Full Text

Title
Flexural strengthening of reinforced concrete beams by bolted

side plates
Advisor(s)
Su, KL
Author(s)
Siu, Wing-ho.; .
Citation
Issued Date
URL
Rights
2009
http://hdl.handle.net/10722/56994
The author retains all proprietary rights, (such as patent rights)

and the right to use in future works.
FLEXURAL STRENGTHENING OF
REINFORCED CONCRETE BEAMS BY
BOLTED SIDE PLATES
by
SIU, WING HO
B.Eng., HKU, HKSAR, PRC
PHD THESIS
A thesis submitted in partial fulfillment of the requirements

for the degree of Doctor of Philosophy
at the University of Hong Kong
August, 2009
Abstract of thesis entitled
Flexural Strengthening of Reinforced Concrete Beams by Bolted

Side-Plates
Submitted by
SIU, Wing Ho
For the degree of Doctor of Philosophy at
The University of Hong Kong in August 2009
The social awareness of the structural deterioration of reinforced concrete (RC)

buildings is raised in the past decade as the dilapidated buildings jeopardize
public safety. Although demolition and reconstruction of these buildings could
solve the deterioration problem, a large amount of construction waste generated
would seriously pollute the environment. Instead of demolishing these
dilapidated buildings, strengthening of deteriorated structures can be a viable
alterative to upgrade these structures to satisfactory conditions. Both the amount
of waste generated and the cost involved are greatly reduced.
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.
Different methods for strengthening existing RC beams have been proposed in

the literature. One of the effective and practical ways is to attach steel plates to
the side faces of RC beams with bolts. The advantages of this bolted sideplated (BSP) strengthening technique are that both the flexure and shear
capacities of existing beams can be enhanced and satisfactory level of ductility
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.
A numerical technique for modeling the responses of BSP beams is then

developed. The RC beam and steel plates were modeled as two different
elements, interacted through discrete groups of bolts. The formulation of the
element stiffness matrix is based on a layered finite element model. To consider
the non-linear behaviour of bolt groups, a non-linear model of bolts under
multi-directional shear together with a procedure to formulate the non-linear
tangential stiffness matrix of bolt groups was adopted. A computer program
SiBAN has been implemented based on the proposed model.
A new concept for quantifying the amount of partial interaction by considering

the deformation of RC beam and steel plates together with the displacement
compatibility of the entire BSP beam is proposed. The strain and curvature
factors which are expressed in terms of the actual bolt-plate arrangement were
introduced. The new approach has been validated by comparing the numerical
results with the experimental results.
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.
Thanks are given to all technicians in the structural engineering laboratory of

the University of Hong Kong for their technical assistance in experimental
works. Without their efforts, the experimental part of this research would not
have been conducted successfully.
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
LIST OF ABBREVIATIONS .. xvi

LIST OF NOTATIONS xvii
CHAPTER 1 INTRODUCTION
1.1 Overview
1.2 Objective of the study
1
1
5
1.3 Scope of thesis ...
CHAPTER 2 LITERATURE REVIEW

2.1 Overview.
9
9
2.2 External strengthening of reinforced concrete beams

2.2.1 Strengthened with adhesively bonded steel plates ..
2.2.2 Strengthened with mechanically anchored steel plates ...
2.2.3 Strengthened with adhesively bonded fibre reinforced plastic
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
2.3.2 Shear interaction between RC beam and steel plates .

2.3.2.1 Shear interaction in standard composite beams ...
2.3.2.2 Shear interaction in BSP beams ...
2.3.3 Buckling of strengthening steel plates
2.4 Numerical studies on composite beams with partial shear connection
and interaction
16
16
18
19
2.5 Rigid plastics analysis of composite beam sections ..

2.6 Concluding remarks ...
22
25
20
CHAPTER 3 EXPERIMENTAL STUDY ON BOLTED SIDE-PLATED
27
BEAMS WITH PARTIAL INTERACTION

3.1 Overview
3.2 Bolted side-plated beam tests
3.2.1 Test setup
27
28
28
3.2.2 Specimen detail ...

3.2.3 Material properties ..
3.2.4 Specimen fabrication ..
3.2.5 Loading history ...
29
33
38
40
3.2.6 Instrumentation ...

3.2.6.1 Strain gauge arrangements ...
3.2.6.2 LVDT and LDT arrangements .
40
40
41
3.2.6.3 Rhombic arrangements of LVDTs ...

3.3 Shear test of HAS-E anchor rods ...
3.3.1 Test program ...
42
46
46
3.3.2 Specimen design and fabrication

3.3.3 Test setup
3.3.4 Loading history and instrumentation ...
3.3.5 Result analysis
46
48
49
CHAPTER 4 RESULT AND ANALYSIS OF EXPERIMENTAL
55
STUDY ON BOLTED SIDE-PLATED BEAMS

4.1 Overview
4.2 Failure modes .
4.3 Moment-deformation response ...
4.4 Strength and ductility..
4.5 Moment-curvature response ..
55
55
58
60
63
50
54
4.6 Various design controlling factors of BSP beams .

4.7 Reduction of moment resistance due to partial shear connection and
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
4.10 Longitudinal slip profile ... 85

4.11 Moment contribution by various components .. 88
4.12 Concluding remarks . 92
vi
CHAPTER 5 NON-LINEAR MODEL FOR BOLT GROUPS WITH
93
LINEAR-ELASTIC-PLASTIC BOLTS UNDER COMBINED INPLANE MOMENT AND SHEAR

5.1 Overview
5.2 Non-linear theory of bolt groups
93
95
5.2.1 Basic assumptions ...

5.2.2 Classic elastic theory on bolt groups ...
5.2.3 Non-linear theory on bolt groups
5.3 Implementation of non-linear bolt group theory
95
97
99
102
5.3.1 Description of the methodology .

5.3.2 Steepest descent method for solving the instantaneous centre
of rotation .
102
5.4.2 Simulation by BOGA ..

112
112
CHAPTER 6 NON-LINEAR MODEL FOR BOLT GROUPS WITH
115
NON-LINEAR KINEMATIC-HARDENED BOLTS UNDER

COMBINED IN-PLANE MOMENT AND SHEAR
6.1 Overview
6.2 Non-linear load-slip response of bolt groups .
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
6.4 Numerical simulation by BOGAN . 128

6.4.1 Experimental investigation .. 128
6.4.2 Simulation by BOGAN ... 129
6.4.3 Comparison of the numerical results with the experimental
results 130
6.5 Concluding remarks ... 134
vii
CHAPTER 7 NON-LINEAR ANALYSIS OF BOLTED SIDE-PLATED
135
BEAMS
7.1 Overview
7.2 Formulation
7.2.1 General
135
137
137
7.2.2 Formulation of the non-linear RC beam and steel plate

elements
7.2.3 Formulation of the stiffness matrix of connecting bolt groups
7.3 Assemblage of the structural stiffness matrix
7.4 Implementation of the structural model .
7.5 Experimental verification ...
138
143
145
147
153
7.6 Slip responses of BSP beams .

7.6.1 Numerical model .
7.6.2 Longitudinal and transverse slips
159
159
160
7.6.3 Rotational slips

161
165
CHAPTER 8 TWO-ALPHA APPROACH FOR ANALYSING OF
167
BOLTED SIDE-PLATED BEAMS WITH PARTIAL INTERACTION

8.1 Overview 167
8.2 Two-Alpha approach .. 168
8.3 New parameters quantifying the level of longitudinal and transverse
interaction .
8.3.1 Partial interaction response of BSP beams ..
8.3.2 New parameters quantifying the level of interaction ..
8.3.3 Strain factor .
8.3.3.1 Normalized longitudinal slip profile of BSP beams
8.3.3.2 Mathematical derivation of strain factor ..
169
169
170
173
173
179
8.3.3.3 Strain factor of simple loading case .

8.3.4 Curvature factor ..
181
183
8.3.4.1 Idealized transverse force model ..

8.3.4.2 Transverse displacement compatibility
8.3.4.3 Curvature factor of simple loading cases .
8.4 Modified moment-curvature analysis
8.4.1 Overview .
183
184
186
191
191
8.4.2 Strain profile of section ... 192

8.4.3 Modified section analysis 193
8.4.4 Program implementation . 196
viii
8.4.4.1 Material model .
196
8.4.4.2 Program algorithm ...

8.5 Numerical verification ...
8.6 Design example ..
198
202
207
210
CHAPTER 9 CONCLUSIONS
9.1 Summary
9.2 Conclusions
213
213
215
9.3 Recommendations for future research ...
216
APPENDIX 1 Evaluation of slip using rhombic arrangement of LVDTs 219

APPENDIX 2 Non-linear stress-strain model of concrete .
221
APPENDIX 3 Explicit form of flexural RC beam and steel plate element . 222
APPENDIX 4 Derivation of equations governing longitudinal slip profile
along beam span ...
APPENDIX 5 Longitudinal slip profile of simple cases .
APPENDIX 6 Derivation of curvature factor of simple cases .
REFERENCES .
224
227
233
237
LIST OF PUBLICATIONS ..
243
ix
LIST OF FIGURES
Figure 1.1
Number of buildings aged over 30 years in Hong Kong 1
Figure 1.2
Mei Ho House, Shek Kip Mei Estate . 3
Figure 1.3
Typical arrangement of BSP beams a) Front view, and b)

Cross-sectional view .. 4
Figure 2.1
Full shear connection full interaction analysis ... 24
Figure 2.2
Partial shear connection partial interaction analysis .. 24
Figure 3.1
Test setup a) Schematic diagram, and b) Actual arrangement 29
Figure 3.2
RC section detail of specimens .. 30
Figure 3.3
Bolt-plate arrangements of strengthened specimens .. 32
Figure 3.4
Measured stress-strain relationship of steel plates . 36
Figure 3.5
Dynamic set washer a) Diagrammatic illustration, and b)

Actual arrangement .... 37
Figure 3.6
Load-slip relationship of HAS-E 12 anchor bolts 37
Figure 3.7
Cross-section detail of anchor bolts ... 39
Figure 3.8
Cast-in aluminium tube .. 40
Figure 3.9
Reinforcement cage 40
Figure 3.10 Strain gauges arrangement of specimens a) In the RC beam,

and b) On the steel plates ... 41
Figure 3.11 LVDT and LDT arrangements of specimens . 42
Figure 3.12 Arrangement of rhombic set of LVDTs a) Instrumentation
of a LVDT, b) Schematic diagram, and c) Actual
arrangement 45
Figure 3.13 Reinforcement detail ...... 47
Figure 3.14 Test specimen . 47
Figure 3.15 Test setup .... 49
Figure 3.16 Specimen with loading plates and LVDTs before testing ...... 49
Figure 3.17 Load-slip responses of HAS-E bolts with diameter a) 10mm,
b) 12mm, and c) 16mm .. 51
Figure 3.18 Bi-linear model ....... 52
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
Figure 4.1
Failure modes of specimens a) SBSP and b) SBWP, c)
56
WBSP, and d) WBWP ...

Figure 4.2
Tilting of anchor bolts a) Deformation mode, and b) Shear
57
transfer mechanism through the action of bolts .

Figure 4.3
Outward deflection of the steel plate in Specimen WBSP . 57
Figure 4.4
Moment-deformation responses of all specimens .......... 60
Figure 4.5
Definitions of displacement ductility factor and post-elastic
62
strength enhancement factor .......

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
Figure 4.7
Idealized load-deflection curves of a) concrete-controlled, b)

bolt-controlled and c) plate-controlled BSP beams 71
Figure 4.8
Generic classification of BSP beams according to the

strength of bolts and plates . 73
Figure 4.9
Definitions of strain factor and curvature factor (at the

section subjected to the critical moment) ... 77
Figure 4.10 Variation of strain factor with mid-span deflection of RC

beams .. 80
Figure 4.11 Variation of curvature factor with mid-span deflection of
RC beams ... 82
Figure 4.12 Longitudinal slip responses against mid-span deflection a)
Specimen SBSP, b) Specimen WBSP, c) Specimen WBWP,
and d) Specimen SBWP . 84
Figure 4.13 Longitudinal slip responses along beam spans a) Specimen
SBSP, b) Specimen WBSP, c) Specimen WBWP, and d)
Specimen SBWP 87
Figure 4.14 Moment contributions from various components a)
Figure 5.1
General arrangement of a bolt group .. 97
Figure 5.2
Rigid body movement of a bolt group ... 99
Figure 5.3
Post-yielded response of bolts 101
xi
Figure 5.4
Schematic diagram of the iteration process 105
Figure 5.5
Iterative scheme using steepest descent method 107
Figure 5.6
Detail of the numerical example a) Load-slip relationship of 102

bolt, b) Geometry of bolt group, and c) Applied load
Figure 5.7
Mesh detail of a single bolt 110
Figure 5.8
Mesh for non-linear finite element analysis ... 110
Figure 5.9
Comparison of finite element analysis with the proposed

analysis in a) Force-displacement in x-direction, b) Forcedisplacement in y-direction and c) Moment-rotation . 111
Figure 6.1
Idealized tri-linear load-slip relationship of bolts .. 117
Figure 6.2
Translation rules of yield surfaces in a) Elastic stage, b) First

stage yielding, and c) Second stage yielding .. 122
Figure 6.3
Rigid body movement of a bolt group ... 123
Figure 6.4
Post-yield load-slip relationship of bolts 124
Figure 6.5
Schematic diagram of the computer program BOGAN . 127
Figure 6.6
Experimental setup in Kulak and Crawfords study .. 128
Figure 6.7
Load-slip relationship of bolts adopted in the simulations . 130
Figure 6.8
Comparison of moment rotation curves with experimental

results a) Specimen B1, b) Specimen B4, and c) Specimen
B7 ... 132
Figure 6.9
Load-rotation response of B7 simulated using different load

increments .. 133
Figure 7.1
Structural model of a BSP beam a) Prototype, b) Structural

model, and c) Typical RC beam & steel plate elements 138
Figure 7.2
Sign conventions for a) Flexural elements, and b) RC beam

elements .. 139
Figure 7.3
Layered models of a) Steel plate sections, and b) RC beam

sections ... 141
Figure 7.4
Parameters defining the current state of a bolt ... 143
Figure 7.5
Force interaction between the bolt group, the steel plate

elements and the RC beam elements at the ith node ... 147
Figure 7.6
Stress-strain relationship of concrete adopted in SiBAN ... 139
Figure 7.7
Stress-strain relationship of a) Reinforcement, and b) Steel
xii
plates adopted in SiBAN 148

Figure 7.8
Stress-strain relationship of bolts adopted in SiBAN . 149
Figure 7.9
Solution algorithm of SiBAN . 151
Figure 7.10 Bi-linear load-slip model of anchor bolts ... 153

Figure 7.11 Finite element models used in the computer program SiBAN 154
Figure 7.12 Comparison of mid-span load-deformation responses of
a) Specimen SBSP, b) Specimen WBSP, c) Specimen
WBWP, and d) Specimen SBWP ... 157
Figure 7.13 Comparison of longitudinal slip responses of a) Specimen
Specimen SBWP 158
Figure 7.14 Beam model for the investigation of bolt-slips .. 160
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
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
Figure 8.1
Illustration of slips in BSP beams under gravity load in a)

Longitudinal direction, and b) Transverse direction .. 170
Figure 8.2
Strain profiles of the RC beam and the steel plates with

different degree of longitudinal and transverse interaction 170
Figure 8.3
Illustration of strain factor and curvature factor at the section

with critical moment ... 171
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
Figure 8.5
Notations of a span with n segments .. 175
Figure 8.6
Parameters of longitudinal slip profile governing strain

factor .. 181
Figure 8.7
Idealized transverse shear force model exerted on the RC

beam from the steel plates under arbitrary gravity loads ... 184
Figure 8.8
Transverse compatibility consideration in shear transfer
xiii
zone 185
Figure 8.9
Idealized transverse shear force model exerted on the RC

beam from the steel plates under uniformly-distributed load . 189
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
Specimen SBWP 205
Figure 8.15 Comparison of moment contributed by various actions a)
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
Summary of bolt-plate arrangements of strengthened

specimens 33
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
Displacement compatibility based analysis
FRP
Fibre-reinforced plastic
LDT
Linear displacement transducer
LVDT
Linear variable direct transducer
RC
Reinforced concrete
SBSP
Strong bolt strong plate
SBWP
Strong bolt weak plate
WBSP
Weak bolt strong plate
WBWP
Weak bolt weak plate
xvi
LIST OF NOTATIONS
A
: Area of section
Ac
: Area of concrete
Ap
: Area of steel plates
Ast
: Area of tension reinforcement
Asc
: Area of compression reinforcement
: Strain-displacement matrix
Ba
: Axial component of strain-displacement matrix
Bb
: Bending component of strain-displacement matrix
: Breath of section
: cos
dc
cT
cT
cT
: Displacement vector of the RC beam ( d1 , d 2 K d n )
dic
: Displacement vector of the ith node of the RC beam
dip
: Displacement vector of the ith node of the steel plates
dp
: Depth of steel plates
d sc
: Depth of compression reinforcement
d st
: Depth of tension reinforcement
Ec
: Youngs modulus of concrete
Ep
: Youngs modulus of steel plate
Es
: Youngs modulus of steel reinforcement
ET
: Youngs modulus
EAc
: Axial stiffness of the RC beam
EAp
: Axial stiffness of steel plates
xvii
EI c
: Flexural stiffness of the RC beam
EAp
: Axial stiffness of steel plates
: Length of rigid arm in RC beam elements
: Bolt groups load vector
Fi
: Force vector of bolt groups
Fp , Fp ( x)
: Plate force
Fp , fi
: Plate force at peak moment in full interaction analysis
Fx , Fy , M
: Internal force of bolt groups in x-, y- and rotational directions
f ci
: Stress of the inflection point in the stress-strain model of

concrete
f co
: Peak compressive stress of concrete
f cu
: Compressive cube strength of concrete
f yp
: Yield stress of steel plate
f ys
: Yield stress of steel reinforcement
gi
: Constant
: Net horizontal force of section
hc
: Centroidal level of the RC beam
hna
: Depth of neutral axis
hp
: Centroidal level of steel plates
: Separation of centroidal axes between the RC beam and steel

plates
: Second moment of area of section
I ccy
: Entry of tangential stiffness matrix of non-linear bolt group
xviii
I cs
I ssx
I zz
: Entry of elastic stiffness matrix of bolt group
: Stiffness matrix of flexural element
K bi
: Stiffness matrix of the ith bolt group
K ic
: Stiffness matrix of the ith flexural element of the RC beam
K ip
: Stiffness matrix of the ith flexural element of the steel plates
K cc
K cs
K ss
Kx
Ky
ke
: Elastic stiffness of bolts
km ( x)
: Stiffness of the connecting media (kN/m/m) along span of the

beam
kmi
: Stiffness of the connecting media of section i
kp
: Plastic stiffness of bolts
k p
: Plastic stiffness in first stage yielding
k p
: Plastic stiffness in second stage yielding
kr
: Reduction in bolt stiffness ( ke - k p )
kx
: Bolt stiffness in x-direction
ky
: Bolt stiffness in y-direction
: Length of element
xix
LLVDT
: Length of LVDT device for slip measurement
Lrc
: Distance between reference points on the RC beam
Lrs
: Distance between reference points on the steel plate
Ls
: Length of span
lab
: Instantaneous distance between reference points a and b
lab
: Instantaneous readings of LVDT
: Translation direction
: Translation unitary direction
M ( x)
: Total moment of section
M c ( x)
: Moment of the RC beam
M p ( x)
: Moment of steel plates
Mu
: Ultimate moment
: Unitary normal vector
nc
: Total layers of concrete for section analysis
np
: Total layers of steel plates for section analysis
np
: Total layers of reinforcement for section analysis
: Force point of a bolt
: Point on the bounding surface at which the normal is n
: Bolt force
Pn
: Bolt force along direction of normal vector
Psh
: Total strength of bolts on a shear span
Psh , fsc
: Minimum strength of bolts on a shear span required for full

shear connection
xx
Pt
: Bolt force perpendicular to direction of normal vector
Pu
: Ultimate force of anchor bolts
Px
: Bolt force in x-direction
Py
: Bolt force in y-direction
Pyf
: Yield force of elastic-plastic bolt
Py
: Notional yield force of anchor bolts, Radii of yield surface
Py
: Radii of bounding surface
: Moment about origin induced by bolt force
: Constant
Qx , Qy , Q
: Nodal forces in x-, y- and rotational directions
Rccy
Rcsx
Rcsy
Rssx
Rx
Ry
: Rigid body movement (Slip vector) of bolt group
: First moment of area of section
: sin
sl , sv , s
: Rigid body movement of bolt groups in x-, y- and rotational

directions
sl ( x)
: Longitudinal slip profile along span of the beam
sl ( x)
: Slip strain (first derivative of longitudinal slip) along span of
xxi
the beam
sl ,max
: Maximum longitudinal slip
snorm ( x)
: Normalized longitudinal slip profile along span of the beam
su
: Ultimate slip of anchor bolts
sv ( x)
: Transverse slip profile along span of the beam
sy
: Notional yield slip of anchor bolts
: Transformation matrix for addition of rigid arms
ua , ub
: Element nodal displacement in x-direction of node a and node
b
un , ut
: Deformation of bolts in normal and tangential directions to

yield surface
ux , u y
: Deformation of bolts in x- and y- directions
: Volume
V ( x)
: Shear force profile along span of the beam
va , vb
: Element nodal displacement in x-direction of node a and node
b
vc ,mid
: Mid-span vertical deflection
: External load vector
Wc
: External load vector applied on the RC beam
Wp
: External load vector applied on the steel plates
W ,W ( x)
: Applied load
w1 , w2 , w3
: Weighing factors for Three-point Gaussian integration
: Distance from support
xxii
xi , yi
: Coordinates of the ith bolt
xo , yo
: Coordinates of the instantaneous centre of rotation
: Instantaneous centre of yield surface
: Updated centre of yield surface
: Strain factor
: Curvature factor
: Ratio of flexural stiffness EI p / EI c
: Instantaneous centre of bounding surface
: Updated centre of bounding surface
lc, hp
: Longitudinal displacement of the RC beam at centroidal level

