dAdx
a a A T
= L 0
B T B yET dAdx
0 b a A
T
0 Ba Bb A yET dAdx
L
T
2
0 Bb Bb A y ET dAdx
L
(7.9)
where V, A and ET are the volume, sectional area and the equivalent Youngs
modulus of the element, respectively.
that the non-linear material variation of the Youngs modulus can be easily
incorporated into the element stiffness matrix. For steel plate elements, the
values of EkAk, EkSk and EkIk at =k may be approximated by the following
summation series,
140
np
Ek Ak = E p j Ap j ,
(7.10)
j =1
np
Ek S k = E p j Ap j y p j ,
(7.11)
j =1
np
Ek I k = E p j Ap j y p j 2 ,
(7.12)
j =1
where Epj, Apj, and ypj are, respectively, the Youngs modulus, the sectional
areas and the depth to the centroid of the steel plates evaluated at the jth layer,
and np is the total layers of the layered model (Figure 7.3a). Meanwhile, the
integration along the length of the element is solved numerically using threepoint Gaussian integration. The integration points k are at 1 = 0.1127, 2 = 0.5
and 3 = 0.8873, and the corresponding weighting factors wk are w1 = 5/9, w2 =
8/9 and w3 = 5/9. With the use of numerical integration, the stiffness matrix of
steel plate Ki,p can be obtained using Equation (7.9), and the explicit form of the
matrix is shown in Appendix 7.1.
nc
hc/2
ns
jth
ysj
ycj
hp/2
hp/2
(a)
np
jth
ypj
hc/2
2nd
1st
2nd
1st
1st
(b)
Figure 7.3 Layered models for a) Steel plate sections and b) RC beam sections
141
According to Figure 7.3b, the values of EkAk, EkSk and EkIk of the RC beam
elements at =k may be expressed as
nc
ns
j =1
j =1
Ek Ak = Ec j Ac j + Es j As j ,
(7.13)
ns
nc
Ek S k = Ec j Ac j yc j + Es j As j ys j ,
j =1
j =1
nc
ns
j =1
j =1
Ek I k = Ec j Ac j yc j 2 + Es j As j ys j 2 ,
(7.14)
(7.15)
where Ecj, Acj, and ycj are, respectively, the Youngs modulus, the sectional
areas and the depth to the centroid of the concrete; Esj, Asj, and ysj are,
respectively, the Youngs modulus and the sectional areas of the steel bars, and
the depth from the steel bars to the centroid of the concrete; and nc and ns are
the total layers of the layered models for concrete and steel bars (Figure 7.3b).
For the RC elements, further transformation is required since two vertical rigid
arms of length e (equal to the separation of the centroidal axes of the concrete
and steel plates) are connected to the ends of the flexural element (Figure 7.2b).
The transformation matrix for the flexural element connected with two vertical
rigid arms of length e is given by
1
0
0
T=
0
0
0
1
0
0
0
0
e
0
1
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
.
e
0
(7.16)
142
Yield
surfaces
(x, y)
(x, y)
Px
143
0 sn
,
ke st
(7.17)
where the subscripts n and t denote the normal and tangential directions of the
yield surface at the loading point, respectively.
Q Rccy Rcsx
K cs
K ss
Rssx + Rcsy
K cc = k e ci2 k
where
Rccy Rcsx sl
Rssx + Rcsy sv ,
I ccy + I ssx + 2 I cs s
K ss = k e si2 k
(7.18)
K cs = ci s i k
Rssx = (k e s i2 k ) xi
Rccy = (k e ci2 k ) yi
Rcsx = ci si xi k
Rcsy = ci s i y i k
i
I ccy = (k e ci2 k ) y i2
Under the current formulation, and refer to Figure 1b, the deformation of
the ith bolt group (sl , sv , s ) is equal to the nodal displacement difference
between steel plates and RC beams at the nodes which the bolt group connects,
which can be expressed as,
p
sl ui ui
s = v v .
v i i
s i i i
(7.19)
Substituting equation (7.19) into (7.18), the equilibrium of the bolt group can be
represented by equation (7.20) as
144
K cc
Qx
Q =
y K cs
Q Rccy Rcsx
K cs
K ss
Rssx + Rcsy
p
c
Rccy Rcsx ui ui
Rssx + Rcsy vi vi
I ccy + I ssx + 2 I cs i i
(7.20)
Equation (7.20) relates the in-plane internal forces within the bolt group with
the nodal displacement of the structural model used for the current analysis and
can be used to assemble the structural stiffness matrix of the system, which will
be discussed in the next section.
Fxb Fxa
Qx
F + F + Q = 0 ,
yb ya
y
M b i M a i +1 Q i +1
p
(7.21)
Fxb Fxa
Qx
F + F Q = 0 .
yb ya
y
M b i M a i +1 Q i +1
(7.22)
ui p ui c
Qx
Q = [ K ] v v .
i b i
y
i
Q i +1
i i
(7.23)
By defining
[Q]i = [Qx , Qy , Q ]iT ,
145
(7.24)
and assembling the stiffness matrix equations of all elements in the beam, the
equilibrium equation can be represented by
[K ]c [d ]c = [Q] + [ F ]c ,
(7.25)
[K ] p [d ] p = [Q] + [ F ] p ,
(7.26)
[Q] = [K ]b [d ] p [d ]c .
(7.27)
Eliminating the bolt force vector [Q] in Equations (7.25) to (7.27), the structural
stiffness matrix can be obtained as follows
Kc + Kb
K
b
K b [ d ]c Fc
,
=
K p + K b [ d ] Fp
p
(7.28)
146
ith element
Qx
-Qx
Steel plate
-Q
ith element
i-1th element
-Qy
Figure 7.5 Force interaction between the bolt group, the steel plate elements
and the RC beam elements at the ith node
2
= f c 2 for < 0 ,
0 0
(7.29)
= f c for 0 ,
(7.30)
147
fc
= Es
(7.31)
= Es ys for ys ,
(7.32)
= Es ys for ys ,
(7.33)
where ys is the yield stress of steel reinforcement. For steel plate, they are
= E p
(7.34)
= E p yp for yp ,
(7.35)
= E p yp for yp ,
(7.36)
148
fs
fp
yp
Es
Ep
yp
ys
fp
P
Pu
Py
kp
kp
Py
ke
smax
149
150
Start
Read and transform input data
Apply incremental load at jth
step
All layers
n=n+1
no
yes
All elements
no
yes
Calculate the internal force
and residual force
no
Convergence
yes
Update stiffness matrices of
bolt groups with subroutine
BOGAN
no
Elements or
bolt groups reach the
corresponding
capacity
yes
Stop
151
The subroutine in SiBAN for modeling the partial interaction due to the
use of bolt groups to transfer shear forces between the RC beam and the steel
plates was based on and modified from the computer program BOGAN
developed in Chapter 6. In each load step, the solution displacement was input
into the subroutine and the yield conditions and yield surfaces of bolts are
updated. Using on the updated properties of bolts, the element stiffness matrices
for the bolt groups are updated and used in the next load step to formulate the
structural stiffness matrices.
The error term of iteration is defined as the ratio between root-meansquare (RMS) incremental displacements in the corresponding iteration and the
RMS value of the accumulated displacement increment of all iterations in each
load step. The convergence criteria are satisfied when the error term is less than
1%. In general, the convergence criteria for the jth iteration of a loading step can
be expressed as
DOF
( d )
i, j
i =1
di ,k
i =1 k =1
DOF
0.01 ,
(7.37)
where d i ,k is the incremental displacement of the ith DOF in the kth iteration.
