Page 1
Copyright Undertaking
This thesis is protected by copyright, with all rights reserved.
By reading and using the thesis, the reader understands and agrees to the following terms:
1. The reader will abide by the rules and legal ordinances governing copyright regarding the use of the thesis.
2. The reader will use the thesis for the purpose of research or private study only and not for distribution or further reproduction or any other purpose.
3. The reader agrees to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
IMPORTANT
If you have reasons to believe that any materials in this thesis are deemed not suitable to be distributed in this form, or a copyright owner having difficulty with the material being included in our database, please contact [email protected] providing details. The Library will look into your claim and consider taking remedial action upon receipt of the written requests.
Pao Yue-kong Library, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong
http://www.lib.polyu.edu.hk
Page 2
RC BEAM-COLUMN JOINTS SEISMICALLY
RETROFITTED WITH SELECTIVE BEAM
WEAKENING AND LOCAL FRP STRENGTHENING
QINGKAI WANG
PhD
The Hong Kong Polytechnic University
2019
Page 4
The Hong Kong Polytechnic University
Department of Civil and Environmental Engineering
RC BEAM-COLUMN JOINTS SEISMICALLY
RETROFITTED WITH SELECTIVE BEAM
WEAKENING AND LOCAL FRP STRENGTHENING
QINGKAI WANG
A Thesis Submitted in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
September 2018
Page 6
I
CERTIFICATE OF ORIGINALITY
I hereby declare that this thesis is my own work and that, to the best of my
knowledge and belief, it reproduces no material previously published or written,
nor material that has been accepted for the award of any other degree or diploma,
except where due acknowledgement has been made in the text.
(Signed)
Qingkai WANG (Name of student)
Page 8
III
ABSTRACT
The capacity design method has been widely accepted in the design of
reinforced concrete (RC) frames to ensure that they have sufficient plastic
deformation capacities when subjected to a seismic attack. The key of this
method is to make columns stronger than the connected beams at a joint, thus
realizing the so-called Strong-Column-Weak-Beam (SCWB) hierarchy.
However, many existing RC frames, especially those designed according to
outdated design codes; do not meet the SCWB principle. Inadequate
consideration of the contribution of a cast-in-place slab to the hogging moment
capacity of a beam is a common cause. Conventional seismic retrofit methods
have limited effects in enhancing the seismic safety of these RC frames. Against
the above background, this thesis presents a systematic study into a new seismic
retrofit method that involves beam weakening and FRP (fibre reinforced
polymer) strengthening (the BWFS method). This method was proposed by the
author’s research group.
In the experimental part of the study, three different beam weakening
techniques were studied by testing 9 full-scale RC beam-column joints (under a
combination of constant axial loading and cyclic lateral loading), including: (a)
the slab slit (SS) technique, in which a transverse slit is cut in the slab at each
beam end; (b) the beam web opening (BO) technique, in which an opening is
cut in the beam web; and (c) the beam section reduction (SR) technique, in
which a deep transverse groove is cut on the soffit of the beam near the joint.
Page 9
IV
The test results show that (a) the SR technique reduces both the strength and
ductility of the specimen; (b) the BO technique leads to a ductile failure mode if
the opening size is sufficiently large; and (c) the SS technique has a small
negative effect on specimen ductility although it can effectively reduce the beam
strength. A combination of the SS and the BO techniques is shown to be an
efficient retrofit method in reducing the strength of a T-beam and enhancing the
ductility of the joint.
In the finite element (FE) analysis part, accurate three-dimensional (3D)
FE models for predicting the behaviour of T-beams with an opening and FRP
strengthening were developed. 3D FE models using either solid or shell
elements were both developed. The static analysis problem was regarded as a
dynamic problem and solved using the explicit centre-difference-method
(CDM). A few significant issues, such as the loading time, the damping scheme,
the computational time and the accuracy associated with the explicit dynamic
method, are discussed in depth. The 3D FE model built using shell elements was
then applied to study the issue of effective slab width of T-beams in a hogging
moment zone. Based on the results of parametric studies, new models for the
effective slab widths of T-beams of both interior and exterior beam-column
joints were proposed. Finally, 3D FE models for retrofitted RC joints were
developed and substantiated with test results; the verified FE model can be used
in further investigations of such seismically retrofitted RC beam-column joints.
Page 10
V
ACKNOWLEDGEMENTS
My PhD research could not have been successfully completed with only
my own effort. I am so lucky and grateful to have received a great deal of help
and support from many people. In this regard, I would like to express my
sincere thanks to them.
First of all, I wish to express my sincere gratitude to my supervisor,
Professor Jin-Guang Teng, a distinguished scholar, for accepting me as one of
his PhD students six year ago; for his patience, motivation and wisdom. Prof.
Teng’s rigorous attitude towards academic research and creative and unique
insight into many academic problems have demonstrated the essential attributes
that a good researcher should possess. His enlightening guidance helped me
successfully finish my PhD research. The method and attitude for doing
research, which I have learned from him, will definitely benefit me in my future
career and life.
My sincere thanks also go to my two co-supervisors, Prof. Guang-ming
Chen of South China University of Technology (SCUT) and Prof. Shi-shun
Zhang of Huazhong University of Science and Technology (HUST). Prof.
Zhang provided me with much constructive advice on my experimental work.
The FE model for FRP-strengthened RC beams previously developed by Prof.
Chen during his PhD study (also supervised by Prof. J.G. Teng) served as the
foundation of my numerical studies presented in this thesis. Prof. Chen also
gave me a great deal of help with the revision of this PhD thesis, in addition to
Page 11
VI
his enlightening guidance on my numerical research work. Their selfless
assistance helped me overcome many tough difficulties encountered in my PhD
study.
I would also like to express my thanks to both The Hong Kong Polytechnic
University and the Research Grants Council of the Hong Kong Special
Administrative Region for their financial support. Thanks are due to The Hong
Kong Polytechnic University for providing me with the research facilities.
I would like to give my special thanks to Mrs Anson for her great help with
the English languages aspects during the preparation of my thesis; my technical
writing skills have substantially improved as a result.
Special thanks also go to the technical staff of the laboratories of
Guangdong University of Technology (GDUT) and The Hong Kong Polytechnic
University (PolyU). They include but are not limited to Mr. Zhen-xiong Wang,
Mr. K.H. Wong, Mr. Y.H., Yiu and Mr. John Chan, who gave me valuable
assistance and advice for the my experimental work. Great thanks also go to
Prof. Li-juan Li, Dean of the School of Civil and Transportation Engineering for
making the laboratory facilities available for my experimental work.
Special thanks also go to many past and current members of Professor
Teng's research group: Dr. Guan Lin, Dr. Bing Fu, Dr. Yi-nan Yang, Dr.
Qiong-Guan Xiao, Dr. Bing Zhang, Dr. Jun-jie Zeng, Mr. Xue-fei Nie, Ms. Pang
Xie, Mr. Jie-kai Zhou, Mr. Pan Zhang and Prof. Jian-Guo Dai, not only for their
Page 12
VII
discussions and constructive comments, but also for their encouragement during
difficult times of my PhD study.
Last but not least, I would link to thank my family and girlfriend for their
constant understanding, support and encouragement throughout my PhD study.
Page 13
VIII
CONTENTS
CERTIFICATE OF ORIGINALITY ................................................................ I
ABSTRACT .................................................................................................... III
ACKNOWLEDGEMENTS ..............................................................................V
CONTENTS ..................................................................................................VIII
LIST OF NOTATIONS ............................................................................... XVIII
CHAPTER 1 INTRODUCTION ...................................................................... 1
1.1 BACKGROUD .............................................................................................. 1
1.1.1 Strong-Column-Weak-Beam Hierarchy in the Design of RC Frames .... 1
1.1.2 Problems in Realizing the Strong-Column-Weak-Beam Hierarchy ....... 2
1.1.3 Inadequacy of Existing Seismic Retrofit Methods ................................. 3
1.1.4 Effects of a Floor Slabs on the Flexural Capacity of a Supporting Beam
................................................................................................................ 4
1.1.5 Effects of Web Openings on the Performance of an RC Beam .............. 6
1.1.6 Effects of Slab Slits on the Performance of an RC frames ..................... 7
1.1.7 Proposed Seismic Retrofit Techniques ................................................... 7
Page 14
IX
1.2 RESEARCH OBJECTIVES ........................................................................... 8
1.3 OUTLINE OF THE DISSERTATION ........................................................... 9
1.4 REFERENCES ............................................................................................. 12
CHAPTER 2 LITERATURE REVIEW ......................................................... 23
2.1 INTRODUCTION ........................................................................................ 23
2.2 POTENTIAL PROBLEMS AND SEISMIC RETROFIT OF RC FRAMES
WITH CAST-IN-PLACE SLABS ............................................................... 24
2.2.1 Potential Problems in RC Frames with Cast-in-place Slabs ................. 24
2.2.2 Seismic Retrofit of RC Frames ............................................................. 29
2.3 EXPERIMENTAL STUDIES AND FE MODELLING OF RC BEAMS
WITH WEB OPENINGS ............................................................................ 32
2.3.1 Experimental Studies of RC Beams with Web Openings ...................... 32
2.3.1 FE Studies of RC Beams with Web Openings ...................................... 39
2.4 EFFECTIVE SLAB WIDTH OF RC FRAMES .......................................... 42
2.5 CONCLUDING SUMMARY ...................................................................... 44
2.6 REFERENCES ............................................................................................. 46
CHAPTER 3 EXPERIMENTAL PREPARATION AND SPECIMEN
Page 15
X
DETAILS .................................................................................. 67
3.1 INTRODUCTION ....................................................................................... 67
3.2 SPECIMEN DESIGN DETAILS ................................................................. 68
3.3 SPECIMEN CONSTRUCTION .................................................................. 72
3.4 MATERIAL PROPERTIES ......................................................................... 74
3.5 LOADING PROTOCOLS ........................................................................... 76
3.6 TEST SET-UP .............................................................................................. 77
3.7 INSTRUMENTATION ................................................................................ 78
3.7.1 Beam end Reaction Forces, Column Top Axial and Lateral loads ....... 79
3.7.2 Beam, Column and Joint Deformations ................................................ 79
3.7.3 Strain Gauges ........................................................................................ 85
3.8 REFERENCES ............................................................................................ 90
CHAPTER 4 EXPERIMENTAL STUDY OF THE SEISMIC
PERFORMANCE OF RETROFITTED RC
BEAM-COLUMN JOINTS................................................... 159
4.1 INTRODUCTION ..................................................................................... 159
4.2 FAILURE PROCESS AND FAILURE MODE ......................................... 159
Page 16
XI
4.2.1 Cracking of Beams, Columns, Slabs and Joint Panel in the First Stage
Loading ............................................................................................... 161
4.2.2 Yield Point and Peak Point .................................................................. 164
4.2.3 Failure Mode ....................................................................................... 166
4.3 HYSTERETIC BEHAVIOUR AND ENVELOP CURVES ....................... 169
4.3.1 Hysteretic Behaviour ........................................................................... 169
4.3.2 Envelop Curves ................................................................................... 174
4.4 DEFORMATION BEHAVIOUR ............................................................... 175
4.4.1 Strains in the Steel Bars ....................................................................... 175
4.4.2 Rotations of Column Sections ............................................................. 183
4.4.3 Strain in Stirrups at Joint Region ........................................................ 184
4.4.4 Strains in FRP Jackets ......................................................................... 185
4.5 DISCUSSIONS .......................................................................................... 187
4.5.1 Specimen Ductility .............................................................................. 187
4.5.2 Energy Dissipation Capacity ............................................................... 191
4.5.3 Equivalent Viscous Damping Ratio ..................................................... 193
4.5.4 Stiffness Degradation .......................................................................... 196
Page 17
XII
4.5.5 Plastic Hinge Lengths ......................................................................... 197
4.5.6 Deformation Components ................................................................... 201
4.6 CONCLUDING SUMMARY .................................................................... 206
4.7 REFERENCES .......................................................................................... 209
CHAPTER 5 THREE-DIMENSIONAL FINITE ELEMENT
MODELLING OF REINFORCED CONCRETE T-BEAMS
WITH A WEB OPENING WEAKENING AND
WITH/WITHOUT FRP SHEAR STRENGTHENING...... 299
5.1 INTRODUCTION ..................................................................................... 299
5.2 PROPOSED 3D FE MODELS .................................................................. 301
5.2.1 Modelling of concrete ......................................................................... 302
5.2.2 Definition of Damage Evolution ......................................................... 304
5.2.3 Modelling of Steel Reinforcement ...................................................... 305
5.2.4 Modelling of FRP Reinforcement. ...................................................... 306
5.2.5 Modelling of Bond Behaviour ............................................................ 307
5.2.6 Solution Strategy and Boundary Conditions ....................................... 311
5.3 VERIFICATION OF THE PROPOSED 3D FE MODEL ......................... 312
Page 18
XIII
5.3.1 Specimen Dimensions and Material Properties ................................... 312
5.3.2 Load versus Displacement Curves ...................................................... 315
5.3.3 Crack patterns ...................................................................................... 317
5.3.4 Energy Release Behaviour .................................................................. 318
5.4 PARAMETRIC STUDIES ON THE EFFECTS OF KEY FACTORS ...... 320
5.4.1 Element Size ........................................................................................ 320
5.4.2 Loading Duration ................................................................................ 321
5.4.3 Damping Coefficient β ........................................................................ 322
5.4.4 Single versus Double Precision ........................................................... 326
5.4.5 FRP Confinement Effect...................................................................... 327
5.5 COMPARISON BETWEEN 2D AND 3D FE MODELS .......................... 328
5.6 SHEAR DEGRDATION OF CRACKED CONCRETE ............................ 320
5.6.1 The Effect of Number of Critical Cracks ............................................ 320
5.6.2 The Effect of Maximum Cracking Strain ............................................ 321
5.6.3 The Effect of Coefficient n of the Power Law .................................... 321
5.6.4 Comparison between the DP and DP+BC Models .............................. 321
Page 19
XIV
5.7 CONCLUDING SUMMARY .................................................................... 333
5.8 REFERENCES .......................................................................................... 337
CHAPTER 6 3D SHELL MODELS FOR RC T-BEAMS .......................... 393
6.1 INTRODUCTION ..................................................................................... 393
6.2 PROPOSED 3D SHELL MODELS .......................................................... 394
6.2.1 3D Shell Models with a One-shell-layer Slab .................................... 395
6.2.2 3D Shell Models with a Two-shell-layer Slab .................................... 396
6.3 VERIFICATION OF 3D SHELL MODELS ............................................. 398
6.3.1 3D Shell models with One-shell-layer Slab ........................................ 399
6.3.2 3D Shell Models with a Two-shell-layer Slab .................................... 400
6.3.3 Efficiency and Accuracy of the 3D Shell Models ............................... 403
6.4 PARAMETRIC STUDIES ON THE EFFECTS OF SOME KEY FACTORS
.................................................................................................................. 404
6.4.1 Element Size ....................................................................................... 405
6.4.2 Loading Duration ................................................................................ 405
6.4.3 Damping Coefficient β ........................................................................ 406
6.5 3D SHELL MODELS FOR A T-BEAM WITH A WEB OPENING ......... 408
Page 20
XV
6.6 CONCLUDING SUMMARY .................................................................... 410
6.7 REFERENCES ........................................................................................... 411
CHAPTER 7 EFFECTIVE SLAB WIDTH OF REINFORCED
CONCRETE FRAMES WITH SLAB UNDER TENSION 429
7.1 INTRODUCTION ...................................................................................... 429
7.2 PROPOSED SIMPLIFIED 3D FE MODELS ............................................ 432
7.2.1 FE Model Details and Boundary Conditions....................................... 432
7.2.2 Ultimate State in Simulation ............................................................... 434
7.2.3 Equations for Effective Slab Width ..................................................... 434
7.3 PARAMETRIC STUDIES ......................................................................... 435
7.3.1 Effect of Stress-strain Models of Steel ................................................ 435
7.3.2 Effect of Bond Slip Behavior .............................................................. 436
7.3.3 Effect of Yield Stress of Steel Bars ..................................................... 436
7.3.4 Effect of Concrete Strength ................................................................. 437
7.3.5 Effect of Beam Length, Width and Height .......................................... 437
7.3.6 Effect of Beam Reinforcement Ratio .................................................. 438
7.3.7 Effect of Column Width ...................................................................... 438
Page 21
XVI
7.3.8 Effect of Slab Size. ............................................................................. 438
7.3.9 Effect of Slab Reinforcement Ratio .................................................... 439
7.3.10 Effect of Transverse Beam ................................................................ 439
7.4 DISCUSSIONS .......................................................................................... 440
7.5 CONCLUDING SUMMARY .................................................................... 442
7.6 REFERENCES .......................................................................................... 444
CHAPTER 8 THREE-DIMENSIONAL FINITE ELEMENT
MODELLING OF RETROFITTED RC
BEAM-COLUMN-SLAB JOINTS ....................................... 461
8.1 INTRODUCTION ..................................................................................... 461
8.2 THE PROPOSED 3D FE MODELS ......................................................... 462
8.3 PREDICTED RESULTS ............................................................................ 465
8.3.1 Load-displacement Curves.................................................................. 466
8.3.2 Crack Patterns ..................................................................................... 467
8.4 DISCUSSIONS .......................................................................................... 468
8.4.1 Confinement Effect of Joint Steel Stirrups ......................................... 468
8.4.2 Effect of Opening Shape ..................................................................... 469
Page 22
XVII
8.4.3 Effect of Longitudinal Slit Length ...................................................... 470
8.4.4 Prediction Errors .................................................................................. 471
8.5 CONCLUDING SUMMARY .................................................................... 471
8.6 REFERENCES ........................................................................................... 472
CHAPTER 9 CONCLUSIONS ..................................................................... 509
9.1 INTRODUCTION ...................................................................................... 509
9.2 EXPERIMENTAL STUDIES OF RETROFITTED RC JONITS .............. 511
9.3 THREE-DIMENSIONAL FINITE ELEMENT MODELLING OF RC
T-BEAMS AND JOINTS .......................................................................... 514
9.3.1 3D Solid FE Model of T-beams with a Web Opening. ........................ 514
9.3.2 3D Shell FE Models for T-beams ........................................................ 517
9.3.3 Effective Slab Width of RC Frames .................................................... 519
9.3.4 3D FE Modelling of Retrofitted RC Joints ......................................... 521
9.4 FURTHER STUDIES ................................................................................. 522
9.5 REFERENCES ........................................................................................... 524
Page 23
XVIII
LIST OF NOTATIONS
EDA : Dissipated energy of one hysteresis cycle;
effb : Effective slab width;
fb : Width of FRP;
jb : Joint width;
,rb ib : Length of right hand beam respective ith region measured by LVDTs;
wb : Beam web width;
D: Steel bar diameter;
ad : Maximum aggregate size;
td : Damage factor;
t bd : Vertical distance separating the two layers of 3D shell-2 models;
cE : Concrete elastic modulus;
fE : Elastic modulus of FRP sheet;
2sE : Modulus of hardening portion of steel stress-strain curve;
secE : Secant modulus of the compressive stress-strain curve of confined concrete;
totalE : Total energy for dynamic method;
IE : Internal energy;
Page 24
XIX
VE : Energy dissipated by viscous effects;
FDE : Energy dissipated through frictional effects;
KEE : Kinetic energy;
WE : External work applied to the structural system;
'cf : Concrete cylinder strength;
'ccf : Compressive strength of confined concrete;
ckf : Concrete standard compressive strength;
cuf : Cube compressive strength of concrete;
yf : Bar yield stress;
uf : Bar ultimate stress;
frpf : Tensile strength of FRP sheet;
syf : Yield stress of slab bars;
cF : Column shear force;
b rightF : Beam tip load of the right hand T-beam;
b leftF : Beam tip load of the left hand T-beam;
1mF : Peak positive load of a hysteresis cycle;
2mF : Peak negative load of a hysteresis cycle;
Page 25
XX
FG : Concrete fracture energy calculated by the area bounded by the strain-stress curve
of concrete under uni-axial tension;
fG : Interfacial fracture energy;
bh : Beam height;
jh : Joint height;
ch : Crack band width;
.rb ih : Respective distances between the top and bottom LVDTs for the right hand beam
respective ith region;
tc.1h : Respective distances between the right and left LVDTs for the upper column
respective ith region;
H : Distance between the inflection points of upper and lower columns;
cH : Length between the column-beam interface and the point of inflection.
c,elasticH : length of the column’s elastic region;
,e bI : Effective moment of inertia of cracked concrete area;
,g bI : Moment of inertia of the gross uncracked concrete area;
K : Stiffness matrix;
0l : Effective span of beam;
bL : Length of the beams;
Page 26
XXI
,b elasticL : Length of beam elastic part;
bL : Length of a beam from the centre of the joint panel;
cL : Length of a column from the centre of the joint panel;
CM : Sum of column moment capacities;
BM : Sum of beam moment capacities;
topcM : Upper column moment capacity of a beam-column joint;
bottomcM : Lower column moment capacity a beam-column joint;
leftr bM : Left R-section beam moment capacity of a beam-column joint;
rightr bM : Left R-section beam moment capacity of a beam-column joint;
M : Mass matrix;
n : A parameter for power law damage model;
1n : A parameter controlling the transition from the elastic branch to the plastic branch;
ss : Slip between steel bar and concrete;
s : Slip between FRP sheet and concrete;
0s : Slip when the bond stress reached max ;
ns : Clear distance between neighbouring beams;
bs : Distance between slab bars;
S(y) :Area moment under the y point towards the neutral axis;
Page 27
XXII
t : Slab thickness;
cV : Horizontal load acting on the end of the upper column;
tw : Crack opening displacement;
crw : Crack opening displacement at the point of concrete stress complete release;
tx : Line displacement along the X-direction of a node of the top slab shell layer of 3D
shell-2 models;
bx : Line displacement along the X-direction of a node of the bottom slab shell layer of
3D shell-2 models;
ty : Line displacement along the Y-direction of a node of the top slab shell layer of 3D
shell-2 models;
by : Line displacement along the Y-direction of a node of the bottom slab shell layer of
3D shell-2 models;
bz : Line displacement along the Z-direction of a node of the bottom slab shell layer of
3D shell-2 models;
tz : Line displacement along the Z-direction of a node of the top slab shell layer of 3D
shell-2 models;
s : Yield offset of steel stress-strain curve
: Mass-proportional damping coefficient;
c : Coefficient representing the initial tangent modulus of concrete;
: Stiffness-proportional damping coefficient;
Page 28
XXIII
w : Width ratio factor;
: Axial compressive strain;
1 : Strain measured by the a inclined LVDTs for joint panel;
2 : Strain measured by the a inclined LVDTs for joint panel;
c : Strain corresponding to cylinder strength;
cc : Strain at compressive strength of confined concrete;
y : Bar yield strain;
u : Bar strain at ultimate strain
p : Strain at the peak stress of concrete stress-strain curve;
plt : Equivalent plastic strain;
h : hardening beginning strain;
s : Steel axial strain;
x : Strain in the joint width direction;
y : Strain in the joint height direction;
: Axial compressive stress;
p : Peak stress of concrete stress-strain curve;
bi : Stress in the i slab bar;
Page 29
XXIV
maxb : Max stress of slab bars;
s : Steel axial stress;
0s : Yield stress of Ramberg Osgood (1943) model;
j : Angle of inclination of LVDT to the joint width direction;
,x t : Angular displacement along the X-direction of a node of the top slab shell layer of
3D shell-2 models;
,y t : Angular displacement along the Y-direction of a node of the top slab shell layer of
3D shell-2 models;
,z t : Angular displacement along the Z-direction of a node of the top slab shell layer of
3D shell-2 models;
,x b : Angular displacement along the X-direction of a node of the bottom slab shell layer
of 3D shell-2 models;
,y b : Angular displacement along the Y-direction of a node of the bottom slab shell layer
of 3D shell-2 models;
,z b : Angular displacement along the Z-direction of a node of the bottom slab shell layer
of 3D shell-2 models;
RBi : Average curvature of a beam ith region;
RBTi : Deformations measured by the LVDT located on the top surface of right hand
beam respective ith region;
Page 30
XXV
RBBi :Deformations measured by the LVDT located on the bottom surface of right hand
beam respective ith region;
1TCR : Deformations measured by the LVDT located on the right hand surface of right
upper column respective ith region;
1TCL : Deformations measured by the LVDT located on the left hand surface of right
upper column respective ith region;
C : Flexural strength ratio;
: Coefficient of effective flange width to real flange width;
e sbys : A parameter accounting for slab bar strength effect on the effective slab width of
a T-beam at a exterior joint;
e cs : A parameter accounting for concrete compression strength effect on the effective
slab width of a T-beam at a exterior joint;
e fbs : A parameter accounting for flange bar spacing effect on the effective slab width
of a T-beam at a exterior joint;
e tbw : A parameter accounting for transverse beam width on the effective slab width of a
T-beam at a exterior joint;
i sbys : A parameter accounting for slab bar strength effect on the effective slab width of
a T-beam at a interior joint;
i bbys : A parameter accounting for slab bar strength effect on the effective slab width of
a T-beam at a interior joint;
Page 31
XXVI
i cs : A parameter accounting for concrete compression strength effect on the effective
slab width of a T-beam at a interior joint;
i bh : A parameter accounting for slab beam height effect on the effective slab width of a
T-beam at a interior joint;
i bw : A parameter accounting for beam width effect on the effective slab width of a
T-beam at a interior joint;
i fbs : A parameter accounting for flange bar spacing effect on the effective slab width of
a T-beam at a interior joint;
i c : A parameter accounting for column weakening effect on the effective slab width of
a T-beam at a interior joint;
e c : A parameter accounting for column weakening effect on the effective slab width of
a T-beam at a exterior joint;
elastic beam , : Lateral displacement component due to beam elastic deformations;
,elastic column : Lateral displacement component due to column elastic deformations;
beam fixe end : Lateral displacement component due to tensile bar bond failure at
beam-column interface;
,hingebeam : Lateral displacement component due to beam hinge rotation;
,hingecol : Lateral displacement component due to column hinge rotation;
int jo : Lateral displacement component due to joint shear deformation;
Page 32
1
CHAPTER 1
INTRODUCTION
1.1 BACKGROUND
1.1.1 Strong-Column-Weak-Beam Hierarchy in the Design of RC Frames
The capacity design method, which aims to ensure there is sufficient plastic
deformation capacity in reinforced concrete (RC) structures, was originally
proposed by Park and Paulay (1975) in New Zealand. The key principle of this
method is to make the beams framing into a joint, weaker than the columns. The
objective is to prevent failure by means of the storey sway mechanism (i.e.
when plastic hinges occur at the column ends), commonly known as the
strong-beam-weak-column (SBWC) mechanism. In general, the success of this
mechanism is ensured if the flexural strength ratio C , which is the sum of the
designed column moment capacities, CM , divided by
BM , the sum of
the designed beam moment capacities at a same joint, is greater than 1, as
express below:
C C BM M (1.1)
In China, this mechanism was first applied in 1989 using code GBJ11-89
for the seismic deign of buildings. The flexural strength ratio C was initially
Page 33
2
set equal to 1.21 for RC frames of seismic grade 1. This value was later
increased to 1.7 in the recently updated edition of the code (GB-50011 2010).
In America, the SCWB mechanism was first adopted in the code ACI-318
1983: specifying the flexural strength ratio value to be 1.2. This value remains
the same today, but after 2002, ACI-318 (e.g. ACI-318 2014) requirements
replaced nominal values with design ones when calculating the moment
capacities of beams and columns.
The flexural strength ratio specified in Eurocode 8 (ENV 1998-1.3: 1995)
relates to the structural ductility class of the RC frame concerned. For instance,
the flexural strength ratio for RC frames of the highest ductility demand, is 1.35.
Those values are still the ones recommended today.
In New Zealand, initially, the flexural strength ratio was within the range
1.6 to 2.4 as specified in the design code NZS-3101(NZS-3101 1982). In 2006
(NZS-3101 2006), the value was modified, relating to two parameters and
, both of which are dynamic magnification factors ranging between 1.3 to 1.8
appropriate to the frame concerned.
1.1.2 Problems in Realizing the Strong-Column-Weak-Beam Hierarchy
The SCWB approach is widely adopted in the design of RC frame
structures subjected to seismic loading, because of its efficient
energy-dissipating capacity. However, many existing RC frames do not meet
Page 34
3
SCWB criterion as most were designed using outdated codes, and have been
found to fail in adequately estimating the moment capacity of beams. A major
reason for this phenomenon lies in the neglecting, or downplaying, of the
contribution from the cast-in-place slab in tension, to the negative moment
capacity of the beam at the joint. This is a common characteristic of existing
Chinese codes concerned with the seismic design of RC structures (Lin et al.
2009; Zhang 2009). The fact that most RC frames failed during the 2008
Wenchuan earthquake (Chinese Academy of Building Research 2008) in
Sichuan Province, China, as a result of the storey sway mechanism (i.e. columns
fail before beams at a frame joint), has been attributed to the inadequacies of the
previous versions of the Chinese code (e.g., GB-50011 2002). Even though the
newly-revised version (GB-50011 2010), which came into practice in Dec. 2010,
takes into account the benefits of a cast-in-place slab to beam flexural capacity,
vague definition of the effective slab width contributing to the negative beam
moment capacity can also lead to an underestimation of slab contribution. Some
factors (e.g. in-fill walls, over-reinforced beam ends) can enhance the moment
capacities of beams and some (e.g. axial compression ratio, bi-directional
seismic action) can decrease the moment capacities of columns. Thus, the
specified flexural strength ratio in earlier codes could be inadequate and fail to
guarantee a beam sway failure mode.
1.1.3 Inadequacy of Existing Seismic Retrofit Methods
Seismic retrofit interventions are in great demand to enhance the seismic
Page 35
4
safety of RC frames, but the SCWB mechanism condition can still be
compromised. To strengthen the columns is an easy and obvious option. Widely
used column strengthening techniques include: (1) concrete jacketing (e.g.
Thermou et al. 2007); (2) steel jacketing (e.g. Xiao and Wu 2003); and (3)
fibre-reinforced polymeric (FRP) jacketing (Teng et al. 2002). Cost-effective
seismic retrofitting is to enhance columns in both strength and ductility with
FRP jackets. However, this process can hardly change a storey sway failure
mechanism to a beam sway mechanism, especially when the columns are
non-circular in sectional shape because the strength improvement in this case
may not be sufficient to meet SCWB criterion. Even when the columns are, in
fact, sufficiently enhanced, failure may possibly occur at the beam-column
connections and the foundations, which are more complicated to be
reinforced/strengthened. Thus, a more effective seismic retrofit method is to
weaken the flexural capacities of T-shape beams, especially when under
negative bending.
1.1.4 Effects of a Floor Slabs on the Flexural Capacity of a Supporting
Beam
Numerous experimentally based research has conclusively shown that a
cast-in-place slab in tension significantly increases the negative flexural
capacity of a beam (e.g. Ehsani and Wight 1985; Durrani and Wight 1987;
Pantazopoulou and Moehle 1990; Pantazopoulou and French 2001; Zerbe and
Durrani 1990; Guimaraes et al 1992; Siao 1994; LaFave and Wight 1999; Shin
Page 36
5
and LaFave 2004a, 2004b; and Canbolat and Wight 2008). The extent of the
contribution increases with increased lateral drift ratio (Durrani and Wight
1987). The contribution of a cast-in-place slab significantly reduce flexural
strength ratio (e.g. by 30%). Thus, the flexural strength ratios specified in many
existing design codes are insufficient for ensuring a beam sway mechanism.
Extensive FE studies carried out by Chinese scholars (e.g., Guan and Du
2005, Lin et al. 2009, Gao and Ma 2009, Tao 2010, Yang 2010, Chen 2010, Guo
2012) have investigated the effect of cast-in-place floor slabs on beam moment
capacity. In these studies, the RC frame structures simulated followed the
requirements specified in the Chinese code at the time when the studies were
carried out. Pushover analyses (Guan and Du 2005, Gao and Ma 2009, Yang
2010, Guo 2012) or elastic-plastic time history analyses (Lin et al. 2009, Tao
2010, Yang 2010, Chen 2010) were conducted to evaluate the effects of floor
slabs (Guan and Du 2005, Lin et al. 2009, Gao and Ma 2009, Tao 2010, Chen
2010, Guo 2012) and of different effective flange widths on the overall response
of RC frames. The simulation results indicated that cast-in place floor slabs
significantly improved the negative moment capacities of beams and frames,
designed according to Chinese seismic design code, all probably fail as a
non-ductile storey sway mechanism.
Various design codes take the effects of a cast-in-place slab in tension into
account by specifying an effective slab width in calculating the T-shaped beam
capacity (ACI-318 2008; NZS-3101 2006; Eurocode-8 2005). For instance, in
Page 37
6
ACI-318 2008 and the New Zealand code NZS-3101 2006, the effective width
is specified as the smallest value of 1/4 of the beam span, 16 times the flange
thickness plus the beam web width, or the beam web plus half of the clear
distance between two adjacent beam webs.
1.1.5 Effects of Web Openings on the Performance of an RC Beam
Many studies since 1967 have explored the behaviour and design of a
rectangular or T-beam with circular or rectangular openings in the web for the
passage of utility ducts and pipes (e.g., Nasser et al. 1967; Mansur et al. 1999;
Pool and Lopes 1986; Hasnat and Akhtaruzzaman 1987; Kennedy and Abdalla
1992; Mansur 1998; Tan et al. 2001). These studies were concerned with
minimizing the effects of web openings in beams by surrounding the openings
with steel reinforcement strengthening. Their findings revealed that web
openings can significantly reduce the shear and flexural capacities of beams.
Five studies (Mansur et al. 1999; Abdalla et al. 2003; Maaddawy and
Sherif 2009; Pimanmas 2010, Nie 2018) studied the effects of creating an
opening in an existing beam and explored the efficiency of using bonded FRP as
a strengthening measure. The study of Mansur et al. (1999) and Nie (2018) were
concerned with T-beams. All these studies found that FRP strengthening, either
by the bonded FRP U jackets/full wraps (Abdalla et al. 2003; Maaddawy and
Sherif 2009, Nie 2018) or near-surface mounted FRP bars at corners (Pimanmas
2010), could offset the significant shear strength reduction caused by the
Page 38
7
openings and effectively control shear
1.1.6 Effects of Slab Slits on the Performance of an RC frames
Zhang et al. (2011), Wang et al. (2012) and Zhang (2013) investigated the
effects of slab slits (referred to as SS hereafter) on the overall performance of
RC frames. Through numerical analysis of RC joints and frames, it was found
that the SS technique increased the probability of achieving an SCWB
mechanism (Zhang et al. 2011; Wang et al. 2012; Zhang 2013). In addition, this
retrofit measure performed better than FRP column strengthening. (Zhang
2013).
1.1.7 Proposed Seismic Retrofit Techniques
Local weakening or retrofitting for seismic design purposes is commonly
practised. For steel structures, a dog-bone design is adopted as a typical
weakening technique for new structures and, for existing structures, a seismic
retrofit to ensure the SCWB strength criterion (Popov et al. 1998). For RC
structures, local weakening by material removal to meet seismic concerns was
discussed in a preliminary and general manner at FEMA-356 (2000). Cutting
out the bottom longitudinal steel reinforcements, as a seismic retrofit technique
for exterior beam-column joints, was investigated by Pampanin (2006) and Kam
et al. (2009). However, cutting out bottom bars is insufficient to offset the
contribution of a cast-in-place slab in tension. Recently, a novel seismic retrofit
Page 39
8
method called “beam-end weakening in combination with FRP strengthening”
method (BWFS method), was proposed by Teng et al. (2013). To weaken the
flexural capacities of the T-beams at a joint, particularly when the flange (i.e. the
cast-in-place slab) is in tension, is a primary measure.
To explore more suitable seismic retrofit techniques, three different
techniques as shown in Figure 1.1 proposed by Teng et al. (2013) were used for
this study: (a) separation of the slabs in the corner region from each supporting
beam by cutting a slit (including severing the steel bars crossing the slit)
between them (i.e. the slab slit (SS) technique); (b) drilling a web opening at the
beam end (i.e. the beam opening (BO) technique); (c) the beam section
reduction (SR) technique (e.g., creating a deep transverse groove (TG) on the
soffit of the beam near the joint). The latter two techniques are also combined
with associated strengthening measures (i.e., bonded FRP) to satisfy
serviceability and limit state requirements. The overall strategy of the proposed
method thus either applies the SS technique or the BWFS method when
retrofitting existing RC frames.
1.2 RESEARCH OBJECTIVES
This PhD research study aims to evaluate the effects of the retrofit
techniques proposed by Teng et al. (2013) on the achievement of ductile beam
sway mechanisms for RC frames subjected to seismic loading. The main
objectives of this study are as follows:
Page 40
9
(1) To evaluate the efficiency of the three retrofit techniques by testing RC
beam-column joints under combined vertically constant axial loading and
horizontally cyclic loading. Although much research has concerned with the
local weakening of steel structures (i.e., dog-bone design ) in ensuring
SCWB hierarchy (Popov et al. 1998), few experiments relating to beam
local weakening, especially the newly proposed BWFS method, have been
conducted on RC joints,
(2) To develop advanced 3D finite element (FE) models for RC T-beams with a
web opening, with or without FRP strengthening, also for RC beam-column
joints retrofitted by the three proposed techniques.
(3) To conduct in-depth study on the issues of effective slab width of RC frames
based on extensive numerical investigations using a developed advanced FE
model.
1.3 OUTLINE OF THE DISSERTATION
The PhD dissertation is composed of the following chapters.
Chapter 1 presents the general background and research scope of the
present study.
Chapter 2 is an in-depth literature review of seismic
performance/behaviour of RC frames, RC beam-column joints with a
Page 41
10
cast-in-place slab, beams with web openings. Both experimental studies and
numerical investigations were covered by the review.
In Chapter 3, details of the test program, and the process of preparing the
test specimens are presented. A total of 9, RC beam-column joints, divided into
two groups, one including a control RC joint and four retrofitted ones and
another one including a control RC joint and three retrofitted ones, were tested.
Test setup and boundary conditions for test specimens are presented. The layout
and functions of strain gauges and linear variable differential transformers
(LVDTs) are also described in detail. The retrofit procedure, including the
processes of creating web openings, transverse grooves and slab slits, and
applying FRP strengthening, is also introduced in detail.
Chapter 4 presents the experimental results of the 9 beam-column joints.
The failure process and failure modes, hysteretic behaviour are presented and
compared. Energy dissipation capacity, ductility, equivalent visual damping
ratio, stiffness degradation, plastic hinge length, beams and column deformation
contributions of the RC joints are discussed.
In Chapter 5, a 3D FE model using solid elements is developed for
T-beams with a web opening. A dynamic approach, using explicit time
integration method is adopted and extensive parametric studies are carried out to
investigate the effects of key factors of the dynamic approach (i.e. loading time
duration and damping schemes). The proposed 3D FE model is verified by the
Page 42
11
RC specimens tested by Nie (2018), and its advantages over a 2D FE model are
discussed.
In Chapter 6, several 3D shell models (using shell elements for concrete) of
T-beams are proposed and assessed by the proposed 3D solid mode (using solid
elements for concrete) and test results. The predicted load-displacement curves
and crack patterns are used for in the assessment. The computational time is also
taken into consideration. The most efficient one is suggested based on accuracy
and computational efficiency.
In Chapter 7, the most effective 3D shell model is used in studying
effective slab width. Parameters including beam length, width, height, bar
reinforcement ratio, slab width, thickness, bar spacing, yield stress of steel bars,
transverse beam height, width, stirrup spacing and column width, are considered
in parametric studies. The effects of the bond-slip relationship between
longitudinal bars and concrete are also assessed. Two types of stress-strain
models for steel bars are used and assessed. Finally, two formulas are proposed
for the effective slab widths at the interior and exterior joints of RC frames,
respectively.
Chapter 8 presents a numerical investigation of RC beam-column joints.
Based on the 3D FE models developed in Chapter 5, 3D FE models for RC
joints retrofitted by the proposed techniques are proposed and accessed.
Load-displacement responses and crack patterns of the RC joints are compared
Page 43
12
and discussed.
Finally, conclusions of this dissertation and an outline of the further
research needed are presented in Chapter 9.
1.4 REFERENCES
ABAQUS (2012). ABAQUS Analysis User's Manual (Version 6.12), Dassault
Systems SIMULIA Corporation, Providence, Rhode Island, USA.
ACI 318 (1983). Building Code Requirements for Structural Concrete and
Commentary (ACI 318-83), ACI Committee 318, American Concrete
Institute, Farmington Hills, MI.
ACI 318 (2014). Building Code Requirements for Structural Concrete and
Commentary (ACI 318-14), ACI Committee 318, American Concrete
Institute, Farmington Hills, MI.
Canbolat, B.B. and Wight, J.K. (2008). “Experimental investigation on seismic
behavior of eccentric reinforced concrete beam-column-slab connections”,
ACI Structural Journal, 105(2), 154-162.
Chen, X.B. (2010). The Effect of Floor Slab and Infill Walls on the Seismic
Behavior of Reinforced Concrete Frames, Master degree thesis: Fuzhou
University, China. (in Chinese)
Page 44
13
Chinese Academy of Building Research (2008). Photo Collection of 2008
Wenchuan Earthquake Damage to Buildings, China Architectural &
Building Press, Beijing, China (in Chinese).
Durrani, A.J. and Wight, J.K. (1987). “Earthquake resistance of reinforced
concrete interior connections including a floor slab”, ACI Journal, 84(5),
400-406.
Ehsani, M.R. and Wight, J.K. (1985). “Effect of transverse beams and slab on
behavior of reinforced concrete beam to column connections”, ACI Journal,
82(2), 188-195.
Eurocode 8 (2004). Design of Structures for Earthquake Resistance – Part 1:
General Rules, Seismic Actions and Rules for Buildings (EN 1998-1: 2004),
CEN, Brussels.
FEMA-356 (2000). Pre-Standard and Commentary for the Seismic
Rehabilitation of Buildings. Federal Emergency Management Agency,
Washington, D.C.