of steel plates
lp,hp
: Longitudinal displacement of RC beam at centroidal level of

steel plates
cv , stz
: Transverse displacement of the RC beam of the shear transfer

zone
vp, stz
: Transverse displacement of the steel plates at the shear

transfer zone
: Strain
axial
: Strain due to axial force
bending
: Strain due to bending
ci
: Strain corresponding to f ci in stress-strain model of concrete
co
: Strain corresponding to peak stress f co in stress-strain model

of concrete
xxiii
hc
: Strain of concrete at centroidal level of steel plates
hp
: Strain of steel plate at centroidal level of steel plates
hslip
: Slip strain at centroidal level of steel plates
: Strain at level y
, (x)
: Curvature
c ( x)
: Curvature of the RC beam
p ( x)
: Curvature of steel plates
it
: The ith trial curvature
: Argument of the unitary normal vector n
: Degree of shear connection
: Angles
a , b
: Element nodal displacement in rotational direction of node a
and node b
: Stress
: Stress of concrete
: Stress of steel plates
sc
: Stress of compression reinforcement
st
: Stress of tension reinforcement
: Maximum transverse shear flow between the RC beam and

steel plates
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
Aged over 40 years
14000
Aged from 30 to 39 years
12000
10000
8000
6000
4000
2000
0
1998
2000
2002
2004
2006
2008
2010
2012
2014
2016
Year
Figure 1.1 Number of buildings aged over 30 years in Hong Kong
One of the trivial solutions to the problem is to demolish these old

buildings. However, huge quantities of demolition wastes are produced in the
process, leading to severe social and environmental problems. Currently there is
an acute shortage of landfill space in Hong Kong. Studies carried out by the
Government show that the public landfill sites will be filled up by 5-6 years and
it is desirable to minimize the waste production. Therefore, instead of

demolishing old buildings, it is better to upgrade and strengthen existing
buildings when their structural performance falls below standard.
Apart from the aging problem of buildings, the problem of vacant

residential and industrial buildings is also getting more serious in Hong Kong.
Due to the migration of industrial activity to the Pearl River Delta, a large
amount of industrial buildings are left vacant nowadays. According to the Royal
Institution of Chartered Surveyors (2009), some 1.07 million sq. m. of
accommodation of industrial use was standing vacant at the end of 2007. And to
utilize such tremendous amount of vacant space, the concept of adaptive re-use
are advocated by professional parties.
Adaptive re-use is the adaption process of existing buildings for

purposes other than their original uses. By changing the functions of buildings,
revival of existing vacant buildings can be achieved, bringing additional value
to them. In European countries and the United States, adaptive re-use is often
used and there are lots of successful cases which redundant industrial and
warehouse
buildings
were
converted
into
residential,
office,
retail,
entertainment and hospitality accommodations. Some famous cases include,

converting the Bankside Power Station in London into the Tate Modern Art
Gallery and the Western Metal Building in San Diego to part of a baseball
stadium. In the adaption process, alternation of the floor plan and rearrangement of the loading paths can be involved. This results in the need to
strengthen existing structures when additional load is taken up by some of the
structural members.
In Hong Kong, it is recognized that heritage bears witness to the

dramatic development of the City from a small village to a global financial hub
and is a valuable asset to the community. To preserve old buildings which
symbolize the development of Hong Kong, the government is taking up the
initiative and a revitalization scheme for preserving historic buildings with new
and innovative use. One of the recent cases in the scheme is the Mei Ho House
of the Shek Kip Mei Estate. Being one of the earliest public housing blocks in
Hong Kong, Mei Ho House symbolizes the beginning of public housing and is
of great heritage value. The building has been evacuated since year 2000 due to
poor living conditions and the government has confirmed to convert Mei Ho
House into an integrated youth hostel and museum in 2009. Different elements,
including restaurant and art gallery will also be included in the re-vitalized Mei
Ho House. During the conversion process, strengthening of structural elements
will be required since the building is dilapidated and designed for residential use
originally.
Figure 1.2 Mei Ho House, Shek Kip Mei Estate
To upgrade and strengthen existing reinforce concrete (RC) beam

structures, external strengthening is often used as it is recognized as one of the
most convenient way to retrofit existing RC beams. Different structural
components, like steel plates, channels or fibre reinforced plastic (FRP) can be
anchored by bolts, or bonded by adhesive glues to the surface of existing beams
in order to provide additional strength.
One of the effective ways to strengthen existing RC beams is by bolting

steel plates to the side faces of beams. The strengthened beam is named as
bolted side-plated (BSP) beams and the detail of the strengthened beam is
illustrated in Figure 1.3. Compared with other strengthening schemes, the use of
BSP beams can preserve the ductility of the beam, preventing the steel
congestion problem near the soffit of the beam and providing space for
installation of propping below the beam during construction process. Despite of
the benefits brought by BSP beams, only limited studies have been carried out.
The understanding to this type of strengthening methods, especially on the
influence due to the mechanical slip of bolts on the structural performance of
beam is insufficient. In light of the lack of understanding of this strengthening
Column/Wall pier
Column/Wall pier
method, a detailed study on the behaviour of BSP beam is carried out.
Bolts
Steel plates
b)
a)
Figure 1.3 Typical arrangement of BSP beams a) Front view, and

b) Cross-sectional view
1.2 OBJECTIVE OF THE STUDY

Concerning the insufficient understanding to BSP beams and their
effectiveness in strengthening existing RC beams, the current study is carried
out. The objectives of this thesis include:
a) To investigate the performance of BSP beams with incomplete interaction
between the RC beam and the strengthening plates and how the bolt-plate
combination adopted for strengthening affects the behaviour of the
strengthened beams
b) To model the response of BSP beams under bending, in particular the slip
between the RC beam and the steel plates and the post-yield strength
enhancement with the presence of incomplete interaction
c) To develop analysis being able to consider the effect of incomplete
interaction into the design of BSP beam structures so as to provide an
accurate and safe prediction to the response of BSP beams
To achieve the above objectives, the following methodology, including all

experimental, numerical and theoretical studies are adopted.
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
1.3 SCOPE OF THESIS

This thesis consists of 9 chapters. This chapter introduces the need of
strengthening, the objective of the current studies and presents an outline on the
remaining chapters of this thesis.
Chapter 2 presents a literature review on different strengthening

techniques for RC beams. Previous studies on BSP beams and the effect of
mechanical slip of bolts on the performance of the BSP beams are also reviewed.
Chapter 3 presents the experimental program of BSP beam tests.

Experimental details including detailed arrangement of specimens, fabrication
processes and instrumentation are included. The arrangement and results of the
accompanied shear test of anchor bolts are also included.
Chapter 4 summarizes the behaviour and responses of BSP beams by

presenting the load-deformation, moment-curvature, strength and ductility
responses of specimens. A categorization for BSP beams according to the boltplate arrangement of specimens is established. The partial interaction response
of BSP beams are investigated by comparing the ultimate strength with the
theoretical capacity with full interaction, presenting the full-range longitudinal
slip responses and slip profiles along the beam span.
Chapter 5 presents the development of a numerical procedure to solve

the non-linear response of bolt groups consisting of elastic-plastic bolts under
in-plane shear. The numerical procedure can be applied to groups of bolts
subjected to in-plane eccentric shear.
Chapter 6 presents the extension of the model developed in Chapter 5 to

consider the general case where a bolt group is consist of non-linear bolts. A
multi-linear kinematic hardening model is developed for the load-deformation
response of bolts under multidirectional shear. The tangential stiffness matrix of
bolt groups consisting of non-linear bolts is developed.
A non-linear finite element model is developed in Chapter 7 to evaluate

the response of BSP beams. The mechanical slips of connector used between
the RC beam and the strengthening plate are modelled using the numerical
model developed in Chapter 6. The numerical model is verified with the
experimental results and further simulations to illustrate the slip response of

BSP beams are carried out.
Chapter 8 presents a new analytical procedure on BSP beams. The

procedure can be applied easily using spreadsheet for quick estimation on the
key parameters in design like the flexural strength, stiffness and internal load
distribution of the beam. It offers a useful tool for structural engineers to
quantify the response of BSP beams.
Finally, conclusions are drawn and recommendations for future research

are given in Chapter 9.
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.
2.2 EXTERNAL STRENGTHENING OF REINFORCED CONCRETE

BEAMS
2.2.1 Strengthened with adhesively bonded steel plates
The most classic way to strengthen RC beams is to attach steel plates to
the tension face of beams with adhesives. The additional steel plates serve as
additional reinforcement of the RC beam. By maximizing the moment arm
between concrete in compression and the strengthening plate, this composite
system is very effective in taking up bending moments. A notable experimental
investigation into the flexural behaviour of plated the strengthened RC beams
was carried out by Jones et al. (1982), in which simply-supported RC beams
strengthened with adhesively bonded steel plates on tension face were tested.
Significant increase in flexural strength, ranging from 62% to 196% was
observed in the tests, showing that the use of adhesively bonded plates is
effective in strengthening existing RC beams. Though the use of adhesively
bonded plate to increase the flexural strength of RC beams is effective, the

ductility of the strengthened beam is greatly reduced when the additional plate
area is relatively large. This is obvious as the steel plates serve as additional
reinforcement and will convert the strengthened beam from an under-reinforced
beam to an over-reinforced beam when the plate area increases.
Other than the dramatic decrease in ductility, the strengthened system is

also prone to premature failure like debonding of steel plates from concrete and
shear peeling failure of concrete from the structural core which is bounded by
the stirrup. Focusing on the plate debonding problem, experimental studies were
carried out by Jones et al. (1988) in which the plate anchorage was strengthened
by bolts or L-shaped anchoring plates extending to the side faces of the beams.
While the flexural capacity is improved as compared with the case when normal
anchorage is used, the debonding of steel plates cannot be prevented.
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.
2.2.2 Strengthened with mechanically anchored steel plates
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.
Experiments were carried out by a number of researchers to investigate the

effectiveness of increasing the flexural capacity of RC beams by bolting steel
components to their tension face. Roberts and Haji-Kazemi (1989) carried out
an experimental study of under-reinforced concrete beams strengthened by
bolting thin steel plates to their tension face. It was observed that both the
flexural stiffness and flexural strength of the strengthened beams were improved.
Foley and Buckhouse (1999) proposed using steel U-channels to strengthen RC
beams and as expected the flexural strength was improved. In their studies, the
problem of pre-mature problem was not observed as the bolts used were
relatively ductile. However, the ductility of the strengthened beam was greatly
reduced.
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
plate arrangements were tested. Most of the specimens showed a significant

increase in strength with a minor loss in the displacement ductility. It appeared
that the bolted side-plated beam is a good choice when ductility is also put into
consideration.
2.2.3 Strengthened with adhesively bonded fibre reinforced plastic plates

Another common way of strengthening RC beams in flexure is to
adhesively bond fibre reinforced plastic (FRP) plates to the faces of RC beams.
The first investigation on the use of FRP plates to strengthen RC beams was in
the 1990s (Meier et al., 1993). The investigation aimed to observe the behaviour
of FRP-strengthened beams under flexure. Many experimental and theoretical
studies were carried out and a detailed summary of these research can be found
elsewhere (Teng et al., 2003). It was generally accepted that FRP is effective in
strengthening because of its high strength to weight ratio which leads to the ease
of handling on sites and a reduction of labour cost.
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.
2.3 CHARACTERISTICS OF RC BEAMS STRENGTHENED WITH

BOLTED SIDE PLATES UNDER FLEXURE
There are only a few researches on flexural behaviour of BSP beams as the
first research on BSP beam was in the mid-1990s. However, when we consider
from a broader perspective, the BSP beams, being formed by attaching external
steel plates to RC beams with bolts, is different, yet similar to standard steelconcrete composite beams, where concrete slabs and steel beams are connected
with the use of shear studs and can be considered as a special form of steelconcrete beam. In both cases, the flexural resistance of the composite beam is
provided by the flexure of the RC beam and steel plate, as well as the composite
action between the RC beam in compression and the steel plate in tension. By
reviewing relevant literature on standard composite beams, one can have a
better understanding on the behaviour of BSP beams.
2.3.1 Shear connection between RC beam and steel plates

The research on flexural behaviour of standard composite beams usually
focused on how the properties of the shear connection between the steel and
concrete components affect different behavioural aspects like the flexural
strength, flexural stiffness and the failure mode of the composite beam. And it
was general accepted that the strength and stiffness of the shear connection
dominates the flexural behaviour of standard composite beams.
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
shear connection is sufficient, the failure of the beam is independent on the

shear connection, but the ultimate strength of concrete or steel. On the contrary,
when the strength of shear connection is insufficient, the failure of the
composite beam can be controlled by one or more components, including
concrete, steel or shear connections.
The degree of shear connection (Yam and Chapman 1968) was

commonly adopted as a parameter to describe the strength of shear connection
between different components. It is defined as the ratio between the actual
strength of shear connection Psh and the minimum strength of shear connection
required for full shear connection Psh,fsc, mathematically expressed as,
Psh
.
Psh , fsc
(2.1)
When the degree of shear connection is greater than 1, full shear

connection exists, meaning that the strength of shear connection is sufficient for
mature failure of the composite beam. On the contrary, partial shear connection
occurs when the degree of shear connection is smaller than 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
Following the experimental study by Chapman and Balakrishnan (1964),

Yam and Chapman (1968) simulated the response of composite beams using the
method of finite difference. Bi-linear model was adopted for concrete and steel,
while a non-linear exponential load-slip relationship was used for shear
connectors. An extensive parametric study was carried out and it was found out
that the behaviour of composite beams with partial shear connection is more
complicated. While the strength of the composite beam is limited by the
strength of the shear connection, the failure mode is controlled by the slip
capacity of shear connectors. Fracture of shear connectors occurred only when
the slip capacity of shear connectors is less than a certain slip demand, which
depended on a number of factors including the degree of shear connection.
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.
Johnson and Molenstra (1991) carried out a detailed investigation on the

behaviour of composite beams with partial shear connection. The slip response
of composite beams with partial shear connection is concerned as in the case of
partial shear connection, the failure mode is controlled by the slip responses of
the beam. Program based on finite difference method, similar to that of Yam
and Chapman (1968) was developed. An extensive parametric study was carried
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.
2.3.2 Shear interaction between RC beam and steel plates

2.3.2.1 Shear interaction in standard composite beams
Other than the strength of the shear connectors, the finite stiffness of the
shear connectors also affects the behaviour of the composite beam. Since the
stiffness of the shear connection is finite, deformation is necessary to mobilize
the shear connection to resist the longitudinal shear force transferred between
RC beam and steel plates. This deformation of the shear connection is often
regarded as interlayer slip and it results in the difference in deformation
between steel and concrete. This phenomenon is known as partial interaction
and affects the behaviour of composite beams
The studies of partial interaction can be traced back to 1950s and a

remarkable investigation, including both experimental and analytical works to
the problem was carried out by Newmark et al. (1951). In the experimental part,
a series of full scale composite beam tests was carried out and incomplete
interactions between the RC slab and the steel beam was observed. The steel
plates slipped inwards towards the mid-span relative to the RC beam, leading to
strain incompatibility between the two components at the critical section.
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
by a second order differential equation with longitudinal slip as the dependent

variable. According to the differential equation, partial interaction in composite
beams depends on a number of geometric properties and mechanical properties
of the RC beam, steel plates and the connecting bolts and is a complicated
global compatibility of the whole beam. The close form solution for simply
supported beam subjected to mid-point load with uniform geometric
arrangement along the whole beam was solved. This close form solution
provides a quick estimate of longitudinal slip response of the beam.
Other than longitudinal partial interaction, transverse partial interaction,

that is the vertical deformation difference between steel and concrete, may
occur as the uplift of the concrete slab from the steel beam is only restricted by
the tension stiffness and strength of shear studs. A number of analytical studies
focused on the transverse partial interaction of composite beams were carried
out (Roberts 1985; Robinson and Naraine, 1988). Same conclusion was drawn
from these studies that the vertical uplift of concrete slab from the steel plate is
only a few percent of the longitudinal slip and can be neglected in most
practical applications.
Oehlers et al. (1997a) investigated the effect of partial interaction on the

flexural strength of standard composite beams analytically. The investigation
was based on consideration of longitudinal force equilibrium and strain profile
at the section with critical moments. It was concluded that, while in most case
the flexural strength is not affected by the partial interaction of the beam, the
flexural strength of composite beams is reduced when the axial strength of the
steel beam is greater than that of the concrete slab.
17
2.3.2.2 Shear interaction in BSP beams

The partial interaction behaviour in BSP beams is more complicated as
partial interaction occurs simultaneously in both longitudinal and transverse
directions. Unlike standard composite beams, where the longitudinal shear flow
is resisted by the shear action of bolts and the transverse shear is transferred by
the bearing between the components, both the transverse and longitudinal shear
forces are resisted by bolts acting in shear. As a result, the steel plate would slip
both upwards in inwards relative to concrete, and results in a reduction in
curvature and strain in the steel plate.
The transverse partial interaction in BSP beams was first demonstrated

experimentally by Ahmed et al. (2000). Simply-supported RC beams,
strengthened by different bolt plate arrangements were tested under four point
bending with monotonically applied load. The strain and curvature reduction
was captured and it was observed that a large portion or even the whole
strengthening plates remained un-yield at the ultimate state. This violated the
rigid plastic assumption that is commonly adopted in the strength estimation of
composite beams, meaning that the strength of the actual beam can be
significantly lower than the strength estimated by the rigid plastic analysis.
To describe the behaviour of transverse partial interaction in BSP beams, a

mathematical model, based on highly simplified assumptions on mechanical
properties like elastic assumption of RC beam and steel plates, and fully-loaded
condition of connecting bolts were developed (Oehlers et al. 1997b). Based on
an idealized internal force distribution, the moment resisted by the components
and parameters like difference in curvature and maximum transverse slip are
18
expressed in closed form solution. A force-based term, the degree of transverse

partial interaction was introduced. Yet no guidelines or studies were carried out
to relate the degree of transverse partial interaction with actual beam
arrangements.
The mathematical model was extended to consider longitudinal and

transverse partial interaction in BSP beams simultaneously (Nguyen et al 2001).
The degree of longitudinal partial interaction (Nguyen et al 1998), previously
developed for standard composite beams was introduced and incorporated into
the model for transverse partial interaction analysis. While this mathematical
model is simple and applicable in practical design, the validity of assuming both
materials being elastic and connecting bolts being plastic is questionable in
general situations where full shear connection exists and plastic hinge forms at
critical section.
2.3.3 Buckling of strengthening steel plates

The flexural behaviour of BSP beams is also controlled by the buckling of
steel plates. When BSP beam is under bending and relatively deep plates are
used, the top part of the steel plates will be in compression. As the steel plate is
only constrained at discrete points, such compression may lead to the problem
of buckling. The problem of plate buckling was observed in the BSP beam tests
(Ahmed et al., 2000) where plates with the same depth as the RC beam were
used. Because of such findings, a comprehensive study on the plate buckling
problem of BSP beams was carried out by Smith et al. (1999a, 1999b, 2000,
2001).
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.
2.4 NUMERICAL STUDIES ON COMPOSITE BEAMS WITH PARTIAL

SHEAR CONNECTION AND INTERACTION
The first analytical study on partial interaction in composite beam
structures, carried out by Newmark et al. (1951) revealed that the problem of
partial interaction is complicated. Even for simple loading cases like mid-point
load and uniformly-distributed loads, the longitudinal slip profile follows a
complicated exponential function and involves solutions of differential
equations. Because of such complexity, much research in the topic of partial
interaction relied on numerical techniques, which provide highly non-linear
solutions and systems involving many governing equations.
Some of the early research works include Arizumi and Hamada (1981) and
Daniels and Crisinel (1993) in which the composite beam is divided into finite
composite beam elements. The composite beam elements are assembled with
concrete elements and steel beam elements, so that the slip between the
20
elements was allowed for. Based on assumed displacement fields which

satisfied displacement and strain continuity, the stiffness matrix can be
formulated. Partial interaction between the components is modelled by elastic
springs connecting nodes of the two basic elements. The incorporation of slip
into the system by considering the structural elements with separate nodes was
developed and illustrated to be an effective way to model the behavior of partial
interaction.
Other research works were also evolved afterwards with different

techniques in finite element method applied. Ayoub (2000) adopted a forcedisplacement mixed-formulation method and the stiffness matrix of the system
is derived based on assumed force field and displacement field. Gattesco (1999),
while adopting the displacement-based formulation, considered the nonlinearity of materials and the connecting media and developed a non-linear
analysis capable of tracing the beam responses up to the peak load. DallAsta
and Zona (2004) used a three field mix formulation. All the displacement, strain
and stress field are assumed and applied in the forumulation of element stiffness
matrix. Faella et al. (2002) and Ranzi et al. (2004) used a new formulation
which includes both slips and displacements as the degrees of freedom and the
exact solution of slip was employed in their formulation.
The method of finite difference was also applied by various researchers to

determine the non-linear partial interaction behaviour in standard composite
beam (Yam and Chapman, 1968; Johnson and Molenstra, 1991). The whole
beam is divided into finite longitudinal segments and the solution is solved by
considering force equilibrium and the compatibility of strain between the
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.
2.5 RIGID PLASTIC ANALYSIS OF COMPOSITE BEAM SECTIONS