152
Reinforcement
steel plate
fcu (MPa)
Es (GPa)
fys (MPa)
Ep (GPa)
fyp (MPa)
35
187
537
208
337
Bolt force
P2
P1
Slip
Slip (mm)
P1
0.625
50
P2
4.0
76
153
The structural models for all specimens are shown in Figure 7.11. Entire
specimens were divided into five to nine macro-finite-elements depending on
the number of bolt groups used in the specimens. The centroidal axes for the RC
in all simulations were set at the mid-depth, i.e., the separation between the
centroidal axes of the RC beam and the steel plates is 75mm.
Unit SBSP
Unit WBSP
Unit WBWP
Unit SBWP
Figure 7.11 Finite element models used in the computer program SiBAN
154
(7.38)
Figure 7.12 shows that the numerical results are consistent with the
experimental results up to the peak loads. The initial stiffness of the
experimental results is slightly higher than the numerical results. This is because
the stiffness contributions from concrete in tension were ignored in the
numerical simulation. The experimental and numerical results agree closely
with each other as further loads were applied. The turning point in the ascending
branch of the curves corresponds to the stage where the tension reinforcement
began to yield, which was successfully predicted by the numerical simulations.
155
Table 7.2 Comparison of peak loads obtained from the present numerical
analysis and previous experiments
Specimen
% error
Experiment
SiBAN
SBSP
162
160
-1.2
WBSP
149
146
-2.0
WBWP
133
132
-0.8
SBWP
145
134
-7.6
156
Moment (kNm)
180
120
60
Experimental
SiBAN
0
(a)
10
20
30
40
50
Moment (kNm)
180
120
60
Experimental
SiBAN
(b)
0
0
10
20
30
40
50
Moment (kNm)
150
100
50
Experimental
SiBAN
(c)
0
0
10
20
30
40
50
60
Moment (kNm)
180
120
60
Experimental
SiBAN
(d)
0
0
10
20
30
40
50
157
(a)
SiBAN 1000
0
0
10
20
30
40
6
Experiment 200 (a)
Longitudinal slip (mm)
(b)
0
0
10
20
30
40
50
(c)
10
20
30
40
50
60
(d)
SiBAN 1025
0
0
10
20
30
40
50
158
The beam considered here has the same RC section as that of the specimen
adopted in the previous experimental study. The strengthened plate arrangement
is the same as Specimen B3, while uniform bolt arrangements are adopted
across the entire span of the beam (Figure 7.14). The beam was loaded with an
incremental mid-point load. As there is no node at the mid-span, the mid-point
load was transformed into the equivalent load vectors and applied at the
corresponding nodes. Simulations were carried out and the bolt slips along the
beam span at 50% and 100% of the peak load were extracted to illustrate the
non-linear slip responses of partially interacted BSP beams in the elastic and
inelastic stages. The corresponding variations are presented in Figures 7.15 and
7.16.
159
225
2T10
75
350
250
3600
3T16
Bolts
Steel plates
As the beam approached the inelastic stage (Figure 7.16a), the slip strain
(i.e., the 1st derivative of slip about x) increased faster at the mid-span and
corresponded to a change in the slip profile. This change was mainly due to the
gradual formation of a plastic hinge at the mid-span. When the plastic hinge
gradually formed, the flexural stiffness of the section dropped, causing the
strain of the concrete at the centroidal axis of plate ( c ,hp ) to increase more
quickly. As the slip strain is equal to p ,hp c ,hp , the slip strain would also
160
increase faster accordingly. Such a higher rate of increase in slip strain at the
mid-span (or at the location of critical sections) reduces the partial interaction
between the RC and steel plate, which further reduces the ultimate strength of
the member. Hence, the non-linear effect due to the occurrence of the plastic
hinge should be considered in the analysis of bolted sided-plated RC beams.
The transverse slip profiles along the span in both the elastic and inelastic
stages (Figures 7.15b and 7.16b, respectively) were quite similar. Being
symmetric about the mid-span, the transverse slip is positive at regions near the
mid-span and negative at regions near the two supports. In other words, the steel
plate moves upward relative to the RC at the mid-span and downward near the
supports. The total bolt slips in the positive zone and negative zones are equal in
the elastic stage so that the vertical force equilibrium of the steel plates is
maintained.
161
plates has not been investigated. The rotational slip is the relative rotation
between the cross sections of the RC and the steel plates, and will occur
simultaneously with the presence transverse partial interaction.
When the transverse partial interaction is present, due to the vertical force
equilibrium, the transverse slip varies from negative at supports to positive
values at the mid-span, as described in the previous section. This implies that
the first derivative of the transverse slip with respect to x is non-zero. This nonzero derivative is equal to the rotational slip in mathematical terms, since the
rotation of a member is equal to the first derivative of the vertical displacement.
The rotational slip affects the slip responses of bolts when more than one
row of bolts is installed, or the bolts are not installed along the centroidal axis of
the plate. Excess rotation between the two media will result in an uneven
distribution of bolt forces on the same transverse plane.
Figures 7.15c and 7.16c show that the inelastic rotational slips near the
mid-span increase to ten times the elastic value at the ultimate stage, but only
about three times for the other parts of the beam. The results indicate that the
rotational slip concentrates at the regions close to the plastic hinges in the
162
0.5
0.25
0
0
0.2
0.4
0.6
0.8
1x /L
0.2
0.4
0.6
0.8
1
x /L
0.2
0.4
0.6
0.8
-0.25
-0.5
(a)
0.06
0.03
-0.03
-0.06
(b)
12
0
0
x /L
1
-6
-12
(c)
Figure 7.15 Slip responses in the elastic stage (i.e., at 50% of the peak
load) a) Longitudinal slip profile, b) Transverse slip profile, and c)
Rotational slip profile
163
0
0
0.2
0.4
0.6
0.8
1x /L
-1
(a)
-2
0.2
0.1
0
0
0.2
0.4
0.6
0.8
x /L
1
0.2
0.4
0.6
0.8
x /L
1
-0.1
(b)
-0.2
120
(c)
60
-60
-120
Figure 7.16 Slip responses in the inelastic stage (i.e., at the peak load) a)
Longitudinal slip profile, b) Transverse slip profile, and c) Rotational slip
profile
164
165
166
CHAPTER 8
TWO-ALPHA APPROACH FOR THE ANALYSIS OF
BOLTED SIDED-PLATED REINFORCED
CONCRETE BEAMS
8.1 OVERVIEW
From the results obtained in previous chapters, it can be concluded that the
flexural strength of BSP beams is greatly affected by partial interaction between
different components of the beam. Due to partial interaction, the additional
strength provided by the bolt-plate system can be more than 30% lower than the
theoretical values when full interaction is assumed even if sufficient strength is
provided by the shear connectors. In order to prevent over-estimation of flexural
strength of BSP beams, analysis able to consider the effects of partial
interaction is necessary.
167
168
of the section can be solved. In the coming sections, the detail of 2-Alpha
approach will be described.
169
sl
sl
sv
sv
b)
a)
a) full interaction
in both directions
b) no interaction
in both directions
Figure 8.2 Strain profiles of the RC beam and the steel plates with different
level of longitudinal and transverse interaction
170
between the longitudinal strains of the steel plates and the RC beam of the
section with maximum moment at the centroidal level of the steel plates
( = p ,hp / c ,hp ). It is introduced to consider the reduction of axial strain of the
steel plates due to longitudinal partial interaction. On the other hand, the
curvature factor, expressed by the symbol , is defined as the ratio between
curvatures in the steel plates and the RC beam ( = p / c ), and is used to
consider the effect of transverse partial interaction on the section with
maximum moment.
Strain factor =
hp
p, h
c,h
Curvature factor =
Strain profile of RC
p ,h
p
c
c,h
Figure 8.3 Illustration of strain factor and curvature factor at the section
with critical moment
These two parameters range from 0 to 1 as the axial strain and curvature of
the steel plates are smaller than but of the same sign as that of the RC beam. In
the full interaction case, both strain factor and curvature factor are equal to 1.