Gao, Z.R. and Ma, Q.L. (2009). “Effect of cast-in-place slab on column-beam
strength ratio in frame structure”, Shanxi Architecture, 35(17), 1-3 (in
Chinese).
GBJ11-89 (1989). Code for Seismic Design of Buildings, Architectural &
Page 45
14
Building Press, Beijing, China. (in Chinese)
GB-50011 (2001). Code for Seismic Design of Buildings, Architectural &
Building Press, Beijing, China (in Chinese).
GB-50011 (2010). Code for Seismic Design of Buildings, Architectural &
Building Press, Beijing, China (in Chinese).
Guan, M.S. and Du, H.B. (2005). “Pushover analysis of effect of casting slab on
RC frame structure”, Earthquake engineering and engineering vibration,
5(5), 117-123. (in Chinese)
Guimaraes, G.N., Kreger, M.E. and Jirsa, J.O. (1992). “Evaluation of joint-shear
provisions for interior beam-column-slab connections using high-strength
materials”, ACI Structural Journal, 89(1), 89-98.
Guo, L. (2012). Research on RC Structures with Cast-in-Place Slab, Master
degree thesis: Beijing University of Technology, China. (in Chinese)
Hasnat, A., and Akhtanizzamam, A. A. (1987), "Beams with small rectangular
opening under torsion, bending, and shear", Journal of Structural
Engineering, 113(10), 2253-2270.
Hawileh, R., El-Maaddawy, T., and Naser, M. (2012). “Nonlinear finite element
modeling of concrete deep beams with openings strengthened with
externally-bonded composites”, Materials & Design (DOI:
Page 46
15
10.1016/j.matdes.2012.06.004, 378-387).
Kennedy, J.B. and Abdalla, H. (1992). “Static response of prestressed girders
with openings”, Journal of Structural Engineering, ASCE, 118(2), 488-504.
LaFave, J.M. and Wight, J.K. (1999). “Reinforced concrete exterior wide
beam-column-slab connections subjected to lateral earthquake loading”,
ACI Structural Journal, 96(4), 577-587.
Lin, X., Pan, P., Ye, L., Lu, X. and Zhao, S. (2009). “Analysis of the damage
mechanism of a typical RC frame in Wenchuan earthquake”, China Civil
Engineering Journal, 42(5), 13-20 (in Chinese).
Lu, X.Z., J.G. Teng, Ye, L.P. and Jiang, J.J. (2005). “Bond-slip models for FRP
sheets /plates bonded to concrete”, Engineering Structures, 27(6), 920-937.
Maaddawy, T. and Sherif, S. (2009). “FRP composite for shear strengthening of
reinforced concrete deep beams with openings”, Composite Structures,
89(1), 60-69.
Maaddawy, T. and El-Ariss, B. (2012). “Behavior of concrete beams with short
shear span and web opening strengthened in shear with CFRP composites”,
Journal of Composites for Construction, ASCE, 16(1), 47–59.
Mansur, M.A. (1998). “Effect of opening on the behaviour and strength of R/C
beams in shear”, Cement and Concrete Composites, 20(6), 477-486.
Page 47
16
Mansur, M.A., Tan K.H. and Wei, W. (1999). “Effects of creating an opening in
existing beams”, ACI Structural Journal, 96(6), 899-906.
Nasser, K. W., Acavalos, A., and Daniel, H. R. (1967), "Behavior and design of
large openings in reinforced concrete beams", In Journal Proceedings (Vol.
64, No. 1, pp. 25-33).
Nie, X.F. (2018). Behavior and Modeling OF RC Beams with an
FRP-Strengthened Web Opening, Doctoral degree thesis: The Hong Kong
Polytechnic University.
NZS-3101 (1982). Code of Practice for the Design of Concrete Structures,
Standards New Zealand, Wellington, New Zealand.
NZS-3101 (2006). Concrete Structures Standard, Standards New Zealand,
Wellington, New Zealand.
OpenSees (2009). Open System for Earthquake Engineering Simulation, Pacific
Earthquake Engineering Research Center, University of California at
Berkeley, http://opensees.berkeley.edu.
Pantazopoulou, S.J. and Moehle J.P. (1990). “Identification of effect of slabs on
flexural behavior of beams”, Journal of Engineering Mechanics, ASCE,
116(1), 91-106.
Pantazopoulou, S.J. and French, C.W. (2001). “Slab participation in practical
Page 48
17
earthquake design of reinforced concrete frames”, ACI Structural Journal,
98(4), 479-489.
Pimanmas, A. (2010). “Strengthening R/C beams with opening by externally
installed FRP rods: Behavior and analysis”, Composite Structures, 92(8),
1957-1976.
Pool, R. S., and Lopes, R. (1986), "Cyclically loaded concrete beams with web
openings", In Journal Proceedings, Vol. 83, No. 5, pp. 757-763.
Popov, E.P., Yang, T.S. and Chang, S.P. (1998). “Design of steel MRF
connections before and after 1994 Northridge earthquake”, Engineering
Structures, 20(12), 1030-1038.
Shin, M. and LaFave, J.M. (2004a). “Seismic performance of reinforced
concrete eccentric beam-column connections with floor slabs”, ACI
Structural Journal, 101(3), 403-412.
Shin, M. and LaFave, J.M. (2004b). “Reinforced concrete edge
beam-column-slab connections subjected to earthquake loading”, Magazine
of Concrete Research, 55(6), 273-291.
Siao, W.B. (1994). “Reinforced concrete column strength at beam/slab and
column intersection”, ACI Structural Journal, 91(1), 3-8.
Tan, K.H., Mansur, M.A., and Wei, W. (2001). “Design of reinforced concrete
Page 49
18
beams with circular openings”, ACI Structural Journal, 98(3), 407-415.
Teng, J.G., Chen, J.F., Smith S.T. and Lam L. (2002). FRP-Strengthened RC
Structures, John Wiley and Sons Ltd, UK, November, 245pp.
Teng, J.G., Zhang, S.S., Jing, D.H., Nie, X.F. and Chen, G.M. (2013). “Seismic
retrofit of RC frames through beam-end weakening in conjunction with
FRP strengthening”, Proceedings of the 4th Asia-Pacific Conference on
FRP in Structures (APFIS 2013), 1-8.
Thermou, G.E., Pantazopoulou, S.J., Einashai, A.S. (2007). “Flexural behavior
of brittle RC members rehabilitated with concrete jacketing”, Journal of
Structural Engineering, ASCE, 133(10), 1373-1384.
Wang, X.G., Shan, M.Y., Ge, N. and Shu, Y.P. (2012). “Finite element analysis
of efficiency of slot-cutting around RC frame joint for ‘strong column and
weak beam’”, Journal of Earthquake Engineering and Engineering
Vibration, 32(1), 121-127 (in Chinese).
Xiao, Y. and Wu, H. (2003). “Retrofit of reinforced concrete columns using
partially stiffened steel jackets”, Journal of Structural Engineering, ASCE,
129(6), 725-732.
Yan, Y.L. (2010). A Study on Eurocodes about the RC Elements Design and
Comparison between Eurocodes and Chinese Codes, Master degree thesis:
Page 50
19
China Academy of Building Research, China. (in Chinese).
Ye, L.P., Qu, Z., Ma, Q.L., Lin, X.C, Lu, X.Z. and Pan, P. (2008). “Study on
ensuring the strong column-weak beam mechanism for RC frames based on
the damage analysis in the Wenchuan earthquake”, Building Structure,
38(11), 52-67 (in Chinese).
Zerbe, H.E. and Durrani, A.J. (1990). “Seismic response of connections in
two-bay reinforced concrete frame subassemblies with a floor slab”, ACI
Structural Journal, 87(4), 406-415.
Zhang, J. (2013). Research on Efficiency of Slot-cutting around Frame Joint for
“Strong Column and Weak Beam" under Earthquake Action, Master degree
thesis: Hunan University, China (in Chinese).
Zhang, Y.P., Hao, Z.J., Shan, M.Y. and Ge, N. (2011). “Research on anti-seismic
performance for reinforced concrete frame joint with slot around”, Building
Science, 27(9), 7-11 (in Chinese).
Page 51
20
(a) The slab slit (SS) technique
(b) The beam opening(BO) technique
lslsls
ls
sl l s
l s slSlitsColumn
Beam
ld ldComplete FRP wrap
NSM FRP strips
FRP U-jacket
Opening
ol
ho
FRP U-jacket
Page 52
21
(c) The section reduction (SG) technique
Figure 1.1 The three seismic retrofit techniques (Teng et al. 2013)
ld ld
Gap
ol
ho
Gap
olFRP U-jacket
Remaining steel bars Remaining steel bars
ho
Groove Groove
Page 54
23
CHAPTER 2
LITERATURE REVIEW
2.1 INTRODUCTION
This chapter presents literature review on relevant topic of the present PhD
research program. The chapter is organized as follows: experimental and finite
element (FE) studies on RC frames/joints with cast-in-place slabs are first
presented with the aim of pointing out the potential problems existing in those
reinforced concrete (RC) frames. Corresponding seismic retrofit methods for
those RC frames are then presented. As given in Chapter 1: creating a beam
opening is a proposed retrofit technique for RC frames with cast-in-place slabs,
hence existing experimental and numerical RC beam studies, in particular, RC
T-section beams (T-beams) with a web opening are reviewed. To better
seismically retrofit those RC frames, the contribution of cast-in-place slabs to a
beam negative moment capacity need to be quantitatively determined. Finally
effective slab width studies are reviewed, both experimentally and by means of
FE analysis
It should be noted that if not otherwise stated, all equations presented in
this chapter and the remainders of this thesis are written in terms of standard SI
unit designations (i.e. MPa and mm)
Page 55
24
2.2 POTENTIAL PROBLEMS AND SEISMIC RETROFIT OF
RC FRAMES WITH CAST-IN-PLACE SLABS
The strong-column-weak-beam (SCWB) mechanism requires the flexural
strength ratio C larger than 1.0. However, the ignorance or underestimation of
the cast-in-place slabs contribution to the negative moment capacity of beams of
RC frames, designed according to previous seismic design codes, might lead to
storey sway failure when subjected to seismic loading. A review of existing
studies, including both experimental and numerical investigations, on the effect
of cast-in-place slabs on the capacity of RC beams and corresponding seismic
retrofit techniques are given below.
2.2.1 Potential Problems in RC Frames with Cast-in-place Slabs
In the early 1980s, researchers from the United States and Japan (Durrani
and Wight 1982, Otani et al. 1984, JTCC 1988) conducted experimental studies
on a full-scale 7-storey RC frame. Test results indicated that, the existence of
cast-in-place floor slabs in RC frames, when under lateral loading, could greatly
increase the flexural capacity of beams. Such an increase was, however, not
taken into consideration when designing the frame. This was because at that
time, such a consideration was not available in design provisions. Failure of the
test frame was consequently controlled by joints shear failure.
Qi and Pantazopoulou (1991) subjected a 1/4-scale single-story RC frame
with cast-in-place floor slabs to cyclic lateral loading. The test results indicated
Page 56
25
that the contribution of cast-in-place slabs to the flexural capacity of the beams,
especially at the interior support, was considerable.
Ning et al. (2016) tested two spatial RC frames: a control specimen
without cast-in-place slabs and one with cast-in-place slabs. The specimens
were subjected to cyclic loading and tests results indicated that the existence of
cast-in-place slabs changed the failure mode of a RC frame from a typical
SCWB mechanism to the strong-beam-weak-column (SBWC) mechanism. The
slab reinforcement contribution to the negative moment capacity of the beams
was the main reason accounting for the occurrence of SBWC failure.
In addition to experimental RC frame tests, massive experimental research
on RC beam-column-slab subassemblies were conducted to assess the
cast-in-place slab effect on RC joints subjected to cyclic loading. Experimental
investigations of both exterior and interior RC joints with cast-in-slab are given
blow.
Ehsani and Wight (1985) conducted tests on six beam-column-slab
subassemblies, subjected to cyclic loading. Test variables including the ratio of
the column flexural capacity to those of the beam with slabs (i.e. flexural
strength ratio), the joint shear stress, and joint transverse reinforcement.
Specimens with cast-in-place slabs and transverse beams were compared with
those control examples without transverse beams and slab (referred as a plane
joint hereafter). Flexural strength ratios of 1.1, 1.5 and 2.0 were included in the
Page 57
26
investigation. Only two longitudinal slab reinforcing bars, adjacent to the main
beam, were included in the calculation. All the slab longitudinal reinforcing bars
had yielded during test. Thus the flexural strength ratio was in fact lower than
anticipated. Specimens with flexural strength ratio equal to 1.1 failed by column
concrete crushing.
Durrani and Zerbe (1987) tested a total of six 3/4-scale joints under cyclic
lateral loading to study the effect of cast-in-place slabs on the behavior of
exterior joints. The test results showed that the cast-in-place floor slab had a
significant effect on the strength, stiffness and energy dissipation characteristics
of the joints. It was thus strongly suggested that the effect of cast-in-place floor
slabs be considered in the design of joints.
Durrani and Wight (1987) tested three interior RC beam-column-slab joints
with variables including joint shear stress level and an amount of joint hoop
reinforcement. All slab bars yielded at 4% storey drift ratio and the inclusion of
slab contribution to beam moment capacity in seismic design was
recommended.
Zerbe and Durrani (1990) tested two-bay frame subassemblies to study the
effect of slab on the behavior of beam-columns-slab connections. Each
subassembly was composed of two exterior joints and one interior joint. Three
subassemblies including a plane specimen C, specimen CTB with transverse
beams only and the remaining one CS1 consisting of both transverse beams and
Page 58
27
slab were considered. Test results showed that continuous joints increased the
contribution of slab to the hogging moment of beams and introduced more shear
force in the joints than the individual joints did.
French (1991) collected test data of 20 beam-slab-column joints (13
interior joints and 7 exterior joints). The average underestimation of predicted
strength of interior joints when the cast-in-place floor slab was ignored was 25%
less than the test result, which is 17% for exterior joints.
A 1/4-scale single-story RC frame with cast-in-place floor slabs subjected
to cyclic lateral loading was tested by Qi and Pantazopoulou (1991). The test
results showed that the flexural capacity of the beams, especially at the interior
support greatly increased due to the existence of cast-in-place slabs.
Jiang et al. (1994) tested two specimens: the plane joint and one with a
cast-in-place slab. The test results showed that the cast-in-place floor slab
increased the beam negative flexural capacity by as much as 30%.
Shin and LaFave (2004a, b) tested four 2/3 scale RC edge
beam–column–slab subassemblies (i.e. two concentric and two eccentric
connections). The slab and transverse beam were only on one side of beams.
The test results revealed that the effective slab width, at peak storey shear force,
was actually wider than design recommendation.
Zhen et al. (2009) tested three groups of RC joints with different
Page 59
28
reinforcement schemes under cyclic lateral loading. Each group was composed
of a plane joint and one/four/two joints with a cast-in-place floor slab. Test
results showed that the existence of a cast-in-place floor slab resulted in
specimen strengths, in group 1, 2 and 3, respectively about 1.6, 2.0 and 2.3 that
of those of the corresponding plane joints
Jiang et al. (2009) tested eight weak reinforced beam-column connections
and two strong ones under cyclic lateral loading. The experimental result
showed that due to the increased joint shear stress introduced by the
cast-in-place slab, the existence of the slab transformed the failure mode of a
weak joint from beam flexural failure to joint shear failure. The specimens with
a cast-in-place slab suffered more damage in the joint region than that suffered
by the corresponding plane specimens.
Li et al. (2012) tested five 3/4-scaled interior beam-column joints to
investigate the influence of floor slabs and column orientation on the seismic
performance of lightly-reinforced concrete beam-column joint subjected to
seismic loading. Test results show that the strengths of interior joints by about
11% to 27% increased by floor slabs
The effect on the seismic performance of RC frames with cast-in-place
floor slabs has been studies by many researchers (Guan and Du 2005, Lin et al.
2009, Gao and Ma 2009, Yang 2010, Chen 2010, and Guo 2012) using
numerical approaches as detailed in Table 2.1. All numerical results indicated
Page 60
29
that cast-in place floor slabs could significantly increase the negative moment
capacity of the beams and thus lead to the SBWC mechanism in RC frames.
2.2.2 Seismic Retrofit of RC Frames
The literature review of experimental and FE analysis of RC frames/joints
with cast-in-place slabs given above indicate that the ignorance or
underestimation of the cast-in-place slabs contribution to a beam negative
moment strength would probably lead to a SBWC failure mode in existing RC
fames, subjected to seismic loading. To avoid such inductile failure mode,
seismic retrofitting of RC frames is in great need of further study.
As stated in Chapter 1, to strengthen the columns is an easy and obvious
option while cost-effective seismic retrofitting can enhance columns in both
strength and ductility with FRP jackets. This process however, changes a storey
sway failure mechanism to a beam sway mechanism, very little, especially when
the columns are non-circular in sectional shape because the strength
improvement may not be sufficient to meet SCWB criteria. Even when the
columns are, in fact, sufficiently enhanced, failure may possibly occur at the
beam-column connections and the foundations, the latter being more
complicated to reinforce. Thus, the benefit of a more effective seismic retrofit
method is that it weakens the flexural capacities of T-beams, especially when
under negative bending.
Page 61
30
Recently, Teng et al. (2013) proposed a novel seismic retrofit method for
RC frames with cast-in-place slabs which violate the SCWB hierarchy. This
method is based on the concept of Beam-end Weakening in combination with
FRP Strengthening (referred to as the BWFS method hereafter for simplicity), to
implement the SCWB hierarchy. To weaken the flexural capacities of the
T-beams at a joint, particularly when the flange (i.e. the cast-in-place slab) is in
tension, is a primary measure. Three beam local weakening techniques were
also presented by Teng et al. (2013). Local weakening for seismic design
purposes or through retrofitting is more commonly practised. For steel
structures, a dog-bone design is adopted as a typical weakening technique for
new structures and, for existing structures, a seismic retrofit is aimed to ensure
the SCWB strength criteria (Popov et al. 1998). For RC structures, local
weakening by material removal to meet seismic concerns was discussed in a
preliminary and general manner at FEMA-356 (2000). Cutting the bottom
longitudinal steel reinforcements, as a seismic retrofit technique for exterior
beam-column joints, was investigated by Pampanin (2006) and Kam et al.
(2009). However, cutting out bottom bars is insufficient to offset the
contribution of a cast-in-place slab in tension.
As presented in Chapter 1, three different techniques are proposed: (a) the
slab slit (SS) technique; (b) the beam opening (BO) technique; (c) the beam
section reduction (SR) technique (e.g., creating a deep transverse groove (TG)
on the soffit of the beam near the joint). The effect of slits on RC joints and
Page 62
31
frames has been studied through FE modelling by Zhang et al. (2011), Wang el
al.(2012) and Zhang (2013). The investigations were FE modelling based.
Conclusion was made by Zhang et al. (2011), Wang el al.(2012) and Zhang
(2013) that the SS technique can effectively change the failure mode from story
sway to beam sway mechanism. The BO technique has been recently
investigated by Nie (2018). Both the experimental test and the 2D FE
modellling of T-beams retrofitted by the BO technique were conducted. Test
results showed that the BO technique effectively decreased the T-beam negative
moment strength. The specimens with both BO weakening and FRP
strengthening had ductile failure modes. To the best of the candidate’s
knowledge, the last mentioned techniques are new and no relevant research
regarding its effectiveness and design methods is available.
More recently, Feng et al. (2017) proposed a novel method to improve the
seismic performance and progressive collapse resistance of RC frame structures
using kinked bar. The kinked bar has locally curved regions, which are usually
placed near the inflection points in beams. The curved region is gradually
straightened when subjected to tension force. The section where kinked bar
curved region is located has the lowest capacity and will firstly yield under
seismic loading. However, the section is of good ductile property as the kinked
bar can keep resisting force when straightened. The seismic performance and
progressive collapse resistance of RC frame structures will therefore be
improved. This novel method can be considered a new retrofit technique for
Page 63
32
exiting RC frames, even though its original proposal was for new construction.
2.3 EXPERIMENTAL STUDIES AND FE MODELLING OF
RC BEAMS WITH WEB OPENINGS
2.3.1 Experimental Studies of RC Beams with Web Openings
Many studies have conducted experimental investigations regarding the
behavior and design of rectangular or T-beams with a rectangular or circular
opening for the passage of utility ducts and pipes since 1967 (e.g. Nasser et al.
1967; Mansur et al. 1985; Tan et al. 1996). Nasser et al. (1967) tested 10 beams
with openings. In the design of beams with openings, two groups of longitudinal
reinforcement were provided in each chord and several specimens were
reinforced with adequate stirrups and longitudinal bars in the chords. The beam
whose cross section was 41.7% (beam B-1) reduced and had an ultimate load 24%
lower than that of the control one. The existence of openings also lowered the
stiffness of beams.
Mansur et al. (1985) designed and tested 12 beams with openings under
concentrated load to verify the validity of the proposed design method for
strengthening beams. The failure mode of a beam with a web opening was chord
end concrete crushing. The existence of a opening of size 1200×180 mm2 and
800×220 mm2 decreased the ultimate load to less than 50% of that of the
control specimen. The application of diagonal bars as corner reinforcement
proved to be more effective in controlling crack propagation and thus the
Page 64
33
increase of the ultimate loads of beams.
Tan et al. (1996) tested 15 T-beams with openings except for the control
beams to investigate the effect of opening size, location on continuous beams.
Two groups of specimens, namely group IT and group T were tested to
investigate the performance of a continuous T-beam under respectively negative
and positive loading. The result showed that the existence of openings decreased
the cracking and ultimate load as well as beam post-cracking stiffness. The
heights of all the openings were not larger than half of a T-beam height. The
openings were all located not closer than one-half the beam depth from the
supports or the concentrated loads. The test result of the IT group revealed that
enlarging the opening length could significantly decrease the post-cracking
stiffness and ultimate capacity of a T-beam and switch the failure mode from
flexural failure at solid section to a failure, due to hinge formation at the
opening corners. The location of opening had limited effect on the ultimate load
of beam. After testing the T group specimens, conclusions could be drawn that
the enlargement of the opening height could slightly lower the positive capacity
of a T-beam.
Conclusions can be drawn from the studies referred to above, that an
opening can significantly reduce the shear and flexural capacities of a beam. All
the above reviewed studies focused on the application of internal steel
reinforcement around the reserved opening to minimize that opening's effect on
its host beam. The effect of drilling an opening in an existing beam and the
Page 65
34
application of external strengthening techniques are reviewed below.
Mansur et al. (1999) tested a series of T-beams to evaluate the effect of
creating beam web circular openings. The reinforcing stirrups crossing openings
were cut and no additional internal reinforcement around the openings was
provided. T-beams, subsequently repaired and strengthened by filling openings
with respective nonshrink construction grout and externally bonded FRP were
tested to evaluate their effects on recovering the T-beam strength degradation.
The result showed that the beneficial effect of grout was limited. The external
bonded FRP could not only fully help a T-beam regain its original stiffness and
ultimate load, but also enhance its ductility. The maximum crack width was also
well controlled.
Abdalla et al. (2003) conducted an experimental program of 10 RC beams
including five strengthened with bonded FRP around the openings, four tested
without FRP strengthening and a control solid beam. Parameters including
opening horizontal location, opening size, configuration and amount of FRP
were studied. The result revealed that those beams with openings weakening but
without FRP strengthening all suffered shear failure at the opening region under
very low loads. When the opening size was relatively small (), the beam
strengthened by bonded FRP could fully regain the original stiffness and
strength as a solid beam. As the opening size kept increasing, the failure mode
switched from mid-span flexure failure to shear failure at the opening region.
Page 66
35
Allam (2005) conducted two series of test to investigate the efficiency of
strengthening schemes when web opening in shear zone of a beam on b. Internal
steel reinforcement, external steel plates and bonded CFRP sheets were all
applied in the test. The results indicated that the web opening decreased the
strength of beam from 167 kN to 105 kN when no strengthening scheme was
applied. Failure mode was also changed to shear failure at the opening zone
rather than flexural failure at the mid-span. When the beam with opening was
strengthened with steel reinforcement, external steel plates or bonded CFRP
sheets, its strength was obviously increased. The specimens strengthened with
external steel plates both outside and inside also failed due to flexural failure at
the mid span.
Maaddawy and Sherif (2009) studied the effect of FRP composites on
shear strengthening of RC deep beams with a group of 13 deep beams tested
under four-point bending. All the specimens were of 80×500 mm cross section
and a total length of 1200 mm. Maaddawy and Sherif (2009) concluded that the
effect of opening was primarily dependent on the degree of the interruption of
natural load path with variation of opening position and size. The strength gain
resulting from CEFP sheets was in a range of 35%-73% and the stiffness was
also upgraded.
Pimanmas (2010) examined the function of externally installed FRP robs
in shear strengthening RC beams with square or circular openings. A total
Page 67
36
number of 13 beams were tested. The test result showed that an opening
significantly reduced the shear capacity of the beams and full depth FRP robs
was most effective in regaining strength and ductility. The flexural failure mode
was also restored with this kind of strengthening method.
Chin et al. (2012) conducted an experimental study on RC beams with
square openings (210×210 mm2). Six beams consisting of control beams,
un-strengthened and strengthened RC beams with square openings. FRP wrap
was employed as a strengthening technique. Experimental results showed that
the existence of openings significantly decreased the beam strength by about
70%. The CFRP wrap could increase the strength of the beam.
Maaddawyand and Ariss (2012) conducted an experimental study on a total
of 15 RC beams with openings, in addition to a solid beam and proposed
analytical formulation of shear resisting capacity for specimens to evaluate the
effect of CFRP composites on shear strengthening RC beams with opening. No
internal web reinforcement was provided to strengthen the opening region to
resemble the case of drilling an opening on a solid beam web. Variables
included the opening size and the amount of CFRP sheets. It is of note that for
FRP strengthened RC beams with web openings, vertical and horizontal FRP
sheets were arranged around the openings. Except that the bottom chords were
strengthened by CFRP wrapping, the top chords and the both sides of openings
were provided with U-shaped CFRP. The result indicated that the shear capacity
and stiffness of a beam were obviously degraded due to the existence of a web
Page 68
37
openings and this weakening effect was effectively offset by the externally
boned CFRP sheets. The increase of shear capacity was not in the same scale as
the amount of CFRP applied. Increasing the opening size lowered the efficiency
of CFRP sheets as regards upgrading the shear resistance and stiffness of the
beam. The proposed analytical approach, used to predict the shear capacity of an
RC beam weakened by openings and shear strengthened by CFRP indicated a
good agreement with the test results.
Suresh and Prabhavathy (2015) tested 14 beams to investigate the
behaviour of steel fibre RC beams with opening strengthened by steel plates.
The results showed that the existence of web opening in shear zone decreased
the strength of beams by 55%-70%. Strengthening the beams with web opening
using steel plates considerably increased their strength and ultimate deflection.
Chin et al. (2016) investigated the behaviour of RC with a web opening,
with/without CFRP strengthening by testing 6 simply supported beams. The
result indicated that a web opening at the mid-span decreased the beam strength
to about 50% and externally bonded CFRP steel lead to 80-90% strength regain.
The experimental studies (Mansur et al. 1999; Abdalla et al. 2003; Yang et
al. 2006; Chin et al. 2012; Maaddawyand and Ariss (2012)) aimed to investigate
the effect of drilling a web opening in an existing beam. Design methods were
proposed for associated strengthening measures to regain the strength, reduced
due to the exiting of openings. Conclusions drawn suggest that the existence of
Page 69
38
a web opening decreased the strength of RC beams and that strengthening with
bonded FRP reinforcement or steel plates could effectively enable the reduced
strength to be regained.
To investigate the proposed BO technique, Jing et al. (2013) tested two RC
T-beams weakened by rectangular opening: one beam (O-500×150) had a 150
mm (height) × 500 mm (length) opening but not FRP strengthening while the
other (FRP-500×220) had a 220 mm (height) × 500 mm (length) opening and
additional FRP strengthening. The result indicated that larger opening size lead
to more strength decrease and the bonded FRP reinforcement avoided
undesirable failure mode caused by the existence of web opening.
Recently, with the aim to reduce a T-beam strength while preserving its
ductility, Nie (2018) tested T-beams to investigate the effect of the BO
technique regarding decreasing the moment capacities of T-beams in his PhD
thesis. Except for a solid R-beam and T-beam, a total 12 T-beams with BO were
tested. Ten of the 12 T-beams were tested under negative loading and the
remaining two were tested under positive loading. Four of the ten specimens
were only weakened by opening and the remaining six were subjected to
additional FRP strengthening. Test results showed that the existence of an
opening effectively decreased T-beam strength. The specimens with additional
FRP strengthening had ductile failure modes. Nie (2018) also proposed a
method to predict the strength of RC T-beams with a web opening.
Page 70
39
2.3.1 FE Studies of RC Beams with Web Openings
Several researchers (Madkour 2009; Pimanmas (2010); Chin et al. 2012;
Hawileh et al. 2012, Nie 2018) proposed FE models to simulate RC beams as
summarized in Table 2.3.
Madkour (2009) applied a damage-non-linear elastic model to conduct RC
beam 3D simulations. The isoperimetric quadratic element (20-node element),
elastic bar element (2-node element), and shell element were used for concrete,
reinforcing steel and FRP. The bond slip relation between FRP/steel
reinforcement and concrete was not mentioned by Madkour (2009). The
predicted crack patterns were poor.
Pimanmas (2010) examined the function of externally installed FRP bars in
shear strengthening of RC beams weakened by square or circular openings. In
addition to an experimental program, a 2D nonlinear FE analysis based on
smeared crack approach was conducted to investigate a preferable arrangement
of FRP robs round openings. A perfect bond between FRP robs and substrate
concrete was assumed by Pimanmas (2010). The predicted load-displacement
curves were comparable to the test results. The predicted crack patterns were
poor as the cracks were not clearly revealed.
Chin et al. (2012) conducted a numerical analysis of RC beams with large
square openings. 2D FE modelling experimental specimens with software
Page 71
40
ATENA was carried on. The tensile behavior of concrete was modeled with a
combination of nonlinear fracture mechanics and crack band method, in which
the smeared crack concept was adopted. A displacement-controlled procedure
based on Newton-Raphson iterative method was applied in analysis. A Bond
slip relation between reinforcing steel and concrete was ignored and that
between CFRP and concrete was considered by applying the bond slip model
developed by Lu et al. (2005). Even though the predicted ultimate strength was
relatively close to the experimental result, the predicted stiffness was much
higher than that of test. Chin et al. (2012) assumed that the omission of bond
slip relation between reinforcing steel and concrete was believed to be the
cause.
Hawileh et al. (2012) carried out a 3D FE analysis on RC deep beams
tested by Maaddawy and Sherif (2009). A total of 12 specimens were modeled.
They included one solid beam, strengthened and un-strengthened beams
weakened by opening. Solid elements for concrete, multi-layer shell elements
for CFRP and link elements for steel reinforcement were applied. A special
interface element was used to represent the bond-slip behaviour between FRP
and substrate concrete. A model with perfect bond for the interface between FRP
and substrate concrete was also modelled for comparison. The models
incorporating bone-slip behavior showed great agreement with test results with
an average of 3.2% error for strengths and an average 14% error for ultimate
deflections. The omission of bond slip behavior led to higher predicted strengths
Page 72
41
and stiffnesses.
Only Nie (2018) has conducted a 2D FE analysis on RC T-beams
weakened by opening using ABAQUS (2012). An explicit dynamic method was
applied to avoid numerical convergence problems. Several FE approaches, with
major difference in the applied models for concrete compression property and
damage under tension, were proposed to model a beam with web opening. The
damage plasticity (DP) approach with power law (PL) tension damage model
was recommended for beams with a flexural failure mode, while the brittle
cracking (BC) approach with secant modulus was recommended for beams with
a shear failure mode. The DP approach with PL tension damage model was most
suitable for the test T-beams with both opening weakening and associated FRP
strengthening, as they all had ductile failure mode, while the BC approach with
secant modulus was suggested for the T-beams weakened by a opening only.
The peak loads of T-beams were predicted with a relatively small error.
However, the 2D model generally overestimated specimen stiffness. Nie (2018)
stated that this may be due to the non-uniform distribution of longitudinal
compressive stresses in the concrete and the reinforcement of the flange along
the width direction, known as shear lag effect. Of interest is that the potential
out-of-plane deformation could not be captured by the 2D models. Nie (2018)
mentioned that a more accurate 3D FE model was needed to take the shear lag
effect into consideration. The cracks on the slab could not be directly presented
by the 2D models. Thus, 3D FE models of RC T-beam weakened by opening
Page 73
42
should be proposed to better predict the behaviour of RC T-beams weakened by
opening.
2.4 EFFECTIVE SLAB WIDTH OF RC FRAMES
Extensive experimental studies (e.g. Jiang 1994; Bijan and Aalami 2001;
Huang et al. 2001) have indicated that the strains of steel bars in a cast-in-place
slab flange are not evenly distributed along the beam width direction. Instead,
the strain in a slab steel bar deceases with the increased distance between the
steel bar and the beam, due to the well-known shear lag effect. Thus, only steel
bars within a limited range of width away from the beam can reach their yield
stress at the failure of the beam (Wu et al. 2002; Wang et al. 2009; Zhen et al.
2009). In order to quantify the contribution of a cast-in-place slab to the beam
flexural capacity, an effective flange width ( effb ) has been proposed by previous
researchers (Wu et al. 2002; Wang et al. 2009). All the slab longitudinal steel
bars within the effective flange width are assumed to be equally strained in the
bending of the beam.
However, currently, a uniform method, to determine the effective width of
a slab, especially for an RC structure with cast-in-place slabs does not exist. The
effective widths of slabs stated in codes (GB 50010-2010, ACI318-05 &
FEMA-356, EC8, NZS-3101:2006), as indicated in Table 2.4, are also different.
Existing research in this area is experimentally based (Ehsani and Wigh 1982, T
Pantazopotrlou et al. 1988, Durrani and Zerbe 1990, French 1991, Li 1994,
Page 74
43
Jiang et al. 1994, Wu et al. 2002, Wang et al. 2009, Zhen et al. 2009, Yang 2010,
Sun et al. 2010, Qi et al. 2010, He 2010, Ning et al. 2016). Table 2.5 presents a
summary of the suggested values of effective slab widths given by the above
researchers based on their experimental results. It is obvious that the effective
slab width beff is mainly related to beam width bw, beam height hb, effective span
of beam l0, slab thickness t, and clear distance between beams sn. Most
researchers determine the effective slab width when the storey drift rato ( ) is
equal to 1/50.
Some researchers, however, also proposed equations, based on their
simulation data, to determine the effective slab width for RC frames. Ning et al.
(2016) studied the effective slab width of an RC frame both experimentally and
by means of FE simulation. The FE model was first verified through
comparison with test result. Ning et al. (2016) then conducted parametric
studies on the axial compression ratio, concrete strength, reinforcement ratio of
slabs, thickness of slabs and dimension of the transverse beams. An equation
involving only the main beam width and height to predict the effective slab
width with a 95% guaranteed accuracy was proposed. However, Ning et al.
(2016) did not study the effect of beam length on beff, which is a parameter
considered in all design codes (e.g. GB 50010-2010, ACI318-05 & FEMA-356,
NZS-3101:2006) except for EC8. In addition, Ning et al. (2016) did not study
the effect of slab width on beff, a parameter also considered by all design codes
(e.g.GB 50010-2010, ACI318-05 & FEMA-356, EC8, NZS-3101:2006) except
Page 75
44
EC8.
Many researchers (e.g. Castro et al. 2006, Nie and Tao 2012, Methee et al
2016, Galuppi and Royer-Carfagni 2016) have studied the effective slab width
of a composite structure under positive loading through FE modelling. Some,
such as Nie and Tao (2012), also studied the effective slab width of composite
structures under negative loading. By conducting parametric studies through FE
simulation on column dimensions, steel beam height, RC slab width and
thickness, transverse beam, and yield stress of the longitudinal reinforcement,
Nie and Tao (2012) found that the main factors influencing the negative
effective flange width, included the column dimensions, the steel beam height,
the flange width of the transverse beam, and the yield stress of the longitudinal
reinforcement. The slab width was found to be less influential.
2.5 CONCLUDING SUMMARY
This chapter has presented a literature review covering the potential
problems existing in RC frames and corresponding retrofit methods,
experimental and FE analysis on RC beams weakened by opening, effective slab
width of RC frames. The following conclusions can be drawn:
(1) Existing RC frames with cast-in-place slabs might fail by the
Strong-Beam-Weak-Column (SBWC) mechanism, rather than
Strong-Column-Weak-Beam (SCWB) mechanism, as the contribution from
cast-in-place slabs are ignored or underestimated.
Page 76
45
(2) Three retrofit techniques, namely the slab slit (SS) technique, the beam
opening (BO) technique and the beam section reduction (SR) technique are
proposed based on the concept of Beam-end Weakening in combination with
FRP Strengthening, proposed by Teng et al. (2013). SS technique has only
been studied through FE modelling and no experimental tests and therefore
results are available. The BO technique has recently been investigated by
Nie (2018) through experimental tests on RC T-beams. The SR technique
for RC frames is new and no relevant research is available.
(3) The existence of web opening can decrease a RC beam shear and flexural
strengths. External bonded FRP is an efficient strengthening technique for a
RC beam with web opening. The BO technique for RC T-beams can
efficiently decrease the T-beam negative moment strength, proved by the
experiments tested by Nie (2018). The ductility of T-beams is good if
additional FRP strengthening is provided.
(4) Nie (2018) proposed 2D FE models for RC beams weakened by web
openings. Negative strength of T-beams with a web opening was well
predicted with a relatively small error. The 2D model, however, generally
overestimated the stiffness of specimens, possibly because shear lag and the
potential out-of-plane deformation could not be captured by the 2D models.
A more accurate 3D FE model for RC T-beams with a web opening is
needed.
Page 77
46
(5) No uniform method to determine the effective width of a slab under tension
exists. Most research on effective slab width of RC frames is experimentally
based. Only Ning et al. (2016) conducted additional parametric studies
through FE modelling to determine the key factors affecting the effective
slab width of RC frames. Some factors, however, such as beam length and
slab width, were not considered by Ning et al. (2016) in their FE analysis. A
parametric study on the effective slab width of RC frames covering a wider
range of factors should be conducted.
2.6 REFERENCES
ABAQUS (2012). ABAQUS Analysis User's Manual (Version 6.12), Dassault
Systems SIMULIA Corporation, Providence, Rhode Island, USA.
ACI 318 (1983). Building Code Requirements for Structural Concrete and
Commentary (ACI 318-83), ACI Committee 318, American Concrete
Institute, Farmington Hills, MI.
ACI 318 (2014). Building Code Requirements for Structural Concrete and
Commentary (ACI 318-14), ACI Committee 318, American Concrete
Institute, Farmington Hills, MI.
Canbolat, B.B. and Wight, J.K. (2008). “Experimental investigation on seismic
behavior of eccentric reinforced concrete beam-column-slab connections”,
ACI Structural Journal, 105(2), 154-162.
Page 78
47
Castro, J. M., Elghazouli, A. Y., and Izzuddin, B. A. (2007), "Assessment of
effective slab widths in composite beams", Journal of Constructional Steel
Research, 63(10), 1317-1327.
Chen, X.B. (2010). The Effect of Floor Slab and Infill Walls on the Seismic
Behavior of Reinforced Concrete Frames, Master degree thesis: Fuzhou
University, China. (in Chinese)
Chinese Academy of Building Research (2008). Photo Collection of 2008
Wenchuan Earthquake Damage to Buildings, China Architectural &
Building Press, Beijing, China (in Chinese).
Chiewanichakorn, M., Aref, A. J., Chen, S. S., and Ahn, I. S. (2004), " Effective
flange width definition for steel–concrete composite bridge girder", Journal
of Structural Engineering, 130(12), 2016-2031.
Durrani, A.J. and Wight, J.K. (1987). “Earthquake resistance of reinforced
concrete interior connections including a floor slab”, ACI Journal, 84(5),
400-406.
Ehsani, M.R. and Wight, J.K. (1985). “Effect of transverse beams and slab on
behavior of reinforced concrete beam to column connections”, ACI Journal,
82(2), 188-195.
Eurocode 8 (2004). Design of Structures for Earthquake Resistance – Part 1:
Page 79
48
General Rules, Seismic Actions and Rules for Buildings (EN 1998-1: 2004),
CEN, Brussels.
FEMA-356 (2000). Pre-Standard and Commentary for the Seismic
Rehabilitation of Buildings. Federal Emergency Management Agency,
Washington, D.C.
Feng, P., Qiang, H., Qin, W. and Gao, M. (2017). “A novel kinked bar
configuration for simultaneously improving the seismic performance and
progressive collapse resistance of RC frame structures”, Engineering
Structures, 147, 752-767.
Gao, Z.R. and Ma, Q.L. (2009). “Effect of cast-in-place slab on column-beam
strength ratio in frame structure”, Shanxi Architecture, 35(17), 1-3 (in
Chinese).
Galuppi, L., and Royer-Carfagni, G. (2016), "Effective Width of the Slab in
Composite Beams with Nonlinear Shear Connection", Journal of
Engineering Mechanics, 142(4), 04016001.
GBJ11-89 (1989). Code for Seismic Design of Buildings, Architectural &
Building Press, Beijing, China. (in Chinese)
GB-50011 (2001). Code for Seismic Design of Buildings, Architectural &
Building Press, Beijing, China (in Chinese).
Page 80
49
GB-50011 (2010). Code for Seismic Design of Buildings, Architectural &
Building Press, Beijing, China (in Chinese).
Guan, M.S. and Du, H.B. (2005). “Pushover analysis of effect of casting slab on
RC frame structure”, Earthquake engineering and engineering vibration,
5(5), 117-123. (in Chinese)
Guimaraes, G.N., Kreger, M.E. and Jirsa, J.O. (1992). “Evaluation of joint-shear
provisions for interior beam-column-slab connections using high-strength
materials”, ACI Structural Journal, 89(1), 89-98.