For composite beam with the steel and RC components being connected
by mechnically connectors (e.g. shear studs or bolts), the beam may be subject
to partial shear connection and/or partial interaction, and the full-shearconnection full-interaction and partial-shear-connection partial-interaction
analyses (Ahmed et al., 2000) was proposed to estimate the flexural strength
under different bolt arrangements.
22
The full-shear-connection full-interaction analysis can be used when the

degree of shear connection of the beam is larger or equal to unity. Full
interaction is assumed for the beam, meaning that the strain profile of RC core
and steel plates coincides. The analysis procedure is illustrated in Figure 2.1.
Since full interaction is assumed, the procedure is to move the coincident
neutral axes in Figure 2.1a up and down until equilibrium of the stress blocks,
as shown in Figure 2.1b is satisfied. The full-shear-connection full-interaction
flexural strength of the section can then be solved simply by taking moment
about any convenient level, for example, the top fibre of the beam.
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
b) Resultant stress and force block
Figure 2.1 Full shear connection full interaction analysis
Fc
Fp=Psh
Fs
Fc+ Fs =Psh
Resultant stress and force block
Strain profile
Figure 2.2 Partial shear connection partial interaction analysis
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.
2.6 CONCLUDING REMARKS

Among various types of external strengthening techniques, the addition of
side steel plates bolting on the surface is efficient as it can provide additional
strength to the beams while preserving satisfactory level of ductility. With the
use of anchor bolts, the strengthened beam has a guaranteed integrity between
the RC beam and the steel plates as the anchor bolts is installed into the
structural core of the RC beam so that the chance of premature failures like
detachment of the steel plates from the RC beam is prevented.
Previous experimental studies on BSP beams have investigated the general

behaviour of BSP beams and it was illustrated that partial interaction in BSP
beams is unique and significantly affect the behaviour of BSP beams. However,
more experiments on BSP beams should be carried out to deduce the structural
behaviour of BSP beams, in particular those relating to the partial interaction
like the slip between the RC beam and steel plates and the separation of
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.
Currently no well-developed models, either analytical or numerical, are

available to estimate the flexural strength of BSP beams under the effects of
partial interaction and non-linearity of materials as partial interaction is a
complicated global compatibility problem of the whole beams, while the
strength of beam is only considered by the section analysis which is based on
consideration on a sectional level. So far there is no methodology developed to
link partial interaction and section analysis for strength estimation 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.
Considering the uncertainties of the behaviour of BSP beams, an

experimental study was carried out. The aims of this study are to investigate,
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
4) the behavioural characteristics of BSP beams with different bolt-plate

combinations.
Four-point bending tests of 5 simply-supported beam specimens were

carried out. These specimens had the same RC beam geometries, but were
strengthened with different properly selected bolt-plate combinations so that the
behavioural characteristics of BSP beams with different bolt-plate combinations
can be revealed. To measure the slip between the RC beam and steel plates at
different sections, a special device which consisted of LVDTs arranged in a
rhombic shape were designed and used to measure the slip between RC and
steel plates. The device measured the distance between points on RC and steel
plates and the slips between them could then be calculated.
Accompanying the experimental study on beams, a series of bolt shear

tests on HAS-E anchor bolts, which were the type of anchor bolts adopted in
beam tests, was conducted. This series of bolt shear tests aims to, 1) provide
necessary information which includes the ultimate strength and the loaddeformation characteristics of bolts for the selection of bolts in the beam tests.
and, 2) reveal the load-deformation relationship of bolts and develop simple
multi-linear model for representing the behaviour of bolts. The developed
model will be used in the analysis described in Chapters 7 and 8.
3.2 BOLTED SIDE-PLATED BEAM TESTS

3.2.1 Test setup
A four-point bending test set-up was designed, as shown in Figure 3.1. The
beams were simply-supported with a clear span of 3600mm and loaded
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
Specimen to be tested (Detail not shown)

Roller
Roller
Strong ground
1200
1200
1200
Figure 3.1 Test setup a) Schematic diagram, and b) Actual arrangement
3.2.2 Specimen detail

Five specimens were fabricated and tested. The RC details of all
specimens are identical. As shown in Figure 3.2, the RC sections were
225mm350mm deep. The specimens were under-reinforced by 3T16 tension
29
reinforcements, contributing to a steel percentage of 0.76%. Transverse

reinforcement of T10-150 was applied throughout the span of the beam. This
amount of longitudinal and transverse reinforcement would be sufficient to
ensure that the specimens would fail in flexure in both the un-strengthened and
strengthened cases. This is desirable since the objective of this experimental
study is to investigate the flexural responses of beams.
2T10
317
350
30
225
3T16
Figure 3.2 RC section detail of specimens
Specimen NBNP is the control specimen without any strengthening

measures. It serves as a control specimen and is used to demonstrate the
structural performance of a RC beam prior to strengthening. The other
specimens are plate strengthened with four different bolt-plate combinations
including Strong-Bolt Strong-Plate (SBSP), Weak-Bolt Strong-Plate
(WBSP), Weak-Bolt Weak-Plate (WBWP) and Strong-Bolt Weak-Plate
(SBWP). The bolt-plate arrangements, having been identified as the major
parameter affecting the structural behaviour of BSP beams, were properly
selected so that the effect of plastic hinges on all bolt-plate combinations could
be covered.
30
The strengthening details are summarized in Table 3.1 and illustrated in

Figure 3.3. Mild steel plates with two different cross sections, 6mm150mm
deep and 6mm75mm deep, were chosen to be the Strong and Weak plate
arrangements respectively. The steel plates were anchored on both sides of the
RC beam. This is equivalent to 2.3% and 1.1% of the gross sectional area of
concrete. The centroidal axis of the plates was set at 250mm from the top of the
beam and in the tension region of the beam. This arrangement can prevent
buckling of steel plates which is out of the scopes of the present study. When
relatively deep plates are used, the top parts of the steel plates would subject to
compression and buckle when the compression exceeds their buckling
resistance. By adopting the current plate arrangement, the entire steel plate was
below the mid-depth of the section and therefore the plate compression can be
minimized and hence plate buckling can be avoided.
In this study, the degree of shear connection is adopted to consider

whether a bolt arrangement is Strong or Weak. The degree of shear
connection is defined mathematically as
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.
According to Equation (3.1), the plate force at ultimate limit state is

required to find out the degree of shear connection. Therefore, full interaction
section analysis was carried out. The analysis adopted non-linear stress-strain
relationship, as described in Appendix 3.2, for concrete and elastic-plastic
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
Unit WBWP and Unit WBSP
75
Unit SBWP
Anchor bolt
Steel plate
All plates are 6mm thick
Figure 3.3 Bolt-plate arrangements of strengthened specimens
32
Table 3.1 Summary of bolt-plate arrangements of strengthened specimens
No. of bolts on shear span

Strength of bolts on shear span Pb
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)
#value refers to the plate on each face of the beam, dp.=deep.
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.
3.2.3 Material properties

A concrete mix containing maximum coarse aggregates of 10mm and with
target mean cube strength of 30 MPa at 28 days after casting was designed for
the specimens. The constituents and the corresponding proportions of the
concrete mix are shown in Table 3.2. For each specimen, three concrete cubes
with dimensions 150mm150mm150mm were cast and compressive tests
were carried out on the test days to obtain the compressive strength of cubes.
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
Table 3.3 Cube compressive strength of concrete

Concrete Cube Compressive Strength (MPa)
Specimen
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
High yield steel reinforcement with characteristic yield strength of

460MPa was used in this study. 16 bars were used as tension reinforcement
while 10 reinforcing bars were used as hangers. Three samples were taken
from each type of reinforcements. Tensile tests were carried out and the yield
strength and Youngs modulus of these samples are summarized in Tables 3.4
and 3.5.
Table 3.4 Material properties of T-16 reinforcements

Yield Strength (MPa)
Youngs Modulus (GPa)
545
189
525
184
540
188
Average
537
187
34
Table 3.5 Material properties of T-10 reinforcements

Yield Strength (MPa)
Youngs Modulus (GPa)
550
185
540
194
Average
545
190
Three strips of dimension 400mm6mm thick were taken from each of

75mm deep and 150mm deep steel plates for tensile tests. The material
properties of steel plates are listed in Table 3.6.
Table 3.6 Material properties of strengthening steel plates

75mm plate
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
HAS-E 12 anchor bolts can be idealized by a bi-linear relationship presented

in Figure 3.6.
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
Figure 3.4 Measured stress-strain relationship of steel plates
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
Figure 3.6 Load-slip relationship of HAS-E 12 anchor bolts
37
3.2.4 Specimen fabrication

The specimens were fabricated in two stages, the RC fabrication stage, in
which the RC component was fabricated and cast, and the strengthening stage,
in which the bolt-plate strengthening system was installed. The steps involved
in the two stages are described in herein.
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
Dynamic set washer
HAS-E anchor rod
HAS-E anchor rod
Aluminium tube
Adhesive RE-500
Figure 3.7 Cross-section detail of anchor bolts
39
Figure 3.8 Cast-in aluminium tube
Figure 3.9 Reinforcement cage
3.2.5 Loading history

In each test, the specimen was first incrementally loaded to 50% of the
theoretical ultimate capacity of the beam as estimated by a full interaction
sectional analysis. After that, the loading process was changed to a
displacement-controlled mode in which incremental displacements were applied.
The loading process was terminated at the instant when the post-peak loading in
the beam was 85% of the peak load, or any of the two bolts in the specimen
were fractured.
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
such as cracking of concrete and buckling of compression reinforcement if only

one set of measurements is used. With the use of three sets of measurements,
the behaviour of the zone can be better represented.
3.2.6.2 LVDT and LDT arrangements

Linear Variable Direct Transducers (LVDTs) and Linear Displacement
Transducers (LDTs) were installed, as illustrated in Figure 3.11. The LVDTs
other than those in the constant moment zone were installed and located along
the centreline of the section. At the constant moment zone, installation along the
centreline of section is not possible, as the space was block by the transfer beam.
In this case, pairs of LDTs, as shown in Figure 3.11, were installed and the
average of the readings was used to represent the vertical displacement of the
beam section.
600
800
400
400
800
600
Linear strain gauges

Section B-B, C-C and D-D
a)
600
Section A-A and E-E
800
400
400
800
50
50
150
600
75
Specimens SBSP and WBSP

600
800
400
400
600
32.5 37.5
32.5
75
b)
800
Specimens WBWP and SBWP

Linear strain gauges
Figure 3.10 Strain gauges arrangement of specimens a) In the RC

beam and b) On the steel plates
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
centroidal level of steel plate
R2, R4
R1, R3
a
Specimens SBSP,
WBSP and WBWP
200
800
Specimen SBWP
175
850
Rhombic arrangement of LVDTs
Figure 3.11 LVDT and LDT arrangements of specimens
3.2.6.3 Rhombic arrangements of LVDTs

One of the primary objectives of this experimental study is to measure the
non-linear slip responses between the RC beam and the steel plates. The major
difficulty in measuring the slip is that the surface of the RC beam is covered by
the steel plate at the locations where the slips are of interest. Furthermore, slips
can be in all directions: longitudinal, transverse and rotational. To overcome the
above difficulties, a new rhombic arrangement of Linear Variable Direct
Transducers (LVDTs), which is able to measure the slips between two target
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).
Fig. 3.12(a) shows the instrumentation of a LVDT to measure the distance

between two reference points. As the reference points (A, B, C and D) moved
relative to each other, a change in the length of the extensible rod would be
recorded by the LVDTs installed adjacent to it. With hinges being installed, any
rotation of the rod could be accommodated.
The rhombic arrangement of LVDTs is illustrated in Figure 3.12(b). Four

reference points, two on the RC beam and the other two on the steel plate, were
used to define the transverse target planes AD of the RC beam and EF of the
steel plate. The reference points of the target plane AD of the RC beam were
located above and below the steel plate so that the reference points were not
covered up by the steel plate.
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
Rhombic arrangements of LVDTs were installed on the specimens to

measure the slip responses at various locations, as illustrated in Figure 3.11. The
same rhombic arrangements of LVDTs were installed at the back face of the
beams. The locations of LVDTs for Specimen SBWP were slightly different
from the other specimens in order to prevent the arrangement from clashing
with the screw heads.
44
To data logger
Hinge with beads inside

(Connected to RC anchorage)
LVDT
Extendible rod
Hinge with beads inside

(Connected to steel plate anchorage)
(a)
Concrete beam
A
E
B
Plate
C
F
Metal
strip
D
(b)
Figure 3.12 Arrangement of rhombic set of LVDTs a) Instrumentation of

a LVDT, b) Schematic diagram, and c) Actual arrangement
45
3.3 SHEAR TEST OF HAS-E ANCHOR RODS

3.3.1 Test program
A total of 6 specimens were cast to investigate the load-slip responses of
bolts with diameters ranging from 10mm to 16mm. The bolts are installed on
the side faces of concrete blocks with target cube compressive strength of
30MPa. 6mm thick plates were used to exert shear loading on the bolts. The
experimental scheme is summarized in Table 3.7 and the test specimen is shown
in Figure 3.14.
Table 3.7 Experimental scheme of bolt shear test

No. of specimens
Bolt
diameter
10
12
16
3.3.2 Specimen Design and Fabrication

The concrete blocks, where the bolts were installed, were cast in wooden
formworks. The concrete blocks for each bolt size were cast at the same time
(i.e. 3 specimens each time) and 3 cubes were cast and tested to obtain the
strength of concrete. The cube strength of specimens on the date of test was
summarized in Table 3.8.
46
Figure 3.14 Test specimen
Figure 3.13 Reinforcement detail
Table 3.8 Cube compressive strength of concrete in bolt shear tests

Concrete cube compressive strength (MPa)
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 reinforcement detail of specimens is shown in Figure 3.13. The

reinforcement was arranged such that the directions of deformation of bolts
were the same as the case when the bolts were subjected to longitudinal slip,
which is the dominant direction of slip, in the beam specimens. And the stirrup
separation was also the same as that of the beam specimens to simulate the
actual cases in this experimental study.
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.
3.3.3 Test setup

Compression shear tests were carried out as illustrated in Figure 3.15 and
3.16. When the platform connected to the hydraulic jack moves upwards, the
reaction force would be exerted on the loading plate and evenly distributed to
the two steel plates, which would subsequently load the bolts to be tested in
shear. The shear force on bolts was balanced by the force applied from the
moving platform to the concrete block
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
top of plates were laterally restrained by stiffeners in order to increase the

buckling resistance.
Firm Plate for reaction
Firm Plate for reaction

Loading plate
Bolts being
tested
Concrete block
Movable platform
connected to hydraulic jack
Movable platform
connected to hydraulic jack
Reference point of LVDT

Displacement to be measured
Figure 3.15 Test setup
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.
3.3.5 Result analysis

The load-slip responses of these specimens were plotted in Figure 3.17.
The deformation response of a pair of bolts was equal to the average readings of
the 4 installed LVDTs. According to Figure 3.17, it could be seen that the loadslip relationship was non-linear. When the initial load was applied, the
tangential stiffness of bolts was large and this was due to the existence of
frictional resistance between the steel plates and concrete block. As further
loads were applied, an approximately linear region was observed, after which
the bolt stiffness dropped gradually. Instead of reaching a perfectly plastic
plateau, a hardening plateau was observed when the applied load approached to
the peak value. The descending branch of the bolts was short and the bolt force
dropped rapidly until bolt fractures occurred.
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
Figure 3.18 Bi-linear model

Table 3.9 Values for bi-linear model of HAS-E bolts
Bolt
diameter
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

The detail of the experimental study on BSP beams was reported in this
chapter. This study focused on the aspects related to the partial shear interaction
behaviour of BSP beams like the strength reduction due to the existence of
partial shear interaction and the non-linear slip responses of bolts along the
beam. To cover different behavioural characteristics of BSP beams due to
partial shear interaction, the bolt-plate arrangements were properly selected so
that all different bolt-plate combinations were included. To measure the slip
response between RC and steel plates properly, a new set of rhombic
arrangement of LVDTs was designed and adopted in this study in order to
measure the slip responses between RC and steel plate.
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.
4.2 FAILURE MODES

Both Specimens SBSP and SBWP failed in concrete crushing similar to
the control specimen, as shown in Figure 4.1. The failure of Specimen SBSP is
associated with concrete crushing prior to yielding of steel plates whereas that
of Specimen SBWP is due to excessive deformation of concrete after yielding
of both steel plates and reinforcement in the beam. These failure modes are
similar to the cases of over- and under-reinforced concrete beams respectively.
55
(a)
(b)
(c)
(d)
Figure 4.1 Failure modes of specimens a) SBSP and b) SBWP, c) WBSP,

and d) WBWP
Bolt-fracture was observed in Specimen WBWP and Specimen WBSP
during the tests, as shown in Figure 4.1. The failure processes of Specimen
WBWP and Specimen WBSP were similar. As the applied load increased,
significant tilting of bolts towards the central of beam was observed (see Figure
4.2a). This is due to an extra moment being induced due to eccentricity of shear
exerted by steel plates and RC, as illustrated in Figure 4.2b. The extra moment
induced increases the demand of bolts, which may become critical if partial
shear connection is adopted. Finally, when the combined moment and shear on
bolts reached the capacity of the bolt, sudden fracture occurred which caused
the failure of the specimen.
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.
Motion of steel plate
Bolt rod
Steel plates
Bolt shaft
Reinforced
concrete
Tilting
Stress exerted on bolts
Resultant force exerted on bolts
(a)
(b)
Figure 4.2 Tilting of anchor bolts a) Deformation mode, and b) Shear

transfer mechanism through the action of bolts
Figure 4.3 Outward deflection of the steel plate in Specimen WBSP
57
4.3 MOMENT-DEFORMATION RESPONSE

The total moment of the constant moment zone of different specimens, as
calculated by considering equilibrium of the beam, are plotted against the
corresponding mid-span deflections, as shown in Figure 4.4. For Specimen
SBSP, which is with strong plate and strong bolt arrangements, the initial
deformation response of the beam was elastic. The bottom reinforcement
yielded when the maximum moment reached 154kNm, equal to 95% of the
peak moment (Mu) and the tangential stiffness of the beam dropped abruptly to
15% of the stiffness prior to yielding of the bottom reinforcement. The
remaining stiffness was provided by both the axial interaction between concrete
in compression and steel plate in tension and the flexure of steel plates. The
stiffness remained unchanged up to the ultimate state (M=161kNm). After that,
concrete crushed and the beam degraded gradually when further deformation
was applied and the beam failed in concrete crushing when the mid-span
deflection reached 55mm.
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
Mid-span deflection (mm)
Figure 4.4 Moment-deformation responses of specimens
4.4 STRENGTH AND DUCTILITY

Compared with the control specimen (Specimen NBNP), the strengthened
specimens show various percentage of strengthening ranging from 32% to 59%.
A summary of the strengthening performance of the specimens is given in Table
4.1. Among the four strengthened beams, Specimen SBSP, being strengthened
with Strong bolt and Strong plate, was the strongest with 59% of additional
strength over the control beam. Specimen WBSP, with the same Strong plate
arrangement as Specimen SBSP, but Weak bolts, achieved a 47% increase in
strength. Even with half of the plate size as the Strong plate case, Specimen
SBWP achieved a 43% increase in strength, which is comparable to Specimen
SBSP. Specimen WBWP was the weakest among the four strengthened beams
and achieved a 32% strength increase. This is much lower than that of
Specimens WBSP and SBWP, which have the same bolt and plate arrangements
respectively.
60
The present experimental results indicate that the load-deformation

responses of the specimens, as shown in Figure 4.4, can be idealized by a bilinear curve (see Figure 4.5). The displacement ductility factor , which is
defined as the ratio between the displacement at peak load u and the notional
yield displacement y, was adopted to measure the ductility performance of the
plated beams. As the strengthened beams considered are predominantly
subjected to gravity loads, the strengthened beam would collapse when the
applied gravity load is higher than the load-carrying capacity of the beam. The
post-peak descending branch of the load-deflection curve would alter the rate of
failure but could not avoid collapse of the beam. Because of that, the
displacement at peak load u was chosen to represent the ultimate deformation
of the BSP beams. As shown in Figure 4.5, the notional yield displacement y is
defined as the intersection of the two straight lines associated with the loaddeflection curves at the elastic and post-elastic stages respectively. The
displacement ductility factors of specimens were calculated using the above
definitions and the results are given in Table 4.2. The four strengthened
specimens have displacement ductility, ranging from 1.70 (Specimen SBSP) to
2.57 (Specimen WBWP). The displacement ductility of all specimens were
lower than the un-strengthened specimen as expected since the bolt-plate system
acts as an additional reinforcement to the specimen and hence would reduce the
displacement ductility of the BSP specimens.
Substantial post-elastic strength enhancement could be found in Figure 4.4.