This situation occurs when infinite stiffness of media is present, for example,
the situation when the steel plates are adhesively bonded on the RC beam.
When full interaction exists, the strain profiles of components coincide, as
illustrated in Figure 8.2a, and rigid plastic analysis commonly adopted for RC
beam sections can be carried out since the plane-section-remaining-plane-after-
171
bending is justified. At the same time, the force and moment induced by the
steel plate is maximized in the case of full interaction and the efficiency in
strengthening reaches maximum.
On the other extreme, when the interaction between the RC beam and the
steel plates is weak, the steel plates are only slightly mobilized or deformed.
The strain profiles of the RC beam and the steel plates for the extreme case
when there is no interaction are illustrated in Figure 8.2b. Even if the
deformation of the RC beam reaches the value corresponds to the ultimate
moment, the deformation and hence the force and moment induced in the steel
plates are equal to zero. This implies that the RC beam is actually unstrengthened since no addition strength is provided.
In general, for BSP beams, partial interaction exist in both longitudinal and
transverse directions. This is because both the longitudinal and transverse shear
forces are resisted by the finite shear stiffness of bolts. In this case, both the
axial strain of the steel plates, i.e. the strain at the plates centroidal level, and
the curvature of the steel plates are smaller than the counterparts of the RC
beam, as illustrated in Figure 8.2c. According to Figure 8.2c, the co-existence of
longitudinal and partial interactions significantly reduces the longitudinal strain
of steel plates. This leads to the un-yield condition in the steel plates in the
ultimate state, as observed and described in Chapter 4, and eventually flexural
strength reduction of the beam.
172
side faces of beam just underneath the soffit of slab. In this case, the vertical
deformations of steel plates are restrained by the slab. Thus the vertical
displacement profiles and hence the curvatures of RC and steel plates are the
same. But longitudinal partial interaction is still present as it is not restrained.
The strain profiles of the RC beam and the steel plates in this situation are
illustrated in Figure 8.2d.
173
W(x)
V(x)
a)
M(x)
b)
sl(x)
c)
Figure 8.4 Response of BSP beams under arbitrary gravity point loads in
terms of a) Shear force, b) Bending moment, and c) Longitudinal slip
174
xa1
1st seg.
xb1
xai
ith seg.
xbi
Supported end
nth seg.
k
kmi h 2
Vi h
sl( x) mi + mi +
= 0,
sl ( x) +
( EI ) p + ( EI )c
( EA) p ( EA)c ( EI ) p + ( EI )c
(8.1)
where
( EA)c , ( EI )c , ( EA) p , ( EI ) p are the axial and flexural stiffness of the RC beams
and the steel plates respectively,
kmi is the stiffness of bolt per unit length along the beam,
Vi is the shear force of the ith segment,
h is the separation of centroidal axis of the RC beam and the steel plates
Defining pi =
kmi
k
kmi h 2
+ mi +
, the longitudinal slip profile of
( EA) p ( EA)c ( EI ) p + ( EI )c
for
(8.2)
which depends on the value of x, while the second term Ci is the particular
solution which can be expressed as
Ci =
Vi h
pi
(( EI ) p + ( EI )c )
(8.3)
For the special case, where the shear connection stiffness kmi is zero, pi will also
become zero and Equation (8.3) is reduced to the following second order
differential equation,
sl( x) +
Vi h
= 0.
( EI ) p + ( EI )c
(8.4)
The solution of longitudinal slip for this special case has the form
sl ( x) = Ci x 2 + Ai x + Bi
for
(8.5)
Vi h
2 ( EI ) p + ( EI )c
(8.6)
It can be observed in Equations (8.3) and (8.5) that for each segment, three
conditions are required to solve the arbitrary constants. Thus to solve the
longitudinal slip profile of a span with n segments, a total of 3n conditions are
required. These 3n conditions include boundary conditions at the two ends of
spans, and the continuity of slip and slip strain at the boundary of segments.
176
(8.7)
where lp,hp ( x) and lc,hp ( x) are the longitudinal displacement of the steel plates
and the RC beam respectively, at the centroidal level of the steel plates.
lp,hp ( x) and lc,hp ( x) are, respectively, the longitudinal strain of the steel plates
and the RC beam at the centroidal level of steel plates at the section considered.
At the supported end, the longitudinal strain is zero as no bending moment or
axial force is induced. Hence the values lp,hp (0) and lc,hp (0) are equal to zero.
Further applying Equation (8.7), the first boundary condition of the longitudinal
slip profile can be expressed as,
sl (0) = 0
(8.8)
The maximum slip, sl,max, is at the supported end. Also the longitudinal slip
at the maximum moment end is equal to zero. Expressing these two conditions
mathematically, we have,
sl (0) = sl ,max ,
(8.9)
sl ( Ls ) = 0 .
(8.10)
As both the RC beam and the steel plates are treated as a continue media in
the present study, the longitudinal slip, which is equal to the difference between
177
for i = 1...n 1 .
(8.11)
Besides the continuity of longitudinal slip, the slip strain is also continuous
along the beam, provided that no external moments or axial force are applied in
the middle of the span. The continuity of slip strain can be expressed
mathematically as
sl ( xib ) = sl ( xia+1 )
for i = 1...n 1
(8.12)
for i = 2...n
(8.13)
Va p
where gi = 1a i are known once the geometry and load acting on the
Vi p1
178
beams. The longitudinal slip profiles of simple loading cases are given in
Appendices 8.1 and 8.2.
hc = hp + hslip .
p
(8.14)
The slip strain is equal to the first derivative of the longitudinal slip profile
along the span of the beam. As the slip strain at the section level is defined as
the strain of concrete subtracted by the strain of plate, the slip strain has the
opposite sign comparing to the first derivative of the slip profile,
hslip = sl( Ls ) .
(8.15)
The plate strain is equal to the plate force divided by the axial stiffness of the
plate
hp =
p
Fp ( Ls )
( EA) p
(8.16)
Consider the internal force equilibrium of the steel plates, the plate force at
the maximum moment end can be expressed as a function of stiffness of the
connecting media and the longitudinal slip profile as,
Ls
Fp ( Ls ) = km ( x) sl ( x)dx
(8.17)
179
Substituting Equations (8.15), (8.16) and (8.17) into (8.14), the strain of
concrete can be expressed as
hc =
Ls
km ( x) sl ( x)dx
( EA ) p
sl ( Ls )
(8.18)
Using Equations (8.16) and (8.18), the strain factor can be expressed as
hp
c
hp
Ls
km ( x) sl ( x)dx
( EA) p
Ls
km ( x) sl ( x)dx
( EA) p
=
1+
+ ( sl( Ls ))
1
( EA) p ( s l ( Ls ))
Ls
(8.19)
km ( x) sl ( x)dx
Equation (8.19) is the closed form solution form of the strain factor. The
strain factor is readily solved once the longitudinal slip profile along the span is
solved using the procedures described in Section 8.3.3.1. If elastic behaviour is
assumed for the shear connection, the expression of strain factor in Equation
(8.19) can be further simplified to
1
=
1+
( EA) p
km
sl( Ls )
Ls
(8.20)
sl ( x)dx
180
( EA) p sl( Ls )
term
is always positive as the first derivative of the
Ls
km
s ( x)dx
longitudinal slip profile is always negative. When the term is equal to zero, then
strain factor is equal to unity and full interaction exists. On the contrary when
the term is equal to infinity, thats the case when the stiffness of the connecting
media km approaches to zero, the strain factor would be equal to zero and no
interaction exists between steel plates and the RC beam.