Guo, L. (2012). Research on RC Structures with Cast-in-Place Slab, Master
degree thesis: Beijing University of Technology, China. (in Chinese)
Hasnat, A., and Akhtanizzamam, A. A. (1987), "Beams with small rectangular
opening under torsion, bending, and shear", Journal of Structural
Engineering, 113(10), 2253-2270.
Hawileh, R., El-Maaddawy, T., and Naser, M. (2012). “Nonlinear finite element
modeling of concrete deep beams with openings strengthened with
externally-bonded composites”, Materials & Design (DOI:
10.1016/j.matdes.2012.06.004, 378-387).
JTCC (1988). “U.S.-Japan research: seismic design implications”, Journal of
Structural Engineering, ASCE, 114(9), 2000-2016.
Page 81
50
Kennedy, J.B. and Abdalla, H. (1992). “Static response of prestressed girders
with openings”, Journal of Structural Engineering, ASCE, 118(2), 488-504.
LaFave, J.M. and Wight, J.K. (1999). “Reinforced concrete exterior wide
beam-column-slab connections subjected to lateral earthquake loading”,
ACI Structural Journal, 96(4), 577-587.
Meas, K., Li, B., and Imran, I. (2012), "eismic performance of lightly reinforced
concrete exterior beam-column joints", Advances in Structural
Engineering, 15(10), 1765-1780.
Lin, X., Pan, P., Ye, L., Lu, X. and Zhao, S. (2009). “Analysis of the damage
mechanism of a typical RC frame in Wenchuan earthquake”, China Civil
Engineering Journal, 42(5), 13-20 (in Chinese).
Lu, X.Z., J.G. Teng, Ye, L.P. and Jiang, J.J. (2005). “Bond-slip models for FRP
sheets /plates bonded to concrete”, Engineering Structures, 27(6), 920-937.
Maaddawy, T. and Sherif, S. (2009). “FRP composite for shear strengthening of
reinforced concrete deep beams with openings”, Composite Structures,
89(1), 60-69.
Maaddawy, T. and El-Ariss, B. (2012). “Behavior of concrete beams with short
shear span and web opening strengthened in shear with CFRP composites”,
Journal of Composites for Construction, ASCE, 16(1), 47–59.
Page 82
51
Mansur, M.A. (1998). “Effect of opening on the behaviour and strength of R/C
beams in shear”, Cement and Concrete Composites, 20(6), 477-486.
Mansur, M.A., Tan K.H. and Wei, W. (1999). “Effects of creating an opening in
existing beams”, ACI Structural Journal, 96(6), 899-906.
Nasser, K. W., Acavalos, A., and Daniel, H. R. (1967), "Behavior and design of
large openings in reinforced concrete beams", In Journal Proceedings (Vol.
64, No. 1, pp. 25-33).
Nie, J. G., and Tao, M. X. (2012),"Slab spatial composite effect in composite
frame systems. I: Effective width for ultimate loading capacity",
Engineering structures, 38, 171-184.
Nie, X.F. (2018). Behavior and Modeling OF RC Beams with an
FRP-Strengthened Web Opening, Doctoral degree thesis: The Hong Kong
Polytechnic University.
Ning, N., Qu, W., and Zhu, P. (2014), "Role of cast-in situ slabs in RC frames
under low frequency cyclic load", Engineering Structures, 59, 28-38.
Ning, N., Qu, W., and Ma, Z. J. (2016), "Design recommendations for achieving
“strong column-weak beam” in RC frames", Engineering Structures, 126,
343-352.
NZS-3101 (1982). Code of Practice for the Design of Concrete Structures,
Page 83
52
Standards New Zealand, Wellington, New Zealand.
NZS-3101 (2006). Concrete Structures Standard, Standards New Zealand,
Wellington, New Zealand.
OpenSees (2009). Open System for Earthquake Engineering Simulation, Pacific
Earthquake Engineering Research Center, University of California at
Berkeley, http://opensees.berkeley.edu.
Otani, S., Kabeyasawa, T., Shiohara, H. et al. (1984). “Analysis of the full-scale
seven-story reinforced concrete test structure”. ACI SP-84, 203-239.
Pantazopoulou, S.J. and Moehle J.P. (1990). “Identification of effect of slabs on
flexural behavior of beams”, Journal of Engineering Mechanics, ASCE,
116(1), 91-106.
Pantazopoulou, S.J. and French, C.W. (2001). “Slab participation in practical
earthquake design of reinforced concrete frames”, ACI Structural Journal,
98(4), 479-489.
Pimanmas, A. (2010). “Strengthening R/C beams with opening by externally
installed FRP rods: Behavior and analysis”, Composite Structures, 92(8),
1957-1976.
Pool, R. S., and Lopes, R. (1986), "Cyclically loaded concrete beams with web
openings", In Journal Proceedings, Vol. 83, No. 5, pp. 757-763.
Page 84
53
Popov, E.P., Yang, T.S. and Chang, S.P. (1998). “Design of steel MRF
connections before and after 1994 Northridge earthquake”, Engineering
Structures, 20(12), 1030-1038.
Qi, X. and Pantazopoulou, S.J. (1991). “Response of RC frames under lateral
loads”, Journal of Structural Engineering, ASCE, 117(4), 1167-1188.
Shin, M. and LaFave, J.M. (2004a). “Seismic performance of reinforced
concrete eccentric beam-column connections with floor slabs”, ACI
Structural Journal, 101(3), 403-412.
Shin, M. and LaFave, J.M. (2004b). “Reinforced concrete edge
beam-column-slab connections subjected to earthquake loading”, Magazine
of Concrete Research, 55(6), 273-291.
Siao, W.B. (1994). “Reinforced concrete column strength at beam/slab and
column intersection”, ACI Structural Journal, 91(1), 3-8.
Tan, K.H., Mansur, M.A., and Wei, W. (2001). “Design of reinforced concrete
beams with circular openings”, ACI Structural Journal, 98(3), 407-415.
Teng, J.G., Chen, J.F., Smith S.T. and Lam L. (2002). FRP-Strengthened RC
Structures, John Wiley and Sons Ltd, UK, November, 245pp.
Teng, J.G., Zhang, S.S., Jing, D.H., Nie, X.F. and Chen, G.M. (2013). “Seismic
retrofit of RC frames through beam-end weakening in conjunction with
Page 85
54
FRP strengthening”, Proceedings of the 4th Asia-Pacific Conference on
FRP in Structures (APFIS 2013), 1-8.
Thermou, G.E., Pantazopoulou, S.J., Einashai, A.S. (2007). “Flexural behavior
of brittle RC members rehabilitated with concrete jacketing”, Journal of
Structural Engineering, ASCE, 133(10), 1373-1384.
Wang, X.G., Shan, M.Y., Ge, N. and Shu, Y.P. (2012). “Finite element analysis
of efficiency of slot-cutting around RC frame joint for ‘strong column and
weak beam’”, Journal of Earthquake Engineering and Engineering
Vibration, 32(1), 121-127 (in Chinese).
Xiao, Y. and Wu, H. (2003). “Retrofit of reinforced concrete columns using
partially stiffened steel jackets”, Journal of Structural Engineering, ASCE,
129(6), 725-732.
Yan, Y.L. (2010). A Study on Eurocodes about the RC Elements Design and
Comparison between Eurocodes and Chinese Codes, Master degree thesis:
China Academy of Building Research, China. (in Chinese).
Ye, L.P., Qu, Z., Ma, Q.L., Lin, X.C, Lu, X.Z. and Pan, P. (2008). “Study on
ensuring the strong column-weak beam mechanism for RC frames based on
the damage analysis in the Wenchuan earthquake”, Building Structure,
38(11), 52-67 (in Chinese).
Page 86
55
Zerbe, H.E. and Durrani, A.J. (1990). “Seismic response of connections in
two-bay reinforced concrete frame subassemblies with a floor slab”, ACI
Structural Journal, 87(4), 406-415.
Zhang, J. (2013). Research on Efficiency of Slot-cutting around Frame Joint for
“Strong Column and Weak Beam" under Earthquake Action, Master degree
thesis: Hunan University, China (in Chinese).
Zhang, Y.P., Hao, Z.J., Shan, M.Y. and Ge, N. (2011). “Research on anti-seismic
performance for reinforced concrete frame joint with slot around”, Building
Science, 27(9), 7-11 (in Chinese).
Page 87
56
Table 2.1 Numerical studies on the effect of cast-in-place floor slabs on the
behaviour of RC frames
Source Analysis type and FE models Software Guan and Du
(2005) Pushover analysis of a 3-storey-3-span RC frame.
SAP2000 (1998)
Lin et al. (2009
Elastic-plastic time history analysis of a 6-storey RC frame structure. Two 3D models for a plane frame and a frame with cast-in-place slabs was compared.
MSC.Marc (2005)
Gao and Ma (2009) Pushover analyses of two RC frames, with and without respective cast-in-place floor slabs.
SAP2000 (1998)
Yang (2010) Pushover analyses of five 6-storey 4x4-span RC frames with different slab widths.
SAP2000 (1998)
Chen (2010)
Elastic-plastic time history analyses of two 6-storey 6x3-span RC frames (one with floor slabs and one without floor slabs).
SAP2000 (1998)
Guo (2012)
Pushover analyses of established three FE models, one with cast-in-place slabs and slab reinforcement, one with cast-in-place slabs but without slab reinforcement and one without slabs.
SAP2000 (1998)
Page 88
57
Table 2.2 Summary of experimental studies on RC beams with a web opening
Source
Beam dimensions Web opening
size Strengthening
method
Observed failure mode
Span/clear span Width Height Length×height
(mm×mm) Without strengthening With strengthening
(mm) (mm) (mm)
Nasser et al.
(1967) 3962.4/3657.6 228.6 457.2
1219.2× 190.5
Internal stirrups
and longitudinal
bars
No unstrengthened specimen
B-1:corner cracking
1219.2×203.2 B-2:shear failure away from the opening
Two 762×203.2 C-1:corner cracking; C2 and C3: flexural
failure away from the opening
762×203.2 D-1: diagonal tension in chords; D2-D4:
flexural failure away from the opening
Mansur et al.
(1985) 3300/3000 200 400
400 × 180
Internal stirrups
or combined use
of stirrups and
diagonal
reinforcement
No unstrengthened specimen Chord ends concrete crushing
600 × 180
800 × 180
1000 × 180
1200 × 180
800 × 140
800 × 220
Page 89
58
Table 2.2 (cont.)
Tan et al. (1996) 3300/3000 200 500
Two 400 × 200
Internal stirrups
and longitudinal
bars
No unstrengthened specimen
Flexural failure failure away from the
opening
Two 600 × 200
IT2, IT7 :flexural failure failure away from the opening; IT5: local flexural failure at the chord ends; IT6: flexural failure in the
opening region
Two 800 × 200 Local flexural failure at the chord ends
Two 1000 ×
200 Local flexural failure at the chord ends
Four 400 × 200 Flexural failure away from the opening
600 × 120 Flexural failure away from the opening
600 × 200 TI,T3:Flexural failure failure away from the opening; T2: flexural failure in the
opening region
Mansur et al.
(1999) 2900/2600 200(c) 500
r=100 NA
Shear failure in the opening region Flexural failure away from the opening r=150 Bonded FRP
plates
r=200 NA
Page 90
59
Table 2.2 (cont.)
Abdalla et al.
(2003) -/2000 100 250
100 × 100
Bonded FRP
sheets and wraps Shear failure in the opening region
Flexural failure away from the opening
200 × 100 Shear failure in the opening region
300 × 100 Shear failure in the opening region
300 × 150 Shear failure in the opening region
Allam (2005) 3200/3000 150 400 450 × 150
Internal steel
reinforcement
Shear failure in the opening region
Shear failure in the opening region
External Steel
plates: outside Shear failure in the opening region
External Steel
plates: outside
and inside
Flexural failure away from the opening
Bonded FRP
sheets and
U-jackets
Shear failure in the opening region
Page 91
60
Table 2.2 (cont.)
Maaddawy and
Sherif (2009) 1200/1000 80 500
150 × 150 NA Two independent diagonal cracks in
the chords
No strengthened specimen
200 × 200 Bonded FRP
sheets and wraps
Two independent diagonal cracks in the
chords; 250 × 250 Relative rotation of three distinct
segments
Pimanmas
(2010) -/2100 400 160
r=150 Near-surface
mounted FRP
rods
Shear failure in the opening region C-rob2: Flexural failure away from the opening; The rest: shear failure in the
opening region 150 × 150
Chin et al. (2012) 2000/1800 300 120 210 × 210 Bonded FRP
sheets and wraps Shear failure in the opening region Shear failure in the opening region
210 × 210
Page 92
61
Table 2.2 (cont.)
Maaddawy and
Ariss (2012) 2600/2400 85 400
200 × 200
Bonded FRP
sheets and
U-jackets
Shear failure in the opening region
Shear failure in opening region after
debonding and rupture of FRP
350 × 200
500 × 120
Shear failure in the opening region
500 × 160
500 × 200
Jing et al. (2013) 3300/3000 250 500 500 × 150 Bonded FRP
plate, U-jackets
and wraps
Shear failure in the opening region Shear failure in the opening region
500 × 220
Suresh and
Prabhavathy
(2015)
2200/2000 150 300
150 × 150
Steel fibers and
steel plates Shear failure in the opening region Shear failure in the opening region
200 × 150
250 × 150
300 × 150
Page 93
62
Table 2.2 (cont.)
Chin et al. (2016) 2000/1800 120 300 800 × 140 Bonded FRP
sheets Shear failure in the opening region
Shear failure in the opening region after
debonding of FRP
Nie (2018) 3300/3000 250 500
700 × 300
Bonded FRP
plate, U-jackets
and wraps
Mixed local flexural and shear
failure at the chord ends Local flexural failure at the chord ends
800 × 280 Mixed local flexural and shear
failure at the chord ends Local flexural failure at the chord ends
600 × 220 Web chord shear failure Local flexural failure at the chord ends
700 × 200 Web chord shear failure Local flexural failure at the chord ends
600 × 280 No unstrengthened specimen Local flexural failure at the chord ends
700 × 260 No unstrengthened specimen Local flexural failure at the chord ends
Page 94
63
Table 2.3 Summary of FE studies on RC beam with a web opening
Source Model
dimension
Flange width of one side ( mm)
Modelling of bond behaviour
Steel-to-concrete FRP-to-concrete
Madkour ( 2009 ) 3D NA Not mentioned Not mentioned Pimanmas (2010) 2D NA Perfect bond Perfect bond
Chin et al. (2012) 2D NA Perfect bond Bond-slip model
developed by Lu et al. (2005)
Hawileh et al. (2012)
3D NA Perfect bond Bond-slip relationship proposed by Xu and Needleman (1994)
Nie (2018) 2D 600 CEB-FIP (1993) Bond-slip relationship proposed by Lu et al.
(2005)
Page 95
64
Table 2.4 Effective slab width definitions in negative moment of interior and exterior joints in different codes
Code Effective slab width of interior joints Effective slab width of exterior joints
GB 50010-2010 { , }eff o w n wb = min l / 3,b +s b +12t { , }eff o w n wb = min l / 3,b +s b +12t
ACI318-05 & FEMA-356 { 2 16 , }eff w o nb = b min l / , t s { 6 12 , }eff w o nb = b min l / , t s
EC8 8eff cb = b t 4eff cb = b t
NZS-3101:2006 1 21 2
1 2 2 3
{6 ,2 2 ,
}
eff w b Trans beam f NTB
b bn n
b b b b
b = b min h w bh hs s
h h h h
1 21 2
1 2 2 3
{6 ,2 , }b beff w b f NTB n n
b b b b
h hb = b min h b s sh h h h
Note: wb is the beam web width, ol is the effective span of beam, ns is the clear spacing between beams, t is the slab thickness, cb is the column width,
bh is the beam height, Trans beamw is web width of transverse beam, f NTBb is the distance at the critical section of the potential plastic region in the beam
between the web and a line drawn at 45⁰ from the intersection of a line drawn parallel to the web and touching the side of the column and the edge of the slab
Page 96
65
Table 2.5 Suggested values of effective slab width by previous researchers
Time Researchers Value of beff Applicable conditions 1982 Ehsani and Wight 7bw Interior joints
1988 Pantazopotrlou et
al. bw+ 5hb
Interior joints under large deformation
1990 Durrani and Zerbe bw + 4 hb bw + 2 hb
Interior joints
1/ 33
Exterior joints
1/ 33
1991 French min{l0/4, bw + 16t, sn} Interior joints
1/ 50
1994 Li bw + 8t Interior joints 1994 Jiang et al. bw + 12t Interior joints
2002 Wu et al. bw + 12t Interior joints
1/ 50
2009 Wang et al. bw + 2t Interior joints
1/ 50
2009 Zhen et al.
min{ bw + 4hb, 0.4l0, sn}
Interior joints
1/ 50
min{ bw + 2 hb, 0.2 l0, sn}
Exterior joints
1/ 50
2010 Yang bw + (12-16)t Interior joints
2010 Sun
bw + min{max (l0/4, 2 hb), 1/2 sn }
Interior joints
1/ 50
bw + min{max (l0/5, 1.5 hb), 1/2 sn }
Exterior joints
1/ 50
Page 97
66
Table 2.5 (cont.)
2010 Qi et al.
bw + min{ l0/4, 12t, sn } Interior joints
bw + min{l0/5, 8t, sn} Exterior joints
2010 He bw + 12t Interior joints
2016 Ning et al.
bw+ 6.4 hb Interior joints
bw + 5.4 hb Exterior joints
Note: beff: effective slab width; bw: beam width; hb: beam height; l0: effective span of
beam; t: slab thickness; sn: clear distance of beams.
1/ 50
1/ 50
1/ 50
1/ 50
1/ 50
Page 98
67
CHAPTER 3
EXPERIMENTAL PREPARATION AND SPECIMEN
DETAILS
3.1 INTRODUCTION
The two groups of specimens shown in Table 3.1 were tested in order to
verify the effect of the proposed weakening techniques when realizing the
SCWB hierarchy. All the reinforced concrete (RC) joints of the two groups were
designed according to an earlier version of the Chinese code (GB-50011 2001)
to reflect the design philosophy of the current RC joint. The key design
principles are summarized below:
(1) For the control 3D beam-column joints, the flexural strength ratios of
column to R-beam (i.e. left( ) / ( )top bottom rightc c r b r bM M M M ) were more than 1.2
and those for the columns to T-section beams (T-beams) (i.e.
left( ) / ( )top bottom rightc c t b t bM M M M ) were less than 1.0.
(2) For retrofitted seismic resisting joints, the flexural strength ratios of columns
to T-beams (i.e. left/ top bottom rightc c t b t bM M M M ) were higher than 1.0.
(3) Flexural failure precedes failure due to shear.
Page 99
68
(4) Joint shear failures should not occur before a beam or column failure.
(5) The column axial compression ratio was set to be 0.2.
(6) Openings or gaps are located at regions critical in flexure.
Except for the beam-column joint control specimens, the specimens of
each test group were weakened by one of web openings, transverse grooves on
the beam soffit, slits at the slab corner regions, or a combination of the above.
To avoid potential non-ductile shear failure due to the removal of concrete when
creating openings or transverse grooves, those weakened beams were
strengthened in shear using FRP U-jackets or combined FRP U-jackets and FRP
wrap.
In addition to the differences in the size as well as position and shape of
openings, grooves and slits, there are some other differences between the two
groups of specimens: (a) the flexural capacity of longitudinal beams of the
second test group was increased; (b) the joint region was provided with more
transverse reinforcement; (c) the loading protocols of the two series were
different. The second test group was designed based on the test results of the
first one to better fulfill the test objectives.
3.2 SPECIMEN DESIGN DETAILS
The second-degree earthquake-resisting rankings specified in the Chinese
Page 100
69
code (GB-50011 2002) were adopted for the design and the corresponding
design strengths for both concrete and reinforcement. When naming these
specimens, the first letter of the labels, namely F and S, stands for first and
second group respectively. The letters O, G, S in the middle of the labels stand
for opening, groove, and slit, respectively. The number after these letters strands
for the dimensions of the opening, groove and slit. For the first test group,
16-mm-diameter deformed steel bars of 360 MPa design yield stress were used
for longitudinal bars in beams and columns. 8-mm-diameter plain steel bars of
300 MPa design yield stress were used for stirrups in beams, columns and slabs.
Joint regions were reinforced with 10-mm-diameter plain steel bars of 300 MPa
design yield stress. The elastic modulus of steel, sE was set to be 200kN/mm2.
C30 concrete with a target cube strength of 30.0 MPa, was adopted for the test
specimens. The beam shear span was 1400 mm and the inflection point of
column to the joint region was 1000 mm. The design details for the first test
group are listed in Table 3.1 and presented in Figure 3.1. Reinforcing and
geometrical details for all specimens are identical except for the stirrups and
slab longitudinal bars at the openings or cut slit positions. Eight column
longitudinal bars were uniformly distributed on each side of the columns. Four
16-mm-diameter bars were located in the tops of beams, close to the slab flange
and three were located in the lower part of the beams. A total of 16
8-mm-diameter bars were used to reinforce the slabs. Beams and column
stirrups were distributed at 80-mm centre-to-centre spacing along the length of
the beams to ensure that shear capacity of the beams is much higher than the
Page 101
70
shear force corresponding to the moment capacity.
In the first test group, the specimen F-O-450-150 had two 450 mm long
and 150 mm high openings in the beam web adjacent to the slab bottom surface
and the beam-column connection. To avoid potential shear failure due to the
removal of concrete and stirrups, FRP U-jackets and FRP wraps were both used
to strengthen the weakened region. FRP fibre anchors were carefully designed
to guarantee that the U-jackets fail by FRP rupture.
Specimen F-G-50-200 had two 50 mm long, 200 mm deep grooves
adjacent to the beam-column connection on the beam soffits. To control crack
growth, FRP U-jackets were used to strengthen the regions around gaps. FRP
anchors were also used.
Specimen F-S-450-450 contained four 450-mm-long longitudinal slits and
four 450-mm-long transverse slits in the corner region to separate the slab from
the major and minor beams. Slab bars which crossed the slit locations were cut.
Specimen F-O-500-180 was weakened by including two 500 mm long and
180 mm high openings which were similarly strengthened against shear using
FRP U-jackets as in specimen F-O-450-150. Unlike the other weakened
specimens of this group, the openings in this specimen were formed by drilling
after concrete casting. The dimensions of these openings were determined after
the testing of other specimens in the first test group.
Page 102
71
The actual materials properties, especially of concrete, of the first test
group were found to be higher than the expect values of concrete of 30 MPa
cylinder strength. The second test group was re-designed using 40 MPa cylinder
strength concrete. The beams in the second group were reinforced with six
20-mm-diameter deformed steel bars (instead of seven 16-mm-diameter steel
bars) with a 360 MPa design yield stress in the longitudinal direction. This
modification increased the beam strength to better realize column end failure of
the control specimen, which did not happen in the control specimen of the first
test group. To avoid premature joint failure in shear as observed when testing
specimen F-Control, the joint regions of the specimens in the second test group
were reinforced with 2 more 10-mm-diameter transverse reinforcing bars. The
design details of this second test group are given in Table 3.1.
Specimen S-G-50-200-100 contained grooves identical in size with those
of specimen F-G-50-200, but the grooves were located 100 mm distant from the
beam-column interface to avoid slippage of the beam bottom longitudinal bars
in the joint region. Specimen S-O-500-180 had slot-shaped openings, which
were different in shape from that of the first test group, as shown in Figure 3.2.
The creating of slot-shaped opening aims to weaken the T-beam like rectangular
one but with less concrete to be removed. Specimen S-O-500-180-S-300-300
was weakened with a combination of 500×180 mm2 web openings and 300
mm long slab slits.
Page 103
72
3.3 SPECIMEN CONSTRUCTION
Because of the limited laboratory space in the HK Polytechnic University,
all specimens were constructed in the structural laboratory at the Guangdong
University of Technology. Specimen construction progress is illustrated in
Figure 3.3. The specimen formwork was constructed first. Due to the existence
of slab, the formwork was constructed vertically, in two phases. In the first
phase, formwork for the beam, slab, and bottom column parts was built. The
formwork for the second phase, for the top column frames, was erected after the
reinforcement cages had been placed in the lower formwork mould. The gaps in
the formwork, where wooden plates abutted were sealed with waterproof glue to
prevent mortar/water leakage during specimen casting. The formwork was
strong enough to avoid any obvious deformation during and after concrete
casting. To create openings, transverse grooves or slab slits, wooden boxes or
plates of the required size were placed at the appropriate locations in the
formwork prior to casting. Finished beam and column reinforcement cages were
placed in the formwork first. The placing of slab reinforcement followed.
Strain gauges were bonded to reinforcing bars and stirrups at critical
positions. To avoid damage during concrete casting, the strain gauges were
covered with waterproof and quakeproof glue. Each strain gauge wire was given
a unique label name to indicate the locations of the strain gauges
Mortar blocks with the required thickness were applied to guarantee the
Page 104
73
thicknesses of concrete cover at different locations, as indicated in Table 3.2.
The specimens were cast when all the above processes had been completed.
C30 (the actual cylinder strength were about 40 MPa, which was not as
designed) and C40 commercial concrete with at least a 120-mm concrete slump
were used for the first and second batch specimen respectively because of the
high density of the steel reinforcements. All specimens were cast simultaneously
to guarantee concrete uniformity. At least 1 concrete cube and 3 cylinders were
prepared for each specimen.
After concrete casting, the specimens were thoroughly wetted three times
each day and the specimen surfaces covered with plastic sheeting to slow down
water evaporation. Two weeks after casting, the formwork was dismantled and
the curing of the whole specimens continued.
The smaller beam chord under the opening (referred to as the web chord
hereafter) was strengthened by FRP wrapping. The FRP splicing length was
greater than 150 mm to avoid failure of the splicing. As FRP U-jackets are
vulnerable to FRP debonding at the free end which lowers the strengthening
efficiency of the U-jackets, FRP fibre anchors were used to prevent that
debonding. The key external bonding processes of CFRP sheets to substrate
concrete and of the installation of FRP anchors in specimens are illustrated in
Figure 3.4. The radius of the rounded beam corners was 25 mm to allow for a
more effective FRP confinement. The surfaces of strengthened regions were
Page 105
74
polished to remove external mortar before the strengthening. The FRP anchors
were prepared first. The drilled holes for FRP anchors were firstly filled with
Sikadur 330 and the straight part of FRP anchors, which was already saturated
with Sikadur 300, was then placed into the hole. After the anchors were installed,
CFRP sheets saturated with Sikadur 300 were bonded to the prepared beam
surfaces following a wet lay-up process, also saturated with Sikadur 330. Some
details of the FRP shear strengthening scheme are shown in Figure 3.5. The
anchor fan had a fan angle of approximately 36°. After saturation with Sikadur
300, the anchor was pressed/bonded to the surface of the corresponding FRP
sheet to anchor the FRP sheet.
The opening of specimen F-O-500-180 and the weakened specimens of the
second test group were created at least 28 days after concrete casting. The
procedure for drilling an opening in an existing RC structure is illustrated in
Figure 3.6. An opening was first created and then polished. The corners of the
critical beams were rounded and polished before application of the externally
bonded FRP.
3.4 MATERIAL PROPERTIES
The average concrete cylinder strength 'cf and corresponding strain o
were obtained by testing three concrete cylinders. According to GB/T
50081-2002, for C30 concrete, the loading rate is suggested to be in the range
0.3-0.5 MPa/s. When C30-C60 concrete is used, the loading rate is increased to
Page 106
75
0.5-0.8 MPa/s. If the concrete grade is higher than C60, an 0.8-1.0 MPa/s
loading rate should be used. The cylinders were capped with plaster of 80 MPa
compressive strength before testing. The stress-strain curves for concrete
cylinders under uni-axial compression of the two groups are shown in Figure
3.8. The summarized data of the first and second test group are presented in
Table 3.3.
For each type of steel bar, at least 3 coupons were tested to obtain an
average yield stress yf and a corresponding yield strain y , an ultimate stress
uf and a corresponding ultimate strain u . Both strain gauges and the particle
image velocimetry (PIV) method were used to obtain steel deformations. The
deformation measured by strain gauges was used for the calculation of the
elastic modulus and the PIV method was used to obtain the hardening range of
the stress-strain curve, as most strain gauges fail at a strain of about 0.02.
However, it is of interest to note that the PIV method works beyond a strain of
0.2 and well into the large deformation range. For steel bars with an obvious
yield plateau, the initial elastic part of stress-strain curve was applied to
calculate elastic modulus. For those steel bars with no yield plateau, a “0.2%
off-set” criterion was adopted as the yielding point. The steel properties of the
two test groups are tabulated in Tables 3.4. The stress-strain curves obtained by
the PIV method are shown in Figures 3.8 for the two groups of specimens. Thus
it appears certain that the PIV method performs well in capturing the
stress-strain curves of steel bars over the complete range.
Page 107
76
Seven coupons of CFRP sheet were tested to determine the material
properties according to ASTM-3039 (2008). The width and length of the coupon
test region were 25 mm and 250 mm respectively. The nominal thickness of
CFRP sheet provided by the manufacturer (i.e., 0.334 mm per ply) was used to
calculate the tensile strength and elastic modulus. The obtained tensile strength
and elastic modulus were 2820 MPa and 227 GPa respectively.
3.5 LOADING PROTOCOLS
Combined vertical axial load and lateral cyclic load were applied at the top
of the upper columns. The axial compression ratio was set to be 0.2. This value
fluctuated to a small extent during loading, due to the limitations of the test
equipment.
The cyclic load was quasi-statically applied at the top of the upper column
by an MTS machine. The displacement history applied was defined by the
storey drift ratio (ratio of the horizontal displacement to the story height, which
is 2400 mm) rather than the yield deformation as it is difficult to predict the
yield deformation of a T-beam, especially when it is weakened by an opening, a
transverse groove or slab slits. The loading protocol as illustrated in Figure 3.9(a)
is defined according to that used in Canbolat and Wight (2008). For each
inelastic cycle, two reverse cycles were applied in both the push and pull
directions as many researchers did (e.g. Shin and LaFave 2004, Canbolat and
Wight 2008, Park and Mosalam 2012). Between some inelastic cycles, a
Page 108
77
reversed cycle of 0.5% storey drift ratio was inserted to quantify the stiffness
degradation (Canbolat and Wight 2008). Besides, in many seismic scenarios,
small cycles precede the main shock which can be represented by the inserted
low-level cycles. (ACI T1.1-01). The loading speed was slow to avoid
unnecessary additional inertia forces and to keep the axial load constant. The
loading sequence and protocol are depicted in Figure 3.9(a) and the applied
horizontal displacement magnitudes, durations and speeds are tabulated in Table
3.5. In the second test group, the first loading cycle was decreased to a 0.25%
storey drift ratio to better estimate specimen initial stiffness. There were no
inserted reversed cycles of 0.5% storey drift ratio as the inelastic cycles could be
used to assess stiffness degradation. Besides, the loading speed was very slow
and the loading was stopped at the peak displacement of each loading cycle for
recording. The test results of the first group showed that the inserted small
cycles did not have any significant effect, so they were not used in the second
group of tests. The loading protocol of the second test group is presented in
Figure 3.9(b) and tabulated in Table 3.5.
3.6 TEST SET-UP
As shown in Figure 3.10, a set of experimental devices had been designed
to provide the required boundary conditions. The hinges were used to represent
the inflection points of beams and columns. The beam ends were connected to
the supports using holes through the beam web. The foot of the lower column
was connected to a supporting hinge and the top of upper column was connected
Page 109
78
to the MTS head.
The horizontal cyclic and axial loads were applied by a 50-ton MTS
actuator and a 320-ton hydraulic jack respectively. The jack was pulled and
pushed horizontally by the MTS and slid with the help of a sliding plate with a
maximum sliding displacement of 150 mm. Load cells measured the axial load
and the beam-end reaction forces.
To measure the external deformation of beams and columns, screw rods
were used to locate linear variable differential transformers (LVDTs). For the
first test group, the rod screws were placed at pre-defined locations of formwork
before concrete casting. However, some pre-located screw rods were damaged
during casting concrete. Thus, for the second test group, the rod screws rods
were placed and fixed at the designed positions through holes drilled after
concrete casting.
3.7 INSTRUMENTATION
During testing, external and internal instrumentation procedures were used
to monitor the following parameters: (1) beam-end reaction forces,
top-of-column axial and horizontal loads; (2) the deformations of beams,
columns and joints; (3) strains in the steel reinforcements and strengthening FRP.
Load cells, LVDTs and strain gauges were used to measure the load,
displacement and strain respectively.
Page 110
79
3.7.1 Beam end Reaction Forces, Column Top Axial and Lateral loads
A 50-ton MTS actuator and a 320-ton jack applied the cyclic lateral load
and top-of-column axial load respectively. The horizontal load was recorded by
the MTS control system. The load cell connected to the hydraulic jack was used
to monitor axial load. The beam-end reaction forces were measured by the two
load cells above the beam-end supports. All load cells were calibrated before
testing.
3.7.2 Beam, Column and Joint Deformations
Similar to Kam (2010), a beam was divided into the three regions of elastic
region, plastic hinge region and bond failure region. The deformations of each
region under loading were measured by LVDTs located on the beam surface.
The arrangement and labelling of the LVDTs are shown in Figure 3.12. The
lateral displacement drift , of a beam-column joint was decomposed into
components expressed in the following equation:
, , , intdrift drift beam drift column drift jo (3.1)
, , ,hinge+drift beam elastic beam beam fixe end beam (3.2)
, , ,hingedrift column elastic column col (3.3)
where ,elastic beam and ,elastic column are the top-of-column lateral displacement
Page 111
80
components due to beam and column elastic deformations respectively.
beam fixe end is caused by a tensile bar bond failure at the beam-column
interface. ,hingebeam and ,hingecol are the components resulting from the
rotations of left and right beams, upper and lower columns plastic hinges
respectively, and int jo is caused by the joint shear deformation.
3.7.2.1 Beam deformation
The average curvature RBi of a beam ith region was determined using the
following equation:
. ,
= RBTi RBBiRBi
rb i rb ih b
(3.4)
where RBTi and RBTi are the total deformations measured by the top and
bottom LVDTs located on the respective ith region, .rb ih and ,rb ib are the
respective distances between the top and bottom LVDTs and the length of the
measured region, as shown in Figures 3.12.
For the specimens of the first test group, instrumented with 14 LVDTs, the
top-of-column lateral displacement components drift,beam , originating from the
beam drift beam , were calculated as follows:
Page 112
81
2drift,beam beamb
HL
(3.5)
1 1,1
.1
1 2 1 2,2 ,1
.2
1 1,1 ,2 ,1
.1
( 0.5 )
( ) ( ) ( 0.5 )
( 0.5 )
RBT RBBbeam b rb
rb
RBT RBT RBB RBBb rb rb
rb
LBT LBBb lb rb rb
lb
L bh
L b bh
L b b bh
(3.6)
where H is the distance between the upper and lower column inflection points ,
bL is the length of the beams. The first part of Equation (3.6) is the beam drift
component resulting from bar slippage. The second part accounts for plastic
hinge deformation and the third part represents the deformation in the elastic
region.
Due to the limited number of LVDTs, only two LVDTs were used to
measure the plastic hinge deformation of the left hand beam. The sum of the
bond failure deformation and the plastic hinge deformation were measured by
only one pair of LVDTs.
According to Kam (2010), the lateral drift component originating from
deformation of the beam elastic region can be determined using the following
basic structural mechanics equation:
3
, ,
,
2( )3 0.2 2b b elastic b b elastic
elastic,beamc e b c w b b
V L qV L HE I E b h L
(3.7)
Page 113
82
where bV is the shear force in the beam, measured by the load cell at the
beam-end support, ,b elasticL is the length of the elastic part of the beam, wb and
bh are beam web width and beam height respectively, ,c e bE I is the effective
flexural stiffness of the prismatic element taking into account the concrete
cracking effect. The effective moment of inertia, ,e bI , given by equation (3.6),
allows for the possible double counting of elastic deformations involved in the
measurement of average curvatures across the beam plastic hinge region; q is
the T-beam shape factor calculated using equation (3.7) according to mechanics
of materials theory:
, ,0.5e b g bI I (3.8)
2
,
( )bb
g b
A S yq dAI b
(3.9)
where ,g bI is the moment of inertia of the gross uncracked concrete area, Ab is
the total area of that concrete, S(y) is the area moment under the y point towards
the neutral axis and b is the section width (i.e. bw for the beam web and bf for
the beam flange).
3.7.2.2 Column deformation
Due to the limited number of LVDTs, only deformations in the column
hinge region were measured. The average curvature 1TC was obtained using
Page 114
83
following equation:
1 11
tc.1 tc,1
= TCR TCLTC h b
(3.10)
where 1TCR and 1TCL are the average deformations measured by the right
hand and left hand LVDTs respectively, installed on the upper column, tc.1h
and tc,1b are the distances between the two respective LVDTs and the length of
the region measured , as shown in Figure 3.13 (a).
The lateral drift component drift,column caused by the upper column hinge
deformation ,hinge,col top was determined using the following equation:
1 1 1 1ct,1( 0.5 ) ( 0.5 )TCR TCL BCR BCL
col,hinge,top c c bt,1tc.1 bc.1
H b H bh h
(3.11)
where cH is the length between the column-beam interface and the point of
inflection.
Similar to beams, the elastic deformation of the upper column is calculated
using Equation 3.12:
3
c, c,
,c3 0.2c elastic c elastic
elastic,column,topc e c c c
V H qV HE I E b h
(3.12)
where cV , obtained from the MTS data, is the horizontal load acting on the end
Page 115
84
of the upper column; c,elasticH is the length of the column’s elastic region, q
(=1.2) is the rectangular section shape factor calculated using Equation (3.10)
and cb and ch are column width and height respectively. The effective
moment of inertia, ,e cI , is given by Equation 3.13:
,c ,c0.7e gI I (3.13)
3.7.2.3 Joint shear distortion
If the joint deformation can be measured by two LVDTs based on the plane
strain cycle, then
1 cos(2 ) sin(2 )2 2 2
x y x y jj j
(3.14)
2 cos(2 ) sin(2 )2 2 2
x y x y jj j
(3.15)
( ) tan(2 )j x y j (3.16)
11
jL
(3.17)
22
jL
(3.18)
2 2j j jL h b (3.19)
Page 116
85
where 1 and 2 are the strains measured by the two respective inclined
LVDTs; x and y are the strains in the joint width and height directions
respectively, j is the angle of inclination of LVDT to the joint width direction;
j is the joint shear distortion; jL is the length of joint core along the diagonal
line and jh and jb are joint height and width respectively .
From Equations 3.15-3.20, the joint distortion j can be given by the
following equation:
2 21 1
2j j jj j
h bh b
(3.20)
The ,hingecol can be determined using following equation:
, int ( )cdrift jo j c b c
b
HH h hL
(3.21)
Due to the existence of the floor slab, the joint shear distortion cannot be
directly measured. Therefore, the deformation of a joint was monitored using
the interior strain gauges mounted on the transverse stirrups at the joint.
3.7.3 Strain Gauges
Due to the limited number of data logger channels, strain gauges were
situated on reinforcing bars only at the critical beams, slabs and columns
Page 117
86
positions, the joint transverse reinforcement, and the FRP. Strain gauges were
bonded on the exterior faces of the reinforcing bars. The arrangement and
labelling of the strain gauges on steel reinforcement are shown in Figures 3.13
to 3.21 and those on FRP reinforcement are in Figure 3.22. The meanings of the
labels are also illustrated in these figures.
3.7.3.1 Strain gauges on beam reinforcement
Those strain gauge with labels starting with TB (i.e. Beam Top) or BB (i.e.
Beam Bottom), were placed on reinforcing bars at the critical beam sections (i.e.
the section of maximum moment) and at weakened sections (i.e. those sections
weakened by the inclusion of an opening or gap). The purpose of the layout of
strain gauges chosen along the longitudinal direction was to measure the strain
distribution and the development of plastic hinges. The two specimens with
openings include more strain gauges along the longitudinal bars, covering the
complete lengths of openings so as to examine the effect of openings on strain
distribution.
In the first test group, the two strain gauges with labels starting with TS (i.e.
Transverse beam Stirrup), were symmetrically located on two stirrups of the two
transverse beams so as to monitor the symmetrical behavioural characteristics of
the specimens and the torsion in the transverse beams. The strain gauges were
not deployed in the second test group as they were found to be of little function.
3.7.3.2 Strain gauges on slab reinforcement
Page 118
87
The strain gauges with labels starting with TL (i.e. Top Longitudinal) or BL
(i.e. Bottom Longitudinal), were positioned parallel to the strain gauges on the
main longitudinal beam bars in the transverse direction. In this way, the shear
lag effect and the contribution of slab bars were evaluated. Three strain gauges
with labels starting with TT (i.e. Transverse Top), among the first test group,
were located on the slab transverse bars, with the aim of examining the effects
of longitudinal slits on those bars. After testing the first group of specimens,
these strain gauges were found to be of little value as the strain levels reached in
the transverse bars were very low. Consequently strain gauges were not used in
the second test group.
In the second test group, more strain gauges, instead, were placed on the
longitudinal bars in the right hand slab of specimen S-O-500-180 to estimate the
effects of a large opening on the deformation of the slab. The strain gauges
installed on the slab reinforcement for specimen S-O-500-180-S-300-30, with
slab slits separating the slab from beams, were located identically to those of the
other three specimens in the same group.
3.7.3.3 Strain gauges on column reinforcement
The strain gauge labels on the reinforcements of the top and bottom
columns began with the letters TC (i.e. Top Column) and BC (i.e. Bottom
Column) respectively. The function of the strain gauges was to measure the
plastic hinge length and the strain levels in the steel bars in columns. More
Page 119
88
strain gauges were used for the columns of the control specimens than for the
other specimens tested because the control ones had longer column plastic
hinges owing to the strong-beam-weak-column (SBWC) mechanism.