The degree of post-elastic strength enhancement is paramount important for a
gravity load-resisting structure as the excessive deformation under additional
loading in the post-elastic stage can provides noticeable warning of impending
61
failure of the structure to occupants for evacuation. The post-elastic strength

enhancement factor is defined as the ratio between the peak strength Pu and
the yield strength Py, see Figure 4.5. The post-elastic strength enhancement
factors of the specimens are tabulated in Table 4.5. It can be observed that the
post-elastic strength enhancement factor of Specimen SBWP was 1.17 which is
substantially higher than the corresponding values of other specimens, ranging
from 1.05 to 1.08. From the displacement ductility factor and post-elastic
strengthening factor, it is found that combining the use of Strong bolts with
limited amount of steel plates can effectively enhance the post-elastic strength
of beam with a satisfactory level of ductility, meaning that Specimen SBWP is
the safer strengthening system as the post-elastic branch provides a warning
signal to occupants if the structure is overloaded prior to collapse.
P Displacement ductility factor = u/y

Post-elastic strength enhancement factor = Pu/Py
Pu
Py
0.75Py
Actual behaviour
Bi-linear approx.
Figure 4.5 Definitions of displacement ductility factor and post-elastic strength

enhancement factor
62
Table 4.1 Failure modes and the increases in strength of BSP beams
Failure mode
Mu (kNm)
Increase in strength (%)
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
Table 4.2 Displacement ductility factors and post-elastic strength enhancement

factors
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
= displacement ductility, = post-elastic strength enhancement factor
4.5 MOMENT-CURVATURE RESPONSE

The moment-curvature relationship of the constant moment zones was
captured in order to study the flexural responses of BSP beams. As partial
interaction existed, the curvatures of the RC beam and the steel plates were
different and two different measuring systems were used for the RC beam and
steel plates accordingly. The average curvature of RC beam was measured by
three sets of LVDTs installed at the two loading points and mid-span (see
Figure 3.10) so that the abrupt changes in curvatures due to localized cracks
would be minimized. As the theoretical curvature profile of the constant
moment zone is constant, the average curvature could be estimated from the
deformations of the three LVDT readings. On the other hand, the curvature of
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.
For Specimen WBSP, the moment-curvature relationship of the RC beam

was elastic until the moment and curvature of the RC beam reached 131kNm
and 1410-3 m-1 respectively. Plastic hinges gradually formed in the RC beam
until the beam failed when its curvature reached 3310-3 m-1. Curvature of the
steel plates was almost the same as the RC beam prior to the formation of
plastic hinges, and was slightly larger than that of the beam as further loads
were added. The curvature in the steel plates stopped to increase with the
curvature of the beam since reaching 2310-3 m-1 and it remained constant until
failure.
For Specimen WBWP, the moment-curvature relationship of both the RC

beam and steel plates in Specimen WBWP were similar to Specimen WBSP.
Both curvatures of the RC beam and the steel plates responded identically and
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.
For Specimen SBWP, the moment-curvature relationship of the RC beam

was elastic until the moment reached 129kNm. Plastic hinges formed afterwards
and the beam failed when the curvature of the RC beam was 6110-3m-1. This
value was much larger than the other specimens, showing that the ductility of
the SBWP arrangement can result in a better ductility when the plastic hinge is
considered. The curvature response of the steel plates was about 10% smaller
than that of the RC beam in the elastic stage. As the RC beam became inelastic
due to the yielding of tension reinforcement, the curvatures in the steel plates
dropped gradually from 1410-3m-1 to 610-3m-1 at the ultimate limit when the
peak moment is reached. Then the curvature in the steel plates slightly increased
until failure was reached.
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
140
120
Moment (kNm)
100
80
60
40
RC
20
Steel plate
0
0
(c)
10
15
20
25
30
180
160
Moment (kNm)
140
120
100
80
60
RC
40
Steel plate
20
0
(d)
10
20
30
40
50
60
70
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
4.6 VARIOUS DESIGN CONTROLLING FACTORS OF BSP BEAMS

From the results mentioned in the previous sections, it is observed that the
behaviour of BSP beams could be significantly affected by using different
strengthening schemes. The strength and ductility of BSP beams could be
controlled by concrete components, anchor bolts or steel plates. A new generic
classification of BSP beams is herein established according to the controlling
factors and three groups of beams, namely concrete-controlled, bolt-controlled
and plate-controlled are identified. The structural behaviour of these three
groups of BSP beams are discussed herein.
Case 1: Concrete-controlled BSP beams

In a concrete-controlled case, the structural behaviour of the strengthened
beam is controlled by concrete in compression. The typical load-deflection of a
concrete-controlled case, such as that of Specimen SBSP, is shown in Figure
4.7a. The tension reinforcement yields at the ended of the elastic stage (Segment
OA). The stiffness of the beam drops abruptly after yielding of the tension
reinforcement (Point A) but the beam is still able to take up further loads as
resisted by the axial force couple between the RC beam and the steel plate
(Segment AC). At the ultimate state (Point C), although the steel plate is still
able to provide additional tensile force, concrete is unable to provide the
required compression to give additional moment resistance. So concrete crushes
and leads to brittle failure of the beam (Segment CE).
Case 2: Bolt-controlled BSP beams

Figure 4.7b illustrates the idealized load-deflection curve of a boltcontrolled beam like that of Specimen WBSP or Specimen WBWP. The beam
68
responds elastically (Segment OA) until the yielding of the tension

reinforcement in the RC beam (Point A). When further loads are applied, the
loads are taken up by the axial force couple similar to that in the concrete
controlled case (Segment AC). However, as the stiffness of the bolts
significantly drops, the interaction between the RC beam and steel plates
decreases and results in a gradual drop in the stiffness of the beam (Segment
AC). At point C, the average stiffness of the bolts reaches zero so that the force
induced in the plates goes to its maximum and the beam becomes plastic.
Further deformation applied to the bolt-plate system is taken up by the bolts
alone and the bolts fail when the ultimate deformation is reached (Point D). The
failure path of bolt-controlled beams depends on the post-peak responses of the
bolts, which can be gradual or brittle, as indicated by the path DE1 or DE2
respectively.
Case 3: Plate-controlled BSP beams

The load-deflection response of plate-controlled beams, similar to that of
Specimen SBWP, is illustrated in Figure 4.7c. Similar to the concrete-controlled
case, an elastic stage (Segment OA), followed by an abrupt drop in stiffness
(Point A) and a further linear stage (Segment AB) is present. Instead of concrete
crushing, the steel plate yielded first. The yielding process of the steel plates
occurs gradually as the strain within the plate is non-uniform, and begins from
the bottom fibre to the top fibre when subjected to a sagging moment. This
results in a gradual drop in stiffness (Segment BC). The beam becomes fully
plastic afterwards (Segment CD) and the concrete starts crushing due to
excessive deformation at point D. Concrete crushing then continues until failure
occurred at point E.
69
Among the three cases, the plate-controlled type is recommended as the

elastic and post-elastic strengths can be enhanced, and sufficient ductility can be
maintained. To achieve this, the amount of plate added should be properly
selected and bolts with adequate strength are required. The bolt-controlled case
can also be used in practices if the anchor bolts have a very high plastic
deformation capacity. However, most of the anchor bolts can only provide
limited ductility. The concrete-controlled case results in brittle failures and
should be avoided under any circumstance.
70
(a)
P
C
A
(b)
P
D
C
A
E1
E2
(c)
P
B
Figure 4.7 Idealized load-deflection curves of a) concrete-controlled,

b) bolt-controlled and c) plate-controlled BSP beams
71
Whether BSP beam is controlled by concrete, bolt or plate failure is

actually dependent on the strength of the bolts and the steel plates. The
relationship between the strength of bolts, strength of steel plates and the
corresponding beam classification is illustrated in Figure 4.8. If the strength
provided by bolts and plates is higher than the compressive capacity of the
concrete stress block in the RC beam, the beam is controlled by the concrete.
The failure behaviour of the beam is brittle and is similar to an over-reinforced
concrete beam. If either bolts or steel plates added is of insufficient strength, the
strengthened beam is controlled by the corresponding weaker element (see the
corresponding regions in Figure 4.8). In such cases, the beams would fail in a
more ductile manner which would provide noticeable warning of impending
failure of the structure. The intersection of the three broken lines in Figure 4.8,
corresponds to the balanced failure case and indicates that both bolts, steel
plates and concrete reaches their strength simultaneously. In this ideal case,
crushing of concrete, fracture of bolts and full yielding of steel plates will occur
simultaneously at the ultimate loading stage.
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
Figure 4.8 Generic classification of BSP beams according to the strength

of bolts and plates
4.7 REDUCTION OF MOMENT RESISTANCE DUE TO PARTIAL

SHEAR CONNECTION AND INTERACTION
The moment resistance of any mechanically-connected BSP beam is lower
than that predicted from full-shear-connection full-interaction analysis due to
the existence of partial interaction and/or partial shear connection. All
specimens are affected by the effect of partial interaction, while both full-shearconnection and partial-shear-connection cases are covered in this study. Serving
as a reference, full-shear-connection full-interaction analysis was carried out, as
described in Section 3.2.2, so that the theoretical increase under a fullinteraction full-shear-connection assumption could be solved. The additional
moment Mu provided by various bolt-plate systems are listed in Table 4.3 and
compared with the theoretical additional moment obtained in the full-shearconnection full-interaction analysis Mu,fi. The specimens showed 25% to 51%
reduction in additional moment resistance as compared with the full-shearconnection full-interaction case.
73
Specimens WBSP and WBWP are subject to greater additional moment

reductions of 51% and 44% respectively. This is because both partial shear
connection (insufficient shear strength of the connecting media) and partial
interaction (slip of the connecting media) existed, as the degrees of shear
connections of the two specimens are less than one (see Table 4.3).
Full shear connection existed in Specimens SBSP and SBWP, as the

degrees of shear connection are 1.00 and 1.26, respectively. Therefore, the
reduction in additional moment resistance was solely due to the effect of partial
shear interaction. According to Table 4.3, the reductions in additional moment
resistance for these two specimens are 38% and 25%, respectively. Although
full shear connections were provided in both cases, the reductions in additional
moment resistance are still significant and are much higher than the 15%
reduction, as previously suggested (Oehlers et al., 2000b) for incorporating the
effects of partial interaction. Apparently, this suggestion cannot fully account
for the effect of partial shear interaction on the reduction of moment resistance
of BSP beams. More studies on the strength reduction in BSP beams are
therefore recommended.
74
Table 4.3 Strengthening performance of BSP beams
NBNP SBSP WBSP WBWP SBWP

Degree of shear connection
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
Reduction in the increase in moment

resistance (%)
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
4.8 PARTIAL INTERACTION RESPONSE AT CRITICAL SECTION

In this study, the partial interaction at the peak moment section is
quantified by two factors, the strain factor and the curvature factor of the critical
section (section with the largest moment), as shown in Figure 4.9. These two
factors quantifies the ratio of cross-sectional deformation between the RC beam
and the steel plates and are appropriate for describing the partial interaction
responses,
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.
Strain profile of plate
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)
4.8.1 Variation of strain factor of the critical section

The longitudinal strains of the steel plate at its centroidal level ( hpp ) of the
critical section were measured directly by the strain gauges attached at the
centroidal level of plate hp. Assuming plane sections remaining plane after
bending, the longitudinal strains of RC at the centroidal level of plate ( hcp )
could be expressed mathematically as,
c
c
hc = top
gauge + c ( h p d top gauge ) ,
(4.5)
c
where top
gauge is the average of the strain gauge readings from the top
reinforcement in the constant moment zone, c is the average curvature of

concrete in the constant moment zone as estimated by LVDTs, hp is the depth
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.
Prior to yielding of the bottom reinforcement, a gradual descending trend

for the strain factor of all specimens is observed. The gradual drop in strain
factor is due to the fact that instead of the assumed bi-linear relationship, the
stiffness of bolts and hence the stiffness of the connecting media reduces
gradually once loading has been applied. This results in a drop in both the strain
factor of the critical section and the longitudinal partial interaction of the beam.
As shown in Figure 4.10, the strain factor decreased more significantly as

a plastic hinge began to form in the critical section of the RC beam. According
to Figure 4.10, the strain factor at the critical section of specimens at the
ultimate limit state was reduced to between 44% and 53% of the value when the
bottom reinforcement yielded. This suggests that other than the gradual drop in
bolt stiffness, the formation of plastic hinges also reduces the longitudinal
partial interaction of the beam. When a plastic hinge forms in the critical section
of the RC beam, a large curvature is accumulated in the plastic hinge, as the
beam is subject to further deformation due to the low flexural stiffness of the
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.
The variation of strain factor in specimens can be explained if the relative

stiffness between the connecting media and the steel plate in the axial direction
is considered. As expected, using more bolts has a positive effect on the
longitudinal partial interaction of BSP beams. By comparing Specimens SBWP
and WBWP, as well as Specimens WBSP and SBSP, it can be seen that the
strain factor increases with the number of bolts used. When more bolts are used,
the stiffness of the connecting media is increased, implying that a smaller slip of
bolts is required to transfer axial force to the steel plates. Thus the slip strain, as
well as the denominator of the expressions for strain factor in Equation (4.3),
will be decreased, resulting in a larger strain factor.
79
Meanwhile, from comparing the strain factors of Specimens WBSP and

WBWP, it can be observed that the strain factor decreases as the axial stiffness
of the steel plates increases. When steel plates of higher axial stiffness are used,
a greater axial force will be induced in the plates and a larger bolt slip is needed
to transfer the larger axial force to the steel plates. As a result, a larger slip
strain will be present at the section considered and the strain factor will be
increased accordingly from Equation (4.3).
From the above findings, it can be concluded that longitudinal partial

interaction of a BSP beam is dependent on the relative stiffness between the
connecting media. The partial interaction increases when the relative stiffness
between the connecting media and the steel plate increases. A proper selection
of bolt-plate arrangements is of utmost importance for the design of BSP beams.
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
Figure 4.10 Variations of strain factor with mid-span deflection of the

RC beam
80
4.8.2 Variation of curvature factor of the critical section

The curvatures of the RC beam and steel plates were determined from the
LVDTs and the strain gauge readings respectively. The curvature factor was
solved using Equation (4.4). Figure 4.11 shows the variation of curvature factor
against the curvature of RC in the critical moment zone. As shown in Figure
4.11, the curvature factors were larger than 0.9 in all specimens prior to the
yielding of the bottom reinforcement. This means that the difference in
curvatures between the RC beam and the steel plates at the critical section were
small in all specimens and that there was a strong transverse interaction between
RC and steel plates in the elastic stage. Also, unlike the variation of strain factor
in the elastic stage, the curvature remained almost constant throughout the
elastic stage. This suggests that the transverse partial interaction response is less
sensitive to the gradual drop in bolt stiffness, as compared with the longitudinal
partial interaction response.
After the yielding of the bottom reinforcement, as marked in Figure 4.11,

the curvature factors began to drop gradually. In the ultimate state, the curvature
factors at the critical section of the specimens range from 0.2 to 0.7, which is
significant. The reduction in curvature can be explained by the formation of
plastic hinges at the critical section. When a plastic hinge forms in the RC beam,
the flexural stiffness of the RC beam in the plastic hinge zone drops to almost
zero and thus the curvature of the RC beam increases dramatically when being
deformed further. On the contrary, due to the reduction of strain in steel plate, a
plastic hinge is unlikely to form in the steel plates, and thus the curvature in the
steel plates will not increase as quickly as that of the RC beam. Both effects,
together, resulted in the decreasing trend of curvature factor.
81
For Specimen WBSP, the measured curvature factor was significantly

larger than 1 when the mid-span deflection is between 20mm and 35mm, as
shown in Figure 4.11. This is due to the inadequacy of the arrangement of
measuring devices. During testing of Specimen WBSP, a large crack was
formed under one of the loading point, resulting in a local curvature
accumulation. This effect was unable to be captured by the strain gauges located
600mm away from the supports (see Figure 3.9). Hence the measured average
curvature in the RC beam is lower the curvature in the steel plates. It is believed
that the average curvature of the steel plates should be smaller than that of the
RC beam and a curvature factor smaller than 1 should have been obtained if a
better arrangement of measuring device is used.
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
Figure 4.11 Variations of curvature factor with mid-span deflection of the

RC beam
82
4.9 LONGITUDINAL SLIP RESPONSE

The full range longitudinal slip responses of strengthened specimens were
captured by the rhombic arrangement LVDTs and are plotted in Figure 4.12.
The responses between Strong bolt (Specimen SBSP and Specimen SBWP)
and Weak bolt (Specimen WBSP and Specimen WBWP) arranged specimens
are significantly different. For the Strong bolt arranged specimens, the slips
were very small (<0.2mm) when initial loading was applied as most of the force
transferred to the steel plates was taken up by the friction between the RC beam
and the steel plates. The bolts were mobilized as the applied load increased and
the relationship between the longitudinal slips and the mid-span deflections was
approximately linear at various locations along beam span until the mid-span
deflections reached the deflections corresponding to the peak load (i.e., 35mm
and 44mm for Specimens SBSP and SBWP respectively). In the post-peak stage,
the slips increased more slowly than in the pre-peak stage until failure occurred.
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
Longitudinal slip (mm)
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
Figure 4.12 Longitudinal slip responses against mid-span deflection a)

Specimen SBSP, b) Specimen WBSP, c) Specimen WBWP, and d)
Specimen SBWP
84
4.10 LONGITUDINAL SLIP PROFILE

Bolt slips at the specified points on the shear span have been measured by
rhombic arrangements of LVDTs. However, such an arrangement is not
applicable to the constant moment zone. As cracks formed at the bottom face in
the constant moment zone, the reference points on the concrete side face would
be dislocated, affecting the reliability of the measurement. Thus the slips in the
constant moment zone were revealed by integrating the slip strain between the
RC beam and the steel plates, instead of installing additional sets of LVDTs.
Further assuming that the slip was zero at mid-span and uniformly varying, the
slip in the constant moment zone could be expressed as
sl ( x) = ( Ls x)( hcp hpp ) ,
(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
M = 0.75Mu
M = 0.9Mu
M = 0.95Mu
1
Plastic Hinge
M = Mu
0
0
600
1200
1800
Distance from support (mm)
(a)
5
M = 0.5Mu
4
M = 0.75Mu
M = 0.9Mu
3
M = 0.95Mu
Plastic Hinge
M = Mu
0
0
(b)
600
1200
1800

5
M = 0.5Mu
4
M = 0.75Mu
M = 0.9Mu
3
M = 0.95Mu
Plastic Hinge
M = Mu
0
0
600
1200
1800
(c)
2
M = 0.5Mu
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
4.11 MOMENT CONTRIBUTION BY VARIOUS COMPONENTS

The moment resistance of BSP beams is provided by three actions,
including the RC beam core and the steel plates in flexure, as well as the axial
force couple between concrete in compression and steel plates in tension. By
investigating the moment distributions among these actions, the effectiveness
and behaviour of a strengthening scheme could be revealed.
In general the moment equilibrium of a BSP beam section can be

expressed mathematically as,
M = M c + M p + Fp h ,
(4.7)
where M is the total moment of section, M c is the moment of RC, M p is the

moment of steel plates, and Fp h is the axial force couple between concrete in
compression and steel plates in tension.
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.
The load-deformation relationships of steel plates and reinforcement are

assumed to be bi-linear, with the stiffness of the material after yielding is zero.
Material tests were carried out to obtain the Youngs modulus and yield strength
of the corresponding materials and the material properties have been presented
in Table 3.4.
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.
The moment contributions of the three components, expressed as the

percentage of the applied moment M, are plotted against the applied moment M
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.
Installation of stronger and stiffer bolts has a significant positive effect on

the percentage contributions among the components. Specimen SBSP and
Specimen WBSP installed with similar plates, but the moment contributions of
the axial force couple are on average 35% and 20% respectively. The 15%
difference is purely due to the change in bolt arrangement, which is significant.
On the other hand, the moment contributions of the axial force couple in
Specimen SBWP and Specimen WBWP are 30% and 20% respectively. While
the difference is slightly lower, it is still significant comparing to the overall
strength of the section.
The depth of plates used has a significant effect on the moment

contributions by flexure of steel plates. 10% of the total moments are
contributed by flexure of steel plates in the specimens with Strong plates
(Specimens SBSP and WBSP) and only 1% of the moments are contributed by
steel plates in Weak plate arranged specimens (Specimen WBWP and
Specimen SBWP). This suggests that the moment contribution by flexure of
steel plates is small when they are less than half of the depth of the beam.
90
100%
Concrete flexure
Plate flexure
Moment contribution (%)
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)
Figure 4.14 Moment contributions from various components a) Specimen

SBSP, b) Specimen WBSP, c) Specimen WBWP, and d) Specimen SBWP
91

The results obtained in the experimental study were presented. It was
observed that the structural behaviours of strengthened beams would be
changed when different bolt-plate combinations were used. A new classification
was introduced to divide different bolt-plate arrangements into different
categories. The beams are classified as concrete-, bolt- or plate- controlled
according to the strength of bolts and plates used and the corresponding
structural behaviour were described. Further studies would be required to
investigate the behaviour of beams under this classification.
The effect of partial shear interaction on the flexural strength performance

was investigated in detail. According to the results, the partial shear interaction
alone would have more than 30% reduction of increase in strength to the
specimen. This figure was large and more than the estimated value proposed by
Oehlers et al. (2000b). As the flexural strength was one of the most important
parameters in design, more studies would be carried out in chapter 8 to consider
partial interaction in flexural strength estimation of BSP beams.
The slip responses of specimens between RC and steel plates were

computed and the non-linear slip behaviour was discussed in detail. It was
revealed that the gradual formation of plastic hinge in RC would result in
longitudinal slip strain accumulation in the plastic hinge zone, which would
cause the change in slip profile and would affect the accuracy of the Two Alpha
Approach for analysis of BSP beams, which is going to be introduced in
Chapter 8. This observations support a further study on the non-linear slip
response by a theoretical approach.
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.
This chapter aims to develop an original iterative procedure to compute the

non-linear load-slip relationship of bolt groups, which consist of un-yielded and
yielded bolts, under any combined in-plane moment and shear. Apart from the
normal assumptions used in elastic analysis of bolt groups, it is further
postulated that the direction of force of the yielded bolts depends on the centre
of rotation of the bolt group. By varying the centre of rotation, equilibrium
between yielded bolt forces and applied loads could be sought in each loading
step. The load-slip relationships of individual bolts are idealized as bi-linear
curves. An original computer program has been implemented to illustrate and
validate the procedure.
94
5.2 NON-LINEAR THEORY OF BOLT GROUPS

5.2.1 Basic assumptions
The basic assumptions adopted in the development of the theory include:
1)
The bolts are perfectly fitted and connected by a plate, which acts as if
rigid.
2)
The media to where the bolts connect are frictionless.
3)
The bolts are separated widely such that the interference between them can
be neglected.
4)
The load-deformation behaviour of bolts are assumed to be linear-elasticplastic
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.
The fourth assumption is an extension of the elastic assumption in classic

bolt group theory. In elastic analysis, it is assumed bolts behave elastically until
the yield strength is reached. Such assumption is extended to post-yield stage
that the magnitude of bolt force remains the same i.e. plastic stage. The elastoplastic assumption has been widely accepted for different engineering
applications, including bolt group analysis (Owens and Cheal, 1989; Surtees et
al., 1981) because of its simplicity and readiness to be analyzed in different
circumstances. While the magnitude of bolt force is assumed to remain
unchanged in post-yield stage, it is further postulated that the direction of the
yield bolt force is varied according to the incremental displacement. As the bolt
group is assumed to displace in rigid body manner, the bolt force of any yielded
bolt would be perpendicular to the line joining the instantaneous centre of
rotation (xo, yo) and the location of that bolt as shown in Figure 5.3.
96
5.2.2 Classic elastic theory on bolt groups

For completeness, the classic elastic theory (Harrison, 1980) for analyzing
bolt groups subjected to in-plane loads is reviewed. Based on that, non-linear
bolt group theory will be presented in the next section.
Figure 5.1 shows a general arrangement of a group of bolts subjected to
general external loads (Fx, Fy, M). The lateral stiffness of the ith bolt in x
and y directions are kxi and kyi respectively. The rigid body movements of the
bolt group in x-, y- and rotational directions at the origin of the coordinate
system are denoted by (sl, sv, s). Based on the first assumption, the lateral
movements of the ith bolt in x and y directions (ux, uv)i can be evaluated as,
u xi = sl yi s and u yi = sv + xi s
(5.1)
The ith bolt (xi, yi)
Fy, sv
O
Fx, sl
M, s
Figure 5.1 General arrangement of a bolt group

By multiplying the deformations of the bolts with its stiffness, the lateral
forces (Pxi, Pyi) attracted to the ith bolt in each direction are equal to,
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)
where K x = k xi , K y = k yi , I zz = k xi yi2 + k yi xi2 , R x = k xi y i and

i
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)
Figure 5.2 Rigid body movement of a bolt group
5.2.3. Non-linear theory on bolt groups
In this section, a non-linear theory for calculating the distributions of loads

and displacements in a bolt group with yielded and un-yielded bolts will be
presented. It can also be employed to find the load-slip relationship of general
bolt groups under in-plane loads. Since a non-linear solution of the centre of
rotation is to be sought, incremental loading and iterative procedure have to be
used in the solution procedure. For simplicity, proportional applied load is
considered in the analysis.
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)
In the above equation, only the un-yielded bolts are considered.