sl ( x)
sl( Ls )
x
0
Area =
Ls
sl ( x)dx
e pLs + e pLs e px e px
sl ( x) = sl ,max
.
e pLs + e pLs 2
(8.21)
The general expression of strain factor in terms of the longitudinal sip profile
with elastic bolts is,
181
=
1+
( EA) p
km
sl( Ls )
Ls
(8.22)
sl ( x)dx
Ls
pLs
2 x=L
e +e
s
Ls
(8.23)
(8.24)
The strain factor can be solved by substituting Equations (8.23) and (8.24) into
(8.22) as
1
=
1+
( EA) p
km
p (e pLs e pLs )
pLs (e pLs + e pLs ) e pLs + e pLs
2
(8.25)
182
Ls
(8.26)
p (e p ( x Ls ) + e p ( x Ls ) e pLs e pLs )
sl ( Ls ) = sl ,max pLs
Ls
(8.27)
+
+
2
p
(
e
e
pL
(
e
e
))
(8.28)
Similarly, the strain factor is solved by substituting Equations (8.27) and (8.28)
into (8.22) as
1
=
1+
( EA) p
km
2 p (2 e pLs e pLs )
2e p ( x Ls ) + 2e p ( x Ls ) + p 2 x(2 Ls x)(e pLs + e pLs )
2
(8.29)
183
part of the total internal shear is transferred from the RC beam to the steel plates
near mid-span, and transferred back to concrete near the supports, through the
bolt connections. In this study, a simplified model, as shown in Figure 8.7, is
adopted for shear transfer between the RC beam and the steel plates. Shear
transfer zones are first identified on each shear spans. They are parts of the
shear spans where bolts are applied so that transverse shear flow is allowed. In
each shear transfer zone, the load varies linearly from at zone boundary near
support to - at the other zone boundary. Such model is adopted as it satisfies 1)
zero net force on plate, 2) symmetry about mid-span and 3) similarity to the
transverse slip profile between concrete and plate.
W(x)
a)
b)
Figure 8.7 Idealized transverse shear force model exerted on the RC beam
from the steel plates under arbitrary gravity loads
184
boundaries of the transfer shear, as illustrated in Figure 8.8. Denoting the zone
boundary at support and mid-span as a and b respectively, we have,
cv , stz = vp, stz + sv , xa + sv , xb .
(8.30)
The slip at the two zone boundaries is the same as it has been assumed that the
transverse shear flows at the two boundaries are both equal to and the
connecting media remaining elastic. So the transverse slips at the two
boundaries sv , xa and sv , xb are equal to the half of the total transverse slip at the
boundaries of the shear transfer zone sv , stz Thus Equation (8.30) can be
expressed as
cv , stz = vp, stz + sv , stz .
(8.31)
Zone boundary a
Zone boundary b
sv , xa
vp, stz
cv , stz
(1)
sv , xa
(2)
In the following section, the general procedure and the solution for some simple
loading cases and bolt arrangements are solved.
M p ( x) =
x2
2
x2
2
x3
3Ls
x3
3Ls
(8.32)
(8.33)
The curvature at the maximum moment point (mid-span in this case) can be
solved by substituting x=Ls into Equations (8.32) and (8.33) as
c ( Ls ) =
p ( Ls ) =
M c ( Ls )
PLs
Ls 2
=
,
( EI )c ( EI )c 6 ( EI )c
M p ( Ls )
( EI ) p
Ls 2
6 ( EI ) p
(8.34)
(8.35)
Integrating Equations (8.34) and (8.35) twice, the total deflections of the RC
beam and steel plates of the shear transfer zone can be solved and expressed
respectively as
cv , stz =
PLs 3
7 Ls 4
3 ( EI )c 120 ( EI )c
PLs
Ls 2
=
( EI ) 6 ( EI )
c
c
Ls 2 Ls 2
3 6 ( EI )
p
186
L 2
s
60
L2
L 2
= s c ( Ls ) s p ( Ls ) ,
3
60
vp, stz =
(8.36)
7 Ls 4
120 ( EI ) p
L 2
s
=
6 ( EI ) p
Ls 2
20
= Ls 2 p ( Ls ) .
20
(8.37)
The total transverse slip of the shear transfer shear zone at boundaries can be
expressed as
sv , stz =
2
km
12 ( EI ) p
=
km Ls 2
p ( Ls ) .
(8.38)
Substituting Equations (8.34), (8.35) and (8.36) into the transverse displacement
compatibility Equation (8.31), we have
12 ( EI ) p
Ls 2
Ls 2
7 2
(
L
)
p ( Ls ) .
c s
p ( Ls ) = Ls p ( Ls ) +
2
k
L
20
3
60
m
s
(8.39)
p ( Ls )
=
c ( Ls )
1
1 + 36 ( EI ) p
1+
+
4
20 km Ls
(8.40)
From the above expression, it can be seen that when the stiffness of the
shear connection is very small, i.e. km approaches to zero, the curvature factor
187
will also approach to zero. This is the case when there is no interaction between
the RC beam and the steel plate and the steel plate is not mobilized. On the
contrary, when km is infinitely large, Equation (8.40) becomes
1
1+
1+
20
(8.41)
In real case, the curvature factor should approach to unity when km is infinitely
1+
large. The
is the error term which is present due to the linear
20
approximation of the transverse shear transfer between the RC beam and the
steel plates. The error term will be reduced when a more accurate transverse
shear transfer model is adopted.
188
a)
b)
Figure 8.9 Idealized transverse shear force models exerted on the RC beam
from steel plates under uniformly-distributed load
By adopting the structural model as previously described in Figure 1 and
applying Euler beam theory, the general expression for the moment of the RC
beam and the steel plates can be expressed as,
M c ( x) = Ls x
M p ( x) =
x2
2
x2
x3
3Ls
x2
2
x3
3Ls
(8.42)
(8.43)
The curvature at the maximum moment point (mid-span in this case) can be
solved by substituting x=Ls into Equations (8.42) and (8.43) as
c ( Ls ) =
p ( Ls ) =
M c ( Ls )
Ls 2
Ls 2
=
,
( EI )c 2 ( EI )c 6 ( EI )c
M p ( Ls )
( EI ) p
Ls 2
6 ( EI ) p
(8.42)
(8.43)
Integrating Equations (8.42) and (8.43) twice, the total deflection of the RC and
plate of the shear transfer zone can be solved and expressed as
189
c
v , stz
5 Ls 4
7 Ls 4
=
24 ( EI )c 120 ( EI )c
Ls 2
Ls 2
=
2 ( EI ) 6 ( EI )
c
c
5L 2 Ls 2 Ls 2
s +
6 ( EI ) p 15
12
5L 2
L 2
= s c ( Ls ) + s p ( Ls ) ,
12
15
p
stz
(8.44)
7 Ls 4
=
120 ( EI ) p
L 2
s
=
6 ( EI ) p
Ls 2
20
= Ls 2 p ,max .
20
(8.45)
The total transverse slip of the shear transfer shear zone at boundaries can be
expressed as
sv , stz =
2
km
12 ( EI ) p
=
km Ls 2
p ( Ls ) .
(8.46)
Substituting Equations (8.44), (8.45) and (8.46) into the transverse displacement
compatibility Equation (8.31), we have
12 ( EI ) p
5Ls 2
Ls 2
7 2
c ( Ls ) +
p ( Ls ) = Ls p ( Ls ) +
2
20
12
15
km Ls
p ( Ls ) .
(8.47)
190
p ( Ls )
=
c ( Ls )
4 12 144 ( EI ) p
1 +
+
4
25 75 5km Ls
4 12
where +
25 75
(8.48)
The methodology for solving the curvature factor has been illustrated by
the two examples above. Similar procedures can be applied to find out the
curvature factors of different loading conditions and bolt arrangements.