3.7.3.4 Strain gauges on joint transverse reinforcement
The strain gauges mounted on the joint stirrups were used to evaluate the
shear deformation of the joints, to assess the strain levels in the stirrups and
hence find whether a joint panel is sufficiently confined.
3.7.3.5 Strain gauges on FRP jackets
The strain gauges mounted on the FRP were located in areas of potentially
large tensile strains. According to an FEM prediction using ABAQUS, the FRP
U Jackets would experience maximum tensile strain in those areas of wide crack
propagation, whereas FRP wraps experience maximum tensile strain in areas
where concrete bulges under compression. The strain gauge labels had initial
letters U or W, to indicate FRP U-jackets or FRP Wraps. In the first test group,
strain gauges were only arranged on the FRP of the right beam. Thus, only U
and W were used to name these strain gauges. In the second test group, strain
gauges were pasted on both left and right beams and R and L were placed in
front of U and W to name these strain gauges.
3.7.3.6 The coordinate system
For the easy analysis of the test results, a Cartesion coordinate has to be
Page 120
89
defined. The directions of the coordinates are shown in Figures 3.13 to 3.21.
The X-axis is in the longitudinal direction of the main beams and the Y-axis is
parallel to the transverse beams. The Z-axis follows the longitudinal direction of
the columns. The origin of the coordinate system is located at the geometrical
centre of the joint regions.
3.7.3.7 Sign of top-of-column lateral displacement
When the specimens experienced a “push” action, the lateral displacement
was in the positive X direction and most of the slab to the right hand side of the
column was, therefore, under compression. When the lateral displacement was
in the opposite direction, the slab to the right side of column was in tension.
3.8 CONCLUDING SUMMARY
This chapter has presented the details of the experimental program. Two
test groups were designed and prepared. The following conclusions can be
drawn:
(1) The specimens were designed according to an old version of the Chinese
seismic design code (GB-50011 2001 Code for Seismic Design of
Buildings). The expected failure mode for control specimens was column
end failure. In terms of those retrofitted by the three proposed techniques,
the expected failure model was beam end flexural failure.
Page 121
90
(2) The control specimen of the second test group had stronger beams and joint
region than the control specimen F-Control to better realize column end
failure.
(3) The loading protocols were determined based on the storey drift ratio as the
yield displacement is hard to predict for specimens with a cast-in-place slab.
3.9 REFERENCES
ACI T1. 1–01. (2001). Acceptance Criteria for Moment Frames Based on
Structural Testing.
Canbolat, B. B., and Wight, J. K. (2008). "Experimental investigation on
seismic behavior of eccentric reinforced concrete beam-column-slab
connections", ACI structural journal, 105(2), 154.
GB-50011 (2001). Code for Seismic Design of Buildings, Architectural &
Building Press, Beijing, China (in Chinese).
Kam, W. Y. (2010). Selective weakening and post-tensioning for the seismic
retrofit of non-ductile RC frames, Doctoral degree thesis: The Hong Kong
Polytechnic University, University of Canterbury.
Park, S., and Mosalam, K. M. (2012). "Experimental investigation of nonductile
RC corner beam-column joints with floor slabs", Journal of Structural
Engineering, 139(1), 1-14.
Page 122
91
Shin, M., and LaFave, J. M. (2004). "Seismic performance of reinforced
concrete eccentric beam-column connections with floor slabs", Structural
Journal, 101(3), 403-412.
Page 123
92
(a) Elevation of specimen F-Control and F-S-450-450
(b) Elevation of specimen F-O-450-150
Openings
Page 124
93
(c) Elevation of specimen F-O-500-180
(d) Elevation of specimen F-G-50-200
Grooves
Openings
Page 125
94
(e) Elevation of specimen S-Control
(f) Elevation of specimen S-G-50-200-100
Grooves
Page 126
95
(g) Elevation of specimens S-O-500-180 and S-O-500-180-S-300-300
Page 127
96
(h) Plan view of specimens F-control, F-G-50-200, F-O-450-150 and F-O-500-180
Page 128
97
(i) Plan view of specimen F-S-450-450
Page 129
98
(j) Plan view of specimens S-Control, S-G-50-200-100 and S-O-500-180
Page 130
99
(k) Plan view of specimen S-O-500-180-S-300-300
Figure 3.1 Specimen reinforcement details
Page 131
100
(a) 50×200 mm2 groove, 100-mm away from beam-column interface of
specimen S-G-50-200-100
(a) Slot-shape opening of specimen S-O-500-180
Figure 3.2 The new groove position and opening shape
100 mm
180 mm
500 mm
Page 132
101
(a) Specimens formworks (b) Steel cages
(c) Placing steel cages into formworks (d) Specimens ready for casting
(e) Casting specimens (f) Concrete cubes and cylinders
Figure 3.3 Specimens construction process
Wooden boxes creating openings
Page 133
102
(a) Drilling holes for FRP anchors
(b) Applying primer-Sikadur 330 for anchors
Page 134
103
(c) Installation of FRP anchors with straight part saturated by Sikadur 300
(d) Applying primer-Sikadur 330 for CFRP sheets
Page 135
104
(e) Saturating CFRP sheet with Sikadur 300
(f) Attaching CFRP sheet to the specimen surface
(g) Saturating FRP anchor fan with Sikadur 300
Page 136
105
(h) Pressing FRP anchor fan on the surface of CFRP sheet
(i) Curing CFRP U-jacket, CFRP wrap and FRP anchors
Figure 3.4 Installing FRP anchors and externally bonding CFRP sheets to
substrate concrete
Page 137
106
CFRP Sheet
200
10090
80
(a) Details of FRP anchor
(b) Details of FRP shear strengthening scheme of specimen F-O-450-150
(c) Details of FRP shear strengthening scheme of specimen F-O-500-180
Page 138
107
(d) Details of FRP shear strengthening scheme of specimen F-G-50-200
(e) Details of FRP shear strengthening scheme of specimens S-O-500-180
and S-O-500-180-S-300-300
Page 139
108
(f) Details of FRP shear strengthening scheme of specimen S-G-50-200-100
(g) FRP anchor components
Figure 3.5 Details of FRP shear strengthening scheme (all units in mm)
Page 140
109
(a) Creating opening
(b) Grinding the inner surfaces of opening and rounding the corners of
chord and beam
Page 141
110
(c) Roughing surfaces around opening
(d) Opening ready for shear strengthening
Figure 3.6 Procedure of creating an opening in existing specimen
Page 142
111
(a) 66-days-old concrete of the first test group
(b) 116-days-old concrete of the first test group
0
5
10
15
20
25
30
35
40
45
0 500 1000 1500 2000 2500 3000
Stre
ss (M
Pa)
Strain (με)
Cylinder-1
Cylinder-2
Cylinder-3
0
5
10
15
20
25
30
35
40
45
0 500 1000 1500 2000 2500 3000
Stre
ss (M
Pa)
Strain (με)
Cylinder-1
Cylinder-2
Cylinder-3
Page 143
112
(c) Concrete of F-O-500-180
(d) 33-days-old concrete of the second test group
0
5
10
15
20
25
30
35
40
0 500 1000 1500 2000 2500 3000 3500 4000
Stre
ss (M
Pa)
Strain (με)
Cylinder-1
Cylinder-2
Cylinder-3
0
5
10
15
20
25
30
35
40
45
0 500 1000 1500 2000 2500 3000
Stre
ss (M
Pa)
Strain (με)
Cylinder-1
Cylinder-2
Cylinder-3
Page 144
113
(e) 73-days-old concrete the second test group
(f) 108-days-old concrete the second test group
Figure 3.7 Concrete stress-strain curves under uni-axial compression test
0
5
10
15
20
25
30
35
40
45
0 500 1000 1500 2000 2500 3000 3500 4000
Stre
ss (M
Pa)
Strain (με)
Cylinder-1
Cylinder-2
Cylinder-3
0
5
10
15
20
25
30
35
40
45
0 500 1000 1500 2000 2500 3000 3500
Stre
ss (M
Pa)
Strain (με)
Cylinder-1
Cylinder-2
Cylinder-3
Page 145
114
(a) D8 plain bars of the first test group
(b) D16 deformed bars of the first test group
0
50
100
150
200
250
300
350
400
450
500
0 50000 100000 150000 200000 250000 300000
Stre
ss (M
Pa)
Strain (με)
HPB335-D8-1
HPB335-D8-2
HPB335-D8-3
0
100
200
300
400
500
600
700
0 50000 100000 150000 200000 250000
Stre
ss (M
Pa)
Strain (με)
HRB400-D16-1
HRB400-D16-2
HRB400-D16-3
Page 146
115
(c) D8 plain bars of the second test group
(d) D10 plain bars of the second test group
0
100
200
300
400
500
600
0 50000 100000 150000 200000 250000 300000
Stre
ss (M
Pa)
Strain (με)
HPB335-D8-1
HPB335-D8-2
HPB335-D8-3
0
100
200
300
400
500
600
0 50000 100000 150000 200000 250000 300000
Stre
ss (M
Pa)
Strain (με)
HPB335-D10-1
HPB335-D10-2
HPB335-D10-3
Page 147
116
(e) D16 deformed bars of the second test group
(f) D20 deformed bars of the second test group
Figure 3.8 Stress-stain curves of steel bars obtained by PIV method
0
100
200
300
400
500
600
700
0 50000 100000 150000 200000 250000 300000
Stre
ss (M
Pa)
Strain (με)
HRB400-D16-1
HRB400-D16-2
HRB400-D16-3
0
100
200
300
400
500
600
700
0 50000 100000 150000 200000 250000 300000
Stre
ss (M
Pa)
Strain (με)
HRB400-D20-1
HRB400-D20-2
HRB400-D20-3
Page 148
117
(a) The first test group
(b) The second test group
Figure 3.9 Loading protocol for two test groups
Page 149
118
(a) Loading and measuring devices
(b) Reaction wall
Figure 3.10 Experimental setup
Page 150
119
Figure 3.11 Deformation disassembling of specimens
Page 151
120
(a) LVDTs arrangement for specimens F-Control, F-G-50-200, F-S-450-450
Labeling guide:
T: LVDT at the column Top
M: LVDT at the Middle
T/BCL/R: LVDT on the Top/Bottom
Column Left/ Right side
R/LBT/B: LVDT on the Right/Left
Beam Top/Bottom side
Page 152
121
(b) LVDTs arrangement for specimens F-O-450-150 and F-O-500-180
Page 153
122
(c) LVDTs arrangement for specimens of the second test group
Figure 3.12 LVDTs arrangement
Page 154
123
(a) Strain gauges distribution from front view and sectional view
Labeling guide: TC: Top Column bars BC: Bottom Column bars TB: Top Beam bars BB: Bottom Beam bars CS: Connection Stirrups
Labeling guide: TL: Top Longitudinal bars on slab BL: Bottom Longitudinal bars on slab
X
Z
Page 155
124
(b) Strain gauges distribution from top view
Labeling guide: TS: Transverse beam Stirrups TT: Top Transverse bars on slab
Push Direction
Pull Direction
X
Y
Page 156
125
(c) Strain gauges distribution on different bars
Figure 3. 13 The labels and distribution of strain gauges on steel reinforcement of specimen F-Control (all units in mm)
Page 157
126
(a) Strain gauges distribution from front view and sectional view
Labeling guide: TC: Top Column bars BC: Bottom Column bars TB: Top Beam bars BB: Bottom Beam bars CS: Connection Stirrups
Labeling guide: TL: Top Longitudinal bars on slab BL: Bottom Longitudinal bars on slab
X
Z
Page 158
127
(b) Strain gauges distribution from top view
Labeling guide: TS: Transverse beam Stirrups TT: Top Transverse bars on slab
X
Y
Push Direction Pull Direction
Page 159
128
(c) Strain gauges distribution on different bars
Figure 3. 14 The labels and distribution of strain gauges on steel reinforcement of specimen F-G-50-200 (all units in mm)
Page 160
129
(a) Strain gauges distribution from front view and sectional view
Labeling guide: TC: Top Column bars BC: Bottom Column bars TB: Top Beam bars BB: Bottom Beam bars CS: Connection Stirrups
Labeling guide: TL: Top Longitudinal bars on slab BL: Bottom Longitudinal bars on slab
X
Z
Page 161
130
(b) Strain gauges distribution from top view
Labeling guide: TS: Transverse beam Stirrups TT: Top Transverse bars on slab
X
Y
Push Direction Pull Direction
Page 162
131
(c) Strain gauges distribution on different bars
Figure 3. 15 The labels and distribution of strain gauges on steel reinforcement of specimen F-S-450-450 (all units in mm)
Page 163
132
(a) Strain gauges distribution from front view and sectional view
Labeling guide: TC: Top Column bars BC: Bottom Column bars TB: Top Beam bars BB: Bottom Beam bars CS: Connection Stirrups
Labeling guide: TL: Top Longitudinal bars on slab BL: Bottom Longitudinal bars on slab
X
Z
Page 164
133
(b) Strain gauges distribution from top view
Labeling guide: TS: Transverse beam Stirrups TT: Top Transverse bars on slab
X
Y
Push Direction Pull Direction
Page 165
134
(c) Strain gauges distribution on different bars
Figure 3. 16 The labels and distribution of strain gauges on steel reinforcement of specimen F-O-450-150 (all units in mm)
Page 166
135
(a) Strain gauges distribution from front view and sectional view
Labeling guide: TC: Top Column bars BC: Bottom Column bars TB: Top Beam bars BB: Bottom Beam bars CS: Connection Stirrups
Labeling guide: TL: Top Longitudinal bars on slab BL: Bottom Longitudinal bars on slab
X
Z
Page 167
136
(b) Strain gauges distribution from top view
Labeling guide: TS: Transverse beam Stirrups TT: Top Transverse bars on slab
X
Y
Push Direction Pull Direction
Page 168
137
(c) Strain gauges distribution on different bars
Figure 3. 17 The labels and distribution of strain gauges on steel reinforcement of specimen F-O-500-180 (all units in mm)
Page 169
138
(a) Strain gauges distribution from front view and sectional view
Labeling guide: TC: Top Column bars BC: Bottom Column bars CC: Central Column bars TB: Top Beam bars BB: Bottom Beam bars CS: Connection Stirrups
Labeling guide: TL: Top Longitudinal bars on slab BL: Bottom Longitudinal bars on slab
X
Z
Page 170
139
(b) Strain gauges distribution from top view
X
Y
Push Direction Pull Direction
Page 171
140
(c) Strain gauges distribution on different bars
Figure 3.18 The labels and distribution of strain gauges on steel reinforcement of specimen S-Control (all units in mm)
Page 172
141
(a) Strain gauges distribution from front view and sectional view
Labeling guide: TL: Top Longitudinal bars on slab BL: Bottom Longitudinal bars on slab
X
Z
Labeling guide: TC: Top Column bars BC: Bottom Column bars CC: Central Column bars TB: Top Beam bars BB: Bottom Beam bars CS: Connection Stirrups
Page 173
142
(b) Strain gauges distribution from top view
X
Y
Push Direction Pull Direction
Page 174
143
(c) Strain gauges distribution on different bars
Figure 3.19 The labels and distribution of strain gauges on steel reinforcement of specimen S-G-50-200-100 (all units in mm)
Page 175
144
(a) Strain gauges distribution from front view and sectional view
Labeling guide: TL: Top Longitudinal bars on slab BL: Bottom Longitudinal bars on slab
X
Z
Labeling guide: TC: Top Column bars BC: Bottom Column bars CC: Central Column bars TB: Top Beam bars BB: Bottom Beam bars CS: Connection Stirrups
Page 176
145
(b) Strain gauges distribution from top view
X
Y
Push Direction Pull Direction
Page 177
146
(c) Strain gauges distribution on different bars
Figure 3.20 The labels and distribution of strain gauges on steel reinforcement of specimen S-O-500-180 (all units in mm)
Page 178
147
(a) Strain gauges distribution from front view and sectional view
Labeling guide: TL: Top Longitudinal bars on slab BL: Bottom Longitudinal bars on slab
X
Z
Labeling guide: TC: Top Column bars BC: Bottom Column bars CC: Central Column bars TB: Top Beam bars BB: Bottom Beam bars CS: Connection Stirrups
Page 179
148
(b) Strain gauges distribution from top view
X
Y
Push Direction Pull Direction
Page 180
149
(c) Strain gauges distribution on different bars
Figure 3.21 The labels and distribution of strain gauges on steel reinforcement of specimen S-O-500-180-S-300-300 (all units in mm)
Page 181
150
(a) Strain gauges on the FRP of specimen F-O-450-150
(b) Strain gauges on the FRP of specimen F-G-50-200
(c) Strain gauges on the FRP of specimen F-O-500-180
Page 182
151
(d) Strain gauges on the FRP of specimen S-G-50-200-100
(e) Strain gauges on the FRP of specimens S-O-500-180 and
S-O-500-180-S-300-300
Figure 3.22 Strain gauges on FRP (all units in mm)
Page 183
152
Table 3.1 Specimens details of the two test groups
Test Unit
Beam width
× height
(mm×mm)
Column
width ×
height
(mm×mm)
Slab width
× height
(mm×mm)
Top
reinforceme
nt in beam
Bottom
reinforceme
nt in beam
Joint
transverse
reinforceme
nt
Top &
Bottom
reinforceme
nt in slab
Reinforcem
ent in
column
Stirrups Weakening
method
FRP
Strengtheni
ng method
First test
group
F-Control
400×200 320×320 650×100 4C16 3C16 B10@80 B8@150 8C16 B8@80
Control
specimen /
F-O-450-150 450 mm × 150
mm opening
FRP
U-jacket &
FRP wrap
F-G-50-200 50 mm × 200
mm groove
FRP
U-jacket
F-S-450-450 450-450 slits /
F-O-500-180 500 mm × 180
mm opening
FRP
U-jacket &
FRP wrap
Page 184
153
Table 3.1 (cont.)
Second
test group
S-Control
400×200 320×320 650×100 3C20 3C20 B10@60 B8@150 8C16 B8@80
Control
specimen /
S-G-50-200-1
00
50 mm × 200
mm groove
100 mm away
from column
face
Two-layer
FRP
U-jacket
S-O-500-180
500 mm × 180
mm
slot-shaped
opening
FRP
U-jacket &
FRP wrap
S-O-500-180-
S-300-300
500 mm × 180
mm opening
and 300-300
slits
FRP
U-jacket &
FRP wrap
Page 185
154
Table 3.2 Cover concrete thickness
Items
Longitudinal bars to each
sides of column
Longitudinal bars to
top/bottom side of beam
Longitudinal bars to
left/right side of beam
Longitudinal bars to
top/bottom side of slab
Cover concrete thickness (mm)
40 40 30 15
Table 3.3 Strengths of concrete cylinders of the two test groups
Test group
Days or Specimen
Items Cylinder-1 Cylinder-2 Cylinder-3 Average
First 66
Strength (MPa)
39.3 38.2 40.5 39.3 166 40.3 41.2 41.5 41
F-O-500-180 39.3 38.2 40.5 39.3
Second 33 36.9 40.9 35.2 37.7 73 38.6 40.5 39 39.3 108 40.9 38.2 40.3 39.8
Page 186
155
Table 3.4 Steel reinforcement properties of the two test groups
Test group
Type Number Weight
(g) Length (mm)
Density (kg/m3)
Area (mm2)
Original Gauge length
Lo (mm)
Ultimate Gauge length
Lu (mm)
Elongation after
Fracture
Yield Stress (MPa)
Average (MPa)
E (MPa) by Strain
Gages
Average (MPa)
First
D8 1 216.5 555 0.00785 49.69 40 51 27.50% 348
340 2.03E+05
2.00E+05 2 217.6 554 0.00785 50.04 40 50 25.00% 334 2.08E+05 3 217.1 552 0.00785 50.1 40 48 20.00% 338 1.88E+05
D10 1 298.2 482 0.00785 78.81 50 65 30.00% /
364.5 /
2.11E+05 2 308.9 499 0.00785 78.86 50 65.5 31.00% 369 2.10E+05 3 311 501 0.00785 79.08 50 65 30.00% 360 2.12E+05
D16 1 851.4 554 0.00785 195.77 80 93 16.25% 491
483.3 2.07E+05
2.00E+05 2 846.5 552 0.00785 195.35 80 97.5 21.88% 483 1.96E+05 3 847 556 0.00785 194.06 80 94.8 18.50% 476 1.97E+05
Page 187
156
Table 3.4 (cont.)
Second
D8 1 163 403 0.00785 51.52 40 50.7 26.75% 352.3
348.55 2.04E+05
2.04E+05 2 168 402 0.00785 53.24 40 50.7 26.75% 353.8 2.04E+05 3 162 400 0.00785 51.59 40 50.6 26.50% 344.8 2.03E+05
D10 1 254 403 0.00785 80.29 50 65.8 31.60% 354.5
353.5 2.11E+05
2.10E+05 2 252 400 0.00785 80.25 50 64.3 28.60% 350.7 2.08E+05 3 257 401 0.00785 81.64 50 / / 352.5 2.07E+05
D16 1 633 405 0.00785 199.1 80 / / 470.6
477.7 2.00E+05
1.99E+05 2 615 400 0.00785 195.86 80 100.7 25.88% 435.5 1.99E+05 3 619 402 0.00785 196.15 80 101.3 26.63% 484.8 1.98E+05
D20 1 970 401 0.00785 308.15 100 / / 460.3
453.6 1.96E+05
1.96E+05 2 979 405 0.00785 307.93 100 / / 441.2 1.96E+05 3 981 406 0.00785 307.8 100 / / 446.9 1.96E+05
Page 188
157
Table 3.5 Loading schemes of the two test groups
`
Drift ratio
Direction 0.25% 0.50% 1% 1.50% 2% 2.50% 3% 4% 5%
Push(mm) 6 12 24 36 48 60 72 96 120
Pull (mm)
First
Cyclic times
/ 1 2 2 2 2 2 2 2
Speed (mm/s)
0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4
Second
Cyclic times
1 1 2 2 2 2 2 2 2
Speed (mm/s)
0.1 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4
Page 190
159
CHAPTER 4
EXPERIMENTAL STUDY OF THE SEISMIC
PERFORMANCE OF RETROFITTED RC
BEAM-COLUMN JOINTS
4.1 INTRODUCTION
The experimental results for the two test groups introduced in Chapter 3
will be presented in this Chapter. The general test observations including the
failure process and failure mode, hysteretic performance and envelope curves,
and interior deformation of the test specimens are presented. Also included for
discussion to further illustrate the effect of the proposed retrofit techniques on
the seismic performance of the specimens are ductility properties, energy
dissipation characteristics, equivalent viscous damping, stiffness degradation,
plastic hinge lengths and deformation components. Conclusions are then drawn
and presented.
4.2 FAILURE PROCESS AND FAILURE MODE
The two test group were to investigate the effect of the three proposed
retrofit techniques, the slab slit (SS) technique, the beam opening (BO)
technique and the beam section reduction (SR) technique (e.g., creating a deep
transverse groove (TG)), on the performance of RC beam-column-slab joints
Page 191
160
(RC 3D joints) subjected to a combination of horizontal cyclic and axial loads.
The failure process/failure modes (including crack propagation process, steel
yielding) are presented below. Among the five specimens of the first group, the
control specimen unexpectedly failed due to shear failure near the supports at
the beam ends. Thus, the beam ends of remaining specimens in the first test
group were all strengthened, in this regard, with FRP sheets, to prevent this
unwanted failure mode from interfering with the purposes of the study. The
failed control specimen beam ends were strengthened by recasting the beam
ends with fresh concrete and strengthened by FRP sheets as shown in Figure 4.1.
The strengthened control specimen was retested until the lateral storey drift ratio
reached 5%. In the first test group, the BO technique was investigated by testing
specimens F-O-450-150 and F-O-500-180, the TG and SS techniques by
specimens F-G-50-200 and F-S-450-450 respectively.
The specimens of the second test group were designed based on the
experimental results/observations of the first test group. As the joint region of
the specimens in the first test group was not strong enough to ensure that the
failure of the beams and/or columns occurs first, the joint region of the second
test group was provided with 50% more transverse stirrups. The
moment-capacity of T-beams was also enhanced by increasing the area of
longitudinal bars. With these enhancements, the design principles mentioned in
the former chapter can be better followed.
In the second test group, except for one control specimen, the other three
Page 192
161
were retrofitted using one of the proposed techniques or a combination. For
specimen S-O-500-180, a new slot-shaped opening was tried. The slot-shaped
opening was 500 mm in length and 180 mm in height. The obvious advantage of
the slot-shaped opening over the rectangular is its smaller open area. For
specimen S-G-50-200-100, 50x200mm transverse grooves were placed 100 mm
distant from the beam-column interface to avoid beam bottom bars slipping
failure. For the third retrofitted specimen S-O-500-180-S-300-300, a
combination of the WB and SS techniques was applied for retrofitting purposes.
The experimental results are given below.
As mentioned in the previous chapter, during the testing, an axial load was
first applied at the end of the upper column first. When the axial load ratio
reached 0.2, the horizontal load was applied. Failure process and crack
behaviour evolution with the increase of displacement, are summarized in Table
4.1. Selected photographs taken during the testing are presented in Figures 4.2
to 4.10 to better indicate the cracking behaviour and failure mode as well. It
should be noted that the blue and red lines on the specimens identify the cracks
which arose when the specimen had undergone particular numbers of loading
cycles.
4.2.1 Cracking of Beams, Columns, Slabs and Joint Panel in the First Stage
Loading
Drift ratios when cracks first appeared on the beams, columns and joint
Page 193
162
panel of specimens are tabulated in Table 4.1. The crack distributions of
different stages are shown in Figures 4.2 to 4.10. Except for specimen F-Control,
whose flexural cracks initiating were first observed at the bottom of the beam
when the axial compression ratio reached 0.2, due to axial deformation of the
column, specimens of the first test group got beam flexural cracks during the
first cycle of 0.5% drift ratio. Figure 4.2(a) shows the cracks on beams of
specimen F-Control at 0.5% drift ratio. Shear cracks on the joint panel and
flexural cracks on the bottom column of specimen F-Control happened at 1.0%
drift ratio, as shown in Figure 4.2(b). For specimen F-G-50-200, by the end of
the cycle of 0.5% drift ratio, many flexural cracks had developed at the bottom
of the beam, with fewer at the top of slab, as shown in Figure 4.3(a). A crack
also showed on the bottom. At 1.0% drift ratio, cracks appeared on the joint
region of specimen F-G-50-200 as shown in Figure 4.3(b). As shown in Figure
4.4(a), flexural cracks arose on the slab and beams around the corners of the
beam openings of specimen F-O-450-150 at 0.5% drift ratio. Flexural cracks
emerged in the bottom column but fewer in the top column at 1.0% drift ratio.
Shear cracks arose on the specimen F-O-450-150 joint panel at 1.5% drift ratio,
as indicated in Figure 4.4(c). For specimen F-O-500-180, some flexural cracks
arose from the beam bottom surface and extended towards the FRP anchors
during the cycle 1 of 0.5% drift ratio as indicated in Figure 4.5(a). Cracks in the
slab emerged from the top corners of the openings. Meanwhile, one flexural
crack showed on the lower column. During the following two cycles of 1.0%
drift ratio, new cracks, close to the transverse beam, appeared on the specimen
Page 194
163
F-O-500-180 bottom column and joint panel, as shown in Figure 4.5(b). As
shown in Figure 4.6(a), many flexural cracks arose on the specimen
F-S-450-450 beam and some on its lower column and the slab during cycle 1.
The cracks on the slab emerged from the end of transverse slits and extended to
the slab edge.
In the second test group, the right and left beam tip loads were released to
zero after applying an axial load of 0.2 axial compression ratio. The drift ratio of
the initial cycle was decreased to 0.25%, rather than 0.5% in the first test group.
For control specimen S-Control, some flexural cracks arose from the beam
bottom surface during cycle 1 of 0.25% drift ratio, shown in Figure 4.7(a). A
flexural crack showed on its bottom column and one on its left slab. At 1.0%
drift ratio, several shear cracks showed on the joint panel around the transverse
beam, as indicated in Figure 4.7(b). As shown in Figure 4.8(a), a flexural crack
arose on the top surface of the specimen S-G-50-200-100 left slab during cycle
1 of 0.25% drift ratio. Flexural cracks showed on the bottom column of
specimen S-G-50-200-100 at 0.5% drift ratio. At the following drift ratio, shear
cracks arose on the joint panel as presented in Figure 4.8(c). As shown in
Figures 4.9(a) and 4.10(a), flexural cracks on the slab and beams of specimens
S-O-500-180 and S-O-500-180-S-300-300 arose during the second cycle of 0.5%
drift ratio. At the 1.0% drift ratio, several flexural and shear cracks arose on the
columns and joint panel respectively of the two specimens as presented in
Figures 4.9(b) and 4.10(b).
Page 195
164
4.2.2 Yield Point and Peak Point
As indicated in Table 4.1, for the first test group, specimens F-Control and
F-O-500-180 reached their strengths at 2.0% drift ratio, Specimens F-G-50-200
and F-S-450-450 at 1.5% and specimens F-O-450-150 at 2.5% drift ratio. For
the second test group, all specimens reached their strength at 2.0% drift ratio.
All specimens of the two test groups, except for specimen S-G-50-200-100, had
all their beam and slab bars yielded before they reached their peak loads.
Specimen S-G-50-200-100 had all its slab bars yielded at 4.0% drift ratio. This
is probably because the positions of strain gauges on its slab bars were not
transversely parallel to the beam groove region where beam flexural
deformation was concentrated. Except for specimens F-G-50-200 and
S-G-50-200-100, in which joint stirrups did not yield, all specimens had a port
of or all joint stirrups yielded during test. Specimens F-Control, F-O-450-50,
S-Control and S-O-500-180 also have obviously column bars obviously yielded.
For specimen F-Control, shear cracks arising near the beam pinned region
were wide at 2.0% drift ratio as indicated in Figure 4.2(c). The joint panel also
had many shear cracks at this drift ratio. For specimen F-G-50-200, when the
drift ratio was increased to 1.5%, cracks due to beam bottom bars slippage were
clear as indicated in Figure 4.3(c). Specimen F-O-450-150 got many cracks
arose on the bottom column due to the pull force provided by the web chords at
2.5% drift ratio, as shown in Figure 4.4.(c). The main beam longitudinal bars of
specimen F-O-500-180 yielded under negative bending at the drift ratio of 1.5%.
Page 196
165
A joint stirrup yielded, as indicated by readings of strain gauges at this drift
ratio. At the 2.0% drift ratio, more cracks appeared near the FRP anchors due to
the pull-out action of the FRP anchors as shown in Figure 4.5(c). The beam
longitudinal bars of specimen F-S-450-450 yielded under negative bending
when the drift ratio was increased to 1.5%. At the drift ratio of 2.0%, two
middle stirrups in the joint region yielded. The slab cracks initiating from the
transverse slit ends were widely open and the number of cracks on the joints
panel were obviously increased, as indicated in Figure 4.6(c). At a drift ratio of
2.0%, many cracks arose around the specimen S-Control transverse beam as
shown in Figure 4.7(c). As the drift ratio grew to 2.5%, some joint stirrups close
to its bottom column yielded. The specimen S-G-50-200-100 beam longitudinal
bars yielded under negative bending at 1.5% drift ratio. At the drift ratio of
2.0%, many cracks arose around its transverse beam and bottom column as
shown in Figure 4.8(c). The specimen S-O-500-180 beam top bars and slab bars
yielded under negative bending at 1.5% drift ratio. As the drift ratio was
increased to 2.0%, several cracks originating from the anchor positions showed
on the top and bottom surfaces of slabs as shown in Figure 4.9(c). Many cracks
showed on the top and bottom columns at this drift ratio. At the drift ratio of
2.5%, two specimen S-O-500-180 joint middle stirrups yielded. The top beam
bars and the uncut slab bars of specimen S-0-500-180-S-300-300 yielded at the
1.0% drift ratio. Afterwards, at the drift ratio of 2.0%, many cracks arose on the
top, bottom columns and around the transverse beam.
Page 197
166
4.2.3 Failure Mode
For specimen F-Control, due to shear failure caused by the very large
openings originating within the beam pinned region as shown in Figure 4.2(d),
the test was stopped at a drift ratio of 4.0% during push loading. At the 4.0%
drift ratio, the beams of this specimen had entered their yield state. The failed
control specimen beam ends was then strengthened as shown in Figure 4.1. Old
concrete was removed and replaced with new concrete of 40 MPa cylinder
strength. To avoid shear failure, steel bars were used to strengthen the pinned
region. A wooden mould was built to protect the original shape of the specimen.
The strengthened control specimen was then further tested starting from the drift
ratio reached at the test stoppage point. The strengthened control specimen
finally failed by joint shear failure as shown in Figure 4.2(e), rather than column
end failure. This unexpected failure mode was due to actual strengths of the
steel reinforcement and concrete involved were not as designed. The
deformation of specimen S-Control was concentrated at the bottom column
hinge region at the later drift ratio. As shown in Figure 4.7(e), the bottom
column concrete at the column-beam intersection was crushed. Specimen
S-Control failed because the bottom column bars buckled and were unable to
sustain the axial load. The buckling of bottom column bars is clearly shown in
Figure 4.7(e).
The failure mode of specimen F-G-50-200 is shown in Figure 4.3(d).The
beam bottom bars slippage in the joint region was the cause of the failure.
Page 198
167
Deformation of specimen S-G-50-200-100 was also clearly concentrated at the
beam hinges formed at the beam groove region. Concrete was pulled out when
the beam bottom bars slipped as shown in Figure 4.8(e). The failure mode of
specimen S-G-50-200-100 is similar to that of specimen F-G-50-200, even
though the grooves were placed 100 mm away from the beam-column interface,
with the aim to avoid beam bottom bars slippage.
In F-S-450-450, a large opening appeared at the beam-to-column interface
and also on the slab, starting from the transverse beam as shown in Figure
4.6(d).The specimen F-S-450-450 achieved the designed failure mode as 75%
slab longitudinal bars were removed.
For specimen F-O-450-150, damage and deformation were concentrated at
the joint region as shown in Figures 4.4-(e) due to joint shear failure. The test
result indicated that the 450x150 mm2 opening size was not sufficiently large to
enable a switch from joint shear failure mode to flexural failure. Thus, a larger
opening size was designed for another specimen F-O-500-180. As shown in
Figure 4.5(d), the existence of a four hinges mechanism (four hinges formed at
the ends of the two chords) was obvious in this specimen. For specimen
S-O-500-180, bottom column concrete at the column-beam intersection was
crushed and the column bars buckled when the specimen failed, as shown in
Figure 4.9(d). Even though the opening size of specimen S-O-500-180 is the
same to that of specimen F-O-500-180, the beams of specimen S-O-500-180
were stronger as they were provided with more longitudinal steel reinforcement.
Page 199
168
The joint region of specimen S-O-500-180 was also provided with more
transverse steel reinforcement. Thus the failure of this specimen happened on
the bottom column. When slab slits were provided to weaken the T-beam
moment capacity in addition to 500×180 mm2 slot-shaped openings, the
specimen S-O-500-180-S-300-300 failed in a ductile manner, as indicated in
Figure 4.10(d). The concrete at the top part of the T-beams of specimen
S-O-500-180-S-300-300 crashed when the specimen failed. The deformation of
main beams obviously followed the four hinges mechanism.
The specimen test results of the two test groups, as presented above,
demonstrated that three proposed retrofit techniques have great effects on the
seismic performance of 3D RC joints if they are well designed. The existence of
a transverse groove can greatly lower the moment-capacity of a T-beam.
However, the failure mode was non-ductile because the main beam bottom
longitudinal bars slipped through the joint region even though the groove was
not placed close to beam-column interface. The slab slitting method can
effectively lower the moment capacity of a T-beam as it removes the
contribution of the cut slab longitudinal bars. Meanwhile, the failure mode is
still ductile. The beam opening size should be well designed (e.g. increasing the
opening size) to lead to an obvious four-hinges mechanism. The slab slits can
help a specimen, not sufficiently weakened by the beam opening technique, to
realize a failure following the four-hinges mechanism.
Page 200
169
4.3 HYSTERETIC BEHAVIOUR AND ENVELOP CURVES
4.3.1 Hysteretic Behaviour
Hysteresis curves of axial compression load are shown in Figure 4.11. The
axial compression load was kept almost unchanged with a relatively small
variation due to limitation of loading equipment. The variation were as big
when lateral drift ratio was close to 5% for specimens S-Control and
S-O-500-180 as at the drift ratio, the two specimens failed due to bottom
column bars buckling and capacity for axial loading dropped obviously.
The hysteretic properties of the beams and the upper column of the
specimens are shown in Figures 4.12 and 4.13. The beam tip loads (the loads at
the beam end support) of the two T-beams were recorded by the load cells
installed in the beam end supports. The figures indicate that the hysteretic
performances of the two T-beams were almost the same. The beam tip loads
were used to calculate the column shear force using Equation (4.1):
(( ) ) /c b right b left b drift cF F F L N d L / 2 (4.1)
where cF is the column shear force and b rightF and b leftF are the beam tip
loads of the right and left T-beams respectively; bL and cL represent the
lengths of beams and columns from the centre of the joint panel respectively;
driftd is the lateral displacement.
Page 201
170
The absolute load value recorded by the MTS machine was higher than
that of the column shear force as the former includes additional friction forces,
as demonstrated in Figure 4.14. As shown in these figures, the effect of friction
was quite significant during the pull action, varying between different test
specimens. This effect could not be accurately measured. The predicted column
strengths shown in Table 4.2 were based on cross-sectional analysis using actual
material properties. Second-order effect was incorporated when calculating the
predicted column strengths. The calculated column shear force was obtained
using Equation (4.1). In Table 4.2, the weakening ratio represents the weakening
degree when comparing the sum of the absolute values of the measured T-beam
negative and positive strengths to that of the control specimens.
The specimen F-control had a gradual yielding load-displacement curve
due to the existence of a wide slab and reached its strength at the drift ratio of
2.0%. The strengths of the left and right T-beams were close to each other as
indicated in Figure 4.12(a).The positive and negative yield loads of the T-beams
were 87.2 kN and 156.6 kN respectively, close to those derived from sectional
analysis. These values were obtained before the beam end of the control
specimen was strengthened. The strengthened specimen was tested at drift ratios
of 4.0% and above. As shown in Figure 4.13(a), the retrofitted specimens had
relatively higher strengths in the push direction, which might be the result from
a regain of strength after being retrofitted. In the pull action, the strength was
lower due to the damage caused during the push action. Pinching behaviour was
Page 202
171
obvious and the reloading stiffness decreased with increased lateral drift ratio.
Specimen F-G-50-200 had a relatively short yielding process and reached
its strength at 1.5% drift ratio. Obviously different from the other specimens, the
specimen retrofitted by TG technique failed quickly after reaching its strength
as shown in Figures 4.12(b) and 4.13(b). According to damage propagation
observations, the slippage of bottom beam bars in the joint region was the major
cause of this non-ductile failure mode. The measured strength of the T-beams
under positive and negative loadings was 87.4 kN and 85.6 kN respectively.
Though the sum of these values was close to that of the R-beam, such a
non-ductile failure mode is not permitted for seismic design. Hence, the TG
retrofit method should be redesigned to prevent bottom beam bar slipping
failure.
Specimen F-O-450-150 also displayed gradual yielding and reached its
strength at 2.5% drift ratio. The strength measured in the push action indicated
in the right hand part of Figure 4.13(c) was a little higher than those due to pull
action as shown in the left hand part of the figure. The T-beams reached their
negative strengths at 4.0% drift ratio, which is much later than the control case.
The obtained strength of the T-beams in the push action was 88.5 kN and 138.5
kN under positive and negative loadings respectively. Their sum was only 7.0%
lower than that of the control specimen F-Control, which was far from the
requirement to offset the contribution from slab. Thus, the designed opening
size of specimen F-O-450-150 was not big enough to ensure a beam flexural
Page 203
172
failure mode.
This opening size of specimen F-O-500-180 was designed after testing the
specimen F-O-450-150. The specimen F-O-500-180 reached its strength at 2.0%
drift ratio as shown in Figure 4.13(d). The T-beams also obtained their strength
at 2.0% drift ratio, which was quicker than the specimen F-O-450-150 as joint
deformation of the specimen F-O-450-150 was the major contribution to lateral
drift. The measured positive and negative strength of the T-beams was 88.2 kN
and 115.7 kN respectively and the weakening ratio was 16.4%, which was more
than twice that of the specimen F-O-450-150. Even though the weakening ratio
was only half of that of the R-beam, the specimen F-O-500-180 still failed in a
ductile manner.
Specimen F-S-450-450 had a quicker yielding process because 75% of the
longitudinal slab bars had been cut. The specimen reached its strength at 1.5%
drift ratio and had a relative fast post-peak ascending range as indicated in
Figure 4.13(e). The T-beams reached their yield and strength at 1.5% drift ratio,
also earlier than those of control specimen did because of the removal of 75%
slab longitudinal bars. The measured positive and negative strength of the
T-beams were 83.1 kN and 109.4 kN respectively, and the weakening ratio was
21.0%. Even though the strength was still higher than those of the R-beam, the
specimen retrofitted by the SS method was able to perform in a ductile manner
under cyclic loading. In this sense, the method was a good option for retrofitting
a beam-column joint.
Page 204
173
Due to this MTS machine problem, the test on specimen S-control was
stopped several times during the test. The specimen failed due to hinge
formation in the bottom column as indicated in Figure 4.7(e). The test was
stopped at the first cycle of the 5% drift ratio as the column bars buckled. The
specimen reached its strength at 2.0% drift ratio and its two T-beams at 2.5%
drift ratio. The measured negative and positive strength of the T-beams was
144.4 kN and 112.8 kN respectively and the yield load from sectional analysis
was 156.4 kN and 119.6 kN, which meant the T-beams did not reached their
strength during testing. As indicated in Figure 4.13(f), the beam tip load
hysteresis curves were similar except for the final hoop, which was affected
because the bottom column bars buckled leading to a sharp decrease of axial
load.