The bolt forces components of the ith yielded bolt about the origin of the
coordinate system are related to the instantaneous centre of rotation ( xoj , yoj )
and the incremental rotation j , and can be expressed as,
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
P ij = xij Pyij yij Pxij ,
(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
Old bolt force
(5.16)
Change in bolt force due to

the shift of centre of rotation
New bolt force
Old centre of rotation

New centre of rotation
Figure 5.3 Post-yielded response of bolt
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
5.3 IMPLEMENTATION OF NON-LINEAR BOLT GROUP THEORY
To study the inelastic behaviour of bolt groups, a computer program,

namely BOGA, written in FORTRAN language was developed based on the
theory described in the last section. In this program, incremental proportional
load is applied to the bolt group and equilibrium is maintained in each loading
step. The bolt group responses in terms of bolt forces and displacements are
calculated by summing up all the incremental responses from each loading step.
Detailed description of the program has been given below.
5.3.1 Description of the methodology
The algorithm of the computer program for the determination of the

instantaneous centre of rotation and the incremental rotation at each loading
step has been presented in Figure 5.4 and the details are described in this
section.
Step 1: The required parameters of the bolt problem are defined. These consist
of,
1) The geometry of the bolt group (i.e. the number and the locations of
bolts),
2) The load-slip relationship of bolts (i.e. the bolt stiffness, the yield
and ultimate displacements of the bolts),
3) The incremental load applied to the bolt group and the target load
steps.
4) A preset tolerance to stop the iteration process at each loading step.
Step 2: To start with, incremental load (Fx Fy, M) is applied to the bolt
group.
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
displacements ( u ij 1 , vij 1 ) and the locations of bolts are updated

accordingly. Then the yield conditions of bolts are updated by
considering the cumulative displacements of individual bolts.
Step 9: The total number of bolts remaining un-yielded is checked by comparing
the cumulative slip with the yield slip of the bolts. And the slip is also
checked against the ultimate slip of the corresponding bolt. Once there is
only one bolt remaining elastic, or any of the bolts reaches its ultimate
slip, the bolt group is identified as reaching its ultimate capacity and the
iterative procedure is terminated. Otherwise, the program proceeds to
Step 2 of the j+1th loading step and the next incremental load is applied
to the bolt group.
104
Start
Define
1. Load slip relation
2. Applied loading
3. Bolt group geometry
Apply Fx Fy, M to the

bolt group
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
1 bolt remains unyielded

no
Limit deformation
reached
no
yes
Stop
Figure 5.4 Schematic diagram of the iteration process
105
5.3.2 Steepest descent method for solving the instantaneous centre of

rotation
To evaluate the instantaneous centre of rotation at each loading step, an

iterative procedure using the steepest descent method (Hageman and Young,
1981) is developed as illustrated in Figure 5.5. As mentioned previously, the
solution has to satisfy global force equilibrium in x- and y- directions, and hence
the goal of the scheme is to find a centre of rotation satisfying the above
equilibrium conditions. To drive the iteration loop, the centre of rotation of the
previous loading step ( xoj 1 , yoj 1 ) is set as the first trial ( xok , yok ) of which k=1.
In general, the steepest descent method is represented by the following

equation,
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)
The function defining the solution criteria is represented as f(x, y),

f ( x, y ) = h ( x, y ) 2 + g ( x , y ) 2 .
(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,

k = arg min f xok s

s >0
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
Points for searching - f(xo1,yo1)
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
Figure 5.5 Iterative scheme using steepest descent method
107
5.4 NUMERICAL VERIFICATION
The computer program BOGA is implemented to simulate the non-linear

response of bolt groups as given below. The results would be compared to those
from the non-linear finite element analysis in order to validate the developed
program. The bolt group considered in the numerical example consists of six
identical M20 bolts with the idealized load slip relationship as shown in Figure
5.6a. The physical properties of the bolts are as follows: the stiffness is
104kN/mm, the capacity is 260kN, the yield displacement is 2.5mm and the
ultimate displacement is 5mm. The geometric arrangement of the bolts is
depicted in Figure 5.6b. The origin of the coordinates, where the load is being
applied, is located at the middle between Bolt Nos. 3 & 6 as shown in Figure
5.6b. The total load applied to the bolt group includes lateral forces (Fx =
380kN, Fy = 350 kN) and a twisting moment (M = 131kNm), as shown in
Figure 5.6c.
5.4.1 Non-linear finite element analysis
The non-linear finite element package ATENA (Cervenka and Cervenka,

2002) is employed to analyze the numerical example and to validate the result
from BOGA, which will be presented in the next section. In the finite element
analysis, the mesh configuration of a single bolt is shown in Figure 5.7. The
load-slip behaviour of bolts is simulated by a ring of bilinear material which
would deform when subjected to in-plane load. The inner ring of the bilinear
material represents the shank of the bolt, and in the analysis, the inner ring is set
as fixed as if the bolt is anchored in a rigid medium, while the outer ring of the
bilinear material is joined to the steel plate in the analysis. The stress-strain
relationship
of
the
material
was
108
adjusted
so
that
the
Bolt Force (kN)
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)
Figure 5.6 Detail of the numerical example a) Load-slip relationship of bolt,

b) Geometry of bolt group, and c) Applied load
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
Fixed inner ring

Figure 5.7 Mesh detail of a single bolt
Figure 5.8 Mesh for non-linear finite element analysis
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
5.4.2 Simulation by BOGA
The program BOGA developed is also adopted to simulate the non-linear

response of the above example. The external load is applied to the bolt group
incrementally, in 500 equal loading steps and the load to obtain the complete
response of the bolt group. The load-slip relationship obtained by BOGA is
super-imposed in Figure 5.9 to compare the results simulated by non-linear
finite element analysis. It is observed that non-linear finite element analysis
terminates at a load level slightly lower than the predicted ultimate capacity by
the proposed method. However, it can be seen in Figure 5.9 that the results from
the two methods agree closely with each other, with maximum error below 5%.
Both analyses show an initial elastic stage, followed by a gradual decrease in
stiffness until the capacity is reached. From the consistency of the result
obtained from the two analyses, it can be concluded that the proposed numerical
technique can be applied to model the load-deformation relationship of a bolt
group in a simplified way with good accuracy.
An original procedure to deal with non-linear load-slip response of bolt

groups subjected to combined in-plane moment and shear was presented in this
chapter. All the bolts were assumed to behave in an elastic-plastic manner. By
assuming the direction of yielded bolt force is perpendicular to the line joining
the instantaneous centre of rotation to the location of the bolt, the relationship
between the forces of the yielded bolts and the instantaneous centre of rotation
of the bolt groups was derived. After solving the centre of rotation in each load
increment by steepest descent method, the complete response of the bolt group
and the behaviour of individual bolts were evaluated accordingly.
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
In Chapter 5, a numerical model is developed to trace the load-slip

relationship of bolt groups under combined in-plane moment and shear. The key
assumption adopted in the formulation of the model is the linear-elastic-plastic
load-slip behaviour of bolts within the bolt groups. While the stress-strain
relationship of steel is generally linear-elastic-plastic, the load-slip response of
bolts and anchor bolts under in-plane shear is often non-linear with gradual drop
in tangential stiffness up to peak load (Crawford and Kulak, 1971; Yam and
Chapman, 1968). As a result, the elastic-plastic model developed has limited
applicability in practical situations when non-linear bolts are used.
Concerning the above limitation, a non-linear model capable of modelling

bolts with gradual drop in stiffness under in-plane shear is developed in this
chapter. A new model describing the non-linear in-plane behaviour of bolts
when being deformed along any non-linear path is established based on an
analogy from the material plasticity of bolts. Then, the governing equations for
the load-slip relationship of bolt groups under combined in-plane moment and
shear are derived and expressed in an incremental form. The mathematical
model is implemented via an original computer program to simulate the
response of bolt groups and the accuracy of the proposed model is validated
against the available experimental data in literature. The method developed is
115
able to be incorporated into a non-linear program to study the effect of partial

interaction on the reinforced concrete beams strengthened by bolted side plate.
6.2 NON-LINEAR LOAD-SLIP RESPONSE OF BOLT GROUPS
In this section, a kinematic hardening model for simulating a bolt

subjected to a multi-directional shear and deformed along a non-linear path is
established. Based on that, incremental relationships between the responses of
bolt groups and the applied in-plane moment and shear are derived by
considering the force equilibrium in the x- and y- directions as well as the
moment equilibrium of bolt groups. By applying finite incremental loads to the
bolt groups, the complete load-slip response can then be solved by summing up
the incremental responses.
6.2.1 Assumptions
The following assumptions are made within the framework of the present
formulation.
1)
Bolts are connected by a plate that acts as if it is rigid.
2)
The media to where the bolts connect are frictionless.
3)
The bolts are separated widely such that the interference between them can
be neglected.
4)
The load-slip relationship of bolts, including the effect of bolt hole

elongation, is assumed to be tri-linear (see Figure 6.1) and kinematically
hardened.
5)
Failure of individual bolts, and hence the bolt group, is governed by

internal bolt forces.
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
Figure 6.1 Idealized tri-linear load-slip relationship of bolts
117
6.2.2 Numerical model for bolts under multi-directional shear
Prior to solving the non-linear response of bolt groups, the behaviour of

bolts under multi-directional shear has to be defined since the deformation path
of bolts is not straight (Crawford and Kulak, 1971). In this study, the bolt
behaviour, which assumes individual bolts are kinematically hardened when the
applied load exceeds their yield force, is postulated based on multilayer
plasticity as developed by Mroz (1967).
As an analogy to multilayer plasticity, the behaviour of bolts is controlled

by two surfaces on the force plane, namely the yield surface and the bounding
surface. The yield surface represents the occurrence of the plastic stage, while
the bounding surface governs the stage of yielding, which will be explained in
the following section. The two surfaces are both circular and centred at the
origin of the force plane, as illustrated in Figure 6.2a, since bolts under shear
generally behave invariantly with respect to in-plane directions. The radii of the
yield and bounding surfaces are defined as Py and Py, respectively, from the
idealized tri-linear load-slip curve of bolts, as shown in Figure 6.1. According to
Mrozs translation rule (Mroz 1967), the yield surface never overlaps with the
bounding surface, and the rule implies that the yield surface always nests within
the bounding surface.
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.
6.2.2.1 First stage yielding
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)
and is the centre of the yield surface. When an infinitesimal force P is

exerted on the bolt, the centre of the yield surface will translate along the
direction joining P and P , a point on the bounding surface (target surface) such
that the normal to the bounding surface at P equals n , by an amount to the
new centre . Denoting the translation direction as m, we have
119
m = PP,
(6.4)
,
= = m
(6.5)
is the translation unitary direction and is given in Eqaution (6.6),

where m
=
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 .
6.2.2.2 Second stage yielding

Second stage yielding refers to the state when the yield surface translates
outward from the origin and touches the bounding surface, as shown in Figure
6.2c. In this stage, the stiffness of the bolt in the direction normal to the yield
surface is further reduced from the first stage plastic stiffness, kp, to kp (see
Figure 6.1), and P lies on both surfaces. This implies that the two surfaces touch
each other at P. Presented mathematically, we have
P = P.
(6.8)
The common normal to the surfaces at P, is calculated as
n =
P
P
=
,
P
P
(6.9)
where is the centre of the bounding surface.
120
If an infinitesimal force P is applied, the centre of bounding surface translates

from to along the common normal. In mathematical terms,
= = n .
(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
Updated yield surface
Bounding surface
Instantaneous yield surface
b) First stage yielding
Updated yield surface
Updated bounding surface
Instantaneous yield surface
Instantaneous bounding surface
c) Second stage yielding

Figure 6.2 Translation rules of yield surfaces in a) Elastic stage, b) First
stage yielding and c) Second stage yielding
122
6.2.3 Incremental load-slip relationship of bolt groups

Consider a general bolt group, as shown in Figure 6.3, subjected to an
external load ( Fx , Fy , M ) . The external load is divided into equal, infinitesimal
load increments (Fx , Fy , M ) and is applied to the bolt group sequentially.
In a particular loading step j, the elastic and plastic stiffness of the ith bolt are ke
and kp respectively. The lateral movements of the bolt group at the origin of the
coordinate system are denoted by (sl , sv , s ) , as illustrated in Figure 6.3.
sv
s
sl
Figure 6.3 Rigid body movement of a bolt group

By classical bolt group theory, the relationship between the deformation of
the ith bolt ( u x , u y )i and the rigid body movement of the bolt group
(sl , sv , s ) can be expressed as
sl
u x 1 0 yi
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
Applying a coordinate transformation to the global x and y directions, and using

Eqaution (6.14), the following relationship can be obtained,
Px c s k p
P =

y i s c i 0
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
Figure 6.4. Post-yield load-slip relationship of bolts
Expanding Eqaution (6.16) and substituting kr = ke k p , where kr represents the

stiffness degradation, results in
Px ke c 2 kr
=
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)
Substituting Eqaution (6.17) into Eqaution (6.18) and rearranging, we have
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
Rssx = (ke si2 kri ) xi
i
2
i ri
I ccy = (ke c k ) yi2 ,
Rccy = (ke ci2 kri ) yi
Rcsy = ci si yi kri ,
Rcsx = ci si xi kri
I ssx = (ke si2 kri ) xi2
and
I cs = ci si xi yi kri .
i
Equation (6.19) represents the incremental load-slip relationship of the bolt

group under in-plane shear force in any direction and eccentricity at a single
load step. In particular, when kr equals zero (i.e. all bolts are elastic), all nondiagonal terms become zero and it represents the elastic case where the
behaviour of bolt groups can be simulated by independent springs. The rigid
body movement of the general bolt group S after the jth loading step is readily
solved by summing up incremental responses, represented mathematically by
j
S = ( sl , sv , s )n .
(6.20)
n =1
6.3 IMPLEMENTATION OF NON-LINEAR BOLT GROUP MODEL

Following the aforementioned non-linear bolt group theory, a computer
program, BOGAN, written in FORTRAN, was developed to carry out
numerical simulations of bolt groups under in-plane shear. The incremental load
step approach is used in the program; a schematic diagram showing the major
steps of BOGAN is presented in Figure 6.5, and the details are presented herein.
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
Read properties of the bolt

group and external load
Apply load increment
(Fx, Fy, M)j
Assemble stiffness matrix
relating (sl, sv,s) with
applied load
Solve xj, yj, j, Pxij, Pyji
j=j+1
Compute xij, yij

Update yield surfaces and
force vectors of bolts
Update locations of bolts
Ultimate load
reached
No
Yes
Stop
Figure 6.5 Schematic diagram of the computer program BOGAN
127
6.4 NUMERICAL SIMULATION BY BOGAN

In this section, numerical simulations carried out by the program BOGAN
are discussed. The experimental study of non-linear deformations of bolt groups,
performed by Crawford and Kulak (Crawford and Kulak, 1967), was modelled
by BOGAN to verify the applicability of the theory to the practical problem.
6.4.1 Experimental investigation

In Crawford and Kulaks experiments (Crawford and Kulak, 1967), eight
bolt groups were tested by a setup in which two bolt groups were used to
connect the supports to the two ends of a deep beam with the load applied at its
mid-span, as shown in Figure 6.6. The eccentric load was applied to the bolt
groups by the reaction at the supports. The load-rotation relationship and the
ultimate strength of the bolt groups were recorded in the experiments. The
parameters varied in the test included the dimensions of the bolt groups, the
separation between bolts, and the eccentricities of the applied loads. The
detailed arrangements of the specimens, which consist of 1 to 2 columns with 4
to 6 bolts in each column, are given in Table 6.1. More details concerning the
experiment, including the test setup and the shear test of the bolts, can be found
elsewhere (Crawford and Kulak, a).
Load P
Bolt group (a)
Bolt group (b)
Figure 6.6 Experimental setup in Kulak and Crawfords study
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
Table 6.2 Mechanical properties of bolts used in the simulations

k (kN/mm)
P (kN)
ke
kp
kp
Py
Py
Pu
526.7
66.3
5.7
133.8
260.0
297.0
6.4.2 Simulation by BOGAN

To simulate the load-slip response of the bolt groups, the properties of the
bolt groups, including the coordinates of the bolts and the load-slip relationship
of the bolts, have to be defined. In this analysis, the load-slip relationship of
bolts is based on the shear test results obtained by Crawford and Kulak, in
which a compression jig is used. It had been revealed that using a compression
jig in the shear test results in a 10% increase in shear strength with little effect
on initial stiffness (Wallaert and Fisher, 1965), and the failure behaviour of
bolts was better reflected by a tension jig experiment (Kulak et al., 1987); hence,
the load-slip relationship of bolts used for simulations was adjusted accordingly
and idealized as a tri-linear function, as shown in Figure 6.7. It should be noted
that most of the bolts behave with gradual stiffness drops as the deformation
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
Figure 6.7 Load-slip relationship of bolts adopted in the simulations
6.4.3 Comparison of the numerical results with the experimental results

Due to the symmetric arrangements of bolt groups about the y-axis, only
the bolt group on the left hand side of the experimental setup (see Figure 6.6)
was modelled and the applied load acted along the y axis, which is also the axis
of symmetry. Moreover, the applied load P was also reduced by half in the
numerical simulation. The calculated load-rotation curves and the experimental
results of specimen B1, B4 and B7 are plotted in Figures 6.8a to 6.8c. In general,
the measured and calculated results agreed well with each other. However, the
130
simulated elastic stiffness is consistently lower than the experimental results.