191
y = ( y hna )
(8.49)
y = y ( y y0 )
(8.50)
In BSP beam, the strain profiles of sections of the RC beam and the steel
plates do not coincide due to the presence of partial interaction. But the plane
section remaining plane assumption is valid for the RC beam and the steel
plates individually. By setting the pivotal point at the centroidal level of steel
192
plate, the strain of the RC beam and the steel plates at level y can be expressed
as,
yc = hc c ( y hp )
(8.51)
yp = hp p ( y hp )
(8.52)
By substituting the strain factor and the curvature factor introduced earlier, the
strain of steel plate at any level y can be expressed as a function of strain of RC
beam at pivotal point.
yp = hc c ( y hp )
(8.53)
Equations (8.51) and (8.53) express the strain of the whole section in terms of
the concrete strain at plates centroidal level and the curvature of RC beam once
the strain factor and curvature factor are solved. With proper use of these
equations, non-linear moment curvature analysis on BSP beams can be carried
out.
In this modified section analysis, two pivotal points for the strain profiles
of the RC beam and the steel plates respectively at the centroidal level of steel
plates are first fixed, as illustrated in Figure 8.10. With the pivotal points fixed,
193
a trial and error process is carried out to find the curvature of RC beam and steel
plate section, which is related by the curvature factor.
The condition for solving the correct curvature is the one which results in a
net zero horizontal resultant force. Since the stress-strain relationship of the
material properties are non-linear, numerical integration is carried out. The
concrete component is divided into m horizontal layers and the steel plate is
divided into n layers. The area of each layer in concrete and steel plate is Ac and
Ap respectively and the strain is assumed to be constant in each layer. By
summing up the forces contributed by concrete and steel plate layers and the
reinforcements, the net horizontal force H can be found by the following
equation
m
i =1
i =1
H = Ac c + Ast st + Asc sc + Ap p
(8.54)
where c , p are the stresses of concrete and steel plates at the corresponding
layers, st , sc are the stresses of tension and compression reinforcements. The
stress terms in Equation (8.45) can be solved by applying any defined stress
strain relationships like the nonlinear curve suggested by Attard and Stewart
(1998), which is adopted in this study.
194
i =1
i =1
(8.55)
Also, the moment of RC, moment of steel plates and the axial force couple can
also be calculated using Equations (8.56)-(8.58)
m
(8.56)
i =1
n
M p = Api pi ( yi hc ) ,
(8.57)
i =1
n
Fp h = h Api pi .
(8.58)
i =1
Fc
Pivotal
points
Fp
Fs
hc
hc
Strain profile
Figure 8.10 Section analysis with strain at centroidal level of the steel
plates as pivotal points
195
yi
ci
yi
Asc
d
Aci
pi
Ast
Api
RC beam
Steel plates
c / f co =
A( c / co ) + B( c / co ) 2
1 + ( A 2)( c / co ) + ( B + 1)( c / co ) 2
196
(8.59)
where
A=
Ec co
( A 1) 2
; B=
1
f co
0.55
for
c co
(8.60)
A=
f ci ( ci co ) 2
; B=0
co ci ( f co f ci )
for
c > co
(8.61)
(8.62)
(8.63)
(8.64)
(8.65)
197
P
(Pu, su)
(Py, sy)
198
Step 3: The geometric properties of the beams, including the length, the section
dimensions and the arrangement of anchoring bolts on the beams are
defined.
Step 4: The normalized slip profile is solved based on the information defined
in Step 1-3 and based on the theory developed in Section 8.3.3.1.
Step 5: The slip increment sl is applied to drive the load step j.
Step 6: The plate force at the critical section i.e. total bolt force induced on the
span is integrated numerically. With the plate force known, and using
the slip profile developed, the strain factor and curvature factor, and the
strain of RC and steel plate at centroidal level of steel plate are
computed.
Step 7: The kth trial curvature is chosen and the corresponding net horizontal
force is calculated by applying the stress-strain relationship for the
corresponding material.
Step 8: The net horizontal force is checked. If the force is less than the pre-set
tolerance, then horizontal equilibrium is considered to be satisfied. The
trial curvature is set as the solution curvature at load step j and the
program is proceeded to step 9. If horizontal equilibrium is not satisfied,
step 7 will be carried out again with a new trial generated by an
iteration scheme.
Step 9: The total moment of the section, as well as the moment of RC beam Mc,
moment of steel plate Mp and the force couple Fp(Ls)h are computed.
Step 10: The total moment of the section at the current load step j (Mj) is
compared with the moment of the previous load step (Mj-1). When Mj >
Mj-1, the program is proceeded to step 5 and the next load step is
199
carried out. When Mj Mj-1, the maximum moment is reached and the
program terminates.
In each load step, the secant method is adopted as the iterative scheme to
speed up the solution procedure. The secant method is simple, yet efficient in
converging to the numerical solution. The horizontal resultant force of the
section is considered as the error term. According to the secant method, trial
curvature of the kth iteration is related to trials in the k-1th and k-2th iterations by
the following equation,
kt = kt 1
H k 1
(kt 1 kt 2 ) ,
H k 1 H k 2
(8.66)
where H k 1 and H k 2 are the resultant horizontal equilibrium in the k-1th and k2th iterations respectively. And the trial curvature is considered as the solution of
the load step when the net horizontal resultant force is less than 0.01kN.
200
Program start
Input specification
Core program
s = s+s
No
Moment
dropped?
Yes
Program ends
201
202
Error (%)
2-Alpha
approach (D)
Experiment (E)
(D-E)/E*100
Unit SBSP
172.0
161.5
6.5
Unit WBSP
144.1
149.2
-3.4
Unit WBWP
135.3
133.4
1.4
Unit SBWP
147.3
144.6
1.8
203
From the above comparison, it can be seen that the 2-Alpha approach is
capable of solving the full moment curvature response of BSP beams at critical
sections, as well as the internal load resisted by the RC beam and the steel plates.
Using 2-Alpha approach, the structural performance of BSP beams with the
presence of partial interactions can be closely predicted.
204
2-Alpha
160
Moment (kNm)
120
80
40
Experimental
2-Alpha
DCBA
0
0
10
20
30
40
Curvature (0.001m-1)
160
Moment (kNm)
120
80
Experimental
40
2-Alpha
DCBA
0
0
10
20
30
40
Curvature (0.001m-1)
160
Moment (kNm)
120
80
Experimental
40
2-Alpha
DCBA
0
0
10
20
30
40
Curvature (0.001m-1)
205
100%
80%
(2-Alpha)
Plate flexure (DCBA)
60%
40%
20%
0%
0
40
80
120
160
200
Moment (kNm)
100%
80%
60%
40%
20%
0%
0
40
80
120
160
Moment (kNm)
100%
80%
60%
40%
(2-Alpha)
Concrete flexure (DCBA)
20%
0%
0
40
80
120
160
Moment (kNm)
100%
80%
60%
40%
20%
0%
0
40
80
120
160
Moment (kNm)
206
Strengthening plates
2 pieces of steel plates with yield stress fyp=300MPa, 6mm thick and
150mm deep each. Strengthened along the whole beam span
Position of strengthening plates
Centroidal axis of steel plates at 250mm from the top surface of the beam
Bolt arrangement
Two rows of bolts 37.5mm above and below the centroidal axis of the steel
plates. Each row consists of 10 bolts evenly distributed. Longitudinal bolt to
bolt separation equals to 400mm
2T10
P
1200
1200
3600
3T16
207
150mm
250mm
tp
tp=6mm
400
37.5
37.5
3600
The properties of concrete, steel reinforcement, steel plates and anchor bolts are
given below:
Material properties of concrete
fcu: 35 MPa
Material properties of reinforcement
Es: 200 GPa
fys: 460 MPa
Material properties of steel plates
Ep: 200 GPa
fyp: 300 MPa
Mechanical properties of anchor bolts
Bi-linear with yield-point and ultimate point given below:
Yield force: 50kN, Yield slip: 0.63mm
Ultimate force: 76kN, Ultimate slip: 4mm
208
The total strength of bolts on a shear span is equal to 1076 = 760kN. The
maximum tensile force of the steel plates equals to 1800300/1000 = 540kN. So
the degree of shear connection is at least 760/540 = 1.41 > 1. This implies that
full shear connection exists and only partial interaction affects the strength of
beam.