The grooves of Specimen S-G-50-200-100 were 200 mm deep and 50 mm
wide and placed 100 mm away from the beam-column intersection. The
specimen reached its strength at 2.0% drift ratio. The measured maximum
negative and positive beam tip loads were 111.0 kN and 107.3 kN respectively.
The weakening ratio was 15.1 %. The ductility of the specimen was still poor
due to the occurrence of bottom beam bars slippage in the joint region, even
though the grooves were placed 100 mm away from the beam-column
intersection. The negative beam tip load has a sudden drop as demonstrated in
Figure 4.12(g) when the bottom beam bars began slipping in the joint region.
When the specimen S-O-500-180 was under pull action in the last cycle, of
Page 205
174
the 5% drift ratio, the column bars finally buckled because they lacked concrete
cover protection. The specimen and its beams reached their strength at 2.5%
drift ratio. The measured maximum negative and positive beam tip loads were
120.4kN and 119.3 kN respectively. The weakening ratio was 6.8%. The beam
tip loads obtained were similar except for the final hoop during which the
bottom column bars buckled
The opening dimensions of the specimen S-O-500-180-S-450-450 were the
same as those of specimen S-O-500-180. The specimen reached its strength at
2.0% drift ratio but the T-beams reached their negative strength at 4.0% drift
ratio, slower than for the other specimens. The measured maximum negative
and positive beam tip loads were 111.7 kN and 102.6 kN respectively, which are
lower than the measured maximum negative and positive beam tip loads of
specimen F-O-500-180, 120.4 kN and 119.3 kN respectively. The hysteresis
curves for the two T-beams were similar and of good ductile property as
indicated in Figure 4.12(i).
4.3.2 Envelop Curves
Envelop curves of calculated column shear forces and beam tip loads of
specimens are shown in Figures 4.15 and 4.16. It is obvious that the specimens
retrofitted by the proposed techniques had lower stiffness and strength than
those of control specimens. Among the first test group, the specimen
F-G-50-200 had the lowest post-cracking stiffness, in addition to the lowest
Page 206
175
strength. Among the second test group, the specimen S-O-500-180-S-300-300
had the lowest stiffness. As shown in Figure 4.16(a), both the BO and SS
techniques decreased the stiffness of T-beams under both negative and positive
loading. The TG technique only had obvious effect on the stiffness and strength
of a T-beam when it was under negative loading. When the grooves were placed
100 mm away from the beam-column interface, the effect of the TG technique
on T-beam stiffness became weaker, as demonstrated in Figure 4.16(b). The
positive strength of specimens F-O-450-150, F-O-500-180 and S-O-500-180
was very close to that of control specimens, which meant that the positive
strength of the T-beams was slightly affected by the existence of BO. The SS
technique deceased both negative and positive strength of T-beams.
4.4 DEFORMATION BEHAVIOUR
Many strain gauges were mounted on reinforcing steel bars to measure the
interior deformation of the specimens under cyclic loading. The labels and
major functions of the strain gauges were illustrated in the previous Chapter 3.
Axis directions and the distribution of strain gauges were also presented in the
previous chapter. Except for some gauges broken before and during testing, the
remaining strain gauges were used to measure the deformation of steel bars at
the peak displacement for each loading drift ratio.
4.4.1 Strains in the Steel Bars
4.4.1.1 X Direction
Page 207
176
Along the X axis, the strain gauges measured the strain distribution in the
steel bars of the main beam and slab. When a specimen was under a push or a
pull action, the right T-beam experienced a hogging moment or sagging moment
respectively. It should be noted that in Figure 4.17, the strains were normalized
to the yield strains of corresponding steel bars. In these figures, the black dashed
lines indicate the yielding of steel bars.
Figure 4.17(a) shows the strain variation along the X axis in one top
reinforcing bar of the T-beam of the control specimen. The test data for the
retrofitted control specimen were not available because the wiring had all been
cut. When the specimen was under a push action/pull action, the top steel bar
was in compression/in tension respectively. In general, the stain gauges closest
to the column recorded large strains as they were located at the section of
highest moment. Some strain gauges measured tensile strains in the top steel
bars when a specimen was under push action, as indicated by the left hand
figure of Figure 4.17(a). This could be because the plastic extension of a steel
bar under tension could not be offset by the compression strain.
For specimen F-G-50-200, the greatest deformation occurred near the
transverse groove as indicated in Figure 4.17(b). It should be noted that the
grooves were 50 mm wide and located close to the beam-column interface.
Obviously, the plastic hinge length of the T-beam was shortened as the beam
deformation was mostly concentrated near the hinge. The plastic extension
deformation of the top beam bars was concentrated where the transverse groove
Page 208
177
was created. The obtained tension strains were still higher even when the beam
top part was under compression, which was because tension plastic strain was
unrecoverable. Strains measured by strain gauge TB11 dropped after the 2.0%
drift ratio, possibly due to slippage of the beam bottom bars.
When a T-beam was weakened by a 450x150 mm2 opening, the
deformation of the beam was much different from the one without opening, as
indicated in Figure 4.17(c). The right hand side of Figure 4.17(c) displays the
elevated strains of a beam top longitudinal bar with the increased drift ratio
when the specimen was under pull action. In this situation, the middle three
strain gauges measured smaller strains and the two close to the corner of the
opening measured higher strains. Strain gauge TB11 failed at 2.5% drift ratio.
Thus unreasonable tensile strain was obtained by this stain gauge when the
specimen was under push action at 2.5% drift ratio as indicated in the left hand
picture of Figure 4.17(c).
As shown in Figure 4.17(d), a beam top steel bar of the specimen
F-O-500-180 deformed similarly to that of specimen F-O-450-150. As the
opening size of specimen F-O-500-180 was larger, the four hinges mechanism
was more obvious, as demonstrated by the experimental observation that, at the
same drift ratio, the hinges at the chord ends rotated more severely. This
phenomenon became more obvious as the lateral drift ratio increased. As
indicated in the right hand picture of Figure 4.17(d), the strain gauge TB14,
which was closer to one corner of the opening, measured a greater steel
Page 209
178
deformation. Strain gauge TB11 measured increasing strain, before the 1.5%
drift ratio, during pushing action. When the drift ratio was increased to a higher
level, the strain gauge measured a descending deformation trend, perhaps
because the effect of the four hinges mechanism overwhelmed the plane
sectional deformation.
Figure 4.17(e) presents the top beam bar deformation of specimen
F-S-450-450. At lower drift ratios, strain gauge TB11 measured the highest
deformation. In later loading cycles, strain gauge TB12 gave the highest strains,
because it was located near a large flexural crack.
Figure 4.17(f) shows the strain distribution along the X axis of a beam top
reinforcing bar of the specimen S-Control. At first the strain gauges closer to the
column indicated more deformation, as they were located at the section of
highest moment. As the lateral drift ratio grew larger, concrete cracking lead to
strain redistribution and thus, some part of the steel bar was the most strained
even though not the closest to the column.
For specimen S-G-50-200-100, the highest deformation occurred at the
groove position as indicated by the red dashed lines in Figure 4.17(g). The
grooves were located 100 mm away from the beam-to-column interface. The
plastic extension deformation of the top beam bars was also concentrated where
the transverse groove was created, similar to Specimen F-G-50-200. Strain
gauge TB12 became detached from the steel bar after the 2.5% drift ratio. Thus,
Page 210
179
even under push action, the strain gauge still measured a large tensile strain,
which was nearly constant during the later drift ratios.
The beam deformation of specimen S-O-500-180 was very different from
the specimen S-control and the specimen S-G-50-200-100, even though the
opening size was not sufficiently large to lead to ductile failure mode. The left
hand picture of Figure 4.17(h) displays the strain distribution of a top
longitudinal T-beam bar. As mentioned above, when a specimen is under push
action, the right hand beam is subjected to negative bending. The strain gauges
TB11 and TB15 on a top longitudinal bar, were placed near the left and right
hand sides of the opening respectively and their data revealed that the two ends
of the strong chord were deformed to a greater extent than the middle part,
similar to the specimen F-O-450-150 and F-O-500-180 cases.
The T-beam deformation of specimen S-O-500-180-S-300-300 showed no
obvious difference to that of specimen S-O-500-180. The deformation measured
by strain gauge TB14 during push action was larger than that by TB15, which
might be due to the existing of slab slits, as slits separated slabs and main beams
in the longitudinal direction.
4.4.1.2 Y Direction
Along the Y direction, the strain distribution was measured to reveal the
slab contribution to T-beam flexural capacity. Except for the cut slab bars
passing the slab slits, all slab bars of specimens yielded during test.
Page 211
180
For the control specimen F-Control, slab steel bars farther away from main
beam in the transverse direction were recorded less deformation before 1.0%
lateral drift ratio, which was due to shear lag effect. This effect was not
obviously presented by strain gauges data at higher drift ratios. A possible
reason was cracks propagating along the transverse direction were not parallel
to the Y direction. Another possible reason was strain gauges were not longer
tightly attached to slab bars after several cyclic loading cycles. Almost all steel
bars yielded when drift ratio reached 2.0%, as indicated in Figure 4.18(a). The
strain gauge TL11 might have become detached from the steel bar after the 2.0%
drift ratio as it remained almost constant as drift ratio was increased. Strain
gauge TL21 measured a large deformation, possibly because it bridged a large
crack.
For specimen F-G-50-200, all slab top bars yielded at the 1.0% drift ratio,
as indicated in Figure 4.18(b). The strain gauges TL11 and TL21 might be
detached from the steel bars after 1.0% and 2.0% drift ratios respectively thus
the recorded strains by them were strange afterwards.
The shear lag effect on slab for specimen F-O-450-150 was obvious before
2.0% lateral drift ratio as slab steel bars farther away from main beam in the
transverse direction were recorded less deformation, indicated in Figure 4.18(c).
All slab top bars yielded at 2.5% drift ratio and strain gauge TL11 also failed at
this drift ratio.
Page 212
181
As shown in Figure 4.18(d), the slab top bars of specimen F-O-500-180
yielded at 1.5% drift ratio, less than that in specimen F-O-450-150, as the
specimen F-O-500-180 chord end hinges rotated more severely.
As strain gauge TL41 of specimen F-S-450-450 broke before the test,
unlike the other specimens, the strain distribution on the left hand side of slab
was plotted in Figure 4.18(e) instead. As expected, the cut slab bars experienced
very low strain levels under conditions of both positive and negative loading.
For the control specimen S-control, there was shear lag effect in the slab
transverse direction and almost all the steel bars yielded when the lateral drift
ratio reached 2.5%, as indicated in Figure 4.18(f),
For specimen S-G-50-200-100, the shear lag effect is also obvious in the
right hand picture of Figure 4.18(g).
Figures 4.18(h) and (i) show the strain distribution in the slabs of
specimens S-O-500-180 and S-O-500-180-S-300-300 in the transverse direction.
For the specimen weakened by openings only, the strain distribution of the top
slab bars in the slab transverse direction indicated the shear lag effect at a low
lateral drift ratio (lower than 1.0%). Crack propagation with increased drift ratio
might result in some strain gauges capturing much higher deformations (more
than 2 times the yield strain). For the specimen weakened by additional slab
slits, the cut slab bars contributed marginally to the beam moment capacity at
Page 213
182
the monitored section.
4.4.1.3 Z Direction
Along the Z axis direction, only data from strain gauges on column
longitudinal bars were plotted. As measurements of strain in the two column
bars were nearly symmetrical about the z axis. The strain distribution in one
column bar is given in Figure 4.19.
Except for the specimens F-G-50-200, F-O-500-180, S-G-50-200-100,
whose column longitudinal bars kept elastic until the end of the test as indicated
in Figures 4.19(b).(e) and (g), all specimens had yielding longitudinal bars in
the upper column or bottom column during test.
As shown in Figure 4.19(a), the bars in the upper column of specimen
F-Control stayed elastic during the test and those in the lower column, reached
yield load after a 2.5% drift ratio. It was clear that strain gauges closest to the
joint region measured more deformation. As shown in Figure 4.19(c), the
specimen F-O-450-150 bottom column bars compression yielding occurred after
2.5% drift ratio as strain gauge BC32 recorded large compression deformation.
As shown in Figure 4.19(e), due to the lack of surrounding slab protection, steel
bars in the upper column of specimen F-S-450-450 was recorded more
deformation. Strain gauge TC31 recorded yielding of the corresponding steel
bar in the upper column at 5% drift ratio. As shown in Figure 4.19(f), the upper
column bars stayed elastic during test and the lower bottom column bars
Page 214
183
reached their yield load after a 2.0% drift ratio. At peak displacement of the 4.0%
drift ratio cycle, large deformation existed in the bottom column bars and when
the drift ratio reached 5%, the column bottom bars buckled. The specimen could
not bear any axial loading and thus, the strain variation after ultimate loading is
not plotted in Figure 4.19(f). The openings size of S-O-500-180 was not big
enough to switch the column flexural failure to a four hinges formation failure.
Thus, the bottom column bars yielded after a 2.5% drift ratio and at the second
cycle of the final cycle group, some bottom column bars buckled after spalling
of the concrete cover. Though specimen S-O-500-180-S-300-300 went through
WBSC failure, the monitored bottom column bar reached its compression
yielding strain as demonstrated in Figure 4.19(i).
4.4.2 Rotations of Column Sections
The strain levels of bottom column eight column bars close to the joint
region were plotted in a 3D form (as indicated in Figure 4.20) to show column
section deformation under pull action. The column section deformation of
specimen F-O-50-180 is not presented because most of the relevant strain
gauges broke when creating the openings. As indicated by these 3D figures, the
plane section assumption was basically obeyed at low lateral drift ratios. When
the drift ratio increased, cracks propagated and concrete gradually spalled,
which led to strain redistribution and disturbance of the plane section
assumption. In addition, plastic deformation of the steel bars was irreversible,
which contributed to the breaking of the plane section assumption at high drift
Page 215
184
ratios. For specimens F-S-450-450, S-O-500-180-S-300-300, the column
deformation basically followed the plane section assumption, even at the last
cycle as indicated in Figure 4.20(h), because they failed in WBSC mode. As
indicated in Figures 4.20(b)-(d), the column section deformation of specimens
F-G-50-200, F-O-450-150, F-S-450-450 reached its peak rotation at 2.0%, 2.5%
and 2.5% lateral drift ratio respectively.
4.4.3 Strain in Stirrups at Joint Region
The interior deformation of the joint was measured by the strain gauges
mounted on the joint stirrups. Strain gauge CS1 was located on the uppermost
stirrup and CS4 on the lowermost. As indicated in Figure 4.21, except for the
specimen F-G-50-200, all specimens had one or more yielding joint stirrups
when the drift ratio reached 2.5%. Generally, the control specimen suffered a
severest joint shear deformation. However, CS1 of the specimen F-S-450-450
measured larger strains than the control specimen did. This might be due to the
lack of surrounding slab protection. The strain gauge CS3 of specimen
F-O-500-180 recorded higher deformations than CS4, which might be due to the
removal of beam web concrete.
Though two strain gauges were used to record the deformation of one
stirrup in the specimens' joints region of the second test group, only data from
one strain gauge for each monitored stirrup was plotted. As indicated in Figure
4.22, except for the stirrup monitored by strain gauge CS42, specimen
Page 216
185
S-Control joint stirrups kept elastic during testing. The strain gauge CS42
measured higher deformation as the stirrup it monitored was closest to the
bottom column among the four monitored stirrups. For specimen
S-G-50-200-100, all joint stirrups were elastic during testing as indicated in
Figure 4.22. The specimen S-O-500-180, whose joint middle two stirrups had
more deformation than those of the other specimens, as indicated in Figures
4.22(b) and (c).The reason might due to the middle two lacking the protection of
beam web concrete, which was removed to create openings. The existence of
SSs resulted in smaller joint shear deformation, as indicated by comparing
specimen S-O-500-180-S-300-300 joint stirrups strain level with that of
specimen S-O-500-180.
4.4.4 Strains in FRP Jackets
The strain gauge arrangements on FRP are indicated in Figure 3.22 of
former chapter. The strain gauges on the left hand beam behaved similarly to
those on the right. As shown in Figures 4.23-4.28, all FRP sheets stayed elastic
until the end of the test. Figure 4.25(b) presents the strain distribution of the
FRP sheet at the two ends of the web chord of specimen F-O-500-180. When
the specimen was under pull action, the right hand T-beam was under negative
bending. The bottom surface of the left end and the top surface of the other end
of the chord were under compression. At this moment, RW5 and RW6 obtained
the highest strains, caused by concrete expansion. Similarly, when the specimen
was under push action, strain gauges LW5 and LW6 obtained similar results.
Page 217
186
This phenomenon was not obvious when the T-beam was under positive
bending.
For specimen S-G-50-200-100, as indicated in Figure 4.26, the strain gauge
(RU1) close to the column face and the upper corner of the groove measured the
largest strain after a 3.0% drift ratio, because the cracks propagating from the
corner became larger with the increased drift ratio. Of the strain gauges RU4,5
and 6, on the other sides of the groove, the one closer to the bottom of the
T-beams measured the largest deformation. For the specimens with beam
openings, as shown in Figures 4.27 and 4.28, strain gauges on the U-jackets
displayed lower strain levels, which meant the U-jackets were strong enough.
For strain gauges on the FRP wrap, RW1, RW3, RW5 and RW7 were mounted
on the corners of the web chord. RW4 and RW8 were mounted on the bottom
surface of the chord. When the right hand beam was under negative loading,
RW4 and RW3 measured the largest deformation at the left hand end of the
chord. Because at this scenario, the beam deformed following the four hinges
mechanism. The bottom concrete of the chord left end was under compression
and bulged. When the right hand beam was under positive loading, RW4 and
RW3 still measured the largest deformations as the four hinges mechanism was
not obvious and cracks propagated from the bottom surface. At the other end of
the chord, when the right hand beam was under negative moment (i.e. the
specimen was under pull action), RW5 measured the largest deformation as the
concrete of the top side bulged. When the beam was under positive moment,
Page 218
187
RW5 measured a relatively larger deformation as that part was still under
compression.
4.5 DISCUSSIONS
The original aim of retrofitting RC beam column joints is to avoid the story
sway failure mechanism. The flexural strength ratios of columns to beams with
cast-in-place slab should be increased to more than 1.0. The failure mode should
be ductile. Apart from the beam strength and the specimen failure mode, the
specimen ductility, energy dissipation capacity, equivalent viscous damping
ratio hyst and peak-to-peak stiffness are also important factors which can be
used to assess the seismic performance of these specimens. A retrofitted
specimen should have better ductility and energy dissipation capacity than the
original one which might fail due to storey sway mechanism. The peak-to-peak
stiffness should not be obviously decreased as according to seismic design
codes (e.g. GB-50011 2001), a reliable RC structure should be able to stand
obvious deformation (e.g. 2% drift ratio). The energy dissipation capacity is a
direct index and hyst is a more effective index to assess the seismic
performance of specimens. Thus, these factors are discussed below.
4.5.1 Specimen Ductility
Yielding of the tested specimens was not easily detected due to the
existence of a wide slab. According to Hu (2005), there are four methods for
Page 219
188
defining yield deformation: (1) as the deformation where yielding first occurs in
the system (see Figure 4.29(a)); (2) as the elastic limit of an equivalent
elastic-perfectly plastic curve with the same elastic stiffness and strength as
those of the test curve (see Figure 4.29(b)); (3) as the elastic limit of an
equivalent elastic-perfectly plastic curve which provides an equal area to that of
the test envelope curve before the peak load (strength), with the real test
strength being taken as the equivalent strength (see Figure 4.29(c)); and (4) as
the elastic limit of an equivalent elastic-perfectly plastic curve with a reduced
stiffness which is equal to the secant stiffness at either the first yielding or at a
certain value (e.g. 75%) of the strength, whichever is less (see Figure 4.29(d)).
Among the four definitions, the first is inappropriate for the present study
because first yielding is vague and thus hard to recognize. Of the last three
methods, the second is suitable for materials with an obvious yielding plateau,
such as hot-rolled steel. The third definition is hard to apply. The final definition
provides the most appropriate and general way and has been claimed the most
suitable for various structures such as concrete, masonry, steel as well as timber
structures (Park 1989).
As for the definition of the ultimate state of a structural member, possible
definitions for the ultimate deformation are: (1) as that corresponding to a
particular limit value for the material ultimate strain (e.g. the attainment of a
specified concrete ultimate compressive strain in the case of reinforced concrete
structures (see Figure 4.30(a)); (2) as that corresponding to the ultimate load of
Page 220
189
a test load-displacement curve (see Figure 4.30(b)); (3) as the value where the
load capacity of a test specimen has undergone a small reduction, for example, a
10% to 30% reduction (see Figure 4.30(c)) ; (4) as the value when the material
fractures or elements buckles; for example, the fracture of transverse reinforcing
steel or the buckling of longitudinal reinforcing steel in the case of reinforced
concrete; see Figure 4.30(d).
Definition (1) is inappropriate as it is evident that the maximum available
deformation does not necessarily correspond to a specified extreme concrete
compressive strain. It should be recognized that most structures have some
deformation capacity after achieving their ultimate load without significant
reduction in load capacity. It is reasonable to include this extra deformation
capacity in defining the ultimate deformation. Hence, definition (2) is
conservative. Park (1989) recommended defining the ultimate deformation
using the criteria (3) and (4) together, whichever occurs first.
The ductility parameter is commonly defined by the following
equation:
u
y
(4.2)
where and are the ultimate and yield displacements respectively.
As indicated in Figure 4.29, the definition (4) for the yield deformation
Page 221
190
was applied to determine the yield displacements of specimens and 75% of
strength was used to calculate the secant stiffness. And the definition (3), for
ultimate state, was applied and a 20% reduction of load after the peak load was
chosen as the ultimate state. The results are summarized in Table 4.3.
The positive parts of the envelope curves, as shown in Figure 4.13, were
used to determine the yield and ultimate displacements of tested specimens
except for the specimen F-Control. Specimen F-Control failed at 4.0% lateral
drift ratio group due to shear failure at the beam pinned position. It was then
strengthened with steel reinforcement and FRP sheets. The test was carried out
about 2 months after being retrofitted. Therefore, the strength of the specimen
was slightly increased at the first cycle loading owing to crack closing and
concrete strength increase.
In terms of the yield deformation, the control specimen F-Control had the
smallest yield displacement among the first group as the stiffness degradation
ratio due to retrofitting is higher than that of strength. As for ductility, the
control specimen had good ductility as the beam yielded first before joint shear
failure. The specimen F-G-50-200 was the poorest as indicated in Figure 4.31. It
is obvious from the envelope curve shown in Figure 4.13(b) that specimen
F-G-50-200 strength falls quickly after peak stress. The slits on the slab can
effectively lower the strength specimen F-S-450-450. The ductility was
decreased by the existence of slab slits as the control specimen did not failed by
column end failure. The BO techniques can reduce strength with ductility
Page 222
191
remaining almost unchanged (the ductilities of specimens F-O-450-150 and
F-O-500-180, which are 4.4 and 4.1 respectively, is close to that of specimen
F-control, which is 4.4).
Unlike the first test group, specimen S-G-50-200-100 of the second test
group had the smallest yield displacement among the second group. That's
because specimen S-G-50-200-100 yield load was also obviously lowered but
the stiffness was less affected by the TG technique than in the first test group as
the grooves were placed 100 mm away from the beam-column interface. The
other three specimens had close yield displacements as both strength and
stiffness were lowered by the BO technique and the combination of the BO and
SS techniques. As shown in Figure 4.31, the ductility of specimen S-Control
was obvious lower than that of specimen F-Control, which was mainly because
the specimen S-Control failed by column end failure. The ductility of specimen
S-G-50-200-100 was much higher than that of specimen F-G-50-200 as the
grooves were placed 100 mm distant from the column face, which postponed
the slippage of bottom longitudinal bars. The specimen S-O-500-180 had
slightly higher ductility than the specimen S-Control. The combined use of the
BO and SS techniques in specimen S-O-500-180-S-300-300 obviously reduced
specimen strength but increased ductility.
4.5.2 Energy Dissipation Capacity
As indicated in Figure 4.32, dissipated energy EDA by each cycle is
Page 223
192
defined by the area enclosed by the cycle. Energy dissipation curves for each
individual cycle for each specimen are plotted in Figure 4.33. It is obvious that
almost all the first hysteresis cycle of each drift ratio dissipate more energy than
the second one. That's because crack propagation decreases the stiffness and
strength of a specimen. Except for specimen F-G-50-200 which has poor
ductility and fails quickly after the 2.0% lateral drift ratio, the remaining four
specimens in the first test group possessed increasing energy dissipation
capacity with increased drift ratio. Column bars buckling lead to the first cycle
at the 5% drift ratio of the control specimen S-Control, dissipating much more
energy than the first cycle of the 4.0% drift ratio. That also resulted in a
consumption of a bit more energy in the second cycle the 4.0% drift ratio. than
in the first cycle. Except for specimen S-G-50-200-100, which has poor ductility
and failed quickly after the 2.0% lateral drift ratio, the drift ratio at the peak load,
the remaining three specimens of the second test group displayed increasing
energy dissipation capacity with increased lateral drift ratio.
As shown in Figure 4.34(a), specimen F-G-50-200 dissipates more energy
than the other four before 2.5% lateral drift ratio, as the specimen suffered more
damage at first cycle owing to the existence of grooves. After failure (lower than
80% of the peak load), less energy is dissipated as demonstrated in Figure
4.33(a). Specimen F-S-450-450 dissipates the least amount of cumulative
energy at first but becomes the one consuming the most energy when reaching
the 3.0% drift ratio. Thus, specimen F-S-450-450 has the best energy dissipation
Page 224
193
capacity in the group after entering yielding stage. Among the remaining three
specimens, the control specimen F-Control dissipates more energy than the
other two do on account of its higher strength, but similar hysteresis curves.
As shown in Figure 4.34(b), specimen S-G-50-200-100 dissipates more
energy than the other three specimens before 3.0% lateral drift ratio, because it
suffered more damage. After failure, the specimen dissipates less energy as
demonstrated in Figure 4.33(b). Among the remaining three specimens, the
control specimen S-control dissipates more energy than the other two on
account of its higher strength.
4.5.3 Equivalent Viscous Damping Ratio
Equivalent viscous damping ratio hyst represents the dissipation due to
the nonlinear (hysteretic) behaviour (Blandon and Priestley 2005), which is a
more effective factor to assess the seismic performance of specimen. Figure
4.35 shows the equivalent damping ratio hyst (referred to as EDR hereafter)
development with increasing lateral drift ratio. The EDR was obtained using
Jacobsen's (1930) approach, dividing dissipated energy by the stored elastic
energy, as indicated in Figure 4.32 and expressed in Equation (4.3).
1 1 2 2
12
EDhyst
m m m m
AF F
(4.3)
Page 225
194
where EDA is the energy dissipated by a cycle; 1mF and 1m are the positive
peak load and its corresponding displacement of the cycle; and 2mF are 2m
the negative peak load and its corresponding displacement of the cycle.
As shown in Figure 4.35(a), the EDR of specimen F-G-50-200 is the
second lowest at first, but quickly exceeds those of the remaining specimens.
The reason for this phenomenon is that specimen F-G-50-200 was the least
ductile and failed quickly after peak load, which lead to its EDR increasing
most quickly after the 1.0% drift ratio. Specimen F-S-450-450 possesses the
second highest EDR after the 1.5% drift ratio, because of its better energy
dissipation capacity and lower stored elastic energy. The remaining specimens
display similar EDR curves. Obvious discrepancies are evident as indicated in
Figure 4.35(a), probably due to the different mechanisms of energy dissipation
of different weakening scheme.
Figure 4.35(b) shows the EDR variation with increased lateral drift ratio
for the second test group. As shown in Figure 4.35(b), the EDRs of all
specimens in the second test group show a growing trend except for a drop at
the 1.5% drift ratio, which is not obvious in the first test group. This might be
because the specimen cracking of the second test group, which occurred during
the 1.0% drift ratio, was more serious than for the first test group. After the 1.5%
drift ratio, continuous yielding of steel bars and concrete cracking caused the
damping ratio to increase.
Page 226
195
The EDR of specimen S-G-50-200-100 was the lowest at first but quickly
exceeded those of the other specimens in the second test group. As specimen
S-G-50-200-100 is the least ductile, it fails quickly after yielding, which leads to
its EDR becoming the highest after the 1.0% drift ratio. Specimens
S-O-500-180 and S-O-500-180-S-300-300 have similar EDR curves versus
lateral drift ratio. The EDR of the control specimen greatly increases at 4.0%,
when the plastic region of the lower column suffered serious damage.
As indicated in Figures 4.35(a) and (b), the EDRs of the second cycle are
lower than those of the first, because less energy is dissipated during the second
cycle but the stored elastic energy remains almost unchanged, as indicated by
the close PTPSs of the first and the second cycles as shown in Figure 4.36 in
next section. The EDR of specimen F-G-50-200 at the second cycle of the 2.0 %
drift ratio is higher than that of the first cycle. This may be because beam
bottom longitudinal bars slipped significantly at this drift ratio, also indicated by
the obvious PTPS degradation between the two cycles as shown in Figure
4.36(a). Even though the specimen F-G-50-200 EDR curve of first cycles keeps
increasing after the 2.0% drift ratio, that of the second cycles descends after the
2.0 % drift ratio due to serious slipping problem brought about by the first
cycles. For the control specimen F-Control and the specimens with openings
F-O-450-150 and F-O-500-180, the distance between the EDRs of first and
second cycles is noticeable at first, but gradually becomes smaller and finally
close to zero at 5% drift ratio. The distance between the two EDRs of first and
Page 227
196
second cycles of specimen F-S-450-450, however, is almost constant with
increased drift ratio.
The bottom column concrete of the control specimen S-Control in the
second test group was seriously damaged during the second cycle of the 4.0%
lateral drift ratio, leading to the EDR of the second cycle suddenly higher than
that of the first cycle. The EDRs of second cycles for the second test group
increase with the increasing drift ratio between 1.0% and 1.5% drift ratios,
unlike the case with the first cycles. This might also be due to the serious
cracking which occurred in the first cycle of the 1.0% drift ratio.
4.5.4 Stiffness Degradation
Figure 4.32 illustrates the definition of peak-to-peak stiffness Keff
(referred to as PTPS hereafter). Keff is defined as the slope of the line linking the
peak-to-peak force-displacement points at each imposed displacement cycle. It
can be taken as an average secant stiffness for both positive and negative
displacement peaks as illustrated by the following equation:
1 2eff
1 2
m m
m m
F FK
(4.4)
Figure 4.36 shows the PTPS variation with increasing lateral drift ratio. As
indicated in Figure 4.36(a), the specimen S-G-50-200-100 has the closest initial
PTPS to that of the specimen F-control. In contrast, the BO and SS techniques
Page 228
197
effectively decreased the initial PTPS of specimens. That means the TG
technique has less effect on specimen initial PTPS than other retrofit techniques.
The control specimen and specimen F-G-50-200 have the highest PTPS at first.
However, the PTPS of the specimen F-G-50-200 descends at a faster rate than
the other four and becomes the lowest after the 1.0% drift ratio. Specimen
F-S-450-450 has a slightly higher PTPS than specimen F-O-500-180 at first but
becomes lower after the 1.5% drift ratio. The control specimen F-Control, the
specimens F-O-450-150 and F-o-500-180 have parallel descending PTPS curves
in relation to lateral drift ratio.
Figure 4.36(b) shows PTPS variation with the increased lateral drift ratio
of the second test group. It is obvious that the control specimen has the highest
PTPS at first. The specimen S-G-50-200-100 has a slightly higher PTPS than
for the specimen S-O-500-180. However, the PTPS of the specimen
S-G-50-200-100 decreases at the faster speed than the others and has the lowest
value after the 2.0% drift ratio, when the peak load has been reached. The
control specimen S-Control and specimen S-O-500-180 share similar PTPS
curves. Specimen S-O-500-180-S-300-300 has the lowest PTPS first. Its PTPS
degrades the most slowly and gets much closer to those of the control specimen
S-Control and specimen S-O-500-180 at the 4.0% drift ratio.
4.5.5 Plastic Hinge Lengths
According to Hines et al. (2004), the equivalent plastic hinge length is a
Page 229
198
length of a structural member over which the plastic curvature can be assumed
constant. A physical plastic hinge, within which the plasticity region actually
spreads, does not include the strain penetration. By assuming that the plastic
curvature is linearly distributed over the physical plastic hinge region, Hines et
al.(2004) proposed the following equation to relate the length of the physical
plastic hinge to that of the equivalent plastic hinge:
/ 2p ppl ypL L L (4.5)
where pL is the length of the equivalent plastic zone, pplL is the length of the
physical plastic zone and ypL is the plastic hinge length due to yield
penetration into the base. Figure 4.37 clearly presents the plastic hinge
composition.
Zhao (2012) investigated plastic hinges in RC structures through
experiments and finite element (FE) modelling. According to Zhao (2012), the
plastic hinge of an RC column consists of three physical zones: the bar yielding
zone, concrete crushing zone, and curvature concentration zone. The bar
yielding zone was defined as the region in which the reinforcing steel in tension
had reached or exceeded its yield stress. The strain concentration zone, whose
length was less than or equal to that of the bar yielding zone was the region
where most of the plastic curvature was concentrated. The concrete crushing
zone was defined as the region in which compressive strains greater than 0.002,
Page 230
199
the strain at the peak concrete strength or 0.006, the strain at complete concrete
crushing. Through FE modelling and experimental studies, Zhao (2012) stated
that the lengths of the bar yielding zone and the concrete crushing zone served
as upper and lower bounds of the physical plastic zones respectively.
Jiang et al. (2014) applied a digital image correlation (DIC) method to
continuously record variation of the strain field on the external column face.
Jiang et al. (2014) focused on the curvature concentration zone, which is similar
to the strain concentration zone studied by Zhao (2012), as it has a closer
relationship with the curvature plastic hinge. The curvature concentration zone
was measured with the DIC method and determined using the dividing point
concept. The dividing point is a point on one side of which curvatures keep on
increasing, while on the other side of the point, they remain almost constant as
the applied displacement increases. The dividing point defines the edge of the
curvature concentration zone. Jiang et al. (2014) also mentioned that the lengths
of curvature concentration zones of the tested columns were consistent with the
lengths of bar yielding zones. In addition, Jiang et al. (2014) applied the
relationship proposed by Hines et al. (2004) to relate the physical plastic zone to
the equivalent plastic zone.
In the author's tests, only strain gauges were available for use to determine
the length of the physical plastic region. The plastic hinge length accounting for
yield penetration into the base could not be estimated. The bar yielding region
was taken as the physical plastic region and the strain concentration zone was
Page 231
200
determined by the dividing point method. The strain distribution of a
longitudinal bar under tension is more reliable in determining a bar yielding
region. The lengths of yielding and strain concentration zones measured by
strain gauges through dividing point are summarized in Table 4.4.
As shown in Figure 4.17(a), at 3.0% drift ratio, the x-coordinate value of
the beam bar yielding region of specimen F-control is about 360 mm and half of
the column width is 160 mm , which means the bar yielding region length is
about 200 mm (i.e. 360-160=200mm). This value is smaller than that of
specimen S-Control mainly because specimen F-control failed at 4.0% lateral
drift ratio. At that drift ratio, the length of the yielding zone is not fully
developed.
For specimen F-G-50-200, the strain concentration zone was located at the
groove region. As indicated in Figure 4.17(b), the x-coordinate values of the
beam bar yielding region and the strain concentration zone are about 350 mm
and 280 mm respectively. Thus the lengths of the bar yielding region and the
strain concentration zone are about 190 mm and 120 mm respectively (i.e.
350-160=190, 280-160=120). It can be conclude that the transverse groove
shortened the plastic hinge length.
As for specimen F-S-450-450, the x-coordinate value of the beam bar
yielding region was 470 mm as indicated in Figure 4.12(e), which means that
the length of the bar yielding region was 310 mm. The strain concentration zone
Page 232
201
length was hard to determine as the number of strain gauges was insufficient.
The length of the strain concentration zone is more than 120 mm as indicated in
Figure 4.17(e).
As shown in Figure 4.17(f), the x-coordinate values of the beam bar
yielding region and the strain concentration zone of specimen S-control are
about 520 mm and about 270 mm respectively, which means the lengths of the
bar yielding region and the strain concentration zone are about 360 mm and 110
mm respectively.
For specimen S-G-50-200-100, the length of the bar yielding region is
about 310mm. It is obvious that the strain concentration zone is located within
the groove region within the red dashed line as shown in Figure 4.36(b). The
strain concentration zone length was hard to determine as the number of strain
gauges was insufficient.
The specimens with beam openings had several plastic regions and thus,
are not included in the discussion.
4.5.6 Deformation Components
LVDTs were applied to measure the deformation of T-beams and columns
based on the plane cross-section assumption. Due to the existence of transverse
beams and slab, the joint deformation could not be measured. The arrangement
of LVDTs was described in Chapter 3 and how the deformations of beams and
Page 233
202
columns can be obtained was also explained in Chapter 3.
4.5.6.1 Beam drift versus beam tip load hysteresis curves
Beam tip load versus beam displacement hysteresis curves were plotted to
reveal the beam behaviour under cyclic loading. As mentioned previously, the
beam tip loads of the left and right hand beams were directly measured by load
cells placed on the beam supports. The beam drift was calculated using LVDTs
data and equations in Chapter 3. The elastic range deformations of beams and
columns were calculated using the Equations (3.5) and (3.10) respectively. The
plastic range deformations of beams and columns were determined using
LVDTs data and Equations (3.4) and (3.9) respectively. In Figures 4.38, a
positive displacement or load means the left or right beam is sustaining a
sagging moment where the compression is in the slab. When a specimen was
under push action, the left hand beam sustained a hogging moment and the right
hand beam a sagging moment.
Figure 4.38, giving curves of beam tip load versus calculated beam
displacement, does not include specimens with openings. That's because the
deformation of beams with openings could not be accurately measured by the
LVDTs as the plane cross-section assumption was not satisfied at larger lateral
drifts. The testing of specimen F-Control stopped at the first cycle of 4.0% drift
ratio. The deformations of beams and columns of the strengthened specimen
F-Control were not recorded by LVDTs.
Page 234
203
As indicated in Figure 4.38, the left and right hand beams of specimens
showed similar behaviour under cyclic loading. In the first test group, two sets
of LVDTs, one for beam end rotation and the other for plastic concentration,
were used to measure the plastic deformation of the left hand beam but only one
set was used on the right hand beam due to insufficient LVDTs available. The
LVDTs measured length for the specimens F-control, F-S-450-450 and
F-G-50-200 plastic region was 360mm, which was equal to the effective height
of the main beam. The length for the beam end rotation was 120 mm, which was
the shortest length possible allowing for the placing of LVDTs. If the beam bar
bond-slipping phenomenon was obvious, more plastic deformation in the left
hand beam of the first test group would be measured by LVDTs than for the
right hand one as only the left beam was equipped with LVDTs for the beam end
rotation. As indicated in Figures 4.38(b) and (c), more deformations in the left
hand beams of specimens F-G-50-200 and F-S-450-450 were measured in the
later cycles than in the right hand beams. However, this phenomenon was not
obvious on the specimen F-Control. The reason for this was the existence of
grooves on beams and slab slits, which weakened the bond between concrete
and steel bars at the beam-column interface. Thus, at larger lateral drift ratios,
beam end rotation resulting from beam bar bond slipping was more obvious.
The second test group had only one set of LVDTs available to measure the
deformation of the plastic region, whose length for measurement was 360 mm
for the S-Control specimen and 250 mm for the S-G-50-200-100. As the test on
Page 235
204
the specimen S-Control was twice stopped due to an MTS problem, the
calculated beam deformation were much in error as indicated in Figure 4.38(d).
The red line showed a huge error during the last cycle owing to the bottom
column bars buckling. The beam drift versus beam tip load curve for the
specimen S-G-50-200-100 is shown in Figure 4.38(e). As indicated in Figure
4.38(e), the two beams show similar behaviour under cyclic loading.
4.5.6.2 Deformation contributions of beams and columns to the lateral drift
The deformation contributions of beams and columns to the lateral drift of
the first and second test groups are presented in Figure 4.39. The column plastic
hinge lengths for LVDTs measurement were all equal to 280 mm. As the
deformation of those beams with openings could not be measured by LVDTs,
their deformation contributions could not be determined. The LVDT data
relating to the column plastic zones of specimen F-Control were lost, thus the
columns contributions in the case of the F-Control specimen is also unknown.
As a result, in Figures 4.39(a) and (b), the plotted contribution ratios of beams
and columns are the sums of the left and right hand beams and the top and
bottom columns respectively. Only the first cycle of each drift ratio was
included. Shown in Figure 4.39(a) are the variation curves of contribution ratio
of beam to the lateral drift versus the drift ratio. It can be seen that except for the
two control specimens F-Control and S-Control case, which have a decreasing
contribution ratio, the contribution of beam has a tendency to increase. The
descending scope of specimen S-Control is more obvious than that of the
Page 236
205
specimen F-Control. The beams contribution ratio for specimen S-Control was
lower than for specimen F-Control after the 1.0% drift ratio. This implies the
beams of specimen S-Control were stronger and thus the deformations of joint
panel and columns dominated the specimen's deformation. The beams
contribution ratio for specimen S-G-50-200-100 shows an ascending trend with
increased drift ratio and the rate of ascent increases after the 2.0% drift ratio,
where the specimen reached its peak load and beam bottom bars slipping
became more severe.
Except for the specimens S-Control and S-O-500-180, whose column
contribution ratios drop during the middle cycles and increase after a 3.0% drift
ratio as the two specimens all suffered column bottom bars buckling at the final
drift ratio, the contribution ratios of columns for all the available specimens
show a descending trend with increased drift ratio, as shown in Figure 4.39(b).