The difference may be due to ignoring the frictional force between the
connecting media in the numerical simulation. When preload was applied on
bolts in the experimental study, the external loading would be first resisted by
the friction between the connecting media (Kulak et al. 1987), the magnitude of
which depends on factors like the preload in bolts and the conditions of the
faying surfaces, and results in a higher stiffness for the bolt connection. A
detailed comparison of the calculated and measured elastic stiffness of the bolt
groups is summarized in Table 6.3.
Table 6.3 Comparison of numerical results with experimental results

Stiffness (Secant stiffness at 40%
ultimate loading ) (kN/0.01rad)
Ultimate Load (kN)

Experimental
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
Rotation (0.01 rad)
Figure 6.8 Comparison of moment rotation curves with experimental

results of a) Specimen B1, b) Specimen B4 and c) Specimen B7
132
From the consistent numerical and experimental results of the individual

bolt groups, it can be concluded that the proposed numerical technique is
generally accurate and reliable for determining the load-slip relationship of bolt
groups. However, when a particular focus is on the initial stiffness of the bolt
groups, the effect of the frictional behaviour between the faying surfaces should
be considered.
The load-slip relationship of B7 is simulated using different load

increment sizes ranging from 1% to 7% of the load-carrying capacity of the bolt
group and is plotted in Figure 6.9. It is observed that variations in load
increment size have little effect on the predicted ultimate strength and the shape
of the load-slip curve.
600
500
Load (kN)
400
300
increment=4.448kN
200
increment=8.896kN
increment=17.792kN
100
increment=35.584kN
0
0
Rotation (0.01 rad)
Figure 6.9 Load-rotation response of B7 simulated using different load increments
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.
The computer program BOGAN, which simulates the non-linear responses of

bolt groups under combined in-plane moment and shear, has been implemented
according to the theory developed to evaluate the effectiveness of the theories in
real applications. Its reliability and accuracy have been verified by comparing
the numerical results with those from an experiment in the literature. Being
expressed as an incremental form, the theory and algorithm developed in
BOGAN can be applied to the non-linear analysis of BSP beams, so that the
effect of partial interaction due to the finite stiffness of bolts can be modelled
properly.
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.
To consider and model the behaviour of partial interaction, the non-linear

finite element analysis is often used. A detail review of numerical methods
previously developed was given in Chapter 2 and in these analyses, the
longitudinal partial interaction was considered by allowing nodes in RC beam
and steel plates to have different longitudinal displacement. However,
transverse partial interaction was neglected since vertical displacement profiles
of RC beam and steel plates were assumed to be identical. So these analyses are
inappropriate to model the response of BSP beams where both longitudinal and
transverse partial interaction are present and interacts with each other. To model
the response of BSP beam, a numerical model allowing for both longitudinal
and transverse partial interaction, and the interaction between them is necessary.
135
The major difficulty in allowing for both longitudinal and transverse

interaction in the numerical analysis of BSP beams is that partial interaction in
the two directions interact with each other in the non-linear stage. This chapter
aim to overcome the difficulty of allowing both longitudinal and transverse
interaction and their interaction by developing a new non-linear finite element
model. Reinforced concrete and steel plates are modeled as two separate
elements interacting with each other through discrete bolt groups. The nonlinear model of bolt groups developed in Chapter 6 are used to simulate the nonlinear interaction between combined in-plane shear in longitudinal and
transverse directions, while the formulations of element stiffness matrix is based
on a layered finite element model. By incorporating the non-linear model of bolt
groups into the formulation of structural stiffness matrix, the partial interaction
between longitudinal and transverse interaction can be modeled so that both the
structural response of BSP beams and the shear resisted by the bolts within the
bolt groups can be modeled accurately.
Using the proposed theory, a computer program SiBAN is developed.

Numerical simulations were carried out to illustrate slip responses in the
longitudinal and transverse directions along with a new type of partial
interaction (the rotational partial interaction) for BSP beams. The rotational
partial interaction is significant for the load distribution in connectors when
more than one row of connectors is used. The pre-peak responses of the loaddeformation of beams and the partial interactions are investigated in detail. As
the present side-plate strengthening system is likely to be governed by gravity
loads, a system collapse would take place when the applied load exceeds the
load-carrying capacity. Consequently, the present numerical study focuses on
136
pre-peak behaviour, and a simple load-controlled solution procedure is adopted

to solve the non-linear equations. The post-peak behaviour of BSP beams lies
outside the scope of this study.
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
The ith bolt group
Steel plate
(a)
The ith RC element
1st
2nd
ith
i+1th
The ith node
The ith steel plate element
Interaction through bolts
Flexural elements
n+1th
Rigid arms
(b)
L
a
e
a
(c)
Rigid Arms
b
A reinforced concrete beam

element
A steel plate element
Figure 7.1 Structural model of a BSP beam a) Prototype, b) Structural model,

and c) Typical RC beam & steel plate elements
7.2.2 Formulation of the non-linear RC beam and steel plate elements

The longitudinal and vertical displacement shape functions corresponding
to unit nodal displacements (ua , ub , va , vb , a ,b ) of the flexural element (Figure
7.2a) are assumed to be
u ( x) = [ (1 ), , 0, 0, 0, 0] ,
(7.1)
v( x) = 0, 0, (1 3 2 + 2 3 ), (3 2 2 3 ), L( 2 2 + 3 ), L( 2 + 3 ) ,
(7.2)
where = x/L is the natural coordinate.
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)
Invoking the relationship ( x) = v "( x) and differentiating Equation (7.2) twice

with respect to x, Equation (7.5) becomes
6
2
2
6
y (1 2 ), 2 y (1 + 2 ), y (2 3 ), y (1 3 ) .
2
L
L
L
L
bending ( x, y ) =
(7.6)
Combining Equations (7.4) and (7.6), the strain-displacement matrix B at any

point (x, y) is
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)
Equation (7.7) can be decomposed into the corresponding axial component Ba

and bending component yBb, i.e.,
B = [B a , yBb ] ,
(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
The stiffness matrix of the flexural element may then be expressed as

BT
K i = a T ET [ B a , yBb ] dV
V yB
b
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.
To evaluate the stiffness matrix, the area integrals representing sectional

properties of the elements must be solved, along with an integration, along the
length of the matrix member expressed in Equation (7.9). This study employs a
layered approach to solve for the section properties Ek Ak = ET dA ,
A
Ek Sk = ET ydA and Ek I k = ET y 2 dA . The advantage of this approach is

A
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)
Multiplying the above transformation matrix with the stiffness matrix as

shown in Equation (7.9), i.e., K i , RC = TT K T , the element stiffness matrix for
142
an RC beam element can be obtained. The explicit expression of the matrix is

presented in Appendix 7.2.
7.2.3 Formulation of the stiffness matrix of connecting bolt groups

To model the non-linear load-slip behaviour of the connecting bolt groups,
the kinematic-hardening model developed for bolts and bolt groups in Chapter 6
is adopted. In this section, the principles of the non-linear model are briefly
overviewed and detailed derivation of the stiffness matrix can be found in
Chapter 6.
According to the kinematic hardening model of bolts developed in Chapter

6, the current state of a bolt can be defined once the locations of centres of yield
surfaces ((Px, Py) and (Px, Py)) at various loading stages and the argument of
the normal vector n are known. The physical meanings of the associated
parameters on the force plane are illustrated in Figure 7.4.
Py
Yield
surfaces
(x, y)
(x, y)
Px
Figure 7.4 Parameters defining the current state of a bolt

When an incremental load is applied to the bolt, the stiffness of the bolt is
kp in the normal direction and ke in the tangential direction. The value kp is equal
to one of the stiffnesses ke, kp or kp depending on the current yielding stage.
143
The incremental load-deformation relationship of a bolt can be mathematically

expressed as
Pn k p
P = 0
t
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.
By applying classical bolt group theory and a series of mathematical

derivation, the tangential stiffness of bolt groups can be expressed as
K cc
Qx
Q = K
cs
y
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
I ssx = (k e si2 k ) xi2
I cs = ci s i xi y i k and ci , si represent the values cos i and sin i respectively.

i
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.
7.3 ASSEMBLAGE OF STRUCTURAL STIFFNESS MATRIX

The interaction of bolt groups is illustrated in Figure 7.6. By defining the
deformation of the bolt group as the movement of steel plates relative to the
concrete, the corresponding nodal equilibrium equations for the reinforced
concrete and the steel plates are
c
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)
Considering the force equilibrium of the ith bolt group, we have
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
[Q] = [Q1T , Q2T ...QnT ]T ,

[d i ]c = [ui , vi , i ]cT ,
[d ]c = [d1T , d 2T ...d nT ]Tc ,
[di ] p = [ui , vi ,i ]Tp ,

[d ] p = [d1T , d 2T ...d nT ]Tp ,
(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)
where [K ]b = diag[K 1 , K 2 ...K n ]b , [ F ]c and [ F ] p are the external load vectors

applied to the RC beam and the steel plate.
146
Reinforced concrete beam

i-1th element Qy
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
7.4 IMPLEMENTATION OF THE STRUCTURAL MODEL

A computer program SiBAN based on the above formulation is developed
to simulate the non-linear response of BSP beams. In this program, the stressstrain (-) relationship adopted for concrete consists of a parabolic relationship
up to the peak stress followed by a linear descending relationship in the postpeak region, as illustrated in Figure 7.6. It is mathematically expressed as
2
= f c 2 for < 0 ,
0 0
(7.29)
= f c for 0 ,
(7.30)
where fc is the uni-axial compressive strength of concrete and 0 is chosen as

0.0025 in this study.
147

fc
Figure 7.6 Stress-strain relationship of concrete adopted in SiBAN
The elastic-plastic relationships adopted for both reinforcements and

strengthening plates are shown in Figure 7.7. The mathematical definitions of
the stress-strain relationships for steel reinforcement are
= Es
for ys < < ys ,
(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
for yp < < yp ,
(7.34)
= E p yp for yp ,
(7.35)
= E p yp for yp ,
(7.36)
where yp is the yield stress of steel plate.
148
fs
fp
yp
Es
Ep
yp
ys
fp
Figure 7.7 Stress-strain relationship of a) Reinforcement and b) Steel plates

adopted in SiBAN
As mentioned in Chapter 6, the load-deformation of anchor bolts under
shear is non-linear, with tangential stiffness dropping gradually up to the point
of failure. Such a non-linear response was simplified in SiBAN to a tri-linear
function for the purposes of simulation. The tri-linear function was governed by
the three turning points on the function, as shown in Figure 7.8. Thus, a total of
six parameters (Py, Py, Pu, ke, kp and kp) were required to define the
mechanical properties of the anchor bolts.
P
Pu
Py
kp
kp
Py
ke
smax
Figure 7.8 Stress-strain relationship of bolts adopted in SiBAN
149
The detailed procedure of the computer program is shown in Figure 7.9

and described below:
Step 1: The required information to specify the program is input.
Step 2: The elastic stiffness matrices of bolt groups are formulated by applying
classic theory of bolt groups.
Step 3: An incremental load is applied to drive the load step.
Step 4: The tangential stiffness matrices of the RC and strengthening plates are
formulated using the procedure described in Section 2.2 based on the
current displacement vector. Together with the stiffness matrices of the
bolt groups from the load step, the structural stiffness matrix is
assembled.
Step 5: The incremental displacement vector is solved and added to the
cumulative displacement vector.
Step 6: The nodal forces are computed using the cumulative displacement
vector. The residual force vector is then solved by subtracting the
external force vector from the nodal force vector.
Step 7: Convergence checking is performed by examining the error term. If
convergence is not satisfied, the residual force vector is input to drive
the simulation and Steps 4 to 6 are repeated until convergence is
reached.
Step 8: The subroutine BOGAN is carried out by inputting the cumulative
displacement vector into the subroutine. At the end of the subroutine,
the updated element stiffness matrices of the bolt groups are obtained.
Step 9: The program is proceeds to Step 3, where a further incremental load is
applied.
150
Start
Read and transform input data
Apply incremental load at jth
step
Update element stiffness

matrices of RC and plate
Assemble load matrices and

stiffness matrices
Solve equilibrium equations
Determine the nodal

displacements of flexural
element m
j=j+1
Determine the strain at layer n
Iterations
m=m+1
Determine the layer stress
based on material constitutive
model
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
Figure 7.9 Solution algorithm of SiBAN
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.
An iterative scheme is required to solve the non-linear solution of the BSP

beam under loading. In SiBAN, the modified Newton-Raphson method is
adopted. This method can reduce the number of iterations in each load step by
updating the structural stiffness matrix in every iteration. The total
computational time for the solution procedure can also be reduced.
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
7.5 EXPERIMENTAL VERIFICATION

In order to verify SiBAN, simulations on the specimens in Chapter 3 and 4
were carried out. The numerical results, including the load-deformation
relationships at mid-span and the longitudinal slip responses were calculated
and compared with the experimental results to validate the proposed theory and
the computer program.
The material properties of concrete, reinforcements, steel plates and

anchor bolts required in the numerical simulations are listed in Table 1 and
shown in Figure 7.10, respectively.
Table 7.1 Summary of the material properties adopted in simulations
Concrete
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)
Bolt force (kN)
P1
0.625
50
P2
4.0
76
Figure 7.10 Bi-linear load-deformation model of anchor bolts
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
Reinforced concrete element

Steel plate element
Bolt group element
Figure 7.11 Finite element models used in the computer program SiBAN
154
The peak moments of specimens predicted by SiBAN are listed and

compared with the experimental results in Table 7.2. The numerical simulation
provided close estimations of peak moment responses with errors ranging from
0.8% to 7.6%, showing that the effect of partial interactions is properly
considered in the present analysis.
The mid-span moment-deflection curves of specimens computed by

SiBAN together with the experimental results are plotted in Figure 7.12. As
there was no node at the mid-span of the structural models, the mid-span
deflections were obtained by displacement interpolation. Substituting =0.5 into
Equation (7.2), the mid-span deflection can be expressed in terms of the
corresponding element nodal displacement vector as
v + v L ( a b )
vc ,mid = a b +
.
8
2
(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
The longitudinal slip response of specimens at different sections was

extracted by linear interpolation using nodal slip response and plotted in Figure
7.13. The numerical results computed by SiBAN generally agree well with the
experimental results. We observed that the numerical slip estimated by SiBAN
is slightly higher than in the experimental results when initial loading is applied.
This is due to the fact that the frictional resistance between the surfaces was
ignored when adopting a bi-linear approximation for the uni-directional loaddeformation relationship of bolts. The frictional resistance increases the initial
stiffness of bolts, which reduces the longitudinal slip responses of the specimens.
From the above comparisons, it can be seen that SiBAN is capable of

modelling the overall responses of BSP beams with accurate consideration for
partial interactions. In the following section, SiBAN will be used to determine
the complete bolt slip response along the span of BSP beams.
Table 7.2 Comparison of peak loads obtained from the present numerical
analysis and previous experiments
Specimen
Peak loads (kNm)
% 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
Figure 7.12 Comparison of mid-span load-deformation responses of a)

Specimen SBSP, b) Specimen WBSP, c) Specimen WBWP, and d)
Specimen SBWP
157
Experiment 200 (a)
(a)
Experiment 200 (b)

Experiment 1000 (a)
Experiment 1000 (b)
SiBAN 200
SiBAN 1000
0
0
10
20
30
40
6
Experiment 200 (a)
Experiment 200 (b)
(b)
Experiment 1000 (a)

4
Experiment 1000 (b)

SiBAN 200
SiBAN 1000
0
0
10
20
30
40
50

6
Experiment 200 (a)
Experiment 200 (b)

Experiment 1000 (a)
Experiment 1000 (b)

SiBAN 200
SiBAN 1000
(c)
10
20
30
40
50
60

2
Experiment 175 (a)
(d)
Experiment 175 (b)

Experiment 1025 (a)
Experiment 1025 (b)
SiBAN 175
SiBAN 1025
0
0
10
20
30
40
50
Figure 7.13 A comparison of longitudinal slip responses of a) Specimen

SBSP, b) Specimen WBSP, c) Specimen WBWP, and d) Specimen
SBWP
158
7.6 SLIP RESPONSES OF BSP BEAMS

7.6.1 Numerical model
Slip responses in the longitudinal and transverse directions have been
addressed and described by previous researchers (Oehlers et al., 1997b).
However, comprehensive and quantitative studies have yet to be done to
illustrate the actual slip responses along the span of BSP beams, especially in
the inelastic stage when a plastic hinge is formed in the RC components. Hence,
numerical simulations were carried out using SiBAN to investigate the slip
responses of BSP beams. Besides longitudinal and transverse slips, the
rotational slip (where the RC and plate rotate relative to each other) was also
identified. Variations in slips along the beams were studied in detail.
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
All plates are 6mm thick
Figure 7.14 Beam model for the investigation of bolt-slips
7.6.2. Longitudinal and transverse slips

The longitudinal slip profile along the span of the beam in the elastic stage
is shown in Figure 7.15a. The slip profile is continuous and anti-symmetric
about the mid-span of the beam. The slip is zero at the mid-span and increases
gradually with a decreasing rate to reach a maximum at a position close to, but
not exactly at, the end supports (at the 1st interior bolt group in this example).
While the occurrence of maximum longitudinal slip is not at the end span, the
longitudinal slip is almost constant near the end supports. For simplicity, we
may assume that the position of the maximum longitudinal slips occurred at the
supported ends.
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.
Further considering the relative magnitude between the longitudinal and

transverse slips, it can be seen that the maximum longitudinal slip along the
span is about ten times the maximum transverse slip in both the elastic and
inelastic stage. This was consistent with the findings of Oehlers et al. (1997),
who reported that the slip response of bolts is dominant in the longitudinal
direction.
7.6.3 Rotational slips

While the influence of translational partial interaction in the x(longitudinal) and y- (transverse) directions has been well studied by previous
researchers, the effect of rotational partial interaction between RC and steel
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.
The variation of rotational slip is anti-symmetric about the mid-span. In

the elastic stage, the rotational slip at the supports is almost zero, while the
rotational slips for each of the shear spans are of the same sign. The peak
rotational slips occur near the mid-span of the beam, and drop rapidly to zero at
the mid-span, as shown in Figure 7.15c.
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
ultimate state, adding significant slip demand to the corresponding bolts.

Special consideration should be made for the effects of rotational slip
concentrations when future studies are conducted.
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)
Transverse slip (mm)
0.06
0.03
-0.03
-0.06
(b)
Rotational slip (0.00001 rad)
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
Transverse slip (mm)
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
Rotational slip (0.00001 rad)
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

In this Chapter, a nonlinear analysis for BSP beams was developed. The
analysis is superior to previous numerical analysis developed for composite
beams as both longitudinal and transverse partial interaction are considered. By
incorporating the non-linear model for bolt groups developed in Chapter 6, the
interaction between longitudinal and transverse partial interaction in the nonlinear stage is also considered, leading to a more accurate prediction in both
load-deformation response of beams and the internal shear and slip within bolt
groups. By implementing the theory into the computer program SiBAN and
validating by experimental results, the applicability of the theory in simulating
the response of BSP beams was demonstrated.
Using the validated program, numerical examples were given to illustrate

partial interactions in BSP beams. It was concluded that the longitudinal slip
response is the most important slip among the three types of slips (i.e.,
longitudinal, transverse and rotational). The effect of non-linearity should be
properly considered in the partial interaction analysis in order to determine the
ultimate capacity of BSP beams. The concept of rotational slip was introduced.
Rotational slip may cause additional and significant slip demands on a beam.
Further studies are suggested to investigate the effects of rotational slip on the
structural response of BSP beams.
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.
The flexural strength of RC beam structures is normally regarded as a

sectional property. Once the material and geometric properties of a RC section
is well-defined, the flexural strength can be estimated by carrying a section
analysis. However, in BSP beams, the flexural strength is affected by the
presence of partial interaction. As shown by Newmark et al. (1951), the partial
interaction in BSP beams is a global compatibility problem and governed by a
second order differential equation with the longitudinal slip as the dependent
variables. Being a global compatibility problem, the effects of partial interaction
cannot be directly incorporated into section analysis for flexural strength
estimation of BSP beams. Thus new theory relating the partial interaction
problem and the sectional response at the locations with critical moment is
required.
167
This chapter aims to describe a new analysis, the 2-Alpha approach, to

incorporate the global compatibility problem of BSP beams into section analysis
which is commonly used for flexural strength prediction of RC structures. Two
new parameters, the strain factor () and the curvature factor () are
introduced to reduce the complicated global compatibility problem into a simple
strain separation problem between RC beam and steel plates at the section with
critical moment. On the one hand, the two parameters are related to the slip
response of the connecting media. On the other hand, the parameters describe
the separation of strain profiles between RC beam and steel plates. With the use
of these two parameters, typical section analysis can be modified to predict the
flexural response of BSP beams, which suppose to be a complicated problem.
8.2 TWO-ALPHA APPROACH

The 2-Alpha approach is an analytical procedure developed for solving the
flexural response of BSP beams. The analysis consists of two main parts. In the
first part, two new factors, the strain factors and curvature factors are introduced
to quantify the degree of partial interaction. The closed form solution of the
strain factor is derived by considering the longitudinal slip profile, which can be
solved by considering the longitudinal compatibility of the RC beam and the
steel plates, while the curvature factor is derived by considering the transverse
compatibility based on an assumed transverse shear transfer model. In the
second part, the strain factor and curvature factor are integrated into the section
analysis which is commonly used for tracing moment-curvature response and
estimating flexural strength of RC beam sections. By establishing relationship
between the strain profiles of the RC beam and the steel plates at the critical
section with strain factor and curvature factor, the moment curvature response
168
of the section can be solved. In the coming sections, the detail of 2-Alpha
approach will be described.
8.3 NEW PARAMETERS QUANTIFYING THE LEVEL OF

LONGITUDINAL AND TRANSVERSE INTERACTION
8.3.1 Partial interaction response of BSP beams
When a BSP beam is subjected to a sagging moment, the beam layers
above and below the neutral axis are shortened and elongated respectively.
When full interaction exists, the deformation of the steel plates is equal to that
of the adjacent concrete. In real situations, the deformation of the steel plates is
smaller than the adjacent concrete. Because of the finite stiffness of bolts,
mechanical slip, as illustrated in Figure 8.1a, is required to transfer the
longitudinal shear force to the steel plates. Considering the longitudinal
compatibility of the beam, the total elongations of the steel plates and the slip of
bolts is equal to the elongation of the RC beam, implying that the elongation of
the steel plates is reduced, and smaller than that of the adjacent concrete. This
behaviour is known as longitudinal partial interaction.
Other than the reduction in longitudinal elongation of steel plates, the

transverse deformation of the steel plates will also be reduced. Due to the finite
stiffness of bolts and the need for slip to transfer vertical shear forces, the plate
ends near the supports and the middle region of the steel plates move,
respectively, downwards and upwards relative to the RC beam, as shown in
Figure 8.1b. Thus the vertical deflection and hence the curvature of the steel
plates will also be reduced as compared to the RC beam in the same transverse
plane.
169
sl
sl
sv
sv
b)
a)
Figure 8.1 Illustration of slips in BSP beams under gravity load in

a) Longitudinal direction, and b) Transverse direction
a) full interaction
in both directions
b) no interaction
in both directions
c) partial interaction d) partial-longitudinal

full- transverse interaction
in both directions
Strain profile of the RC beam
Strain profile of the steel plates
Figure 8.2 Strain profiles of the RC beam and the steel plates with different
level of longitudinal and transverse interaction
8.3.2 New parameters quantifying the level of interaction