947583kN
7512kNm2
Centroidal axis of the RC cracked section: 102mm from the top of the
beam
Properties of strengthening steel plates:
Axial stiffness EAp:
360000kN
675kNm2
Centroidal axis of steel plates: 250mm from the top of the beam
Properties of connecting media
Stiffness of the connecting media km: 444kN/mm/m
Using the above information, the longitudinal slip profile can be solved
sl ( x ) = smax ( 0.0855e0.0017 x 0.0855e 0.0017 x + 1.1710 ) for 0 x < 1200
sl ( x ) = smax ( 0.0096e 0.0017 x + 4.4338e 0.0017 x ) for 1200 x 1800
209
Using this set of strain and curvature factors, modified section analysis is
carried out using the non-linear stress-strain relationship given in Appendix 2
for concrete and elastic-plastic relationship for steel plates and reinforcement.
The moment-curvature response of the section is given in Figure 8.18.
According to Figure 8.18, the flexural strength of the BSP beam is 172 kNm
when the curvature of the section is equal to 3510-3 m-1.
210
The analysis was implemented into a spreadsheet and validated using the
experimental results presented in Chapter 3. It was concluded that the analysis
is able to properly consider the effect of partial interaction in predicting the
moment-curvature response and the moment contribution of various
components of BSP beams. A design example was given to demonstrate how
the analysis can be carried out when design schemes for BSP beams are
proposed.
211
212
CHAPTER 9
CONCLUSIONS AND RECOMMENDATIONS
9.1 SUMMARY
The current research aims to promote the use of bolted side plate to
strengthen existing RC beams in order to extend the life span of existing
structures and facilitate the concept of adaptive re-use in buildings with great
cultural and historical values.
Studies in literature also pointed out that BSP beams behave uniquely as
partial interaction exists due to the slip of connecting media, leading to
reduction of stiffness and strength. However, many critical aspects, for example,
the method to quantify partial interaction in BSP beams, the relationship
between strength reduction and partial interaction and how partial interaction
being affected by different geometries of BSP beams are still uncertain.
213
bolt weak-plate were tested in order to reveal the behaviour of BSP beams with
all different bolt-plate arrangements. Their performances were compared and a
classification on different types of BSP beams was proposed.
The focus of the study was then shifted to the partial interaction between
the RC beam and the steel plates due to the mechanical slip of connecting bolt
groups. Two non-linear models, the elastic-plastic model and the kinematic
hardening model, were developed to evaluate the non-linear response of bolt
groups under combined in-plane shear and moment. The kinematic hardening
model, being more appropriate to model the load-deformation characteristics of
anchor bolts, was applied in the macro-finite-element model developed to
model the non-linear response of BSP beams. In the model, the steel plates and
the RC beam were modelled as different elements interacted with each other
through discrete groups of bolts. The model was verified using experimental
data and applied to illustrate the non-linear slip response between the RC beam
and the steel plates in BSP beams.
214
by carrying out the section analysis which is commonly used for estimating the
flexural strength of RC sections.
9.2 CONCLUSIONS
The experimental study conducted has proven that adding steel plates to
the side faces of RC beam with the use of bolts is an effective way to strengthen
them. It was found that the behaviour of BSP beams with different bolt-plate
combinations are quite different and a classification is introduced to divide BSP
beams into different groups according to the bolt-plate arrangements used. The
longitudinal slip between the RC beam and the steel plates was measured by a
rhombic set of LVDTs and it was found that in non-linear stage, slip strain
accumulation in plastic hinges occurs, leading to a reduction in interaction
between the RC beam and the steel plates and eventually a significant strength
reduction of the beam compared with the full interaction strength. It was
concluded that the effects of partial interaction can significantly reduce the
strength of beam for both full shear connection and partial shear connection
cases.
215
RC beam and the steel plates and eventually lowering the strength of BSP
beams. It was also observed that the sections of RC beam and steel plates rotate
relatively to each other. This type of partial interaction, namely rotational partial
interaction, occurred simultaneously with transverse partial interaction and
results in uneven distribution of bolt forces when more than one row of bolts are
used.
In the experimental study, specimens with different plate sizes were used.
However, the plate depth was always smaller than half depth of the RC beam,
which means that the plate depth is relatively small. The behaviour of BSP
beams can be different when deeper steel plates are used. The strengthened BSP
beams may be dominated by the flexural resistance of the steel plates, rather
than the axial force couple. More experimental studies should be carried out for
deep-plated specimens, where the curvature factor is more likely to govern the
behaviour of BSP beams and the effects of buckling of steel plates.
216
The computer program SiBAN has been developed to model the response
of BSP beams. However, simple guidelines have not yet developed to aid
practicing engineers in real design situations. It is recommended that a
sophisticated parametric study should be carried out using SiBAN to investigate
the performance and behavioural change when different bolt-plate arrangements
are used. Simple equations relating the basic performance and the geometries of
the strengthening system, for example, the relationship between the additional
strength required and the geometry of the bolt plate arrangement, should be
developed.
217
218
APPENDICES
APPENDIX 1 Evaluation of slip using rhombic arrangement of LVDTs
Consider an example when a coordinate system (x,y) is set on the side face of
the RC beam. The steel plate slips relatively to the RC beam with an amount
(sl,sv,s), as illustrated in Figure A.I.1. The reference points of the steel plate on
the metal strip move rigidly together with the target plane of the steel plate. The
distances between the reference points are changed and measured by the
LVDTs. The slips could then be solved by considering trigonometry.
The slips at points on the shear span are measured by rhombic arrangements of
LVDTs. The LVDTs measure the time-to-time distances between reference
points as load is applied to the beams. The original length of the LVDT set
(LLVDT) is equal to the initial distance between the reference points on the RC
beam and the steel plate, which can be expressed as
LLVDT =
Lrs 2 + Lrc 2
,
2
(A.1.1)
where Lrs and Lrc are the distances between the reference points on the RC beam
and the steel plate, respectively. The instantaneous distances between the
reference points ab, bd, ac and cd could be expressed as
lab = LLVDT + lab ,
(A.1.2)
(A.1.3)
(A.1.4)
(A.1.5)
219
where lab , lbd , lac , lcd are the readings of LVDTs between the nodes denoted
by the subscripts.
Lrs
sin ( + ) ,
2
(A.1.6)
Lrc
L
( LLVDT + lab ) cos + rs cos ( + ) ,
2
2
(A.1.7)
(A.1.8)
2
2
2
L 2 + lab 2 lbd 2
1 Lrs + lab lac
where = cos 1 rc
and
cos
=
.
2lab Lrc
2lab Lrs
a
LLVDT
lac
lab
c
b
Lrc
sv
s
x
sl
lcd
lbd
Lsp
220
The concrete model used was based on the model proposed by Attard and
Stewart (1998), with minor modifications to account for the local properties of
concrete. In the proposed model, the parameters required are the initial Youngs
modulus of concrete Ec, the peak compressive stress of concrete fco and the
corresponding strain co, and the stress fci and strain ci of the inflection point on
the descending branch of the model. The compressive stress of concrete c can
be related to the compressive strain c by
c / f co =
A( c / co ) + B( c / co ) 2
,
1 + ( A 2)( c / co ) + ( B + 1)( c / co ) 2
(A.2.1)
where
A=
Ec co
( A 1) 2
; B=
1
f co
0.55
A=
f ci ( ci co ) 2
; B=0
co ci ( f co f ci )
for
for
c co ,
c > co .