It should be due to the slab slits existence, the LVDT sets had error in measuring
the column plastic deformation at a higher drift ratio. That is why the column
contribution ratio of specimen F-S-450-450 became close to 0% at the 4.0%
drift ratio. The ratio was accurate before the 1.5% drift ratio after a check of the
obtained data. The decreasing column deformation contribution ratio means the
deformations of specimens were gradually dominated by the deformations of
beams as the drift ratio increased. The column contribution ratio for specimens
F-G-50-200 and S-G-50-200-50 decreases quickly with increased drift ratio and
becomes close to zero at the final drift ratio, because the slipping of beam bars
Page 237
206
dominates the specimen's deformation. The greater column contribution ratio for
the control specimen S-Control than for specimen S-O-500-180 reveals that the
existence of beam openings lessens the column deformation contribution ratio.
In addition, larger web openings weaken the column deformation contribution,
indicated by comparing the column contribution ratio of specimen F-450-150
and F-500-180. The existence of slab slits further lessens the column
deformation contribution ratio as indicated in Figure 4.39(b).
4.6 CONCLUDING SUMMARY
This chapter has presented the results of two test groups of RC
beam-column joints are presented in this chapter. The effects of the proposed
retrofit techniques on the performance of RC beam-column joints (RC 3D joints)
have been explored. Based on the test results and discussion above in this
chapter, the following conclusions can be drawn:
(1)The existence of a transverse groove (TG) can lower the moment-capacity of
a T-beam greatly. However, the failure mode was non-ductile because the
main beam bottom longitudinal bars slipped in the joint region, even though
the groove was not placed close to the beam-column interface. The slab
slitting (SS) method can effectively lower the moment capacity of a T-beam
as it removes the contribution of the cut slab longitudinal bars. Meanwhile,
the failure mode is beam end flexural failure. The beam opening (BO) size
should be well designed (e.g. increasing the opening size)to lead to an
Page 238
207
obvious four-hinges mechanism if the BO technique is applied as a retrofit
measure. The slab slits can help a specimen, not sufficiently weakened by
the BO technique, to realize the four-hinges mechanism. Slot-shaped
openings have similar effect on the seismic behaviour of a RC beam-column
joint as the rectangular one does.
(2) As the horizontal load recorded by the MTS machine included friction force,
which varied during test and among different specimens, the calculated
column shear force was used for discussions instead. The specimens
retrofitted by the proposed techniques had lower stiffness and strength than
the control specimens. Both the BO and SS techniques decrease the stiffness
of T-beams under both negative and positive loading. The TG technique
only has obvious effect on the stiffness and strength of a T-beam when it
was under negative loading. When the groove is not placed close to the
beam-column interface, the effect of the TG technique on T-beam stiffness
becomes weaker. The positive strength of T-beams is slightly affected by the
existence of BO. The SS technique decreases both negative and positive
strength of T-beams.
(3) When the control specimen has good ductility, the TG technique leads to
very poor ductility. The BO technique keeps the specimen ductility almost
unchanged. The SS technique decreases the ductility of a specimen. When
the control specimen has a poor ductility, the TG technique has a small
effect on specimen ductility but leads to a smaller yield displacement than
Page 239
208
that of control specimen. The BO technique slightly increases the specimen
ductility. When the combination of the BO and SS techniques is applied,
specimen ductility can be increased in addition to effectively decreasing the
moment-capacity of T-beams.
(4) The Specimen retrofitted by the SS technique had better energy dissipation
capacity than those by the BO and the TG technique. The specimens
retrofitted the TG technique had best energy dissipation capacity at first but
worse energy dissipation capacity after they reached peak load.
(5) Equivalent viscous damping ratios (EDRs) of specimens basically increased
with increasing drift ratio. The EDRs of those specimens weakened by TGs
were almost the highest before they failed. The EDR of the specimen
F-S-450-450 was the second highest among the first test group. The BO
technique has slight effect on a specimen EDR. Generally, the EDRs of the
second cycle of each drift ratio are lower than those of the first cycle.
(6) Unlike the BO, SS methods and their combination, the TG method has less
effect on the initial PTPS of specimens. In addition, the PTPS of the
specimens weakened by TGs, reduced the most quickly with increasing drift
ratio, as those specimens had poor ductility and failed soon after peak load
had been reached. The specimen retrofitted by a combination of the BO and
SS techniques has the lowest speed of PPTS descending.
Page 240
209
(7) The TG technique shortens plastic hinges length of T-beams as the
deformation of T-beams is concentrated on the groove region.
(8) Except for two control specimens F-Control and S-Control, beams’
deformation contribution to the lateral drift ratio increased with the
increasing lateral drift ratio. Thus the contribution of columns deformation
was decreased with the increasing lateral drift ratio.
4.7 REFERENCES
Blandon, C. A., & Priestley, M. J. N. (2005), "Equivalent viscous damping
equations for direct displacement based design", Journal of earthquake
Engineering, 9(sup2), 257-278.
Hines, E. M., Restrepo, J. I., and Seible, F. (2004), "Force-displacement
characterization of well-confined bridge piers", ACI Structural
Journal, 101(4), 537-548.
Jacobsen, L. S. (1930), "Steady forced vibration as influenced by
damping", Trans. ASME-APM, 52(15), 169-181.
Jiang, C., Wu, Y. F., & Wu, G. (2014), "Plastic hinge length of FRP-confined
square RC columns", Journal of Composites for Construction, 18(4),
04014003.
Zhao, X. (2012). Investigation of plastic hinges in reinforced concrete (RC)
Page 241
210
structures by finite element method and experimental study, Doctoral
dissertation: City University of Hong Kong
Page 242
211
(a) Removing the cracked concrete and clean the exposed surface with clean
water
(b) Shear strengthening the opening region with steel bars
(c) Building up wooden temple for concrete casting;
Page 243
212
(d) Recasting specimen
Figure 4.1 The control specimen strengthening process
(a) Flexural cracks appeared on beam at 0.5% drift ratio
(b) Cracks first appeared on bottom column and joint panel at 1% drift ratio
Page 244
213
(c) Cracks distribution at peak load of 2% drift ratio
(d) Cracks distribution at first failure
Page 245
214
(e) Cracks distribution at 5% after being strengthening
Figure 4.2 Failure mode and cracks distribution of Specimen F-Control
Page 246
215
(a) Flexural cracks appeared on beam, slab and column at 0.5% drift ratio
(b) Cracks appeared on joint panel at 1% drift ratio
(c) Cracks distribution at peak load of 1.5% drift ratio
Page 247
216
(d) Cracks distribution after failure at 5 % drift ratio
Figure 4.3 Failure mode and cracks distribution of specimen F-G-50-200
(a) Flexural cracks first appeared on beam and slab at 0.5% drift ratio
Bottom beam bars slippage
Page 248
217
(b) Cracks appeared on column at 1% drift ratio
(c) Cracks appeared on joint panel at 1.5% drift ratio
(d) Cracks distribution at peak load of 2.5 % drift ratio
Page 249
218
(e) Cracks distribution after failure at 5 % drift ratio
Figure 4.4 Failure mode and cracks distribution of specimen F-O-450-150
(a) Flexural cracks appeared on beam, slab and column at 0.5% drift ratio
Page 250
219
(b) Cracks appeared on joint panel at 1% drift ratio
(c) Cracks distribution at peak load of 2 % drift ratio
Local flexural rotation
Page 251
220
(d) Cracks distribution and deformation shape after failure at 5 % drift ratio
Figure 4.5 Failure mode and cracks distribution of specimen F-O-500-180
Local flexural rotation
Page 252
221
(a) Flexural cracks appeared on beam, slab and column at 0.5% drift ratio
(b) Cracks appeared on joint panel at 1.5% drift ratio
Page 253
222
(c) Cracks distribution at peak load of 2 % drift ratio
Page 254
223
(d) Cracks distribution after failure at 5 % drift ratio
Figure 4.6 Failure mode and cracks distribution of specimen F-S-450-450
(a) Flexural cracks appeared on beam at 0.25% drift ratio
(b) Cracks appeared on slab and column at 0.5% drift ratio
Page 255
224
(c) Cracks appeared on joint panel at 1% drift ratio
(d) Cracks distribution at peak load of 2 % drift ratio
Page 256
225
(e) Column bottom bars buckled at 5% drift ratio
Figure 4.7 Failure mode and crack distributions of specimen S-Control
Page 257
226
(a) A flexural crack appeared on slab at 0.25% drift ratio
(b) Cracks appeared on bottom column at 0.5% drift ratio
(c) Cracks appeared on joint panel at 1% drift ratio
Page 258
227
(d) Cracks distribution at peak load of 2 % drift ratio
(e) Cracks distribution and failure mode at 5 % drift ratio
Figure 4.8 Failure mode and cracks distribution of specimen S-G-50-200-100
Bottom beam bars slippage
Page 259
228
(a) Cracks appeared on beam, slab and column at 0.5% drift ratio
(b) Cracks appeared on joint panel at 1% drift ratio
Page 260
229
(c) Cracks distribution at peak load of 2 % drift ratio
(d) Cracks distribution and failure mode at 5 % drift ratio
Figure 4.9 Failure mode and cracks distribution of specimen S-O-500-180
(a) Cracks appeared on beam and slab at 0.5% drift ratio
Page 261
230
(b) Cracks appeared on column and joint panel 1% drift ratio
(c) Cracks distribution at peak load of 2 % drift ratio
Page 262
231
(d) Cracks distribution and failure mode at 5 % drift ratio
Figure 4.10 Failure mode and crack distribution of specimen S-O-500-180
Local flexural rotation
Page 263
232
(a) Specimen F-Control
(b) Specimen F-G-50-200
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
200
400
600
800
1000
1200
Lateral Drift(%)
Axi
al L
oad(
kN)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
J-Control-Axial Load-Strengthened J-Control-Axial Load
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
200
400
600
800
1000
1200
Lateral Drift (%)
Axi
al L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Axial Load
Page 264
233
(c) Specimen F-O-450-150
(d) Specimen F-O-500-180
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
200
400
600
800
1000
1200
Lateral Drift(%)
Axi
al L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Axial load
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
200
400
600
800
1000
1200
1400
Lateral Drift(%)
Axi
al L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Axial load
Page 265
234
(e) Specimen F-S-450-450
(f) Specimen S-Control
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
200
400
600
800
1000
1200
Lateral Drift (m%)
Axi
al L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Axial load
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
100
200
300
400
500
600
700
800
900
1000
Lateral Drift (%)
Axi
al L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Axial load-1 Axial load-2 Axial load-3
Page 266
235
(g) Specimen S-G-50-200-100
(h) Specimen S-O-500-180
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
100
200
300
400
500
600
700
800
900
1000
Lateral Drift (m%)
Axi
al L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Axial load
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
100
200
300
400
500
600
700
800
900
1000
Lateral Drift (%)
Axi
al L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Axial load
Page 267
236
(i) Specimen S-O-500-180-S-300-300
Figure 4.11 Hysteresis curves of the top-of-column axial loads
(a) Specimen F-Control
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
100
200
300
400
500
600
700
800
900
1000
Lateral Drift (m%)
Axi
al L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Axial load
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-200
-150
-100
-50
0
50
100
150
Lateral Drift(%)
Bea
m T
ip L
oad
Load
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-Control-Left-Strengthened F-Control-Right-Strengthened F-Control-Left F-Control-Right
Page 268
237
(b) Specimen F-G-50-200
(c) Specimen F-O-450-150
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-100
-80
-60
-40
-20
0
20
40
60
80
100
Lateral Drift(%)
Bea
m T
ip L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-G-50-200-Left F-G-50-200-Right
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-O-450-150-Left F-O-450-150-Right
Page 269
238
(d) Specimen F-O-500-180
(e) Specimen F-S-450-450
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-O-500-180-Left F-O-500-180-Right
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-S-450-450-Left F-S-450-450-Right
Page 270
239
(f) Specimen S-Control
(g) Specimen S-G-50-200-100
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144 Lateral Displacement (mm)
S-Control-Left-1 S-Control-Left-2 S-Control-Left-3 S-Control-Right-1 S-Control-Right-2 S-Control-Right-3
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
Load
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
S-G-50-200-100-Left S-G-50-200-100-Right
Page 271
240
(h) Specimen S-O-500-180
(i) Specimen S-O-500-180-S-300-300
Figure 4.12 Hysteresis curves of the beam tip loads
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
S-O-500-180-Left S-O-500-180-Right
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
S-O-500-180-S-300-300-Left S-O-500-180-S-300-300-Right
Page 272
241
(a) Specimen F-Control
(b) Specimen F-G-50-200
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce (k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Envelop Curve F-Control F-Control-Strengthened Yield point Peak point Failure point
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce(k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Envelop Curve F-G-50-200 Yield point Peak point Failure point
0.75Hpeak
Hpeak
δy=-21.8 mm
0.80Hpeak
δu=-96.2 mm
0.75Hpeak
Hpeak 0.80Hpeak
δu=51.4 mm δy=24.3 mm
Page 273
242
(c) Specimen F-O-450-150
(d) Specimen F-O-500-180
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce(k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Envelop Curve F-O-450-150 Yield point Peak point Failure point
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce(k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Envelop Curve F-O-500-180 Yield point Peak point Failure point
δu=111.1 mm δy=25.5mm
0.75Hpeak
Hpeak
0.80Hpeak
δu=101.1 mm δy=24.9 mm
0.75Hpeak
Hpeak
0.80Hpeak
Page 274
243
(e) Specimen F-S-450-450
(f) Specimen S-Control
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce(k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Envelop Curve F-S-450-450 Yield point Peak point Failure point
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce(k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Envelop Curve S-Control Yield point Peak point Failure point
δu=72.8 mm δy=23.7 mm
0.75Hpeak
Hpeak
0.80Hpeak
δu=86.9 δy=28.6 mm
0.75Hpeak
Hpeak
0.80Hpeak
Page 275
244
(g) Specimen S-G-50-200-100
(h) Specimen S-O-500-180
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce(k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Envelop Curve S-G-50-200-100 Yield point Peak point Failure point
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce(k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Envelop Curve S-O-500-180 Yield point Peak point Failure point
δu=61.2 mm δy=20.5 mm
0.75Hpeak
Hpeak
0.80Hpeak
δu=106.00mm
δy=29.88mm
0.75Hpeak
Hpeak
0.80Hpeak
Page 276
245
(i) Specimen S-O-500-180-S-300-300
Figure 4.13 Column shear force versus top-of-column lateral drift ratio hysteresis
and envelop curves
(a) Specimen F-Control
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce(k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
Envelop Curve S-O-500-180-S-300-300 Yield point Peak point Failure point
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-200
-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce (k
N)
F-Control-MTS-Retrofit F-Control-Calculated with Beam Tip Loads-Strengthened F-Control-MTS F-Control-Calculated with Beam Tip Loads
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144 Lateral Displacement (mm)
δu=102.8 mm δy=28.6 mm
0.75Hpeak
Hpeak
0.80Hpeak
Page 277
246
(b) Specimen F-G-50-200
(c) Specimen F-O-450-150
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift(%)
Col
mun
She
ar F
orce
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-G-50-200-MTS F-G-50-200-Calculated with Beam Tip Loads
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Displacement(mm)
Col
umn
Shea
r For
ce (k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-O-450-150-MTS F-O-450-150-Calculated with Beam Tip Loads
Page 278
247
(d) Specimen F-O-500-180
(e) Specimen F-S-450-450
Figure 4.14 Comparison between lateral load applied by MTS and calculated
column shear force
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift(%)
Col
umn
Shea
r For
ce (k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
J-O-500-180-MTS J-O-4500-180-Calculated with Beam Tip Loads
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce (k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-S-450-450-MTS F-S-450-450-Calculated with Beam Tip Loads
Page 279
248
(a) The first test group
(a) The second test group
Figure 4.15 Envelop curves of column shear force versus lateral drift ratio
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-200
-150
-100
-50
0
50
100
150
Lateral Drift (%)
Col
umn
Shea
r For
ce (k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-Control F-G-50-200 F-O-450-150 F-O-500-180 F-S-450-450
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-200
-150
-100
-50
0
50
100
150
200
Lateral Drift (%)
Col
umn
Shea
r For
ce (k
N)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
S-Control S-G-50-200-100 S-O-500-180 S-O-500-180-S-300-300
Page 280
249
(a) The first test group
(b) The second test group
Figure 4.16 Envelop curves of beam tip loads versus lateral drift ratio
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-200
-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-Control F-G-50-200 F-S-450-450 F-O-450-150 F-O-500-180
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-200
-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
S-Control S-G-50-200-100 S-O-500-180 S-O-500-180-S-300-300
Page 281
250
(a) Specimen F-control
(b) Specimen F-G-50-200
200 250 300 350 400 450-0.3
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05 Push-TB11,TB12,TB13
X Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3%
200 250 300 350 400 4500
0.2
0.4
0.6
0.8
1
1.2
1.4 Pull-TB11,TB12,TB13
X Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3%
200 300 400 500 600 700-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4 Push-TB11,TB12,TB13,TB14
X Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
200 300 400 500 600 7000
0.5
1
1.5
2
2.5
3
3.5 Pull-TB11,TB12,TB13,TB14
X Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
Page 282
251
(c) Specimen F-O-450-150
(d) Specimen F-O-500-180
200 400 600 800 1000-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5 Push-TB11,TB12,TB13,TB14,TB15
X Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
200 400 600 800 10000
0.5
1
1.5
2
2.5 Pull-TB11,TB12,TB13,TB14,TB15
X Direction (mm) N
orm
aliz
ed S
train
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
200 300 400 500 600 700-1
-0.5
0
0.5
1
1.5
2 Push-TB11,TB12,TB13,TB14
X Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
200 300 400 500 600 7000
0.5
1
1.5
2
2.5 Pull-TB11,TB12,TB13,TB14
X Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
Page 283
252
(e) Specimen F-S-450-450
(f) Specimen S-control
200 300 400 500 600 700-1
-0.5
0
0.5
1
1.5 Push-TB11,TB12,TB13,TB14
X Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
200 300 400 500 600 7000
0.5
1
1.5
2
2.5
3
3.5
4 Pull-TB11,TB12,TB13,TB14
X Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
200 300 400 500 600-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6 Push-TB11,TB12,TB13,TB14
X Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4%
200 300 400 500 6000
0.5
1
1.5
2
2.5
3 Pull-TB11,TB12,TB13,TB14
X Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4%
Page 284
253
(g) Specimen S-G-50-200-100
(h) Specimen S-O-500-180
200 300 400 500 600-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1 Push-TB11,TB12,TB13,TB14
X Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
200 300 400 500 600-1
-0.5
0
0.5
1
1.5
2 Pull-TB11,TB12,TB13,TB14
X Direction (mm) N
orm
aliz
ed S
train
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
200 300 400 500 600 700-0.4
-0.2
0
0.2
0.4
0.6
0.8
1 Push-TB11,TB12,TB13,TB14,TB15
X Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
200 300 400 500 600 7000
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8 Pull-TB11,TB12,TB13,TB14,TB15
X Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
Page 285
254
(i) Specimen S-O-500-180-S-300-300
Figure 4.17 Strain variation along the X direction at the peak displacement of
each cycle
(a) Specimen F-control
200 300 400 500 600 700-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
3 Push-TB11,TB12,TB13,TB14,TB15
X Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
200 300 400 500 600 700-1
0
1
2
3
4
5
6 Pull-TB11,TB12,TB13,TB14,TB15
X Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-200 -300 -400 -500 -600 -700-0.5
0
0.5
1
1.5
2 Push-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mai
lzed
Stra
in
0.5% 1% 1.5% 2% 2.5% 3%
-200 -300 -400 -500 -600 -7000
0.5
1
1.5
2
2.5
3
3.5 Pull-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3%
Page 286
255
(b) Specimen F-G-50-200
(c) Specimen F-O-450-150
-200 -300 -400 -500 -600 -700-0.5
0
0.5
1
1.5
2
2.5
3 Push-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-200 -300 -400 -500 -600 -7000.5
1
1.5
2
2.5
3
3.5
4 Pull-TL11,TL21,TL31,TL41
Y Direction (mm) N
orm
lized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-200 -300 -400 -500 -600 -700-0.5
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5 Push-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-200 -300 -400 -500 -600 -7000
1
2
3
4
5
6 Pull-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
Page 287
256
(d) Specimen F-O-500-180
(e) Specimen F-S-450-450
-200 -300 -400 -500 -600 -700-0.4
-0.2
0
0.2
0.4
0.6
0.8
1 Push-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4%
-200 -300 -400 -500 -600 -7000.5
1
1.5
2
2.5
3
3.5
4 Pull-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4%
-200 -300 -400 -500 -600 -700-1
0
1
2
3
4
5
6 Push-TL51,TL61,TL71,TL81
Y Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-200 -300 -400 -500 -600 -700-0.5
0
0.5
1
1.5
2
2.5
3
3.5 Pull-TL51,TL61,TL71,TL81
Y Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
Page 288
257
(f) Specimen S-control
(g) Specimen S-G-50-200-100
-200 -300 -400 -500 -600 -700-1
0
1
2
3
4
5
6 Push-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4%
-200 -300 -400 -500 -600 -7000
1
2
3
4
5
6
7
8 Pull-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4%
-200 -300 -400 -500 -600 -700-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6 Push-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-200 -300 -400 -500 -600 -7000
0.5
1
1.5
2
2.5
3
3.5 Pull-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
Page 289
258
(h) Specimen S-O-500-180
(i) Specimen S-O-500-180-S-300-300
Figure 4.18 Strain variation along the Y direction at the peak displacement of
each cycle
-200 -300 -400 -500 -600 -700-0.5
0
0.5
1
1.5
2
2.5
3
3.5
4 Push-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-200 -300 -400 -500 -600 -7000
1
2
3
4
5
6 Pull-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-200 -300 -400 -500 -600 -700-0.2
0
0.2
0.4
0.6
0.8
1
1.2 Push-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-200 -300 -400 -500 -600 -700-0.2
0
0.2
0.4
0.6
0.8
1
1.2 Pull-TL11,TL21,TL31,TL41
Y Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
Page 290
259
(a) Specimen F-Control
(b) Specimen F-G-50-200
-1 -0.5 0 0.5 1 1.5-600
-400
-200
0
200
400
600
800 Push-TC31-TC35,BC31-BC34
Normalized Strain
Z D
irect
ion
(mm
)
0.5% 1% 1.5% 2% 2.5% 3%
-1 -0.5 0 0.5-600
-400
-200
0
200
400
600
800 Pull-TC31-TC35,BC31-BC34
Normalized Strain
Z D
irect
ion
(mm
)
0.5% 1% 1.5% 2% 2.5% 3%
-0.4 -0.2 0 0.2 0.4-600
-400
-200
0
200
400
600
800 Push-TC31-TC35,BC31-BC33
Normalized Strain
Z D
irect
ion
(mm
)
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-0.8 -0.6 -0.4 -0.2 0 0.2-600
-400
-200
0
200
400
600
800 Pull-TC31-TC35,BC31-BC33
Normalized Strain
Z D
irect
ion
(mm
)
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
Page 291
260
(c) Specimen F-O-450-150
(d) Specimen F-O-500-180
-3 -2 -1 0 1-600
-400
-200
0
200
400
600
800 Push-TC31-TC35,BC31-BC33
Normalized Strain
Z D
irect
ion
(mm
)
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-6 -4 -2 0 2-600
-400
-200
0
200
400
600
800 Pull-TC31-TC35,BC31-BC33
Normalized Strain
Z D
irect
ion
(mm
)
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-0.5 0 0.5 1-600
-400
-200
0
200
400
600
800 Push-TC32,TC34,TC35,BC32,BC34
Normalized Strain
Z D
irect
ion
(mm
)
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-1 -0.5 0 0.5-600
-400
-200
0
200
400
600
800 Pull-TC32,TC34,TC35,BC32,BC34
Normalized Strain
Z D
irect
ion
(mm
)
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
Page 292
261
(e) Specimen F-S-450-450
(f) Specimen S-Control
-1.5 -1 -0.5 0 0.5 1-600
-400
-200
0
200
400
600
800 Push-TC31-TC35,BC31-BC34
Normalized Strain
Z D
irect
ion
(mm
)
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-1 -0.5 0 0.5 1-600
-400
-200
0
200
400
600
800 Pull-TC31-TC35,BC31-BC34
Normalized Strain Z
Dire
ctio
n (m
m)
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-6 -4 -2 0 2-600
-400
-200
0
200
400
600 Push-TC71-TC75,CC7,BC71-BC74
Normalized Strain
Z D
irect
ion
(mm
)
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4%
-2 0 2 4 6 8-600
-400
-200
0
200
400
600 Pull-TC71-TC75,CC7,BC71-BC74
Normalized Strain
Z D
irect
ion
(mm
)
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4%
Page 293
262
(g) Specimen S-G-50-200-100
(h) Specimen S-O-500-180
-1 -0.5 0 0.5-400
-300
-200
-100
0
100
200
300
400 Push-TC71-TC73,CC7,BC71-BC72
Normalized Strain
Z D
irect
ion
(mm
)
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-0.5 0 0.5 1-400
-300
-200
-100
0
100
200
300
400 Pull-TC71-TC73,CC7,BC71-BC72
Normalized Strain
Z D
irect
ion
(mm
)
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-4 -3 -2 -1 0 1-400
-300
-200
-100
0
100
200
300
400 Push-TC71-TC72,CC7,BC71,BC73
Normalized Strain
Z D
irect
ion
(mm
)
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-2 0 2 4 6-400
-300
-200
-100
0
100
200
300
400 Pull-TC71-TC72,CC7,BC71,BC73
Normalized Strain
Z D
irect
ion
(mm
)
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
Page 294
263
(i) Specimen S-O-500-180-S-300-300
Figure 4.19 Strain variation along the Z direction at the peak displacement of
each cycle
(a) Specimen F-Control
-1.5 -1 -0.5 0 0.5 1-400
-300
-200
-100
0
100
200
300
400 Push-TC71-TC73,CC7,BC71-BC73
Normalized Strain
Z D
irect
ion
(mm
)
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-1 -0.5 0 0.5 1-400
-300
-200
-100
0
100
200
300
400 Pull-TC71-TC73,CC7,BC71-BC73
Normalized Strain Z
Dire
ctio
n (m
m)
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-150-100
-500
50100
150
-200
-100
0
100
200-1
-0.5
0
0.5
1
1.5
X Direction (mm) Y Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3%
Page 295
264
(b) Specimen F-G-50-200
(c) Specimen F-O-450-150
-150-100
-500
50100
150
-200
-100
0
100
200-1.5
-1
-0.5
0
0.5
1
X Direction (mm) Y Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-150-100
-500
50100
150
-200
-100
0
100
200-1.5
-1
-0.5
0
0.5
1
X Direction (mm) Y Direction (mm)
Nor
mal
ized
Stra
in
0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
Page 296
265
(d )Specimen F-S-450-450
(e) Specimen S-Control
-150-100
-500
50100
150
-200
-100
0
100
200-1
-0.5
0
0.5
1
X Direction (mm) Y Direction (mm)
Nor
mal
ized
Stra
in 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-120-100
-80-60
-40-20
0
-200
-100
0
100
200-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
X Direction (mm) Y Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2%
Page 297
266
(f) Specimen S-G-50-200-100
(g) Specimen S-O-500-180
-150-100
-500
50100
150
-200
-100
0
100
200-2
-1.5
-1
-0.5
0
0.5
1
X Direction (mm) Y Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5%
-150-100
-500
50100
150
-200
-100
0
100
200-8
-6
-4
-2
0
2
4
X Direction (mm) Y Direction (mm)
Nor
mal
ized
Stra
in
0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4%
Page 298
267
(h) Specimen S-O-500-180-S-300-300
Figure 4.20 Rotations of bottom column critical section under push action
(a) CS1 strain gauge
-150-100
-500
50100
150
-200
-100
0
100
200-1.5
-1
-0.5
0
0.5
1
1.5
X Direction (mm) Y Direction (mm)
Nor
mal
ized
Stra
in 0.25% 0.5% 1% 1.5% 2% 2.5% 3% 4% 5%
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
0.2
0.4
0.6
0.8
1
1.2
1.4
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-Control-CS1 F-G-50-200-CS1 F-S-450-450-CS1
Page 299
268
(b) CS2 strain gauge
(c) CS3 strain gauge
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-1
0
1
2
3
4
5
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-Control-CS2 F-S-450-450-CS2
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
0.5
1
1.5
2
2.5
3
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-G-50-200-CS3 F-S-450-450-CS3
Page 300
269
(d) CS4 strain gauge
Figure 4.21 Strain level of joint stirrups at the peak displacement of each cycle of
the first test group
(a) CS12 strain gauge
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
1
2
3
4
5
6
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
F-Control-CS4 F-O-450-150-CS4 F-G-50-200-CS4 F-S-450-450-CS4
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-0.2
0
0.2
0.4
0.6
0.8
1
1.2
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
S-Control-CS12 S-G-50-200-100-CS12 S-O-500-180-CS12
Page 301
270
(b) CS22 strain gauge
(c) CS32 strain gauge
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 60
0.5
1
1.5
2
2.5
3
3.5
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
S-Control-CS22 S-G-50-200-100-CS22 S-O-500-180-CS22 S-O-500-180-S-300-300-CS22
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-0.5
0
0.5
1
1.5
2
2.5
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
S-Control-CS32 S-G-50-200-100-CS32 S-O-500-180-CS32 S-O-500-180-S-300-300-CS32
Page 302
271
(d) CS42 strain gauge
Figure 4.22 Strain level of joint stirrups at the peak displacement of each cycle of
the second test group
Figure 4.23 Strain level of FRP U-jacket at the peak displacement of each cycle of
specimen F-G-50-200
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6-0.5
0
0.5
1
1.5
2
2.5
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -120 -96 -72 -48 -24 0 24 48 72 96 120 144
Lateral Displacement (mm)
S-Control-CS32 S-G-50-200-100-CS32 S-O-500-180-CS32 S-O-500-180-S-300-300-CS32
-6 -4 -2 0 2 4 6-0.02
0
0.02
0.04
0.06
0.08
0.1
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
U1 U2 U3 U4 U5
Page 303
272
(a) Strain gauges mounted on the FRP U-jacket
(b) Strain gauges mounted on the FRP wrap
Figure 4.24 Strain level of FRP at the peak displacement of each cycle of
specimen F-O-450-150
-6 -4 -2 0 2 4 6-0.02
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
U1 U2 U3
-6 -4 -2 0 2 4 6-0.01
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
U1 U4 U5
-6 -4 -2 0 2 4 6-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
W1 W2 W3
Page 304
273
(a) Strain gauges mounted on the FRP U-jacket
(b) Strain gauges mounted on the FRP wrap
Figure 4.25 Strain level of FRP at the peak displacement of each cycle of
specimen F-O-500-180
-6 -4 -2 0 2 4 6-0.05
0
0.05
0.1
0.15
0.2
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144
Lateral Displacement (mm)
U1 U2 U3
-6 -4 -2 0 2 4 60
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144
Lateral Displacement (mm)
U4 U5 U6
-6 -4 -2 0 2 4 60
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144
Lateral Displacement (mm)
U7 U8 U9
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
0.3
0.4
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
RW1 RW2 RW3 RW4 RW5
-6 -4 -2 0 2 4 6
0
0.2
0.4
0.6
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
RW6 RW7 RW8 RW9 RW10
-6 -4 -2 0 2 4 6
0
0.1
0.2
0.3
0.4
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
LW1 LW2 LW3 LW4 LW5
-6 -4 -2 0 2 4 6
0
0.1
0.2
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
LW6 LW7 LW8 LW9 LW10
Page 305
274
Figure 4.26 Strain level of FRP U-jacket at the peak displacement of each cycle of
specimen S-G-50-200-100
(a) Strain gauges mounted on the FRP U-jacket
-6 -4 -2 0 2 4 6-0.05
0
0.05
0.1
0.15
0.2
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
RU1 RU2 RU3
-6 -4 -2 0 2 4 6-0.05
0
0.05
0.1
0.15
0.2
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
RU4 RU5 RU6
-6 -4 -2 0 2 4 6-0.05
0
0.05
0.1
0.15
0.2
Lateral Drift (%) N
omal
ized
Stra
in
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
LU1 LU2 LU3
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
RU1 RU2 RU3
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
RU3 RU4 RU5
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
LU1 LU2 LU3
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
LU3 LU4 LU5
Page 306
275
(b) Strain gauges mounted on the FRP wrap
Figure 4.27 Strain level of FRP at the peak displacement of each cycle of
specimen S-O-500-180
-6 -4 -2 0 2 4 6-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
RW1 RW2 RW3 RW4
-6 -4 -2 0 2 4 6-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
RU9 RU10
-6 -4 -2 0 2 4 6-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
RW5 RW6 RW7 RW8
-6 -4 -2 0 2 4 6-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
LW1 LW2 LW3 LW4
-6 -4 -2 0 2 4 6-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
LU9 LU10
-6 -4 -2 0 2 4 6-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
LW5 LW6 LW7 LW8
Page 307
276
(a) Strain gauges mounted on the FRP U-jacket
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
0.3
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
RU1 RU2 RU3
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
0.3
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
RU3 RU4 RU5
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
0.3
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
LU1 LU2 LU3
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
0.3
Lateral Drift (%)
Nom
aliz
ed S
train
-144 -96 -48 0 48 96 144
Lateral Displacement (mm)
LU3 LU4 LU5
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
0.3
0.4
0.5
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
RW1 RW2 RW3 RW4
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
0.3
0.4
0.5
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
RU9 RU10
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
0.3
0.4
0.5
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
RW5 RW6 RW7 RW8
Page 308
277
(B) Strain gauges mounted on the FRP wrap
Figure 4.28 Strain level of FRP at the peak displacement of each cycle of
specimen S-O-500-180-S-300-300
(a) Definition 1 for yield deformation
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
0.3
0.4
0.5
Lateral Drift (%)
Nom
aliz
ed S
train
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
0.3
0.4
0.5
Lateral Drift (%)
Nom
aliz
ed S
train
-6 -4 -2 0 2 4 6-0.1
0
0.1
0.2
0.3
0.4
0.5
Lateral Drift (%)
Nom
aliz
ed S
train
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
LW1 LW2 LW3 LW4
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
LU9 LU10
-144-96 -48 0 48 96 144 Lateral Displacement (mm)
LW5 LW6 LW7 LW8
Page 309
278
(b) Definition 2 for yield deformation
(c) Definition 3 for yield deformation
Page 310
279
(d) Definition 4 for yield deformation
Figure 4.29 Definitions for the yield deformation (Hu 2005)
(a) Definition 1 for the ultimate state
Page 311
280
(b) Definition 2 for the ultimate state
(c) Definition 3 for the ultimate state
Page 312
281
(d) Definition 4 for the ultimate state
Figure 4.30 Definitions for the ultimate state (Hu 2005)
Figure 4.31 Specimens ductility
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
μ δ
Page 313
282
Figure 4.32 Energy dissipated, equivalent viscous damping ratio and effective
secant stiffness definitions for hysteresis curves
F
△
A2
A1
AED
△
1
△
2
F1
F2
Keff
Fm2
△
m2
Fm1
△
m1
1 21 2 1
3 41 1 2
1 1 2 2
( )2
( )2
12
d i ED
EDhyst
m m m m
F FA
F FA
E A A
AF F
F3
F4
Page 314
283
(a) The first test group
(b) The second test group
Figure 4.33 Dissipated energy of each hysteresis cycle
0 1 2 3 4 5 60
2
4
6
8
10
12
14
Lateral Drift(%)
Dis
sipa
ted
Ener
gy o
f Eac
h H
yste
resi
s C
ycle
(kN
·m)
F-Control F-G-50-200 F-O-450-150 F-O-500-180 F-S-450-450
0 1 2 3 4 5 60
5
10
15
20
25
Lateral Drift (%)
Dis
sipa
ted
Ener
gy o
f Eac
h H
yste
resi
s C
ycle
(kN
·m)
S-Control S-G-50-200-100 S-O-500-180 S-O-500-180-S-300-300
Page 315
284
(a) The first test group
(a) The second test group
Figure 4.34 Development of cumulative dissipated energy
0 1 2 3 4 5 60
10
20
30
40
50
60
70
80
Lateral Drift(%)
Cum
ulat
ive
Dis
sipa
ted
Ener
gy (k
N·m
)
F-Control F-G-50-200 F-O-450-150 F-O-500-180 F-S-450-450
0 1 2 3 4 5 60
10
20
30
40
50
60
70
80
90
Lateral Drift (%)
Cum
ulat
ive
Dis
sipa
ted
Ener
gy (k
N·m
)
S-Control S-G-50-200-100 S-O-500-180 S-O-500-180-S-300-300
Page 316
285
(a) The first test group
(b) The second test group
Figure 4.35 Equivalent viscous damping ratio development
0 1 2 3 4 5 60
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Lateral Drift (%)
Equ
ival
ent V
isco
us D
ampi
ng R
atio
ζhy
st
F-Control F-G-50-200 F-O-450-150 F-O-500-180 F-S-450-450
0 1 2 3 4 5 60
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Lateral Drift (%)
Equ
ival
ent V
isco
us D
ampi
ng R
atio
ζ
S-Control S-G-50-200-100 S-O-500-180 S-O-500-180-S-300-300
Page 317
286
(a) The first test group
(b) The second test group
Figure 4.36 Evolution of the stiffness degradation
0 1 2 3 4 5 60
1000
2000
3000
4000
5000
6000
7000
Lateral Drift (%)
Pea
k-to
-pea
k St
iffne
ss K
eff (
kN/m
)
F-Control F-G-50-200 F-O-450-150 F-O-500-180 F-S-450-450
0 1 2 3 4 5 60
1000
2000
3000
4000
5000
6000
7000
8000
Lateral Drift (%)
Pea
k-to
-pea
k St
iffne
ss K
eff (
kN/m
)
S-Control S-G-50-200-100 S-O-500-180 S-O-500-180-S-300-300
Page 318
287
Figure 4.37 Components of plastic hinge
(a) Specimen F-Control
-50 -40 -30 -20 -10 0 10 20 30 40 50-200
-150
-100
-50
0
50
100
150
Beam Tip Displacement (mm)
Bea
m T
ip L
oad(
kN)
Left beam Right beam
Page 319
288
(b) Specimen F-G-50-200
(c) Specimen F-S-450-450
-80 -60 -40 -20 0 20 40 60 80-100
-80
-60
-40
-20
0
20
40
60
80
100
Beam Tip Displacement (mm)
Bea
m T
ip L
oad(
kN)
Left beam Right beam
-80 -60 -40 -20 0 20 40 60 80-150
-100
-50
0
50
100
Beam Tip Displacment (mm)
Bea
m ti
p lo
ad(k
N)
Left beam Right beam
Page 320
289
(d) Specimen S-Control
(e) Specimen S-G-50-200-100
Figure 4.38 Hysteresis curves of beam tip load versus calculated beam tip
displacement
-50 -40 -30 -20 -10 0 10 20 30 40 50-200
-150
-100
-50
0
50
100
150
Beam Displacement (mm)
Bea
m T
ip L
oad
(kN
)
Left beam Right beam
-80 -60 -40 -20 0 20 40 60 80-150
-100
-50
0
50
100
150
Beam Tip Displacment (mm)
Bea
m T
ip L
oad(
kN)
Left beam Right beam
Page 321
290
(a) Contribution of beam deformation
(b) Contribution of column deformation
Figure 4.39 Deformation contributions of beams and columns to the lateral drift
ratio
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
100.00%
0 1 2 3 4 5 6
Con
trib
utio
n of
bea
m d
efor
mat
ion
to th
e la
tera
l dri
ft
Lateral drift ratio (%)
F-Control F-S-450-450 F-G-50-200 S-Control S-G-50-200-100
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
0 1 2 3 4 5 6
Con
trib
utio
n of
col
umn
defo
rmat
ion
to th
e la
tera
l dri
ft
Lateral drift ratio (%)
F-O-450-150 F-O-500-180 F-S-450-450 F-G-50-200 S-Control S-O-500-180 S-G-50-200-100 S-O-500-180-S-300-300
Page 322
291
Table 4.1 Failure process and mode
Specimen
Drift ratio when beam
concrete first
cracked (%)
Drift ratio when
column concrete
first cracked
(%)
Drift ratio when joint
concrete first
cracked (%)
Drift ratio at peak load (%)
Beam Steel bars
yielding
Slab longitudinal
bars yielding
Joint stirrup
yielding
Column Steel bars yielding
Failure mode
F-Control
During applying 970 kN
axial load.
1.0 1.0 2.0 Yielded at 1.5% drift
ratio
Yielded at 1.5% drift
ratio
One stirrup yielded at 1.5% and all yielded
at 3.0% drift ratio
Steel bars at bottom column
yielded at 2.5% drift ratio under
tension
Before being retrofitted, shear failure
at the beam end pin region at 4% drift ratio. After being retrofitted, shear failure at the joint
panel
Page 323
292
Table 4.2 (cont.)
F-G-50-200 0.5 0.5 1.0 1.5
Beam bottom bars yielded at 1.0% drift ratio under
compression
Some close to the beam yielded at 1.0% drift
ratio and all yielded at 1.5% drift
ratio
No yielding No yielding Beam bottom bars
slipping in the joints region
F-O-450-150 0.5 1.0 1.5 2.5 Yielded at 1.5% drift
ratio
Some close to the beam yielded at 1.0% drift
ratio and all yielded at 1.5% drift
ratio
Stirrups yielded at 2.0% ratio
Steel bars at bottom column
yielded at 2.0% drift ratio under
compression
Shear failure at the joint pane
Page 324
293
Table 4.3 (cont.)
F-O-500-180 0.5 0.5 1.0 2.0 Yielded at 1.5% drift
ratio
Some close to the beam yielded at 1.0% drift
ratio and all yielded at 1.5% drift
ratio
One stirrup yielded at 1.5% and all yielded
at 4.0% drift ratio
No yielding Local flexural failure at
the chord ends
F-S-450-450 0.5 0.5 1.5 1.5 Yielded at 1.5% drift
ratio
The uncut slab bars yielded at 1.0% drift
ratio
Two stirrups
yielded at 2.0% drift
ratio and all yielded at 3% drift
ratio
Steel bars at top column yielded at 5.0% drift ratio under
compression
Beam flexural failure
Page 325
294
Table 4.4 (cont.)