To quantify the effect of longitudinal and transverse partial interactions,
two factors, the strain factor and the curvature factor, are introduced. The two
factors describe the deformation of the section with maximum moment, as
illustrated in Figure 8.3. The strain factor, denoted as , is defined as the ratio
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 profile of plate
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.
In some cases, the level of interaction in transverse and longitudinal

directions may be different, for example, when the steel plate is installed on the
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.
8.3.3 Strain factor

In this section, the derivation of the strain factor is presented. The general
procedures to solve the longitudinal slip profile of BSP beams are first
presented. Then the strain profile is solved in terms of the longitudinal slip
profile of the beam. By applying the solutions of longitudinal slip profile, the
strain factor for simple loading cases can be solved.
8.3.3.1 Normalized longitudinal slip profile of BSP beams

Figure 8.4 shows the general arrangement of a simply supported BSP
beams. Under an arbitrary gravity load W(x), internal shear force V(x) and
sagging moment M(x) are induced. Due to the use of bolts, the steel plates will
slip longitudinally relatively to the RC beam. The whole beam can be divided
into two spans, of which the longitudinal slip is towards the same direction as
shown in Figure 8.4c. On each span, the longitudinal slip reaches maximum at
the support, reduces gradually and becomes zero at the location where the
internal moment is maximum.
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
Considering a beam span of which the longitudinal slip varies from

maximum at the support end to zero at the maximum moment end, we can
divide the span into finite segments of which the shear force and the stiffness of
shear connection is constant, as illustrated in Figure 8.5. The boundaries of the
ith segment closer to the support end and the maximum moment end are denoted
as xai and xbi respectively.
174
xa1
1st seg.
xb1
xai
ith seg.
xbi
Supported end
nth seg.
Maximum moment end
Figure 8.5 Notations of a span with n segments

According to Newmark et al. (1951), the longitudinal slip profile of each
of these segments is governed by a second order differential equation with the
longitudinal slip as the dependent variable. For the ith segment, the governing
differential equation can be expressed mathematically as,
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
the segments can be expressed as

sl ( x) = Ai e pi x + Bi e pi x + Ci
for
xai x < xbi ,
(8.2)
where Ai , Bi , Ci are arbitrary constants depending on boundary conditions. In

the above solution, the first term Ai e pi x + Bi e pi x is the characteristics solution
175
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
xai x < xbi .
(8.5)
By differentiating the expression for normalized longitudinal slip profile twice,

it could be seen that the value Ai is dependent on the constant term of Equation
(8.5) as
Ci =
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
According to the definition of longitudinal slip, the first derivative of the

longitudinal slip (slip strain), is equal to the first derivative of the longitudinal
displacement difference between the steel plates and the RC beam. It can be
expressed mathematically as
sl ( x) = lp, hp ( x) lc, hp ( x) ,
(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
the longitudinal displacements of the components, should be continuous, and

can be expressed mathematically as
sl ( xib ) = sl ( xia+1 )
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)
The coefficients Ci in the expressions of the longitudinal slip profile are

only related to the shear force profile V(x) of the beam and the geometric
properties of the segments. Once the applied load acting on the beam is defined,
the relationship between these coefficients can be established and set as follows,
Ci = gi C1
for i = 2...n
(8.13)
Va p
where gi = 1a i are known once the geometry and load acting on the
Vi p1
beam is well-defined. Similar expressions can be solved when segments with no

interaction are present.
By solving the above 3n simultaneous equations (Equations (8.8)-(8.13)),

the longitudinal slip profile of a span in the BSP beam can be solved and
expressed in terms of the maximum longitudinal slip smax. It will be shown in
the next section that the longitudinal slip profile, being solved by compatibility
consideration between the RC beam, the steel plates and the connecting bolts,
controls the value of the strain factor and the amount of interaction of BSP
178
beams. The longitudinal slip profiles of simple loading cases are given in
Appendices 8.1 and 8.2.
8.3.3.2 Mathematical derivation of strain factor
Consider the case when a simply supported BSP beam is subjected to

arbitrary gravity loading W(x). The longitudinal slip profile due to partial
interaction is denoted as sl(x). At the critical moment section, the concrete strain
is equal to the sum of plate strain and the slip strain, which can be
mathematically expressed as,
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
It can be seen in Equation (8.20) that the strain factor is inversely

( EA) p sl( Ls )
. It should be noted that the
proportional to the function 1 +
Ls
km
s
(
x
)
dx
l
0
( 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
Figure 8.6 Parameters of longitudinal slip profile governing strain factor

8.3.3.3 Strain factor of simple loading cases
Example 1: Simply supported beam under mid-point load with evenly

distributed bolts
As derived in Appendix 8.1, the longitudinal slip profile for simply
supported beam under mid-point load with evenly distributed bolts is
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
The derivative term sl ( Ls ) and integral term
Ls
sl ( x)dx can be solved using
Equation (8.21) as,

p (e px + e px )
sl ( Ls ) = sl ,max pLs
pLs
2 x=L
e +e
s
p(e pLs + e pLs )

= sl ,max pLs
,
pLs
2
e +e
Ls
(8.23)
pL (e pLs + e pLs ) e pLs + e pLs

sl ( x)dx = sl ,max s
p (e pLs + e pLs 2)
(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)
Example 2: Simply supported beam under uniformly-distributed load with

evenly distributed bolts
According to Appendix 8.2, the normalized longitudinal slip profile for simply
supported beam under uniformly-distributed load with evenly distributed bolts
is,
e p ( x Ls ) e p ( x Ls ) + p( Ls x)(e pLs + e pLs )

sl ( x) = sl ,max
.
e pLs e pLs + pLs (e pLs + e pLs )
The derivative term sl ( Ls ) and integral term

Equation (8.26) as
182
Ls
(8.26)
sl ( x)dx can be solved using
p (e p ( x Ls ) + e p ( x Ls ) e pLs e pLs )
sl ( Ls ) = sl ,max pLs
e pLs + pLs (e pLs + e pLs ) x = L

e
s
p(2 e pLs e pLs )

= sl ,max pLs
,
pLs
pLs
pLs
e + pLs (e + e )
e
Ls
(8.27)
2e p ( x Ls ) + 2e p ( x Ls ) + p 2 x(2 Ls x)(e pLs + e pLs )

sl ( x)dx = sl ,max
.
pLs
pLs
pLs
pLs
+
+
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)
8.3.4 Curvature factor

A procedure for solving the curvature factor is described in this section.
The curvature factor is derived based on an idealized transverse force model.
Using the idealized transverse force model and considering transverse
displacement compatibility, the ratio between curvatures of the steel plates and
the RC beam at the section with maximum moment is expressed as a function of
geometric and material properties of the beam and the shear connection.
Examples are given to illustrate how the curvature factor for simple loading
cases can be solved.
8.3.4.1 Idealized transverse force model

The derivation of the curvature factor is based on a simplified model for
the shear transfer between the RC beam and the steel plates. Consider the
simple case when bolts are evenly distributed along the beam span and the beam
is loaded by a symmetric arbitrary gravity load W(x). In the transverse direction,
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)
Shear transfer zone Shear transfer zone
Figure 8.7 Idealized transverse shear force model exerted on the RC beam
from the steel plates under arbitrary gravity loads
8.3.4.2 Transverse displacement compatibility

When transverse partial interaction occurs, the plate deflection is reduced
due to finite stiffness of the connecting media. The total transverse displacement
difference of the RC beam in the shear transfer zone is equal to the sum of
displacement difference of the steel plates and the vertical slip at the two
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)
(1) transverse displacement profile of steel plates

(2) transverse displacement profile of RC beam
Figure 8.8 Transverse compatibility consideration in shear transfer zone
Equation (8.31) is the transverse compatibility equation. By relating the

displacement terms in the compatibility equations as functions of curvatures of
the RC beams and the steel plates, the curvature factor can be solved. As the
relationship between displacement difference of the shear transfer zone and
curvature at section with maximum moment depends on the load acting on the
beam, the curvature factor can only be solved when the applied load is defined.
185
In the following section, the general procedure and the solution for some simple
loading cases and bolt arrangements are solved.
8.3.4.3 Curvature factor of simple loading cases

Example 1. Simply-supported beam under mid-point load with evenly
distributed bolts
By adopting the transfer shear force model illustrated in Figure 8.7 and
applying Euler beam theory, the general expression for the moment of the RC
beam and the steel plates of the left span can be expressed as,
M c ( x) = Px
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)
Rearranging and simplifying Equation (8.39), the curvature factor can be

derived as
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
Example 2. Simply-supported beam under uniformly-distributed load with

evenly distributed bolts
UDL =
a)
b)
Shear transfer zone Shear transfer zone
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)
Rearranging and simplifying Equation (8.47), the curvature factor can be

derived as
190
p ( Ls )
=
c ( Ls )
4 12 144 ( EI ) p
1 +
+
4
25 75 5km Ls
4 12
where +
25 75
(8.48)
is the error term as explained in the previous example.
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.
8.4 MODIFIED MOMENT CURVATURE ANALYSIS

8.4.1 Overview
Moment-curvature analysis has been well recognized as an effective tool
to study the flexural behaviour of beams and columns. Many researchers
(Carreira and Chu, 1986; Skeikh and Yeh, 1992; Ho et al., 2003) employed the
analysis to study the flexural strength, stiffness and more recently the ductility
behaviour of RC structural elements. While different solution techniques were
adopted by these researchers, the underlying philosophies of these analyses are
the same. Firstly, the curvature of the section being analyzed is set. Then the
maximum compressive strain or the depth of neutal axis is moved until the
horizontal equilibrium of the section is satisfied. Further considering moment
equilibrium of the section, the internal moment of the section at the pre-defined
curvature is obtained. By repeating the above procedures with different pre-set
curvatures, the complete moment curvature response of the section can be
solved and the flexural characteristics can be revealed.
191
One of the key assumptions adopted in the moment-curvature analysis is

that plane sections will remain plane after bending. This assumption implies that
the strain profile of the section is linear. This assumption is generally valid in
RC structures. However, this assumption is not valid in BSP beams. As partial
interaction is present, the strain profiles of the RC beam and steel plates do not
coincide. So the general approach adopted in moment-curvature analysis cannot
be used directly for BSP beams. A modified section analysis will be developed
in this section to solve the moment-curvature response of BSP beams.
8.4.2 Strain profile of section

In the nonlinear moment-curvature analysis of RC sections, the plane
section remaining plane assumption is made, such that the strain profile along
the depth of section is linear. By assigning the value of the neutral axis depth as
hna and the curvature as , the strain at any levels of the section can be
expressed in terms of these two parameters as
y = ( y hna )
(8.49)
In general, the strain profile can be expressed as a function of curvature and a

pivotal point ( y0 , y0 ) on the strain profile. Equation (8.49) can be re-written as,
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.
8.4.3 Modified section analysis

As shown in Section 8.3.3.2, the strain factor is dependent on the stiffness
of the connecting media km, which drops with the bolt stiffness as the applied
load acting on the beam increases. So the relationship between the strain
profiles of the RC beam and steel plates is not constant. So the solution
procedure adopted in typical section analysis has to be modified accordingly.
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.
Once the curvature satisfying horizontal equilibrium is solved, the strain

profile of the section is fixed and the resultant moment of the section can be
calculated by considering moment equilibrium, as given in Equation (8.46)
based on the layered model illustrated in Figure 8.11.
194
i =1
i =1
M = Aci ci ( yi hc ) + Ast st (d hc ) + Asc sc (d ' hc ) + Api pi ( yi hc ) .
(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
M c = Aci ci ( yi hc ) + Ast st (d hc ) + Asc sc (d ' hc ) ,
(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
Resultant stress and force block
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
Figure 8.11 Layered model adopted in modified section analysis
8.4.4 Program implementation

In order to verify 2-Alpha approach, a computer program has been
developed using commercial spreadsheet. The program is capable of solving the
moment-curvature response of the critical section of the span, hence estimating
the flexural stiffness and moment capacity of the span, as well as solving the
internal moment distribution between RC beam, steel plates and the axial force
couple. The program first derived the normalized slip profile of the span based
on the input parameters. Then the strain factor and curvature factor are solved
and the modified section analysis is carried out in each load step until the peak
moment is reached.
8.4.4.1 Material model

The compressive stress of concrete was estimated by the model proposed
by Attard and Stewart (1998). 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
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)
To account for local properties of concrete, the following relationships,

expressing the parameters Ec, co, fci, ci in terms of fco, (Su and Cheng, 2008)
were adopted in this study.
Ec = 7200( f cu )1/ 3 7200(1.25 f co )1/ 3
(8.62)
co = 700( f co / Ec ) 2 2.8( f co / Ec ) + 0.0059
(8.63)
f ci / f co = 1.77 0.4 ln( f co )
(8.64)
ci / co = 3.86 0.54 ln( f co )
(8.65)
The tensile stress developed in concrete is neglected in this program. On the

other hand, the typical elastic-plastic model is used for both reinforcement and
steel plates in both tension and compression.
A bi-linear load-slip relationship is adopted to model the anchoring bolts

in this study. Such model is justified by the bolt test conducted and described in
Chapter 3. Furthermore, the bi-linear model is flexible as it can easily be
changed to an elastic or elastic-plastic model by setting the post-yield stiffness
as the elastic stiffness and zero respectively. To specify the model, the notional
yield force and slip, and the ultimate force and slip, as illustrated in Figure 8.12
are required.
197
P
(Pu, su)
(Py, sy)
Figure 8.12 Load slip model of bolts in the current analysis
8.4.4.2 Program algorithm

The whole program can be divided into 2 parts, as illustrated in Figure
8.13. The first part is the input specification, where the information required to
define the case to be analyzed is provided. These include the loading conditions,
the properties of the materials used and the geometric properties of the
structures. The second part is the core program, which is responsible for solving
the outputs. 3 subroutines are included in the core program, they are 1)
Derivation of normalized slip profile, 2) Computation of strain factor and
curvature factor, 3) Modified section analysis.
The detail flow of the program, as illustrated in Figure 8.13, is described

step by step below:
Step 1: The material properties of reinforced concrete, steel plates and bolts are
defined.
Step 2: The loading conditions are defined. Three options, 1) single point load
applied at the middle (3-point-bending), 2) two point loads applied
symmetrically (4-point-bending) and 3) uniformly distributed load, can
be chosen. For the 4 point bending case, further specification on the
distance of loadings from supports is required.
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
Material properties RC, plates

and bolts defined
The geometric properties
along the beam defined
Core program
Deriving the slip profile

shape
The maximum slip for the
step defined
Solving slip profile
Solving strain and

curvature factor
The total bolt force induced

integrated
The total bolt force induced
integrated
Slip strain, plate strain
concrete strain at the
centroidal level of plate
solved
s = s+s
Modified section analysis
Initial trial of curvature of the

RC beam is assumed
Net horizontal force of
section calculated
Iteration for new
trial curvature
Equilibrium?
No
Yes
Moment computation
No
Moment
dropped?
Yes
Program ends
Figure 8.13 Program algorithm
201
8.5 Numerical Verification

The responses of the four strengthened specimens are simulated using the
spreadsheet program developed. The input properties of the specimens,
including the geometric properties of the beam and the material properties of
concrete, steel reinforcement and plates and connecting bolts are obtained from
material tests and are same as those presented in Chapter 3. Furthermore, the
normalized slip profile under the corresponding loading case and with the bolt
arrangement described in Chapter 3 is solved using the technique explained in
Section 8.3 and the detail derivation is given in Appendix 5. The analytical
flexural strength, full moment curvature response of the mid-span section and
the moment contribution by the RC beam, the steel plates and the axial force
couple are extracted and compared with the experimental results reported in
Chapter 4.
The analytical and experimental flexural strength of the four strengthened

specimens are summarized and compared in Table 8.1. According to Table 8.1,
the flexural strength estimated by 2-Alpha approach differs from the
experimental results by -3.4% to 6.5% with an average of 1.6%. This shows that
the program can closely predict the flexural strength of BSP beams with effect
of partial interaction properly considered. The small error is most likely due to
variations of material properties, which can hardly be perfectly controlled in
experiment.
202
Table 8.1 Comparison of predicted strength with experimental results

Flexural strength Mu (kNm)
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
The analytical moment-curvature responses of the four strengthened

specimens, predicted by 2-Alpha approach, are plotted in Figure 8.14. The
experimental results are super-imposed on Figure 8.14 for comparison.
According to Figure 8.14, the 2-Alpha approach provides a very accurate
prediction on the initial moment curvature response in the elastic stage prior to
the yielding of bottom reinforcement with errors less than 1%. The sharp
turning point of the moment curvature responses, corresponding to the instant
where the yielding of bottom reinforcement starts, is captured by the 2-Alpha
approach.
The analytical moment contribution of flexure of the RC beam, flexure of

the steel plates and the axial couple of the RC beam and steel plates, in terms of
percentage of total moment of section are plotted against the mid-span moment
in Figure 8.15. According to Figure 8.15, 2-Alpha approach provided a close
estimation for the percentage of moment taken up by the RC beam, the steel
plates and the axial force couple between the RC beam and the steel plates for
both strong and weak plate cases. This shows that 2-Alpha approach can be
applied to estimate the internal forces resisted by the components of BSP beams.
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
160
Moment (kNm)
120
80
Experimental
40
2-Alpha
DCBA
0
0
10
20
30
40
160
Moment (kNm)
120
80
Experimental
40
2-Alpha
DCBA
0
0
10
20
30
40
Figure 8.14 Comparison of moment curvature responses a) Specimen SBSP,

b) Specimen WBSP, c) Specimen WBWP and d) Specimen SBWP
205
100%
80%
Concrete flexure (Experiment)
Plate flexure (Experiment)
Axial couple (Experiment)
Concrete flexure (2-Alpha)

(DCBA)
(2-Alpha)
Plate flexure (DCBA)
Axial couple (2-Alpha)

(DCBA)
60%
40%
20%
0%
0
40
80
120
160
200
Moment (kNm)
100%
80%
60%
40%

(DCBA)
Plate flexure (2-Alpha)

(DCBA)

(DCBA)
20%
0%
0
40
80
120
160
Moment (kNm)
100%
80%
60%
40%
(2-Alpha)
Concrete flexure (DCBA)

(DCBA)

(DCBA)
20%
0%
0
40
80
120
160
Moment (kNm)
100%
80%
60%
40%

(DCBA)

(DCBA)

(DCBA)
20%
0%
0
40
80
120
160
Moment (kNm)
Figure 8.15 Comparison of moment contributed by various actions

Specimens a) Specimen SBSP, b) Specimen WBSP, c) Specimen WBWP
and d) Specimen SBWP
206
8.6 DESIGN EXAMPLE

Consider a 3600mm long simply-supported beam subjected to 2 equal
point loads, each applied at 1200mm from the support. The cross-section detail
is given in Figure 8.16. It is proposed to add steel plates on both side faces of
the RC beam in order to increase the flexural strength of the beam. A design
scheme, illustrated in Figure 8.17, is proposed. The detail of the scheme is
summarized as follows
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
Figure 8.16 Specimen to be strengthened in the example
207
150mm
250mm
tp
tp=6mm
400
37.5
37.5
3600
Figure 8.17 Bolt-plate arrangement of design example
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.
To analyze the strengthened BSP beam, 2-Alpha approach is carried out.

Properties of the RC cracked section
Axial stiffness EAc:
947583kN
Flexural stiffness EIc:
7512kNm2
Centroidal axis of the RC cracked section: 102mm from the top of the
beam
Properties of strengthening steel plates:
Axial stiffness EAp:
360000kN
Flexural stiffness EIp:
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
Considering that the mid-span (x=1800mm) is the most critical and

assuming that all bolts remain elastic, the strain factor and curvature can be
solved using the above longitudinal slip profile as,
209
Strain factor = 0.67
Curvature factor = 0.98
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.
Figure 8.18 Simulated moment-curvature response of mid-span section
210
A new analytical method has been developed in this chapter to consider

the effect of partial interaction in the analysis of BSP beams. Two new factors,
the strain factor and the curvature factor were introduced to quantify the amount
of interaction between the RC beam and the steel plates in longitudinal and
transverse direction respectively. These factors describe the separation of strain
profiles of the RC beam and steel plates. By establishing the relationship
between these two factors and the displacement compatibility of the RC beam
and the steel plates in longitudinal and transverse directions, the complicated
global compatibility problem can be reduced to a strain separation problem at
the critical section. The section analysis can then be carried out to predict the
response of the BSP beams.
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.
Previous studies on external strengthening techniques were reviewed.