(A.2.2)
(A.2.3)
(A.2.4)
(A.2.5)
(A.2.6)
(A.2.7)
221
Ki, p
c8
c2
c9
c3
c1
c8
c8
c2
c4
c9
c1
c3
c8
c2
sym.
c10
c7
c6
,
c10
c7
c5
where
c1 =
1 3
Ek Ak wk ,
2 L k =1
c2 =
18 3
E I (2 pk 1) 2 wk ,
3 k k
L k =1
c3 =
6
L2
E I
k k
(2 pk 1)(3 pk 2) wk ,
k =1
c4 =
2 3
Ek I k (3 pk 2) 2 wk ,
L k =1
c5 =
2 3
Ek I k (3 pk 1)2 wk ,
L k =1
6 3
c6 = Ek I k (3 pk 1)(3 pk 2) wk ,
L k =1
c7 =
c8 =
6
L2
3
L2
E I
k k
(3 pk 1)(2 pk 1) wk ,
k =1
E S
k
(1 2 pk ) wk ,
k =1
c9 =
1 3
Ek Sk (2 3 pk ) wk ,
L k =1
c10 =
1 3
Ek Sk (1 3 pk ) wk .
L k =1
222
(A.3.1)
K i , RC
a8
a2
a9
a3
a1
a8
a8
a2
a4
a9
a1
a3
a8
a2
sym.
a10
a7
a6
,
a10
a7
a5
(A.3.2)
where
a1 =
1 3
Ek Ak wk ,
2 L k =1
a2 =
18 3
E I (2 pk 1) 2 wk ,
3 k k
L k =1
a3 =
6
L2
a4 =
Ek I k (2 pk 1)(3 pk 2)wk
k =1
3e 3
Ek Sk (1 2 pk ) wk ,
L2 k =1
2 3
2e 3
e2 3
2
(3
2)
(2
3
)
E
I
p
E
S
p
w
+
kk k
k k
Ek Ak wk ,
k
k
k
2 L k =1
L k =1
L k =1
2 3
2e 3
e2 3
2
a5 = Ek I k (3 pk 1) wk + Ek Sk (1 3 pk ) wk +
Ek Ak wk ,
L k =1
L k =1
2 L k =1
e 3
e2 3
6 3
a6 = Ek I k (3 pk 1)(3 pk 2) wk Ek Sk (3 6 pk ) wk +
Ek Ak wk ,
L k =1
L k =1
2 L k =1
a7 =
a8 =
6
L2
E I
3
L2
E S
k k
(3 pk 1)(2 pk 1) wk +
k =1
3e 3
Ek Sk (1 2 pk ) wk ,
L2 k =1
(1 2 pk ) wk ,
k =1
a9 =
1 3
e 3
Ek S k (2 3 pk ) wk
Ek Ak wk ,
L k =1
2 L k =1
a10 =
1 3
e 3
Ek S k (1 3 pk ) wk +
Ek Ak wk .
L k =1
2 L k =1
223
A.4.1 Assumptions:
1) Euler beam theory applies to RC and steel plate sepearately, but not as a
whole. That means both RC and steel plates obey the rule of plane section
remains plane after bending,but the strain profile of them are not necessarily
coincident.
2) The bolts are assumed to be distributed along the centroidal axis of the steel
plate. And instead of discrete point connections, the connecting media is
assumed to be continuous along the beam.
3) The effect of material nonlinearity on the longitudinal slip profile is ignored
224
Va
Vb
Ma
Mb
M c ( xa )
Fp ( x a )
M p (xa )
M c ( xb )
Fp ( x b )
M p ( xb )
Figure A.4.1 Free body diagram and internal force distribution of beam segment
(A.4.1)
(A.4.2)
The relationship between plate force and bolt slip can be represented by
x
Fp ( x) = km s ( x) dx ,
(A.4.3)
s ( x) =
Fp ( x)
( EA) p
(A.4.4)
225
c ( x) =
M c ( x)h Fp ( x)
.
( EI )c ( EA)c
(A.4.5)
(A.4.6)
(A.4.7)
, we have
M p = Mc .
(A.4.8)
1
( M Fp h )x .
1+
(A.4.9)
Substituting (3), (4), (5) and (9) into (2) and integrate, we have
x
x
x
(
)
h
M
x
h
s ( x) dx
k
s
(
x
)
dx
k
s
(
x
)
dx
m
m 0
m 0
0
s ( x) s (0) =
+
0
EA
EA
(1 + ) ( EI ) RC
(
)
(
)
p
RC
) dx .
(A.4.10)
Differentiating Equation (A.4.10) twice and rearranging, we have
k
km
km h2
s( x) m +
+
( EA) p ( EA) RC (1 + ) ( EI ) RC
hM ( x)
s( x) +
= 0.
(1 + ) ( EI ) RC
226
(A.4.11)
sl( x) p 2 sl ( x) +
where p =
V h
=0,
(1 + ) ( EI ) RC
(A.5.1)
km
km
km h 2
+
+
.
( EA) p ( EA) RC (1 + ) ( EI ) RC
sl ( x) = Ae px + Be px + C ,
(A.5.2)
1
p
e pLs
1
p
e
pLs
1 A sl ,max
0 B = 0
1 C 0
(A.5.3)
Solving Equation (A.5.3) and back substitute the solution of A, B and C into
Equation (A.5.2), the longitudinal slip profile can be found as,
227
e pLs + e pLs e px e px
sl ( x) = sl ,max
.
e pLs + e pLs 2
(A.5.4)
Substituting Equation (A.5.4) into Equation (A.5.1), the slip profile can be
expressed in terms of the shear force and the geometric properties of the beam
as
V h
e px + e px
sl ( x) = 2
1
.
p (1 + ) ( EI ) RC e pLs + e pLs
(A.5.5)
A.5.2 Longitudinal slip profile of simply-supported BSP beams with evenlydistributed bolts under uniformly-distributed loading
x 2
2
(A.5.6)
(A.5.7)
228
k
km
km h2
m
s( x)
+
+
( EA) p ( EA) RC (1 + ) ( EI ) RC
h( Ls x)
s( x) +
= 0 (A.5.8)
(1 + ) ( EI ) RC
h( Ls x)
,
p (1 + ) ( EI ) RC
s p ( x) =
sc ( x) = Ae px + Be px .
where p =
(A.5.9)
km
km
km h2
+
+
and A, B are arbitrary constants.
( EA) p ( EA) RC (1 + ) ( EI ) RC
(A.5.10)
1
e
pLs
Ls A sl ,max
0 B = 0 .
1 C 0
(A.5.11)
(A.5.12)
Substituting Equation (A.5.12) into Equation (A.5.8), the slip profile can be
expressed in terms of the shear force and the geometric properties of the beam
as
h ( (e p ( x L ) e p ( x L ) ) + p( Ls x)(e pL + e pL ) )
s
sl ( x) =
p3 (1 + ) ( EI ) RC (e pLs + e pLs )
229
(A.5.13)
A.5.3 Longitudinal slip profile of simply supported BSP beams under two
symmetric point loadings with bolts only on shear spans
Span 2
Span 1
P
P
x
SFD
Segment 1
Segment 2
The span of the beam in this case is defined as half of the total length of beam
and the each span can be divided into 2 segments, the constant shear segment
and the zero shear segment. According to the above bolt and loading
arrangements, and using the theory developed in Section 8.3.3.1, the
longitudinal slip profile of the segments can be summarized in Table A.5.1.