S-Control 0.25 0.5 1.0 2.0 Yielded at 1.5% drift
ratio
Some close to the beam yielded at 1.0% drift
ratio and all yielded at 2% drift
ratio
Some stirrups
yielded at 2.5% drift
ratio
Steel bars at bottom column
yielded at 2.0% drift ratio under
compression
Bottom column bars buckling
S-G-50-200-100 0.25 0.5 1.0 2.0 Yielded at 1.5% drift
ratio
Some close to the beam yielded at 2.5% drift
ratio and all yielded at 4.0% drift
ratio
No yielding No yielding Beam bottom bars
slipping in the joints region
Page 326
295
Table 4.5 (cont.)
S-O-500-180 0.5 0.5 1.0 2.0 Yielded at 1.0% drift
ratio
All yielded at 1.0%
drift ratio
Two middle stirrups in
joint yielded at 2.5% ratio
Steel bars at bottom column
yielded at 2.0% drift ratio under
compression
Bottom column bars buckling
S-O-500-180-S-300-300 0.5 1.0 1.0 2.0 Yielded at 1.0% drift
ratio
50% slab bars yielded
at 2.0% drift ratio
One middle stirrup in
joint yielded at 2.5% ratio and two
middle ones yielded at 4.0% drift
Steel bars at bottom column
yielded at 2.5% drift ratio under
compression
Local flexural failure at the chord ends
Page 327
296
Table 4.2 Strengths and weakening ratios of test and prediction
Test group
Items
Beam Column Strength (kN)
Weakening ratio
Calculated column shear force (kN)
Predicted column shear
force (kN) - + Total
First
Cross-sectional analysis
R-section 95.6 72.2 167.8 -30.7% 101.6 150.1 T-section 154.2 87.8 242 0.0% 146.5 150.1
F-Control 156.6 87.2 243.8 0.0% 130.2 150.1 F-G-50-200 85.6 87.4 173 -29.0% 96.9 150.1 F-O-450-150 138.3 88.5 226.8 -7.0% 120.2 150.1 F-O-500-180 115.7 88.2 203.9 -16.4% 190.4 150.1 F-S-450-450 109.4 83.1 192.5 -21.0% 110.9 150.1
Second
Cross-sectional analysis
R-section 97.8 97.8 195.6 -29.1% 118.2 150.8 T-section 156.4 119.6 276 0.0% 166.7 150.8
S-Control 144.4 112.8 257.2 0.0% 147.0 150.8 S-G-50-200-100 111 107.3 218.3 -15.1% 122.0 150.8
S-O-500-180 120.4 119.3 239.7 -6.8% 132.5 150.8 S-O-500-180-S-300-300 111.7 102.6 214.3 -16.7% 118.6 150.8
Page 328
297
Table 4.3 Specimen ductility
Specimen Calculated peak column strength
(kN)
Displacement at 75% peak load (mm)
Yield displacement
(mm)
Displacement at failure
(mm) Ductility
F-Control 128.5 16.3 21.8 96.2 4.4 F-G-50-200 96.9 18.2 24.3 51.4 2.1 F-O-450-150 120.2 19.1 25.5 111.1 4.4 F-O-500-180 109.4 18.7 24.9 101.3 4.1 F-S-450-450 110.9 17.7 23.7 72.8 3.1
S-Control 147.0 21.5 28.6 86.9 3.0 S-G-50-200-100 122.0 15.4 20.5 61.2 3.0
S-O-500-180 132.5 22.4 29.9 94.3 3.2 S-O-500-180-S-300-300 118.6 21.5 28.6 102.7 3.6
Page 329
298
Table 4.4 The lengths of yielding and strain concentration zones
Items F-Control F-G-50-200 F-S-450-450 S-Control S-G-50-200-100 Yielding zone length (mm) 200 190 310 360 310 Strain concentration zone
length (mm) /* 120
More than 120
110 /**
Note:*: This value could not be determined as the pre-failure of the specimen F-Control.
**: This value could not be determined as limited number of strain gauges.
Page 330
299
CHAPTER 5
THREE-DIMENSIONAL FINITE ELEMENT
MODELLING OF REINFORCED CONCRETE
T-BEAMS WITH A WEB OPENING WEAKENING
AND WITH/WITHOUT FRP SHEAR
STRENGTHENING
5.1 INTRODUCTION
A beam sway (BS) mechanism (i.e. one with plastic hinges at beam ends)
is the most preferred failure mode for a reinforced concrete (RC) frame
subjected to seismic attack, because of its efficient energy-dissipation capacity.
To ensure the BS mechanism failure and enhance the seismic safety of existing
RC frames, appropriate seismic retrofitting is needed for those RC structures
designed according to the out-dated and inadequate design codes. One technique
is to include web openings at beam ends, the consequential reduced shear
capacity can be compensated by shear strengthening (e.g. using FRP). This
approach has been recently proposed by the authors’ research group based on
the previous studies on the behaviour of RC beams containing web openings for
other purposes, e.g. openings for the passing of the pipes or wires of ventilation
and air conditioning systems (Mansur et al. 1999; Abdalla et al. 2003;
Page 331
300
Maaddawy and Sherif 2009; Madkour 2009; Pimanmas 2010; Chin et al. 2012;
Hawileh et al. 2012; Maaddawy and Ariss 2012). These studies showed that
web openings can significantly reduce beam flexural capacities. Recently, this
proposed technique was validated through a series of tests by Nie et al. (2018)
and Nie (2018) on full scale T-beams. A few recent studies (Madkour 2009;
Pimanmas 2010; Chin et al. 2012; Hawileh et al. 2012, Nie 2018) used
two-dimensional (2D) (Pimanmas 2010; Chin et al. 2012; Nie 2018) and
three-dimensional (3D) (Madkour 2009; Hawileh et al. 2012) finite element (FE)
models to simulate RC beams with web openings. The last three of those studies
(Chin et al. 2012; Hawileh et al. 2012, Nie 2018) also modelled bond slip
behaviour between the shear strengthening FRP and the beam concrete. Among
the five studies, only Nie (2018) proposed a 2D FE models concerned
specifically with RC T-beams weakened by web openings. However, Nie’s
(2018) 2D models are subjected to an obvious limit: it can't predict the
non-uniform deformations/stresses in the transverse direction, especially those
in the flange of the T-beam, necessitating the development of a more
accurate/powerful 3-D FE model for RC beams with and without opening in the
present study.
To gain an in-depth understanding of the potential of this retrofit technique,
a 3D finite element (FE) model, using the general purpose FE programme
ABAQUS (2012) for RC T-beams with web openings, was proposed and
substantiated with test results. The applicability of using an explicit dynamic
Page 332
301
method, which is not subject to numerical convergence problems, to the
simulation of RC T-beams with web openings under static load, is also
examined and described below. In particular, the effect of some key parameters
including loading scheme, duration and damping ratio, among others, were
evaluated. The results of the 3D FE model are then compared with the
predictions of the 2D FE models used in Nie (2018) to show the advantages of
the proposed 3-D FE model.
5.2 PROPOSED 3D FE MODEL
The proposed 3D FE models are based on the 2D FE model of Chen et al
(2011), which did overcome the convergence problem and provided accurate
predictions on the test results of RC beam, flexurally strengthened by FRP
sheets and failed by intermediate crack induced debonding (IC debonding)
under monotonic loading. The 2D FE model, which is based on the
smeared-crack approach, employed the crack band model of Bazant and Planas
(1998) to overcome the mesh sensitivity problem. According to Chen et al.
(2011), to precisely predict flexural cracks and the associate FRP debonding
failure, an accurate modelling of concrete cracking behaviour (especially the
post-cracking behaviour of concrete), accurate bond-slip modelling of bond
behaviour between steel reinforcement and concrete and that between FRP and
substrate concrete are needed (Chen et al. 2011). In the proposed 3D FE model,
all these key components are considered and included, as will be detailed below.
Taking advantage of the symmetry of the specimen and the loading pattern, only
Page 333
302
a half width of beam was modelled in the proposed FE model. The 3D model is
meshed with 20 mm elements, which is small enough to obtain reliable results
according to a separate convergence study.
5.2.1 Modelling of concrete
The concrete is represented by the 3D solid elements C3D8R in ABAQUS
(2012). The accurate representation of cracked concrete behaviour is essential
for the accurately modelling of the cracking behaviour of an RC beam. There
are three built-in models of concrete in ABAQUS (2012). These consist of the
smeared concrete cracking in ABAQUS/Standard, the brittle cracking (i.e. BC)
model in ABAQUS/Explicit and the concrete damaged plasticity (i.e. DP)
model in both ABAQUS/Standard and ABAQUS/Explicit. The brittle cracking
model assumes linear behaviour in compression, which is not appropriate when
compression failure dominates the failure mode. Therefore, in the proposed
model, the concrete damaged plasticity model of ABAQUS (2012) was adopted
to model concrete cracking. For modelling compressive behaviour of concrete,
the constitutive equation proposed by Saenz (1964), for the state of uni-axial
compression, was adopted following Chen et al. (2011).
2 1 / 2 / /p p p p
(5.1)
where and are axial compressive stress and strain respectively; p
Page 334
303
and p are the peak stress and its corresponding strain; c is the coefficient
representing the initial tangent modulus of concrete. In the study, c was set
to be equal to the elastic modulus of concrete Ec. The latter three parameters
( p , p and ) were given by the average values of three cylinder tests, as
given in Tables 5.1 and 5.2. When there are no available cylinder test results p
and Ec were set equal to 0.002 and 4730 'cf respectively.
According to Malm (2009), using a stress-strain relationship for tensile
behaviour of cracked concrete introduces considerable mesh sensitivity for the
parts of an RC specimen without steel reinforcement. In those places it is more
appropriate to use fracture energy or stress-crack opening laws for modelling
the tensile behaviour cracked concrete. The tension-softening relationship
proposed by Hordijk (1991), as shown in Equation (5.2) was adopted following
Chen et al. (2011) for its good performance (Malm, 2009).
t
2cr 2
3 w-cw -c3t t t
1 1t cr cr
w w= 1+ c e - 1+c ef w w
(5.2)
Fcr
t
Gw = 5.14f (5.3)
where t is the tensile stress normal to the crack, tw is crack opening
displacement; crw is crack opening displacement at the point of concrete stress
Page 335
304
complete release; FG is the concrete fracture energy calculated by the area
bounded by the strain-stress curve of concrete under uni-axial tension; 1c and
2c are constants determined from tensile tests of the concrete, which can be set
equal to 3.0, 6.93, respectively; for normal concrete. tf and FG can be
estimated from the cylinder compressive strength based on the equations in
CEB-FIP (1993), if no specific test data are provided for determining these two
parameters.
2
' 3c
tf - 8f = 1.410
Mpa (5.4)
0.7'
20.0469 0.5 2610
cF a a
fG d d
(5.5)
where ad represents the maximum aggregate size, which is assumed as 20 mm
if no test data is available.
5.2.2 Definition of Damage Evolution
In the proposed 3D FE model, a damage evolution curve is employed to
describe the progressive degradation of concrete when it enters the softening
phase. The damage model used in Chen el al. (2011) was adopted for simulation
purposes as it takes into account the element size effect. In the damage model,
the unloading path is assumed to return to the original point and therefore the
Page 336
305
equivalent plastic strain plt is set equal to zero. Thus, the damage model is a
elastic (ELA) damage model. The damage factor dt , expressed in terms of
crack-opening displacement tw for cracked concrete as below:
( ) /
tt
t c t c
wdw h E
(5.6)
where ch is crack band width which is defined as the characteristic crack
length of an element in ABAQUS. In the present study, it is equal to the element
size with a reduced integration scheme as recommend in Rot's (1988).
5.2.3 Modelling of Steel Reinforcement
The steel reinforcements in the 3D FE model are modelled with B31 beam
elements. A beam element is more suitable than a truss element for representing
steel reinforcement as it can resist local moments. Two types of steel bar, type
one bar and type two bar are assumed, as indicated in Figure 5.1. The
stress-strain curve of type 1 bar possesses an obvious yielding platform and a
hardening range; whereas, that of Type 2 features a gradual yielding process. To
better capture type 2 bar behaviour under monotonic loading, a model proposed
by Ramberg Osgood (1943) was applied. The equation proposed by Ramberg
Osgood is:
Page 337
306
1
0
= ( )ns ss s
soE
(5.7)
where s is the axial strain and s is the axial stress; s is the yield offset
and 0s is the yield stress. n1 is a parameter controlling the transition from the
elastic branch to the plastic branch. The hardening ratio of the material is
decreased by increasing “n1”. Commonly used value for n1 is 5 or greater. The
first part of the equation at the right hand side represents elastic strain and the
second part, plastic strain. 0.2% is a generally used value for the yield offset s
if the yield stress is hard to determine. This is also the value suggested in the
Chinese code (GB/T 228.1-2010) and the British standard (BS EN ISO
6892-1:2009). Slab steel bars arranged in the transverse direction are assumed
to experience marginal slip during the loading. Thus these steel bars are
perfectly embedded in concrete. For those steel bars in the longitudinal direction,
interfacial cohesive elements are employed to simulate the bond-slip behaviour
between the longitudinal steel bars and concrete, as will be detailed in
Sub-section two .
5.2.4 Modelling of FRP Reinforcement.
In the proposed 3D FE model, the CFRP reinforcement is represented by
S4R 4-node shell elements and treated as a linear elastic brittle material with an
elastic modulus of 227.380 GPa. Only the fibre direction is provided with
stiffness. The Poisson's ratio of the FRP is set as small as possible and as it can't
Page 338
307
be set to be equal to 0, value of 0.001 is used. The FRP reinforcement ruptures
when it reaches its tensile strength of 2820 MPa. It should be noted that the
values of elastic modulus and tensile strength for FRP were calculated based on
the nominal thickness of CFRP sheet coupons according to ASTM-3039(2008).
Bond-slip behaviour between FRP and concrete surface is represented by
cohesive elements, as detailed below.
5.2.5 Modelling of Bond Behaviour
In the proposed 3D FE model, the bond behaviour between steel
reinforcement and concrete and between FRP and concrete is modelled by
employing cohesive elements COH3D8, which are built using the separate
nodes at the same locations of the two adjacent elements, representing the
concrete and the reinforcements (either FRP or steel reinforcement) respectively.
The shear force between the two elements connected via the cohesive element is
equal to the interfacial shear stress (i.e. bond stress) multiplied by the bonded
area represented by the section area of the cohesive element.
For bond-slip behaviour of the steel-concrete interface, the CEB-FIP(1993)
bond-slip model, as expressed by the following Equations (5.8-5.10), were
adopted to define the cohesive elements, following Chen et al. (2012).
Page 339
308
11
2
3 2
( )
max s
max 1 s 2s
sm
s s
s
s s sax m f 3
sf 3
ax 2 s
s
0 s ss
s s s=
s s s s ss s
s > s
s
(5.8)
max max2.0 , 0.15s s sck ff deformed bar (5.9)
max 0.3s sf ckf plain bar (5.10)
where s is the local shear bond stress between steel bars and concrete; ss is
the slip; maxs is the local shear bond strength; s
f is the failure bond stress;
ckf is the concrete standard compressive strength; =0.4 for deformed bars
and 0.5 for plain bars; , for deformed bars;
for plain bars.
No relative displacement is allowed for in the direction normal to the steel
reinforcement and concrete interface, following Chen et al. (2012).
As one B31 element used for modelling the steel bars has only two nodes,
a 3-D prism cohesive element modelling approach using the cohesive element
COH3D8 was proposed to connect the corresponding nodes of concrete and
steel bar, as shown in Figure 5.2. In this modelling approach, the first two nodes,
Nodes 1 and 2 of the cohesive element are connected to one concrete element
node, which means the two nodes are located at the same position. The third
Page 340
309
node, Node 3 is connected to a bar element node. The forth node, Node 4, is
used to create an area for the cohesive section. All the degrees of freedom
(DOFs) of node 4 are set equal to those of Node 3 to secure an unchanged area.
The area of cohesive element is determined by the positions of sectional nodes
and the thickness of cohesive elements and the shear force between the two
elements connected via the cohesive element is equal to the interfacial shear
stress (i.e. bond stress) multiplied by the bonded area represented by the section
area of the cohesive element. Thus, if the sectional area created by the four
nodes is not equal to the circumference of a bar, the shear stress along the
steel-concrete interface direction should be adjusted to keep the shear force
unchanged.
To verify the reliability of the proposed 3D prism cohesive element
modelling approach, a prism cohesive element was built and modelled using
ABAQUS (2012). a assumed circumference of a deformed bar equal to 10 mm
was applied. As the area of the prism cohesive element is only half that of a
rectangular section corresponding to a cuboid cohesive element, which is
usually used, the shear stress was multiplied by 2. Figure 5.3 indicates that the
3D prism cohesive model can accurately reproduce the input data, thus
demonstrating that the proposed model is reliable.
The bond behaviour between FRP and concrete was modelled using
rectangular section cohesive elements. The simplified bond-slip model
developed by Lu et al (2005) was adopted to define the bond behaviour of the
Page 341
310
FRP-concrete interface. The model can be expressed by following equation:
0
0
0
0
( 1)
max
m
ss
ax
0 s s=
ss
e s s
(5.11)
0.0195o w ts f (5.12)
max 1 w tf (5.13)
2 /1 /
f ww
f w
b bb b
(5.14)
0.550.395( )t cuf f (5.15)
max 0
123
fGs
(5.16)
20.308f w tG f (5.17)
where is the local shear bond stress between FRP sheet and concrete; max
is the local bond strength; s is the slip; s0 is the slip when the bond stress
reached max ; w is the width ratio factor; fb is the width of FRP; cb is the
width of beam; cuf is the cube compressive strength of concrete; fG is the
Page 342
311
interfacial fracture energy; 1 =1.5 and / 1f wb b for the case of presence
study.
In the direction normal to the interface, it was assumed there is no relative
displacement between FRP and concrete. That is because the effect of normal
stress on IC de-bonding is insignificant (Chen et al 2011) and besides, IC
de-bonding generally propagates parallel to the FRP- to-concrete interface (e.g.,
Brena et al. 2003; Matthys 2000).
5.2.6 Solution Strategy and Boundary Conditions
Convergence issue is a big problem to be overcome when using a 3D FE
model to simulate structural failure associated with concrete cracking, especially
when bond behaviour is incorporated in the FE model (Chen et al. 2015). The
existence of a web opening further worsens the convergence problem. Thus, the
dynamic explicit approach, which adopts the central difference integration
method, was used to avoid the convergence issue. The validity and accuracy of
the dynamic explicit method have been proved by Chen el al. (2015). In the
proposed a 3D FE model, the densities of materials are set equal to the actual
values, of 2.5e-9 ton/mm3 for concrete and cohesive elements, 7.25e-9 ton/mm3
for steel elements and 1.75e-9 ton/mm3 for FRP elements. The
stiffness-proportional damping coefficient β was used to damp out the dynamic
effect associated with high mode and a value of 0.000002 was adopted based on
parameter study. To minimize the dynamic effect, a displacement-control load is
Page 343
312
applied following the smooth loading scheme shown in Figure 5.5, as suggested
by Chen et al. (2015). The loading duration was set equal to 50 times T0, the
natural period of vibration of simulated specimen. The parameter values were
determined based on a careful parametric study, as will be described below.
The boundary conditions of the 3D models are illustrated in Figure 5.4.
The displacement-control load was applied centrally at the top loading point.
The left hand supporting points are restrained in both the Y and Z directions and
the right hand supporting points are restrained in the Y direction only. None of
the nodes on the section of symmetry are allowed for any displacement in the X
direction.
5.3 VERIFICATION OF THE PROPOSED 3D FE MODEL
5.3.1 Specimen Dimensions and Material Properties
The results for the specimens tested by Nie (2018) were employed to verify
the proposed 3D FE models. Those specimens include T-beams with slab flange
widths satisfying the requirements of the Chinese design code GB-50011 (2010).
The behaviour of T-beams with web openings, with or without FRP
strengthening, was investigated when subjected to monotonic loading. Sufficient
details of the specimens are available in Nie (2018).
5.3.1.1 Specimens' dimensions
Two groups of full-scale RC beams, one including a rectangular beam
Page 344
313
(R-beam) and 7 T-beams, the other 6 T-beams, had been tested by Nie (2018).
The aims of their test program had been to estimate the performance of the
proposed retrofit technique of creating a web opening in existing RC T-beams to
decrease their flexural strength along with associated FRP shear strengthening
measures. The first test group included two control specimens, consisting of a
solid rectangular beam CB-Rec and a solid T-beam CB-T. The remaining
specimens were all weakened with web openings. In the first test group, two
opening sizes (length × depth = 700 mm ×300 mm and 800 mm ×280 mm) were
studied. For each opening size, three specimens were tested, one with a web
opening only, one with a web opening and associated FRP shear strengthening,
and third one also with a web opening and associated FRP strengthening, but
tested under a positive loading condition only (i.e. the slab was in compression).
For the second test group, four opening sizes (length × height=600mm ×
220mm, 700 mm × 200 mm, 600mm × 280 mm and 700mm × 260 mm
respectively) were studied by testing six T-beams. Two of the six T-beams were
weakened by a web opening (opening sizes of length × height=600mm ×
220mm, 700 mm × 200 mm) only. The remaining four specimens have
additional FRP strengthening and their openings are of different opening sizes.
The names of specimens indicated their opening size, bending direction and
having/having not FRP strengthening. FRP strengthening. details of the main
variables of the specimens are listed in Table 5.1.
The length of all specimens was 3500 mm with a clear span of 3300 mm.
Page 345
314
The width of the beam web was 250 mm. Except for the rectangular beam
CB-Rec, all T-beams had two 600-mm-wide and 100-mm-deep flanges on each
side of the beam, leading to a total slab width of 1450 mm. Web openings were
located 250 mm from specimen centres, to simulate the real situation of web
openings adjacent to columns of 500-mm width along the main beam direction.
Web openings were placed adjacent to the flange. The control specimens (i.e.
CB-Rec and CB-T) failed due to concrete compression crushing. For specimens
weakened by openings but without FRP strengthening (i.e.O-700-300-N,
O-800-280-N, O-600-220-N and O-700-200-N), the two have larger web
opening ( specimens O-700-300-N, O-800-280-N) failed by local flexural or
mixed flexural an shear failure at the ends of web and flange chords (i.e. the
chords in the bound of beam web and flange). Specimens O-600-220-N and
O-700-200-N failed due to diagonal cracks developing in span of web chord.
Those with additional FRP strengthening, failed by local flexural rotation at the
two ends of the web chord and flange chord.
5.3.1.2 Material properties
Commercial concrete was used and its strength was obtained by averaging
the strengths of three concrete cylinders. The concrete strength of each
specimen is listed in Table 5.1. Longitudinally placed deformed steel bars of
20-mm diameter (D20 bars) were used for beam tension reinforcements. Four
D20 bars were placed at the top side (i.e. the side with slab flanges) and three
D20 bars at the bottom side. Plain steel bars of 8-mm diameter (D8 bars) were
Page 346
315
used as steel stirrups and slab reinforcements. The spacings for steel stirrups and
slab longitudinal bars were 100 mm and 200 mm respectively. There were two
layers of D8 longitudinal slab reinforcing bars in the slab. For the first test
group, the yield load and ultimate stresses of the D8 steel bars were 307 MPa
and 447 MPa, respectively; and 475 MPa and 625 MPa, respectively for the
D20 steel bars. The yield and ultimate stresses of D8 steel bars were 349 MPa
and 526 MPa respectively for the second test group and those for the D20 bars
were 434 MPa and 559 MPa respectively. The properties of type 1 bar are
tabulated in Table 5.3 based on coupon tests. Only the D8 bar of the first group
is type 2 bar. To better describe the stress-strain curves obtained from tests, the
average stress 0.2% obtained at a yield offset of 0.2%, is applied to 0s and
n1 is set equal to 9.0, which is determined by the obtained peak stress and the
corresponding strain of the D8 steel bars in the first test group. The elastic
modulus for all steel bars was assumed to be 200 GPa. Except for D8 bar of the
first test group, which is a type 2 bar, the material properties of the type 1 bars
are tabulated in Table 5.3. CFRP, which was used in the U-shaped, or wrapping,
form, was of 0.334mm fabric design thickness, 2820 MPa tensile strength and
227 GPa elastic modulus. The bond stress and corresponding slip of Lu etal.'s
(2005) model for specimens with FRP strengthening were tabulated in Table
5.4.
5.3.2 Load versus Displacement Curves
Page 347
316
Figure 5.6 shows the predicted curves for all specimens. The predicted
curves for the control specimens match well with the test results (Figure 5.6(a)).
For those specimens with openings with and without FRP strengthening, the FE
model predicts peak load (load capacity) quite accurately, but higher
post-cracking stiffness (Figures 5.6(b), 5.6(c), 5.6(e) and 5.6(f)) for specimens
without FRP strengthening. This is because shear damage is underestimated by
the proposed 3D models. This is confirmed by the fact that for those models
provided with additional FRP strengthening, there were better predictions of
post-cracking stiffness. Except for the two specimens loaded in positive bending
as indicated in Figure 5.6(d). Almost all cracking loads were slightly
overestimated by the 3D FE Models. For the post-yielding range (i.e. the range
after a beam yields) load-displacement response, the behaviour of the specimens
with web openings but without FRP strengthening was not well predicted by the
3D mode: the predicted load-displacement curve in the post-yielding range
show higher stiffnesses than those in the test curves. This might also because
shear damage is underestimated by the proposed 3D models. In Figure 5.6(c)
and (e), the FE curves for models F-800-200-N and F-600-220-N stop earlier
than the test results. This is because local debonding at the FRP-concrete
interface at the opening’s corner near the loading point leads to great local
vibration and a sudden drop in the predicted curves.
The test and predicted load capacities are tabulated in Table 5.5. For the
CB-R specimen who had a obvious yielding load, its yield load is tabulated as
Page 348
317
specimen predicted load. For specimens with web openings but without FRP
strengthening, the peak loads are tabulated. For models with both web openings
and FRP strengthening, the loads at 33 mm displacement, 2% of beam effective
length, are tabulated. It can be seen from Table 5.4 that the predicted load
capacities are much closer to the test results except for the two specimens
loaded in positive bending. The average of the ratios (including the two
specimens subject to positive bending) of predicted load to test result is 1.03,
with a standard deviation of 0.09.
5.3.3 Crack patterns
The predicted crack patterns are shown in Figures 5.7(a)-5.7(n). In these
Figures, the FRP had been removed to better demonstrate the cracks
propagating in and near the web chords. Except for the beam CB-Rec, the main
test cracks are indicated by red dashed lines, as shown in Figure 5.7(b)-(n).
Figure 5.7(a), predicts many cracks concentrated in the middle of the
specimen at the ultimate state for the R-beam, which is similar to those of the
test result shown in Nie (2018). As for the CB-T beam, the 3D FE model
predicts good crack pattern predictions in terms of the number and distribution
of the major cracks in the beam web and slab flange. For the specimens with
web openings, the major cracks are also well predicted by the 3D FE models.
For those specimens with the larger web openings (e.g. O-700-300-N), the
predicted major cracks are located near the ends of the chords and become
Page 349
318
wider as the four hinges mechanism (i.e. local flexural rotation at the ends of
web and flange chords) becomes more obvious, which is close to test
observation. In comparison, for the specimens with smaller web openings (e.g.
O-600-220-N), additional flexural cracks and shear cracks are predicted to
appear near mid-span of the beams and on the web chord respectively,
consistent with test results. When FRP is provided to control shear crack
propagation (e.g. F-O-600-220-N), more flexural cracks are predicted near
specimen mid-spans, which is consistent with test results. For those specimens
tested in positive bending, many flexural cracks are predicted in the beam webs,
in addition to those near the ends of chords. This is also similar to the test
results.
5.3.4 Energy Release Behaviour
Energy analyses predicted by a 3D FE model can be obtained and used to
better understand and evaluate the reliability of predicted results. As presented
ABAQUS (2012) manual, the total energy totalE is given as:
total I V FD KE WE E E E E E (5.18)
where IE is internal energy, VE is energy dissipated by viscous effects, FDE
is total energy dissipated through frictional effects, KEE is kinetic energy and
WE is external work applied to the structural system. For a structure in a stable
Page 350
319
state, WE is equal to IE . Therefore, when making a quasi-static analysis for
which the static structural response dominates the structural response, the
difference between WE and IE ought to be very small. VE is relatively
small compared to IE . That is because viscous forces aim to damp out local
dynamic effects. A viscous force which is too large leads to obviously higher
predicted load capacity of the structure and an unreliable result. FDE is equal
to 0 in this study as no friction force is applied in the proposed FE model. For a
quasi-static analysis, the dynamic energy of any particular analysis also
accounts for a small portion of the internal energy (normally below 1%).
Figure 5.8 shows the curves of the ratios of external work, viscous energy,
kinetic energy to the internal energy for specimen CB-T. As indicated in Figure
5.8, the external work to internal energy ratio is nearly equal to 100% except at
the very beginning. The energy due to viscous effects is negligible (0.01%) and
decreasing at first but suddenly increases to a value of about 10% upon the
specimen cracking, due to kinetic energy increase; then followed by a gradually
decrease to the value of below 1%. The kinetic energy to internal energy ratio is
relatively small (1%) at first, gradually decreasing as displacement increases,
except that, at about a 3-mm displacement, specimen cracking causes a sudden
dynamic energy increase (to a value of about 2%), then it quickly decreases to a
value below 0.1% , as indicated in Figure 5.8. The above results imply that the
predicted structural responses are essentially static except for a few discrete
Page 351
320
moments, such as at the beginning of loading and at the appearance of
significant cracks.
5.4 PARAMETRIC STUDIES ON THE EFFECTS OF KEY
FACTORS
This section concerns the choice of the values for the key factors, including
element size, loading duration, damping coefficient β and computation precision
setting as they relate to the accuracy of the 3D FE analysis. In addition, as the
web chords of some T-beams with web openings were strengthened by FRP
wrapping, the confinement effect provided by FRP wrapping was also explored
ensure the validity of the 3D FE analysis. It should be noted that in the
following parametric studies, when the value of one parameter is changed, those
of the rest parameters were kept unchanged and equal to those in the reference
case specified in Subsection 5.2.6.
5.4.1 Element Size
Two control specimens, and specimens O-700-300-N, F-700-300-N were
used for a convergence study of element size. In Figure 5.9, the label ELE4050
means the element sizes are either 40 mm or 50 mm, which are dependent on
the dimensions of specimens, spaces between stirrups and bars, etc. ELE20 and
ELE10, indicate that all element sizes are 20 mm or 10 mm respectively. As
shown in Figure 5.9, the element size of 20 mm is small enough for
reliable/accurate results to be obtained as the numerical results obtained using
Page 352
321
the element size of 20 mm are very close to those using element size of 10 mm.
The predicted load-displacement curve for F-700-300-N-ELE10 stops at about
30 mm as from then on, FRP-concrete local interface debonding causes a
sudden drop of load, bringing about a large kinetic energy increase. However,
the obtained numerical results are enough to indicate that a 20 mm element size
is small enough for the proposed 3-D FE models.
5.4.2 Loading Duration
As explicit centre difference method (CDM) is used as time integration
method to solve the 3D FE models, the loading duration should be long enough
to minimize the dynamic effect and inertia forces effects. However, longer
loading durations lead to longer computational times and larger accumulation
errors (Chen et al. 2015). Thus, the loading duration effect was investigated to
determine a suitable value of loading time with both accuracy and
computational time kept in mind.
The FE models for CB-T, O-700-300-N and F-700-300-N mentioned above
were used to study the loading duration effect. The natural period T0 of a 3D
model was first determined by ABAQUS (2012). Then loading durations
ranging from 12.5T0 to 200T0 were applied for specimens CB-T and
O-700-300-N first. Loading durations ranging from 25T0 to 100T0 were applied
to specimen F-700-300-N for further confirmation. As shown in Figure 5.10 (a),
the predicted loading-displacement curve for specimen CB-T with a 50 T0
Page 353
322
loading duration is close to those with longer loading durations. Figure 5.10 (b)
indicates that shorter loading durations lead to higher cracking and peak loads.
For specimen O-700-300-N, the 50 T0 loading duration leads to a slightly higher
peak load than for longer loading durations, but the difference is less than 5%,
which is acceptable. Based on the above discussions, a loading duration of 50 T0
was chosen in this study.
Fig. 5.11 plotted the dynamic energy versus displacement curves on
logarithmic scales. It is obvious that the spacings between two neighbouring
dynamic energy-displacement curves are almost the same. That is because the
loading duration was double. At initial and final states, the dynamic energy is
close to zero, because the loading speed of the smooth loading scheme is zero at
the beginning and the end of loading duration. It can be found from numerical
results not presented here that cracking of concrete and local debonding of the
FRP-concrete interface lead to some local fluctuations of the dynamic
energy-displacement curves. It is of interest to note that the each of the main
local fluctuations of load-displacement curves shown in Figure 5.10 occurs at
the same displacement as that in the dynamic energy-displacement curves,
suggesting that the local fluctuations in the load-displacement response are
associated with the local dynamic effects, which will be damped out by the
damping system of the model.
5.4.3 Damping Coefficient β
Page 354
323
Rayleigh damping of ABAQUS (2012) was adopted to define material
viscous damping. The viscous damping matrix C is expressed as:
C M K (5.19)
where M is the mass matrix and K is the stiffness matrix, and are
mass-proportional and stiffness-proportional damping coefficients respectively.
Thus the damping ratio for the jth model of system can be expressed as :
2 2j
jj
(5.20)
where j is the circular frequency corresponding to the jth model. The
relationship between damping ratio and circular frequency is illustrated in
Figure 5.12. If parameter is equal to zero, the damping ratio is proportional
to the stiffness matrix. Then damping ratio is a linearly increasing function of
circular frequency. Under this circumstance, a higher vibration mode is assigned
with higher viscous damping. If parameter is equal to zero, viscous
damping is proportional to mass, which leads to higher viscous damping at
lower modes of vibration. When both parameters are larger than zero, the
damping ratio will not be lowest at the beginning. In Figures 5.13 to 5.15, the
first five modes of vibration of the models CB-Rec, CB-T and O-700-300-N are
presented. The vibration amplitudes are 100-times enlarged to better present the
results. The corresponding frequencies are tabulated in Table 5.4. As indicated
Page 355
324
in Figures 5.13 to 5.15, the first mode of vibration is the expected deformation
shape. Therefore, to damp out the dynamic effect of higher vibration modes and
minimize dynamic effect associated with the lower mode (e.g. first mode), only
is needed.
To determine the value of to be used in the proposed FE models, CB-T,
O-700-300-N and F-700-300-N were used for a parameter study. The values of
are from zero to 5e-5 and the predicted load-displacement curves are shown
in Figure 5.16. As indicated in Figure 5.16, when is equal to zero or below
5e-7, the predicted curves drop before yield load. When is higher than 1e-5,
the FE models predict higher post-cracking stiffness and load capacity due to
the effect of viscous force associated with damping (Chen et al. 2015).
Furthermore, when explicit CDM is used, the stable time increment largely
decreases due to the increased (ABAQUS 2012), which leads to greatly
increased computational times. Therefore, the with a value between 5e-7
and 1e-5 should be an reasonable value for the proposed FE model. This
conclusion is further confirmed when checking predicted kinetic energy as
shown in Figure 5.17. It is obvious that when is below 5e-7, the predicted
kinetic energy is much higher. When is above 1e-5 (e.g. 5e-5), the predicted
kinetic energy is also higher than for those with between 5e-7 and 1e-5. To
further narrow the range of , the FE models of CB-T, O-700-300-N and
F-700-300-N with equal to 1e-6, 2e-6 and 4e-6 were studied and the results
Page 356
325
are shown in Figure 5.18; for all three values of , similar load-displacement
and kinetic energy-displacement curves were predicted except that for CB-T,
unexpected kinetic energy appears at a displacement around 28 mm when is
equal to 1e-6. As a result, = 2e-6 was chosen for all the specimens for the
proposed 3D FE models.
To find the reason for the early failure of FE models with very small (e.g.
below 5e-7), the predicted crack patterns for model O-700-300-N, with =0,
was studied as shown in Figure 5.19. The positions of a, b, c and d are
illustrated in Figure 5.16(b). Points a and b occur immediately before and after a
local peak load of 81.9 kN; and c and d are the lowest (-17.1kN) and highest
(107.9 kN) points respectively after points a and b. As shown in Figure 5.19 (a),
some cracks emerge at the corners of the opening at moment a. At the
displacement of moment b, corner cracks keep growing and a new flexural
crack propagates on the slab. As the displacement increases to moment c,
numerical local cracks are predicted as indicated by the large number cracked
elements and at d, cracks elements cover the whole specimen. These
strange/doubtful phenomena were actually not observed in tests (Nie 2018),
which can be explained as follows. When is artificially/ideally set to zero,
the local vibrations associated with the initiation and propagation of critical
cracks, which are high vibration mode, can't be efficiently damped out, leading
to the quick and unreal appearance of a large number of cracks, first next to the
critical local crack and then propagate to the whole beams; as a result, global
Page 357
326
dynamic effect appears and the kinetic energy becomes doubtfully large (Figure
5.17(b)). As a result, the related numerical results should be unreliable.
To summarize, in the proposed 3-D FE model, the stiffness-proportional
Rayleigh damping (characterized by stiffness-proportional damping coefficient
) is favoured and should be small enough to avoid the unnecessary viscous
force associated with damping, which may cause the over-shooting of the
predicted load. On the other hand, the damping cannot be such small that the
local dynamic effects (local vibration) associated with the cracking of concrete
cannot be efficiently damped out.
5.4.4 Single versus Double Precision
Even though simulation with single precision requires less computational
time, simulation with double precision is suggested for FE simulation for
improved accuracy, especially when the number of increments is very large,
which is the case of explicit CDM. To investigate the computation precision
effect on the 3D FE modelling, the FE models for CB-Rec, CB-T and
O-700-300-N were used. The predicted load-displacement curves are shown in
Figure 5.20. As indicated in Figure 5.20, the FE models calculated using single
precision predict similar load-displacement curves to those calculated with
double precision, except that the model CB-Rec calculated with single precision
fails earlier than that with double precision. When checking the kinetic energy
of these FE models, it was found that single precision calculation predict a
Page 358
327
slightly higher kinetic energy than those calculated using double precision,
which is more obvious at the beginning and the end of loading, as shown in
Figure 5.21. Figure 5.22 shows the predicted crack patterns for the above three
specimens based on single precision calculation. It is obvious that at the
yielding displacement, the FE models simulated with single precision predict
crack patterns well. As the displacement increases, numerous numerical local
cracks appear, making the crack pattern deviated from that crack pattern
observed in test. Similar phenomena exist with the other two specimens with
less difference between the test observed crack pattern and numerical crack
patterns. The inaccuracy (or errors) in predicting the crack patterns can be
attributed to the error accumulation nature of to the explicit integration method.
The issues can be well solved when the calculations were carried out with
double precision mode. Thus, to obtain reliable 3D FE predictions, calculations
with double precision were adopted for the FE simulations in the present study
if not otherwise specified.
5.4.5 FRP Confinement Effect
To avoid non-ductile shear failure caused by the removal of beam concrete
with opening, FRP is applied to strengthen the beam in the vicinity of the
opening where potential shear failure might occur. The existence of FRP can
both increase the shear capacity of a weakened region and at the same time,
improve the compressive strength of the concrete, especially the web chord
confined with FRP wraps where FRP confinement effect should not be ignored.
Page 359
328
Yu et al.'s (2009a, b) modified Plastic-damage model for a concrete column
with FRP confinement has been proved to be very accurate for modelling the
mechanical properties of FRP-confined concrete under compression. Therefore,
that model was adopted for modelling the concrete of the web chord
strengthened by FRP wrapping, as shown in Figure 5.23.
Figure 5.24 shows the typical numerical results with and without
considering FRP confinement. It is obvious that the inclusion of the
confinement effect on the web chord leads to more accurate predicted load
carrying capacity and post-peak load-displacement response. When the opening
is smaller and the web chord bigger (e.g. 700 × 200 mm2 opening), the
difference between the two FE models with and without the confinement effect,
is bigger, which is understandable because the member section affected by the
FRP is larger for the chord with larger cross section.
5.5 COMPARISON BETWEEN 2D AND 3D FE MODELS
2D FE Modelling using the dynamic method has been proposed by Chen et
al. (2011, 2015) and used by several researchers (Fu 2016, Nie 2018) and the
predicted results match well with test results for RC rectangular beams.
However, when the slab flange is wide, 2D models may do not perform well as
they do on rectangular beams because the uniform distribution of stress/strain
along the width direction cannot be satisfied in this circumstance.
The differences in modelling performance between 2D and 3D FE models
Page 360
329
for T-beams will be examined in this section. As the ELA tension damage model
is used by the author, for comparison convenience, the 2-D modelling results
predicted by the damage plasticity (DP) approach with ELA tension damage
model are used for comparison purpose, even though the results predicted by DP
approach with ELA damage model is not better than those with power law (PL)
tension damage, as stated by Nie (2018). The 2-D FE models were meshed with
10-mm elements. CPS4R elements and T2D2 elements were used to represent
concrete and steel reinforcement respectively. 2D cohesive elements were
applied to model the bond-slip behaviour of the FRP-concrete and the
steel-concrete interfaces.
Figure 5.25 shows the results predicted by the 2D and 3D FE models for
the specimens CB-Rec, CB-T, O-700-300-N, O-700-200-N, F-700-300-N and
F-700-200-N. The 2D CB-T FE model overestimates the post-cracking stiffness
as demonstrated in Figure 5.25(a), though it predicts a yield load which is as
close to the test result as the 3D model. This is because the non-uniform flange
deformation of the T-beam in the direction perpendicular to the longitudinal axis
of the beam cannot be captured by a 2D FE model. Moreover, the 3D FE model
performs better than the 2D in predicting the CB-T beam post-yielding response.
The 2D FE models also behave worse than the 3D FE model when modelling a
T-beam with web opening weakening but no FRP strengthening against shear.