These studies showed that bolted side-plated beam has several advantages over
other strengthening methods including preservation of ductility, guaranteed
structural integrity as a result from the use of mechanical anchors, prevention of
steel congestion and providing sufficient space for temporary support at the
soffit.
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.
In order to deepen the understanding on the behaviour of BSP beams, full

scale tests on four BSP beams were carried out. Beam specimens with strongbolt strong-plate, weak-bolt strong-plate, weak-bolt weak-plate and strong-
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.
To provide a simple and efficient analysis for practical use by engineers,

the 2-Alpha approach was developed. Two factors, the strain factor and
curvature factor were introduced to quantify the separation of strain profile of
the section with the maximum moment. The two factors were expressed in
terms of the geometric arrangements of the beam by compatibility consideration
between the RC beam, the steel plates and the connecting media. With the use
of these two factors, the global compatibility problem resulting from partial
interaction is reduced to the local strain separation problem at the section with
the maximum moment. The local strain separation problem can then be solved
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.
An original macro-finite-element model was developed to predict the

response of BSP beams. The RC beam and the steel plates were modelled as
two different elements interacted through discrete bolt groups, of which the
non-linear load-deformation relationship was established based on kinematic
hardening assumption of individual bolts. The macro-finite-element model was
used to demonstrate the non-linear slip response between the RC beams and the
steel plates. It was found that the longitudinal slip response became severe as
plastic hinge formed in the critical section, reducing the interaction between the
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.
The 2-Alpha approach has been developed to consider partial interaction in

the analysis of BSP beams. The complicated global compatibility problem has
been successfully simplified to a local strain separation problem by the strain
factor and the curvature factor. As a result, the response of BSP beams can be
solved by carrying out the modified section analysis. With the use of the 2Alpha approach, designers can consider the effect of partial interaction by
simple calculation and prevent over-estimation of flexural strength of BSP
beams. The internal force distribution, which reveals the effectiveness of a
strengthening scheme can also predicted accurately.
9.3 RECOMMENDATIONS FOR FUTURE RESEARCH
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.
The 2-Alpha approach has been developed to serve as a quick analysis of

BSP beams. However, its accuracy is dependent on the longitudinal slip profile
used. While the non-linearity of bolts and different loading arrangements are
considered, the gradual formation of plastic hinges is not considered in the
derivation of longitudinal slip profile. As observed, the presence of plastic
hinges results in the slip strain accumulation in the plastic hinges and eventually
lowers the interaction between the RC beam and the steel plates. More
theoretical studies on the effects of the plastic hinges and solution for
longitudinal slip with the presence of plastic hinges are recommended.
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)
lbd = LLVDT + lbd ,
(A.1.3)
lac = LLVDT + lac ,
(A.1.4)
lcd = LLVDT + lcd ,
(A.1.5)
219
where lab , lbd , lac , lcd are the readings of LVDTs between the nodes denoted
by the subscripts.
Using trigonometric relationships, the slips could be expressed as

sx = ( LLVDT + lab ) sin +
sy =
s =
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
Target plane of steel plate

Metal strip
Reference points
Figure A.1.1 Schematic diagram of rhombic arrangement of LVDTs in motion
220
APPENDIX 2 Non-linear stress-strain model of concrete
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)
To account for local properties of concrete, the following relationships,

expressing the parameters Ec, co, fci, ci in terms of fco, (Su and Cheng, 2008)
were adopted in this study.
Ec = 7200( f cu )1/ 3 7200(1.25 f co )1/ 3
(A.2.4)
co = 700( f co / Ec ) 2 2.8( f co / Ec ) + 0.0059
(A.2.5)
f ci / f co = 1.77 0.4 ln( f co )
(A.2.6)
ci / co = 3.86 0.54 ln( f co )
(A.2.7)
221
APPENDIX 3 Explicit form of flexural RC beam and steel plate element
The explicit form of a flexural steel plate element

c1
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)
The explicit form of a flexural RC element

a1
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
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
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
APPENDIX 4 Derivation of equations governing longitudinal slip profile

along beam span
In this section, the governing equations for the slip of bolts along a RC beam
segment, considering the displacement compatibility between RC, steel plate
and bolts are derived. The general solution forms of a beam segment are solved
and will be applied to assemble the normalized slip profile of beam spans
consisting of multiple segments.
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
A.4.2 Case specification

The segment of a span is defined as a length of the span where the geometric
arrangements of components are constant and the bolts are uniformly distributed.
The segment is subjected to an uniformly distributed load . The two boundary
of the segment are denoted as a and b and the dimension along the segment is
defined as x from left to right. The free body diagram and the boundary forces
and deformations are illustrated in Figure A.4.1.
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.4.3 Governing equations

The slip of bolts is equal to the difference between movement of the concrete
and plate component, expressed mathematically as,
sl ( x) = lp,hp ( x) lc,hp ( x) .
(A.4.1)
Differentiating Equation (A.4.1), we have

s( x) = hpp ( x) hcp ( x) .
(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)
where km is the stiffness of the media of connection per unit length.

The strain in steel and concrete at the centroidal level of plate can be expressed
as
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)
Considering moment equilibrium at section, we have

M ( x) = ( M c + M p + Fp h ) .
(A.4.6)
Assuming a linear relationship between the curvature of concrete and plate, we

have
M
M
= ,
EI p
EI c
(A.4.7)
where is the curvature factor and ranges from 0 to 1.

EI
Assuming that equals to unity and further substituting = p
EI c
, we have
M p = Mc .
(A.4.8)
Substituting (7) into (5) and rearranging, we have,

Mc =
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)
APPENDIX 5 Longitudinal slip profile of simple cases

A.5.1 Longitudinal slip profile of simply-supported BSP beams with
uniformly-distributed bolts under mid-point loading
Figure A.6.1 Simply-supported BSP beams with evenly-distributed bolts

under mid-point loading
When loaded by mid-point load, the slip response is symmetric about the midpoint of the beam as illustrated in Figure A.5.1. Consider the left-span of the
beam, the derivative of bending moment M(x) is equal to the shear force V =
0.5P. So Equation (8.1) can be rewritten as
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
The solution of Equation (A.5.1) has the general form
sl ( x) = Ae px + Be px + C ,
(A.5.2)
where A, B and C are arbitrary constants.

With the boundary condition sl (0) = sl ,max , s( Ls ) = 0 , s(0) = 0 , we have
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
Figure A.5.2 Simply-supported BSP beams with evenly-distributed bolts

under uniformly-distributed loading
For a simply supported beam subjected to uniformly distributed load, the

bending moment of the half span is given by
M ( x) = Ls x
x 2
2
(A.5.6)
where is the applied load per unit length.

Differentiate once, we have
M ( x) = Ls x .
(A.5.7)
So the governing equation can be expressed by substituting Equation (A.5.7)

into the governing equation of longitudinal slip as
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
The characteristic and particular solutions of Equation (A.5.8) are,
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
Thus the solution of equation (A.5.8) has the general form,

sl ( x) = Ae px + Be px + C ( Ls x) .
(A.5.10)
With the boundary condition s( Ls ) = 0 , s(0) = 0 , we have

1
e pLs
1
e
pLs
Ls A sl ,max
0 B = 0 .
1 C 0
(A.5.11)
Solving Equation (A.5.11) and substituting the solutions of A, B and C into

Equation (A.5.10), the longitudinal slip profile can be found as,
e p ( x Ls ) e p ( x Ls ) + p( Ls x)(e pLs + e pLs )
sl ( x) = sl ,max
e pLs e pLs + pLs (e pLs + e pLs )
(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
A simply supported beam is loaded by two symmetric point load at a distance X

from the supports as illustrated in the figure below. The beam is strengthened by
steel plates throughout the whole length of beam. Bolts are installed evenly on
the two shear spans, while no bolts are installed in the constant moment zone.
P
Span 2
Span 1
P
P
x
SFD
Segment 1
Segment 2
Figure A.5.3 Simply-supported BSP beams under two symmetric

point loadings with bolts only on shear spans
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.
Table A.5.1 Summary of longitudinal slip profile of span

km
V (x)
Slip profile
Segment 1
Constant
Constant
sl ( x) = A1e p1x + B1e p1x + C1
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)
Figure A.5.4 Longitudinal slip profile of simply-supported BSP

beams under two symmetric point loadings with bolts only on shear
spans
Consider the three boundary conditions of the span
( sl (0) = sl ,max , sl (0) = 0 , sl ( Ls ) = 0 ), together with the two compatibility
equation between the two segments, 5 equations are present.
1. Longitudinal slip at x=0 is equal to 1
A1e p1 (0) + B1e p1 (0) + C1 = 1
A1 + B1 + C1 = 1
(A.5.14)
2. Slip strain at x=0 is equal to 0

A1 p1e p1 (0) B1 p1e p1 (0) = 0
A1 B1 = 0
(A.5.15)
3. Longitudinal slip at x=Ls is equal to 0

A2 Ls + B2 = 0
(A.5.16)
4. Longitudinal slip continuity between segment 1 and segment 2

A1e p1 X + B1e p1 X + C1 A2 X B2 = 0
(A.5.17)
231
5. Slip strain continuity between segment 1 and segment 2

A1 p1e p1x B1 p1e p1 x A2 = 0
(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
APPENDIX 6 Derivation of curvature factor for simple cases (A.8.5)

A.6.1 Derivation of curvature factor for Simply-supported beam under
symmetrical four-point bending with bolts uniformly distributed on shear
spans only
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
Shear transfer zone
Shear transfer zone
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
Substituting Equations (A.6.7), (A.6.8) and (A.6.9) into the compatibility

condition (Equation 8.20), we have
12 ( EI ) p
2 2
13 2
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
Finite Elements in Analysis and Design, 40, pp425-448

13. Daniels, B.J., and Crisinel, M. (1993). Composite slab behavior and
strength analysis. part 1: calculation procedure. Journal of Structural
Engineering ASCE, 119(1), pp16-35.
14. Faella, C., Martinelli, E., and Nigro, E. (2002). Steel and concrete
composite beams with flexible shear connection: exact analytical expression
of the stiffness matrix and applications. Computers and Structures, 80,
pp1001-1009.
15. Fuchs, W., Eligehausen, R., and Breen, J.E. (1995). Concrete capacity
design (CCD) approach for fastening to concrete. ACI Structural Journal,
92(1), pp 73-94.
16. Foley, C.M., and Buckhouse, E.R. (1999). Method to increase capacity
and stiffness of reinforced concrete beams. Practical Periodical on Structural
Design and Construction, 4(1), pp36-42.
17. Foreman, R.T., and Rumpf, J.L. (1961). Static tension tests of compact
bolted joints. Transaction ASCE, 126(2), pp228-254.
18. Fisher, J.W., Ramseier, P., and Beedle, L.S. (1963). Strength of A440
steel joints fastened with A325 bolts. Publications IABSE, 23.
19. Gattesco, N. (1999). Analytical modeling of nonlinear behavior of
composite beams with deformable connection. Journal of Constructional Steel
Research, 52, pp195-218.
20. Harrison, H.B. (1980). Structural analysis and design, parts 1 and 2.
Pergamon Press: Oxford.
21. Hageman, L.A., and Young, D.M. (1981). Applied iterative methods.
Academic Press: New York.
22. Hilti Corporation (2005), Fastening Technology Manual, Issue 2005-06.
23. Ho, J.C.M., Kwan, A.K.H., Pam, H.J. (2003). Theoretical analysis of
post-peak flexural behaviour of normal- and high-strength concrete beams. The
Structural Design of Tall and Special Buildings, 12(2), pp109-125.
24. Johnson, R.P., and May, I.M. (1975). Partial-interaction design of
238
composite beams. The Structural Engineer, 53(8), pp305-311.

25. Johnson, R.P., and Molenstra, N. (1991). Partial shear connection in
composite beams in buildings. Proceedings of the Institution of Civil
Engineers Part 2, 91, pp679-704.
26. Jones, R., Swamy, N., and Ang, T. (1982). Under- and over-reinforced
concrete beams with glued steel plates. International Journal of Cement
Composites and Lightweight Concrete, 4(1), pp19-32.
27. Jones, R., Swamy, R.N., and Charif, A. (1988). Plate separation and
anchorage of reinforced concrete beams strengthened by epoxy-bonded steel
plates. The Structural Engineer, 66, pp85-94.
28. Kulak, G.L., Fisher, J.W., and Struik, J.H.A. (1987). Guide to design
criteria for bolted and riveted joints, second edition. John Wiley & Sons.
29. Lim, J.B.P., Nethercot, D.A. (2004). Stiffness prediction for bolted
moment-connections between cold-formed steel members. Journal of
Constructional Steel Research, 60(1), pp85-107.
30. Meier, U., Deuring, M., Meier, H., and Schwegler, G. (1993) Fibrereinforced-plastic (FRP) Reinforcement for Concrete Structures: Properties
and Applications (Nanni, A. (ed.)). Elsevier Science, The Netherlands.
31. Mroz, Z. (1967). On the description of anisotropic hardening. Journal of
the Mechanics and Physics of Solids, 15(3), pp163-175.
32. Newmark, N.M., Siess, C.P., and Viest, I.M. (1951). Test and analysis of
composite beams with incomplete interaction. Proceedings of the Society for
Experimental Stress Analysis, 9, pp75-92.
33. Nowak, P.S., and Hartmann, T.W. (1993). Eccentric connection design:
geometric approach. Journal of Structural Engineering, ASCE. 119(2), pp606618.
34. Nguyen, N.T., Oehlers, D.J., and Bradford, M.A. (1998). A rational
model for the degree of interaction in composite beams with flexible shear
connectors. Mechanics of Structures and Machines, 26(2), pp175-194.
35. Nguyen, N.T., Oehlers, D.J., and Bradford, M.A. (2001). An analytical
model for reinforced concrete beams with bolted side plates account for
239
longitudinal and transverse partial interaction. International Journal of Solids

and Structures, 38, pp6985-6996.
36. Oehlers, D.J., Nguyen, N.T., Ahmed, M., and Bradford, M.A. (1997a).
Partial interaction in composite steel and concrete beams with full shear
connection. Journal of Constructional Steel Research, 41(2), pp235-248.
37. Oehlers, D.J., Nguyen, N.T, and Bradford, M.A. (1997b). Transverse and
longitudinal partial interaction in composited bolted side-plated reinforced
concrete beams. Structural Engineering and Mechanics, 5(5), pp553-564.
38. Oehlers, D.J., Ahmed, M., Bradford, M.A., and Nguyen, N.T. (2000)
Retrofitting reinforced concrete beams by bolting steel plates to their sides.
Part 2: Transverse interaction and rigid plastic design. Structural Engineering
and Mechanics, 10(3), pp227-243.
39. Oehlers, D.J., and Seracino, R. (2004). Design of FRP and steel plated RC
structures, retrofitting beams and slabs for strength, stiffness and ductility.
Elsevier: Oxford.
40. Owens, G.W., Cheal, B.D. (1989). Structural steelwork connections,
London, Boston: Butterworths.
41. Ranzi, G., Bradford, M.A., and Uy, B. (2004). A direct stiffness analysis
of a composite beam with partial interaction. International Journal for
Numerical Methods in Engineering, 61, pp657-672.
42. Robinson, H., and Naraine, K.S. (1988). Slip and uplift effects in
composite beams. Composite Construction in Steel and Concrete. ASCE, New
York, pp487-497.
43. Roberts, T.M. (1985). Finite difference analysis of composite beams with
partial interaction. Computers and Structures, 21(3), pp469-473.
44. Roberts, T.M., and Haji-Kazemi, H. (1989). Strengthening of underreinforced concrete beams with mechanically attached steel plates.
International Journal of Cement Composites and Lightweight Concrete, 11(1),
pp21-27.
45. Royal Institution of Chartered Surveyors (2009). Research report:
Industrial Buildings Strategic Review of Issues Associated with Conversion
for Adaptive Re-use. Hong Kong.
240
46. Skeikh, S.A., and Yeh, C.C. (1992). Analytical moment-curvature

relations for tied concrete columns. Journal of Structural Engineering, ASCE.
118(2), pp529-544.
47. Smith, S.T., Bradford, M.A., and Oehlers, D.J. (1999a) Local buckling of
side-plated reinforced-concrete beams. I: theoretical study. Journal of
Structural Engineering ASCE, 125(6), pp622-634.
48. Smith, S.T., Bradford, M.A., and Oehlers, D.J. (1999b) Local buckling of
side-plated reinforced-concrete beams. II: experimental study. Journal of
Structural Engineering ASCE, 125(6), pp635-643.
49. Smith, S.T., Bradford, M.A., and Oehlers, D.J. (2000) Unilateral buckling
of elastically restrained rectangular mild steel plates. Computational
Mechanics, 26, pp317-324.
50. Smith, S.T., Bradford, M.A., and Oehlers, D.J. (2001) Buckling tests on
steel plates restrained at discrete points in the retrofit of reinforced concrete
beams. Proceedings of the Institution of Civil Engineers, Structures and
Buildings, 146(2), pp115-127.
51. Su, R.K.L., and Zhu, Y. (2005). Experimental and numerical studies of
external steel plate strengthened reinforced concrete coupling beams.,
Engineering Structures, 27(10), pp1537-1550.
52. Su, R.K.L., and Cheng, B. (2008). The effect of coarse aggregate size on
the stress-strain curves of concrete under uniaxial compression. Transactions
of Hong Kong Institution of Engineers, 15(3), pp33-39.
53. Surtees, J.O., Gildersleeve, C.P., and Watts, C.J. (1981). A general
tabular method for elastic and plastic analysis of eccentrically loaded fastener
groups. The Structural Engineer, 59A(6), pp202-208.
54. Teng, J.G., Chen, J.F., Smith, S.T., and Lam, L. (2002). FRP-strengthened
RC Structures. Wiley, Chichester.
55. Teng, J.G., Chen, J.F., Smith, S.T., and Lam, L. (2003). Behaviour and
strength of FRP-strengthened RC structures: a state-of-the-art review.
Proceedings of the Institution of Civil Engineers, Structures and Buildings,
156(1), pp51-62.
56. Wallaert, J.J., Fisher, J.W. (1965). Shear strength of high-strength bolts.
241
Journal of the Structural Division, ASCE, 91(3), pp99-124.

57. Ueda, T., Stitmannaithum, B., and Matupayont, S. (1991). Experimental
investigation on shear strength of bolt anchorage group. ACI Structural
Journal, 88(3), pp292-300.
58. Yam, L.C.P, and Chapman, J.C. (1968). The inelastic behaviour of
simply supported composite beams of steel and concrete. Proceedings of the
Institution of Civil Engineers, 41, pp651-684.
59. Zhang, S., Raoof, M., and Wood, L.A. (1995). Prediction of peeling
failure of reinforced concrete beams with externally bonded steel plates.
Proceedings of the Institution of Civil Engineers, 110(3), pp257-268.
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.
SIU, W.H., and Su, R.K.L. (2009). Load-deformation prediction for

eccentrically loaded bolt groups by a kinematic hardening approach,
Journal of Constructional Steel Research, 65, pp436-442.
4.
Su, R.K.L, and Siu, W.H. (2008). Displacement-based approach for

analysis of plated RC beams with bolt connections, Proceedings of The
4th International Conference on Advances in Structural Engineering and
Mechanics, 26-28 May 2008, Jeju, Korea, pp2288-2298.
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

Full Text

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Full Text

Hochgeladen von

Copyright:

Verfügbare Formate

Title

Flexural strengthening of reinforced concrete beams by bolted

The author retains all proprietary rights, (such as patent rights)

A thesis submitted in partial fulfillment of the requirements

Abstract of thesis entitled

Flexural Strengthening of Reinforced Concrete Beams by Bolted

The social awareness of the structural deterioration of reinforced concrete (RC)

Different methods for strengthening existing RC beams have been proposed in

A numerical technique for modeling the responses of BSP beams is then

A new concept for quantifying the amount of partial interaction by considering

Thanks are given to all technicians in the structural engineering laboratory of

LIST OF ABBREVIATIONS .. xvi

1.3 Scope of thesis ...

CHAPTER 2 LITERATURE REVIEW

2.2 External strengthening of reinforced concrete beams

2.3.2 Shear interaction between RC beam and steel plates .

2.5 Rigid plastics analysis of composite beam sections ..

CHAPTER 3 EXPERIMENTAL STUDY ON BOLTED SIDE-PLATED

BEAMS WITH PARTIAL INTERACTION

3.2.2 Specimen detail ...

3.2.6 Instrumentation ...

3.2.6.3 Rhombic arrangements of LVDTs ...

3.3.2 Specimen design and fabrication

CHAPTER 4 RESULT AND ANALYSIS OF EXPERIMENTAL

STUDY ON BOLTED SIDE-PLATED BEAMS

4.6 Various design controlling factors of BSP beams .

4.10 Longitudinal slip profile ... 85

CHAPTER 5 NON-LINEAR MODEL FOR BOLT GROUPS WITH

LINEAR-ELASTIC-PLASTIC BOLTS UNDER COMBINED INPLANE MOMENT AND SHEAR

5.2.1 Basic assumptions ...

5.3.1 Description of the methodology .

5.4.2 Simulation by BOGA ..

CHAPTER 6 NON-LINEAR MODEL FOR BOLT GROUPS WITH

NON-LINEAR KINEMATIC-HARDENED BOLTS UNDER

6.4 Numerical simulation by BOGAN . 128

CHAPTER 7 NON-LINEAR ANALYSIS OF BOLTED SIDE-PLATED

7.2.2 Formulation of the non-linear RC beam and steel plate

7.6 Slip responses of BSP beams .

7.6.3 Rotational slips

CHAPTER 8 TWO-ALPHA APPROACH FOR ANALYSING OF

BOLTED SIDE-PLATED BEAMS WITH PARTIAL INTERACTION

8.3.3.3 Strain factor of simple loading case .

8.3.4.1 Idealized transverse force model ..

8.4.2 Strain profile of section ... 192

8.4.4.1 Material model .

8.4.4.2 Program algorithm ...

9.3 Recommendations for future research ...

APPENDIX 1 Evaluation of slip using rhombic arrangement of LVDTs 219

Number of buildings aged over 30 years in Hong Kong 1

Mei Ho House, Shek Kip Mei Estate . 3

Typical arrangement of BSP beams a) Front view, and b)

Full shear connection full interaction analysis ... 24

Partial shear connection partial interaction analysis .. 24

Test setup a) Schematic diagram, and b) Actual arrangement 29

RC section detail of specimens .. 30

Bolt-plate arrangements of strengthened specimens .. 32

Measured stress-strain relationship of steel plates . 36

Dynamic set washer a) Diagrammatic illustration, and b)

Load-slip relationship of HAS-E 12 anchor bolts 37

Cross-section detail of anchor bolts ... 39

Cast-in aluminium tube .. 40

Figure 3.10 Strain gauges arrangement of specimens a) In the RC beam,

Failure modes of specimens a) SBSP and b) SBWP, c)

WBSP, and d) WBWP ...