V (x)
Slip profile
Segment 1
Constant
Constant
Segment 2
sl ( x) = A2 + B2 x
230
The shape of the longitudinal slip profile of the beam is shown in Figure A.5.4
and the normalized slip function of the span will be solved.
L
X
Segment 1
Ls
Segment 2
Span 1
0
Span 2 (Anti-symmetric with span 1)
A1 + B1 + C1 = 1
(A.5.14)
A1 B1 = 0
(A.5.15)
(A.5.16)
(A.5.17)
231
(A.5.18)
Expressing the above 5 equations into matrix form, we have the following
matrix equation
1
1
p1 X
e
p1e p1 X
1
1
p1 X
e
p1e p1 X
1
1
Ls
X
A1 1
B 0
1
1 C1 = 0
1 A2 0
B2 0
(A.5.19)
The above matrix represents the solution for the arbitrary constants required to
define the normalized longitudinal slip profile. Once the material properties and
the value X are known, the arbitrary constants can be solved numerically and the
normalized longitudinal slip profile can be found.
232
For a more complicated case, like the specimens adopted in experimental study,
which symmetrical four point bending were chosen, the derivation is more
complicated. The structural model used for shear transfer in this case is
summarized in Figure A.6.1. The shear transfer zones are the two shear spans,
each with a length of xn. Similar to example 1, the shear transfer follows a linear
profile varying from to -.
P
a)
b)
xn
xn
Figure A.6.1 Idealized transverse shear force models adopted for a) the whole
plated beam, b) transverse shear force exerted on beam from plate
Using similar procedure as that of the central point load case, the moment of the
RC beam and the steel plates in the shear transfer zone can be expressed as
M c = Px
Mp =
x2
2
x2
2
x3
3Ls
x3
3Ls
(A.6.1)
(A.6.2)
233
Using Equations (A.6.1) and (A.6.2), the curvature at constant moment zone,
where the moment is maximum can be solved as
c ,max =
p ,max =
M c ,max
( EI )c
M p ,max
( EI ) p
Pxn
xn 2
=
,
( EI )c 6 ( EI )c
=
xn 2
6 ( EI ) p
(A.6.3)
(A.6.4)
Integrating Equations (A.6.3) and (A.6.4) once and using the boundary
condition that the rotation of RC and steel plates at xn is equal to the product of
the curvature of constant moment zone in the corresponding element and the
length Ls-xn, the rotation of RC and steel plates in the shear transfer zone can be
expressed as
xn xn 2
x
1 Px 2 x3 xn 4
c ( x) =
Ls n
+
Pxn Ls +
6
12 xn
2
6
2
( EI )c 2
p ( x) =
1
( EI )c
x 3 xn 4 xn 2
x
Ls n .
12 xn
6
2
6
(A.6.5)
(A.6.6)
Further integrating Equations (A.6.5) and (A.6.6) once, the deflection of the RC
and plate of the shear transfer zone can be solved and expressed as
Pxn
xn 2
cv , stz =
( EI ) 6 ( EI )
c
c
2 2 xn 2
Ls xn xn
3
6 ( EI ) p
x 2
n
60
(A.6.7)
= Ls xn xn 2 c ,max xn 2 p ,max
3
60
p
v , stz
x 2
n
=
6 ( EI )
p
13 2
Ls xn
xn
20
(A.6.8)
13
= Ls xn xn 2 p ,max
20
234
The total transverse slip of the transfer shear zone at the two boundaries can be
expressed as a function of curvature of steel plate as
sv , stz =
2
km
12 ( EI ) p
=
km xn 2
(A.6.9)
p ,max
xn p ,max +
Ls xn xn c ,max xn p ,max = Ls xn
km xn 2
3
20
60
p ,max
(A.6.10)
Rearranging and simplifying Equation (A.6.10), the curvature factor can be
expressed as
p , mid
=
c ,mid
1
3L
1 +
1 + s 2
xn
20
1
36 ( EI ) p
+
4
km xn
235
(A.6.11)
236
REFERENCES
1. Abolitz, A.L. (1966). Plastic design of eccentrically loaded fasteners.
Engineering Journal, American Institute of Steel Construction, 3(3), pp122-131.
2. ACI Committee 355. State-of-the-art Report on Anchorage to Concrete
(ACI 355.1R-91). American Concrete Institute: Detroit.
3. Ahmed, M., Oehlers, D.J., and Bradford, M.A. (2000) Retrofitting
reinforced concrete beams by bolting steel plates to their sides. Part 1:
Behaviour and experimental work. Structural Engineering and Mechanics,
10(3), pp211-226.
4. Arizumi, Y., and Hamada, S. (1981). Elastic-plastic analysis of composite
beams with incomplete interaction by finite element method. Computers and
Structures, 14, pp453-362.
5. Attard, M.M., and Stewart M.G. (1998). A two parameter stress block for
high-strength concrete, ACI Structural Journal, 95(3), pp305-317.
6. Ayoub, A. (2000). Mixed formulation of nonlinear steel-concrete
composite beam element. Journal of Structural Engineering ASCE, 126(3),
pp371-381.
7. Bendigo, R.A., Hansen, R.M., and Rumpf, J.L. (1963) Long bolted
joints, Journal of the Structural Division ASCE 1963, 89(6).
8. Carreira, D.J., and Chu, K.H. (1986). The moment-curvature relationship
of reinforced concrete members. ACI Journal, 83(2), pp191-198.
9. Chapman, J.C., and Balakrishnan, S. (1964). Experiments on composite
beams. Structural Engineer, 1964, 42(11), pp369-383.
10. Crawford, S.F., Kulak, G.L. (1967). Behaviour of eccentrically loaded
bolted connection. Department of Civil Engineering, Nova Scotia Technical
College, Halifax, Nova Scotia.
11. Crawford, S.F., Kulak, G.L. (1971). Eccentrically loaded bolted
connection. Journal of the Structural Division, ASCE, 97(3), pp765-783.
12. DallAsta, A., and Zona., A. (2004). Three-field mixed formulation for
the non-linear analysis of composite beams with deformable shear connection.
237
238
239
242
LIST OF PUBLICATIONS
1.
SIU, W.H., and Su, R.K.L. (2007). Non-linear behaviour of bolt groups
under eccentric and inclined shear, Proceedings of The Third
International Conference on Structural Engineering, Mechanics and
Computation, Cape Town, South Africa, 10-12 September 2007, CD ROM.
2.
Su, R.K.L., and SIU, W.H. (2007). Nonlinear response of bolt groups
under in-plane loading, Engineering Structures, 29, pp626-634.
3.
4.
5.
SIU, W.H. and Su, R.K.L. (2009). Nonlinear analysis of bolted sideplated reinforced concrete beams, Proceedings of The 1st International
Conference on Computational Technologies in Concrete Structures, 24-27
May 2009, Jeju, Korea, CD ROM.
6.
SIU, W.H., and Su, R.K.L. (2009). Analysis of bolted side plated beam
with partial interaction, Computers and Concrete. (Accepted in July 2009)
7.
Su, R.K.L, Siu, W.H., and Smith, S.T. (2009). Effect of bolt-plate
arrangements on plate strengthened concrete beams, Engineering
Structures. (Under review in June 2009)
8.
SIU, W.H., and Su, R.K.L. (2009). Effects of plastic hinges on partial
interaction behaviour of bolted side-plated beams, Journal of
Constructional Steel Research. (Under review in August 2009)
9.
SIU, W.H., and Su, R.K.L. Moment-curvature analysis of bolted sideplated beam. (Under preparation)
243