As shown in Figure 5.25(b), when the opening size is large (e.g. 700 × 300
mm2), the 2D FE model predicts a appreciably higher post-cracking stiffness
Page 361
330
and lower peak load than the test results, whereas the predictions of the 3D FE
model are much closer to the result in terms of both post-cracking stiffness and
peak load. When the size of the opening is smaller (e.g. 700 × 200 mm2), the 2D
FE model predicts a much lower peak load than the test result, whereas the
prediction of the 3D FE model match well with the test result. When modelling
T-beams with web opening weakening and FRP strengthening, the 2D FE
models predict peak load higher than both test results and the 3D FE model
predictions, especially for the specimen with smaller opening (e.g. 700 × 200
mm2). Post-cracking stiffness are also overestimated by 2D FE models. When
the opening is smaller (e.g. 700×200 mm2), the 2D predicts a hardening
load-displacement curve which is deviated from both test result.
The predicted crack patterns of the specimens CB-R, O-700-200-N,
F-700-200-N of 2D and 3D models are compared in Figure 5.26, which shows
that the 2D and 3D models predicted similar crack patterns, both in the beam
web and slab, though the 2D models were unable to clearly show the cracks
propagating on the slab.
5.6 SHEAR DEGRADATION OF CRACKED CONCRETE
As mentioned earlier, the proposed 3D FE models predicted poorly the
post-yielding load-displacement curves for T-beams with a web opening but
without FRP strengthening, and a possible reason is that the DP model
underestimated the effect of shear resistance degradation of cracked concrete.
Page 362
331
To investigate the effect of shear degradation, a new model using a combination
of the DP model and the BC model (namely the brittle crack model (ABAQUS
2012)) to represent concrete (i.e. a DP+BC model) was adopted, following Chen
et al. (2012). The model is illustrated in Figure 5.27. The concrete above the
opening was modelled with the DP model to capture the compression damage
associated with concrete compression failure and the remaining concrete was
modelled with the BC model considering that this concrete was mainly
subjected to tensile damage associated with shear tension failure. It should be
noted that the BC model assumes the concrete to be elastic under compression.
When using the BC model, the postcracking shear resistance is defined using a
power law as indicated in Equations 5.21 and 5.22:
( )ckc nnG e G (5.21)
( ) (1 )ck
ck nnnnn
max
eee
(5.22)
where G is the shear modulus of uncracked concrete; cG is the postcracking
shear modulus; cknne is the cracking strain; and maxe is the maximum cracking
strain corresponding to the complete loss of aggregate interlock.
As shown in Figure 5.28, as the coefficient n increases, the shear modulus
G decreases at a faster speed with a larger cracking strain. To define the failure
state, a number of critical cracks with the minimum and maximum values to be
Page 363
332
1 and 3, respectively, is also specified. Therefore, three parameters, namely, the
number of critical cracks, the maximum cracking strain and the coefficient n,
may affect the concrete shear degradation behaviour in the BC model. The
specimen O-700-300-N was simulated with the proposed model to examine the
effects of the three parameters. The following reference parameters were
adopted in the simulation if not otherwise specified: maxe =0.0075, n=5 and the
number of critical cracks = 2.
5.6.1 The Effect of Number of Critical Cracks
As indicated in Figure 5.29, a larger number of critical cracks leads to a
more gradual descending post-peak load-displacement response and a higher
strength. The predicted strength increases appreciably when the number of
critical cracks increases from 1 to 2. When the number of critical cracks is equal
to 1, the DP+BC model predicts a critical crack happening at a displacement of
around 24 mm, leading to a sudden increase in the predicted kinetic energy.
5.6.2 The Effect of Maximum Cracking Strain
Maximum cracking strains equal to 0.005, 0.006 and 0.0075 were adopted
in the simulation for comparison. As indicated in Figure 5.30, a lower maximum
cracking strain leads to a lower predicted strength and softer post-peak response.
This is because a lower cracking strain results in a faster shear modulus
degradation (Figure 5.28).
Page 364
333
5.6.3 The Effect of Coefficient of the Power Law
As shown in Figure 5.28, a larger n value results in a faster degradation
speed of the shear resistance. Figure 5.31 shows that when n increases from 1 to
4, the predicted strength becomes lower and the post-peak load-displacement
response becomes softer. When n is larger than 4, the effect of n is marginal,
except that when n = 7, the DP+BC model predicts a premature failure.
5.6.4 Comparison between the DP and DP+BC Models
In this section, the results predicted with the DP+BC model are compared
with those predicted with the proposed DP model presented in section 5.3. The
following values were used for the three parameters in the DP+BC model:
number of critical cracks = 2; maximum cracking strain = 0.006; and n = 5 for
the BC part, since these values provide the best predicted load-displacement
curve compared with the test curve as shown in Figure 5.29. As indicated in
Figure 5.32, the two models predict similar load-displacement curves, with the
DP+BC model predicting a post-peak load-displacement curve slightly closer to
the experimental one. The slightly better performance of the DP+BC model is
due to the fact that the DP+BC model can better consider the effect of shear
degradation of cracked concrete. However, the DP+BC model predicts larger
kinetic energy when the applied displacement is larger than about 12 mm, which
indicates that the DP+BC model may predict more cracks than the DP model.
Page 365
334
The predicted crack patterns by the two FE models are compared in Figure
5.33. It can be seen that the predicted cracks propagating in the web chord are
similar although the DP+BC model predicts slightly more flexural cracks at the
mid-span of the specimen.
The above comparison results show that the DP model can provide
predictions comparable to the DP+BC model for RC T-beams with a web
opening but without FRP strengthening, but it is easier to create an FE model
using the former model; as a result, the DP model will be used in the remainder
of this PhD research if not otherwise specified.
5.7 CONCLUDING SUMMARY
Three-dimensional (3D) finite element (FE) modelling T-beams with web
opening was conducted using ABAQUS (2012) in this chapter. To avoid
convergence problems, a dynamic explicit approach which adopts the central
difference integration method was used. In this study, a 3D prism cohesive
model was proposed for the steel-concrete interface. The effect of the element
size, loading duration, damping ratio, and calculation precision which affect
dynamic explicit approach prediction accuracy, were studied. The confinement
effect provided by the web chord FRP was also investigated. Finally, the
predicted results of the 3D models were compared with the 2D results predicted
by the damage plasticity (DP) approach with elastic (ELA) tension damage
model, which was proposed by Nie (2018). The following conclusions can be
Page 366
335
drawn based on the achieved modelling results:
(1) The proposed 3D FE models simulated well, the behaviour of the specimens
tested by Nie (2018) except for the two under positive loading. The
yield/peak loads and test crack patterns of all specimens were accurately
predicted, though post-yielding curves of T-beams with web opening were
not well captured. A possible reason is shear damage is underestimated by
the proposed models.
(2) The proposed 3D prism cohesive model is verified as reliable for 3D FE
analysis purposes when used for the steel-concrete interface and it was
successfully applied in the proposed 3D FE models for T-beams.
(3) For a quasi-static analysis, the dynamic energy of any particular analysis
should account for a small portion of the internal energy (normally below
1%), except for a few discrete moments, such as at the beginning of loading
and at the appearance of significant cracks.
(4) Element size, loading duration and stiffness damping coefficient were
determined based on convergence studies. A 20-mm element size, 50T0 (T0
is the FE model natural period) and equal to 2e-6 were found
appropriate for the proposed 3D models. Only the value of is needed to
define Rayleigh damping, since the first mode of vibration is the expected
deformation shape.
Page 367
336
(5) Modelling with double precision computation is suggested for 3D FE
simulation. Single precision modelling may cost less in computational time,
but more errors accumulate as the number of increments increases, which
was obviously indicated by the predicted crack patterns.
(6) Yu et al.'s (2010a, b) modified plastic-damage model for a concrete column
confined with FRP was applied for the concrete in web chords with FRP
wrapping. When the confinement effect of the FRP wrapping is simulated,
more accurate results are obtained, especially when modelling a T-beam
with a smaller opening and a bigger web chord.
(7) The proposed 3D FE models are better predictors of reality than the 2D FE
models proposed by Nie (2018), in the modelling of T-beams, especially
those weakened by web openings and with no FRP shear strengthening. The
2D FE models overestimated the post-cracking stiffness of T-beams, which
was better predicted by the proposed 3D FE models.
(8) A DP+BC model, which is capable of modeling the shear degradation effect
of cracked concrete, was examined and the predictions were compared with
those from the DP model. The effects of three parameters, namely, the
number of critical cracks, the maximum cracking strain and the coefficient
of power law, of the BC model, were investigated through parametric
studies. The results indicated that the DP+BC model predicted the
load-displacement curves and crack patterns similar to those predicted with
Page 368
337
the DP model; the DP+BC model predicted slightly better the post-peak
load-displacement response. The DP model is thus recommended for
modelling RC T-beams with a web opening but without FRP strengthening
since it is more convenient to create an FE model using the DP model.
5.8 REFERENCES
ABAQUS (2012). ABAQUS Analysis User's Manual (Version 6.12), Dassault
Systems SIMULIA Corporation, Providence, Rhode Island, USA.
Abdalla, H.A., Torkey, A.M., Haggag, H.A. and Abu-Amira, A.F. (2003).
“Design against cracking at openings in reinforced concrete beams
strengthened with composite sheets”, Composite Structures, 60(2), 197-204.
ACI-318 (2014). Building Code Requirements for Structural Concrete and
Commentary (ACI 318-14), ACI Committee 318, American Concrete
Institute, Farmington Hills, MI.
ATC-40 (1996). Seismic Evaluation and Retrofit of Concrete Buildings, Applied
Technology Council, Redwood City, California.
Bazant, Z. P., and Becq-Giraudon, E. (2002). “Statistical prediction of fracture
parameters of concrete and implications for choice of testing standard”, Cem.
Concr. Res., 32(4), 529–556.
Canbolat, B.B. and Wight, J.K. (2008). “Experimental investigation on seismic
Page 369
338
behavior of eccentric reinforced concrete beam-column-slab connections”,
ACI Structural Journal, 105(2), 154-162.
CEB-FIP. (1993). “CEB-FIP Model Code 90”, Thomas Telford, London.
Chen, G.M. (2010). Shear Behaviour and Strength of RC Beams
Shear-Strengthened with Externally Bonded FRP Reinforcement, PhD
Thesis, Department of Civil and Structural Engineering, The Hong Kong
Polytechnic University, Hong Kong. China.
Chen, G.M., Teng, J.G. and Chen, J.F. (2011). “Finite element modeling of
intermediate crack debonding in FRP-plated RC beams”, Journal of
Composites for Construction, ASCE, 15(3), 399-353.
Chen, G.M., Teng, J.G., Chen, J.F. and Xiao, Q.G. (2015). “Finite element
modeling of debonding failures in FRP-strengthened RC beams: A
dynamic approach”, Computers and Structures, 158, 167–183.
Chin, S.C., Shafiq, N. and Nuruddin, M.F. (2012). “Strengthening of RC Beams
with Large Openings in Shear by CFRP Laminates: Experiment and 2D
Nonlinear Finite Element Analysis”, Research Journal of Applied Sciences
Engineering and Technology, 4(9), 1172-1180.
Chinese Academy of Building Research (2008). Photo Collection of 2008
Wenchuan Earthquake Damage to Buildings, China Architectural &
Page 370
339
Building Press, Beijing, China (in Chinese).
Durrani, A.J. and Wight, J.K. (1987). “Earthquake resistance of reinforced
concrete interior connections including a floor slab”, ACI Journal, 84(5),
400-406.
Ehsani, M.R. and Wight, J.K. (1985). “Effect of transverse beams and slab on
behavior of reinforced concrete beam to column connections”, ACI Journal,
82(2), 188-195.
El-Maaddawy, T. and El-Ariss, B. (2012). “Behavior of Concrete Beams with
Short Shear Span and Web Opening Strengthened in Shear with CFRP
Composites”, Journal of Composites for Construction, Vol. 15(1), 47-59.
Eurocode 8 (2004). Design of Structures for Earthquake Resistance – Part 1:
General Rules, Seismic Actions and Rules for Buildings (EN 1998-1: 2004),
CEN, Brussels.
Fu, B. (2016). Debonding failure in FRP-strengthened RC beams: prediction
and suppression, Doctoral degree thesis: The Hong Kong Polytechnic
University.
GB/T 228.1 (2010). Tensile test of metallic material – part 1: test method at
ambient temperature. China Standard Press, Beijing, China (in Chinese).
GB-50011 (2008). Code for Seismic Design of Buildings, Architectural &
Page 371
340
Building Press, Beijing, China (in Chinese).
GB-50011 (2010). Code for Seismic Design of Buildings, Architectural &
Building Press, Beijing, China (in Chinese).
Guimaraes, G.N., Kreger, M.E. and Jirsa, J.O. (1992). “Evaluation of joint-shear
provisions for interior beam-column-slab connections using high-strength
materials”, ACI Structural Journal, 89(1), 89-98.
Hasnat, A. and Akhtaruzzaman, A.A. (1987). “Beams with small rectangular
opening under torsion, bending, and shear”, Journal of Structural
Engineering, ASCE, 113(10), 2253-2270.
Hawileh, R., El-Maaddawy, T., and Naser, M. (2012). "Nonlinear finite element
modeling of concrete deep beams with openings strengthened with
externally-bonded composites." Materials & Design,
10.1016/j.matdes.2012.06.004, 378-387.
Hordijk, D. A. (1991). “Local approach to fatigue of concrete.” Ph.D. thesis,
Delft Univ. of Technology.
BS EN ISO 6892-1 (2009). Metallic materials – tensile testing – Part 1: Method
of test at room temperature
Jiang, L. X., Zheng, S. J., Zhang, W. P. and Gu, X. L. (2009). "Cyclic Loading
Behavior of Weak Reinforced Concrete Beam-Column Connections", Key
Page 372
341
Engineering Materials, 400, 881-886.
Kennedy, J.B. and Abdalla, H. (1992). “Static response of prestressed girders
with openings”, Journal of Structural Engineering, ASCE, 118(2), 488-504.
LaFave, J.M. and Wight, J.K. (1999). “Reinforced concrete exterior wide
beam-column-slab connections subjected to lateral earthquake loading”, ACI
Structural Journal, 96(4), 577-587.
Lam, L., and Teng, J. G. (2003a). “Design-oriented stress–strain model for
FRP-confined concrete”, Construction and building materials, 17(6),
471-489.
Lam, L., and Teng, J. G. (2003b). “Design-oriented stress-strain model for
FRP-confined concrete in rectangular columns”. Journal of Reinforced
Plastics and Composites, 22(13), 1149-1186.
Lu, X.Z., J.G. Teng, Ye, L.P. and Jiang, J.J. (2005). “Bond-slip models for FRP
sheets /plates bonded to concrete”, Engineering Structures, 27(6), 920-937.
Maaddawy, T. and Sherif, S. (2009). “FRP composite for shear strengthening of
reinforced concrete deep beams with openings”, Composite Structures, 89(1),
60-69.
Maaddawy, T. and El-Ariss, B. (2012). ”Behavior of Concrete Beams with
Short Shear Span and Web Opening Strengthened in Shear with CFRP
Page 373
342
Composites.” J. Compos. Constr., 16(1), 47–59.
Madkour, H. (2009). “Non-linear analysis of strengthened RC beams with web
openings”, Proceedings of the Institution of Civil Engineers-Structures and
Buildings, 162(2), 115-128.
Mansur, M.A. (1998). “Effect of opening on the behaviour and strength of R/C
beams in shear”, Cement and Concrete Composites, 20(6), 477-486.
Mansur, M.A., Tan, K.H., and Lee, S.L. (1985). “Design method for reinforced
concrete beams with large openings”, ACI Journal, 82(4), 517-524.
Mansur, M.A., Tan K.H. and Wei, W. (1999). “Effects of creating an opening in
existing beams”, ACI Structural Journal, 96(6), 899-906.
Nasser, K.W., Acavalos, A. and Daniel, H.R. (1967). “Behavior and design of
large openings in reinforced concrete beams”, ACI Journal, 64(1), 25-33.
Nie, X.F. (2018). Behavior and Modeling OF RC Beams with an
FRP-Strengthened Web Opening, Doctoral degree thesis: The Hong Kong
Polytechnic University.
Pantazopoulou, S. J. and Moehle J. P. (1990). “Identification of effect of slabs
on flexural behavior of beams.” Journal of Engineering Mechanics, ASCE,
116(1), 91-106.
Page 374
343
Pimanmas, A. (2010). “Strengthening R/C beams with opening by externally
installed FRP rods: Behavior and analysis”, Composite Structures, 92(8),
1957-1976.
Pool, R.B. and Lopes, R. (1986). “Cyclically loaded concrete beams with web
openings”, ACI Journal, 83(5), 757-763.
Ramberg, W., and Osgood, W. R. (1943). “ Description of stress-strain curves
by three parameters”.
Rots, J. G. (1988). “Computational modeling of concrete fracture”, Ph.D. thesis,
Delft Univ. of Technology.
Saenz, L. P. (1964). “Discussion of ‘Equation for the stress-strain curve of
concrete’ by P. Desayi and S. Krishan.” J. Am. Concr. Inst., 61(9),
1229–1235.
Shin, M. and LaFave, J.M. (2004a). “Seismic performance of reinforced
concrete eccentric beam-column connections with floor slabs”, ACI
Structural Journal, 101(3), 403-412.
Shin, M. and LaFave, J.M. (2004b). “Reinforced concrete edge
beam-column-slab connections subjected to earthquake loading”, Magazine
of Concrete Research, Vol. 55, No. 6, pp. 273-291.
Siao, W.B. (1994). “Reinforced concrete column strength at beam/slab and
Page 375
344
column intersection”, ACI Structural Journal, 91(1), 3-8.
Tamer El-Maaddawy and Bilal El-Ariss. (2012). “Behavior of Concrete Beams
with Short Shear Span and Web Opening Strengthened in Shear with CFRP
Composites”, Journal of Composites for Construction, Vol. 15(1), 47-59.
Tan, K.H., Mansur, M.A. and Huang, L.M. (1996a). “Reinforced concrete
T-beams with large web openings in positive and negative moment regions”,
ACI Structural Journal, 93(3), 277-289.
Tan, K.H., Mansur, M.A. and Huang, L.M. (1996b). “Design procedure for
reinforced concrete beams with large web openings”, ACI Structural Journal,
93(4), 1-8.
Tan, K.H., Mansur, M.A., and Wei, W. (2001). “Design of reinforced concrete
beams with circular openings”, ACI Structural Journal, 98(3), 407-415.
Teng, J.G., Chen, J.F., Smith S.T. and Lam L. (2002). FRP-Strengthened RC
Structures, John Wiley and Sons Ltd, UK, November, 245pp.
Thermou, G.E., Pantazopoulou, S.J., Einashai, A.S. (2007). “Flexural behavior
of brittle RC members rehabilitated with concrete jacketing”, Journal of
Structural Engineering, ASCE, 133(10), 1373-1384.
Wei, Y. Y., and Wu, Y. F. (2012). “Unified stress–strain model of concrete for
FRP-confined columns”, Construction and Building Materials, 26(1),
Page 376
345
381-392.
Xiao, Y. and Wu, H. (2003). “Retrofit of reinforced concrete columns using
partially stiffened steel jackets”, Journal of Structural Engineering, ASCE,
129(6), 725-732.
Yu, T. T. J. G., Teng, J. G., Wong, Y. L., and Dong, S. L. (2010a), "Finite
element modeling of confined concrete-I: Drucker–Prager type plasticity
model", Engineering Structures, 32(3), 665-679.
Yu, T., Teng, J. G., Wong, Y. L., and Dong, S. L. (2010b), "Finite element
modeling of confined concrete-II: Plastic-damage model", Engineering
structures, 32(3), 680-691.
Zerbe, H.E. and Durrani, A.J. (1990). “Seismic response of connections in
two-bay reinforced concrete frame subassemblies with a floor slab”, ACI
Structural Journal, 87(4), 406-415.
Page 377
346
Figure 5.1 Two kinds of Stress-strain models for steel bar
Figure 5.2 The proposed cohesive element model
0 0.01 0.02 0.03 0.04 0.05 0.06 0.070
50
100
150
200
250
300
350
400
450
500
550
Strain
Sre
ss (M
Pa)
Type 1 Rebar Type 2 Rebar
Bar element node
Nodes 1 and 2
Node 3 Node 4
Yielding point
0.2%
Hardening point
Ultimate point
0.2%
Es
Page 378
347
Figure 5.3 Predicted stress-slip curve of proposed cohesive model versus input
data
(a) CB-Rec
0 0.5 1 1.50
2
4
6
8
10
12
Slip (mm)
She
ar s
tres
s (M
Pa)
Input Prediciton
' 30cf MPa
Deformed bar of 10 mm circumference
Page 379
348
(b) CB-T
(c) F-700-300-N
Figure 5.4 Boundary conditions of 3D FE models
Figure 5.5 Smooth loading scheme
Load
X-symmetric section
Restrained in Y direction Restrained in both Z and Y direction
Y direction
Loading time
Dis
plac
emen
t
Page 380
349
(a) CB-Rec and CB-T
(b) O-700-300-N and F-700-300-N
0 10 20 30 40 50 60 70 800
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
CB-Rec-FE CB-T-FE CB-Rec-Test CB-T-Test
0 10 20 30 40 50 60 70 800
50
100
150
200
250
Displacement (mm)
Loa
d (k
N)
O-700-300-N-FE F-700-300-N-FE O-700-300-N-Test F-700-300-N-Test
Element size=20 mm =2e-6
Loading duration=50T0
Element size=20 mm =2e-6 Loading duration=50T0
Page 381
350
(c) O-800-280-N and F-800-280-N
(d) F-700-300-P and F-800-280-P
0 10 20 30 40 50 60 70 800
50
100
150
200
250
Displacement (mm)
Loa
d (k
N)
O-800-280-N-FE F-800-280-N-FE O-800-280-N-Test F-800-280-N-Test
0 10 20 30 40 50 60 700
50
100
150
200
250
300
350
Displacement (mm)
Loa
d (k
N)
F-700-300-P-FE F-800-280-P-FE F-700-300-P-Test F-800-280-P-Test
Element size=20 mm =2e-6 Loading duration =50T0
Element size=20 mm =2e-6 Loading duration =50T0
Page 382
351
(e) O-600-220-N, F-600-220-N and F-600-280-N
(f) O-700-200-N, F-700-200-N and F-700-260-N
Figure 5.6 Comparisons of predicted load-displacement curves and test results
0 10 20 30 40 50 60 700
50
100
150
200
250
300
350
400
450
Displacement (mm)
Loa
d (k
N)
O-600-220-N-FE F-600-220-N-FE F-600-280-N-FE O-600-220-N-Test F-600-220-N-Test F-600-280-N-Test
0 10 20 30 40 50 60 70 800
50
100
150
200
250
300
350
400
Displacement (mm)
Loa
d (k
N)
O-700-200-N-FE F-700-200-N-FE F-700-260-N-FE O-700-200-N-Test F-700-200-N-Test F-700-260-N-Test
Element size=20 mm =2e-6 Loading duration =50T0
Element size=20 mm =2e-6 Loading time =50T0
Page 383
352
(a) CB-Rec
(b) CB-T
(c) O-700-300-N
Test crack pattern indicated by red dashed lines
Test crack pattern indicated by red dashed lines
Page 384
353
(d) O-800-280-N
(e) O-600-220-N
(f) O-700-200-N
Test crack pattern indicated by red dashed lines
Test crack pattern indicated by red dashed lines
Test crack pattern indicated by red dashed lines
Page 385
354
(g) F-700-300-N
(h) F-800-280-N
(i) F-700-300-P
Test crack pattern indicated by red dashed lines
Test crack pattern indicated by red dashed lines
Page 386
355
(j) F-800-280-P
(k) F-600-220-N
(l) F-700-200-N
Test crack pattern indicated by red dashed lines
Test crack pattern indicated by red dashed lines
Test crack pattern indicated by red dashed lines
Page 387
356
(m) F-600-280-N
(n) F-700-260-N
Figure 5.7 Predicted crack patterns
Figure 5.8 Energy relations of model CB-T
0 10 20 30 40 50 6010-8
10-6
10-4
10-2
100
102
104
Displacement (mm)
Per
cent
(%)
External work/Internal energy Energy by viscous effects/Internal energy Kinetic energy/Internal energy
Test crack pattern indicated by red dashed lines
Test crack pattern indicated by red dashed lines
Page 388
357
(a) CB-Rec
(b) CB-T
0 10 20 30 40 50 600
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test CB-T-ELE4050 CB-T-ELE20 CB-T-ELE10
0 10 20 30 40 50 600
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test CB-T-ELE4050 CB-T-ELE20 CB-T-ELE10
=2e-6 Loading duration =50T0
=2e-6 Loading duration =50T0
Page 389
358
(c) O-700-300-N
(c) F-700-300-N
Figure 5.9 Convergence study on element size
0 10 20 30 40 50 60 700
50
100
150
200
250
Displacement (mm)
Loa
d (k
N)
Test T-700-300-N-ELE4050 T-700-300-N-ELE20 T-700-300-N-ELE10
0 10 20 30 40 50 60 70 800
50
100
150
200
250
300
Displacement (mm)
Loa
d (k
N)
Test F-700-300-N-ELE4050 F-700-300-N-ELE20 F-700-300-N-ELE10
=2e-6 Loading duration=50T0
=2e-6 Loading duration=50T0
Page 390
359
(a) CB-T
(b) O-700-300-N
0 10 20 30 40 50 600
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test CB-T-12.5T0
CB-T-25T0
CB-T-50T0
CB-T-100T0
CB-T-200T0
0 10 20 30 40 50 60 700
20
40
60
80
100
120
140
160
180
200
Displacement (mm)
Loa
d (k
N)
Test O-700-300-N-12.5T0
O-700-300-N-25T0
O-700-300-N-50T0
O-700-300-N-100T0
O-700-300-N-200T0
Element size=20 mm =2e-6
Element size=20 mm =2e-6
Page 391
360
(c) F-700-300-N
Figure 5.10 The loading duration effect on the predicted load-displacement
curves
(a)CB-T
0 10 20 30 40 50 60 70 800
50
100
150
200
250
Displacement (mm)
Loa
d (k
N)
Test F-700-300-N-25T0
F-700-300-N-50T0
F-700-300-N-100T0
0 10 20 30 40 50 6010-8
10-6
10-4
10-2
100
102
104
106
Displacement (mm)
Kin
etic
Ene
rgy
(kN
. mm
)
CB-T-12.5T0
CB-T-25T0
CB-T-50T0
CB-T-100T0
CB-T-200T0
Element size=20 mm =2e-6
Page 392
361
(b) O-700-300-N
(c) F-700-300-N
Figure 5.11The loading duration effect on the dynamic energy
0 10 20 30 40 50 60 7010-8
10-6
10-4
10-2
100
102
104
106
Displacement (mm)
Kin
etic
Ene
rgy
(kN
. mm
)
O-700-300-N-12.5T0
O-700-300-N-25T0
O-700-300-N-50T0
O-700-300-N-100T0
O-700-300-N-200T0
0 10 20 30 40 50 60 70 80
10-6
10-4
10-2
100
102
104
Displacement (mm)
Kin
etic
Ene
rgy
(kN
*mm
)
F-700-300-N-25T0
F-700-300-N-50T0
F-700-300-N-100T0
Page 393
362
Figure 5.12 Relation between Rayleigh damping ratio and frequency
(a) 1st mode of vibration
(b) 2nd mode of vibration
Page 394
363
(c) 3rd mode of vibration
(d) 4th mode of vibration
(e) 5th mode of vibration
Figure 5.13 First 5 modes of vibration of model CB-Rec
Page 395
364
(a) 1st mode of vibration
(b) 2nd mode of vibration
(c) 3rd mode of vibration
Page 396
365
(d) 4th mode of vibration
(e) 5th mode of vibration
Figure 5.14 First 5 modes of vibration of model CB-T
(a) 1st mode of vibration
Page 397
366
(b) 2nd mode of vibration
(c) 3rd mode of vibration
(d) 4th mode of vibration
Page 398
367
(e) 5th mode of vibration
Figure 5.15 First 5 modes of vibration of model O-700-300-N
0 10 20 30 40 50 60
0
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test CB-T-No damping CB-T-Beta=1e-8 CB-T-Beta=5e-8 CB-T-Beta=1e-7 CB-T-Beta=5e-7 CB-T-Beta=1e-6 CB-T-Beta=5e-6 CB-T-Beta=1e-5 CB-T-Beta=5e-5
Element size=20 mm Loading duration=50T0
Page 399
368
000(a) CB-T
(b) O-700-300-N
Figure 5.16 The damping ratio beta effect on the predicted load-displacement
curves
(a) CB-T
0 10 20 30 40 50 60 70-50
0
50
100
150
200
Displacement (mm)
Loa
d (k
N)
Test O-700-300-N-No damping O-700-300-N-Beta=1e-8 O-700-300-N-Beta=5e-8 O-700-300-N-Beta=1e-7 O-700-300-N-Beta=5e-7 O-700-300-N-Beta=1e-6 O-700-300-N-Beta=5e-6 O-700-300-N-Beta=1e-5 O-700-300-N-Beta=5e-5
0 10 20 30 40 50 6010-8
10-6
10-4
10-2
100
102
104
106
Displacement (mm)
Kin
etic
Ene
rgy
(kN
. mm
)
CB-T-No damping CB-T-Beta=1e-8 CB-T-Beta=5e-8 CB-T-Beta=1e-7 CB-T-Beta=5e-7 CB-T-Beta=1e-6 CB-T-Beta=5e-6 CB-T-Beta=1e-5 CB-T-Beta=5e-5
Element size=20 mm Loading duration=50T0
a b c
d
Page 400
369
(b) O-700-300-N
Figure 5.17 The damping ratio beta effect on the dynamic energy
(a) CB-T
0 10 20 30 40 50 60 7010-8
10-6
10-4
10-2
100
102
104
106
Displacement (mm)
Kin
etic
Ene
rgy
(kN
. mm
)
O-700-300-N-No damping O-700-300-N-Beta=1e-8 O-700-300-N-Beta=5e-8 O-700-300-N-Beta=1e-7 O-700-300-N-Beta=5e-7 O-700-300-N-Beta=1e-6 O-700-300-N-Beta=5e-6 O-700-300-N-Beta=1e-5 O-700-300-N-Beta=5e-5
0 10 20 30 40 50 600
200
400
600
Loa
d (k
N)
Displacement (mm)
0 10 20 30 40 50 600
2.5
5
7.5x 104
Kin
etic
Ene
rgy
(kN
. mm
)
0 10 20 30 40 50 600
2.5
5
7.5x 104
0 10 20 30 40 50 600
2.5
5
7.5x 104
Test CB-T-Beta=1e-6 CB-T-Beta=2e-6 CB-T-Beta=4e-6
Element size=20 mm Loading duration=50T0
Page 401
370
(b) O-700-300-N
(c) F-700-300-N
Figure 5.18 The damping ratio beta effect on the predicted load-displacement
curves and dynamic energy
0 10 20 30 40 50 60 700
100
200
Loa
d (k
N)
Displacement (mm)
0 10 20 30 40 50 60 700
2500
5000
7500
10000
Kin
etic
Ene
rgy
(kN
. mm
)
0 10 20 30 40 50 60 700
2500
5000
7500
10000
0 10 20 30 40 50 60 700
2500
5000
7500
10000
Test O-700-300-N-Beta=1e-6 O-700-300-N-Beta=2e-6 O-700-300-N-Beta=4e-6
0 10 20 30 40 50 60 70 800
50
100
150
200
250
Loa
d (k
N)
Displacement (mm)
0 10 20 30 40 50 60 70 800
0.4
0.8
1.2
1.6
2x 104
Kin
etic
Ene
rgy
(kN
. mm
)
0 10 20 30 40 50 60 70 800
0.4
0.8
1.2
1.6
2x 104
0 10 20 30 40 50 60 70 800
0.4
0.8
1.2
1.6
2x 104
Test F-700-300-N-Beta=1e-6 F-700-300-N-Beta=2e-6 F-700-300-N-Beta=4e-6
Element size=20 mm Loading duration=50T0
Element size=20 mm Loading duration=50T0
Page 402
371
(a) Crack pattern at moment a
(b) Crack pattern at moment b
(c) Crack pattern at moment c
Page 403
372
(d) Crack pattern at moment d
Figure 5.19 Crack patterns of model O-700-300-N with damping ratio beta equal
to zero at four specified displacements
Figure 5.20 The computation precision effect on the predicted load-displacement
curves
0 10 20 30 40 50 60 70 80 900
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
CB-Rec-Single CB-Rec-Double CB-T-Single CB-T-Double O-700-300-N-Single O-700-300-N-Double
=2e-6 Loading duration=50T0 Element size=20 mm
a b
Page 404
373
Figure 5.21 The computation precision effect on the dynamic energy
(a) Yielding state of CB-Rec
0 10 20 30 40 50 60 70 80 9010-10
10-8
10-6
10-4
10-2
100
102
104
106
Displacement (mm)
Kin
etic
Ene
rgy
(kN
. mm
)
CB-R-Single CB-R-Double CB-T-Single CB-T-Double O-700-300-N-Single O-700-300-N-Double0 3
10-3
100
104
Single precision
Double precision
Page 405
374
(b) Moment a of CB-Rec
Single precision
Single precision
Double precision
Page 406
375
(c) Moment b of CB-Rec
(d) Ultimate state of CB-Rec
Single precision
Double precision
Double precision
Page 407
376
(e)Yielding state of CB-T
Single precision
Double precision
Page 408
377
(f) Ultimate state of CB-T
Single precision
Double precision
Page 409
378
(g) Peak load state of O-700-300
Single precision
Double precision
Page 410
379
(h) Ultimate state of O-700-300
Figure 5.22 Crack patterns of 3D FE models simulated with single and double
precision
Single precision
Double precision
Page 411
380
Figure 5.23 The part of FE model considering FRP confinement effect
Figure 5.24 FRP confinement effect (CE) on load-displacement curves
0 10 20 30 40 50 60 70 800
50
100
150
200
250
300
350
400
Displacement (mm)
Loa
d (k
N)
F-700-300-N-Test F-700-200-N-Test F-700-300-N-FE-With-CE F-700-300-N-FE-Without-CE F-700-200-N-FE-With-CE F-700-200-N-FE-Without-CE
=2e-6 Loading duration=50T0
Element size=20 mm
Web chord considering FRP confinement effect
Page 412
381
(a) CB-Rec and CB-T
(b) O-700-300-N and O-700-200-N
0 5 10 15 20 25 30 35 40 450
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
CB-Rec-Test CB-Rec-2D CB-Rec-3D CB-T-Test CB-T-2D CB-T-3D
0 10 20 30 40 50 600
50
100
150
200
250
300
350
Displacement (mm)
Loa
d (k
N)
O-700-300-N-Test O-700-300-N-2D O-700-300-N-3D O-700-200-N-Test O-700-200-N-2D O-700-200-N-3D
Page 413
382
(c) F-700-300-N and F-700-200-N
Figure 5.25 Predicted load versus displacement curves of 2D and 3D models
(a) CB-T
0 10 20 30 40 50 600
50
100
150
200
250
300
350
400
450
500
Displacement (mm)
Loa
d (k
N)
F-700-300-N-Test F-700-300-N-2D F-700-300-N-3D F-700-200-N-Test F-700-200-N-2D F-700-200-N-3D
2D
3D
Page 414
383
(b) O-700-200-N
(c) F-O-700-200-N
Figure 5.26 Comparison between crack patterns predicted with 2D and 3D FE
models
2D
3D
2D
3D
Page 415
384
Figure 5.27 Implementation of the BC+DP model in the FE model
Figure 5.28 Shear modulus degradation with different values of coefficient n
0
0.2
0.4
0.6
0.8
1
1.2
0 0.02
ρ=G
c/G
ennck
n=1
n=5
n=2
n=3
n=4
n=6
n=7
DP part
BC part
emaxck
Page 416
385
Figure 5.29 The effect of number of critical cracks
Figure 5.30 The effect of maximum cracking strain
0.1
1
10
100
1000
10000
100000
1000000
10000000
0
25
50
75
100
125
150
175
200
0 5 10 15 20 25 30 35 40 45
Loa
d (k
N)
Inte
rnal
Ene
rgy
(N·m
m)
Displacement (mm)
Test
O-700-300-N-DP+BC-N=5-0.0075-1-TZ
O-700-300-N-DP+BC-N=5-0.0075-2-TZ
O-700-300-N-DP+BC-N=5-0.0075-3-TZ
0.1
1
10
100
1000
10000
100000
1000000
0
25
50
75
100
125
150
175
200
0 5 10 15 20 25 30 35 40 45
Kin
etic
ene
rgy
(N·m
m)
Loa
d (k
N)
Displacement (mm)
Test
O-700-300-N-DP+BC-N=5-0.005-2-TZ
O-700-300-N-DP+BC-N=5-0.006-2-TZ
O-700-300-N-DP+BC-N=5-0.0075-2-TZ
Page 417
386
Figure 5.31 The effect of coefficient of power law
Figure 5.32 Comparison of load-displacement curves predicted with the DP and
the DP+BC models
0
25
50
75
100
125
150
175
200
225
0 10 20 30 40
Loa
d(kN
)
Displacement(mm)
Test O-700-300-N-DP+BC-N=1-0.006-2-TZ O-700-300-N-DP+BC-N=2-0.006-2-TZ O-700-300-N-DP+BC-N=3-0.006-2-TZ O-700-300-N-DP+BC-N=4-0.006-2-TZ O-700-300-N-DP+BC-N=5-0.006-2-TZ O-700-300-N-DP+BC-N=6-0.006-2-TZ O-700-300-N-DP+BC-N=7-0.006-2-TZ
0.1
1
10
100
1000
10000
100000
0
25
50
75
100
125
150
175
200
0 5 10 15 20 25 30 35 40 45
Loa
d (k
N)
Kin
etic
Ene
rgy
(N·m
m)
Displacement (mm)
Test O-700-300-N-DP O-700-300-N-DP+BC-N=5-0.006-2-TZ
Page 418
387
(a) O-700-300-N-DP
(b) O-700-300-N-DP+BC-N=5-0.006-2-TZ
Figure 5.33 Comparison of crack patterns predicted with the DP and the DP+BC
models
Page 419
Table 5.1 Specimen dimensions and test method
Source Specimen Beam dimensions Slab dimensions Opening size
Bending direction l0/s, mm bw, mm hb/ h'b, mm bslab, mm t, mm
Fabrication of web opening / Length/ Height, mm
Nie (2018) Batch 1
CB-Rec
3300/1650 250 500/460
NA Negative(a)
CB-T Negative O-700×300-N
1400 1000
Pre-formed/700/300 Negative
F-700×300-N Negative F-700×300-P Positive(b)
O-800×280-N Pre-formed/800/280
Negative F-800×280-N Negative F-800×280-P Positive
Nie (2018) Batch 2
O-600×220-N
3300/1650 250 500/460 1400 1000
Post-cut/600/220 Negative
F-600×220-N Negative O-700×200-N
Pre-formed/700/200 Negative
F-700×200-N Negative F-600×280-N Post-cut/ 600/280 Negative F-700×260-N Pre-formed/ 700/260 Negative
Note: l0: effective span of beam; s: shear span of beam. bw: beam width; hb: beam height; h'b: beam effective height; bslab: slab width; t: slab thickness; (a) The beam flange was in tension; (b) The beam flange was in compression.
Page 420
389
Table 5.2 Specimen material properties
Source Specimen
Concrete Steel reinforcement CFRP
reinforcement
fc’ (MPa)
Ec (GPa) c
Tension bars (deformed
bars) Number/D/ fy/fu, Mpa
Compression bars
(deformed bars)
Number/D/ fy/fu, Mpa
Stirrups (plain bars) D/spacing/ fv/fv, Mpa
Slab bars (plain bars) Number/D/ fv/fv, Mpa
Es, Gpa
tf ffrp,
Mpa Ef , Gpa
Nie (2018) Batch 1
CB-Rec 42.5 24.9 0.0026
4/20/ 475/625
3/20/ 475/625
8/100/ 307/447
Two layers 6/8/
307/447 200
NA CB-T 55.2 25.1 0.0033 O-700×300-N 42.5 24.9 0.0026 F-700×300-N 41 23.3 0.0028
0.334 2820 227 F-700×300-P 44.1 / / O-800×280-N 42.5 24.9 0.0026 NA F-800×280-N 41 23.3 0.0028
0.334 2820 227 F-800×280-P 44.1 / /
Nie (2018)
O-600×220-N 40.3 / /
4/20/ 434/559
3/20/ 475/625
8/ 100/ 349/526
Two layers 6/8/
200 NA
F-600×220-N 0.334 2820 227
Page 421
390
Batch 2 O-700×200-N 36.2 349/526 NA F-700×200-N 39.6
0.334 2820 227 F-600×280-N 42
F-700×260-N
Note: f'c: concrete cylinder compressive strength; Ec: concrete elastic modulus. c : concrete compressive strain at concrete cylinder strength; D: steel bar
diameter (in mm unit); fy: bar yield stress; fu: bar ultimate stress; Es :steel bar elastic modulus; tf : FRP sheet thickness; ffrp: tensile strength of FRP sheet; Ef:
elastic modulus of FRP sheet.
Page 422
391
Table 5.3 Steel properties of type 1 bar of the two test groups
Source Nie (2018)
Batch 1 Nie (2018)
Batch 2 D/mm 20 20 8
Es (MPa) 200000 200000 200000 fy (MPa) 475 434 349
h 0.03 0.028 0.013
Es2 (MPa) 2680 2450 4490 fu (MPa) 625 559 526
Note: D: steel bar diameter; Es :steel bar elastic modulus; fy: deformed bar yield stress;
h : hardening beginning strain; Es2: modulus of hardening curve; fu: deformed bar
strength.
Table 5.4 Peak bond stress and corresponding slip of Lu et al.'s (2005)model
Source Specimen fc’ (MPa)
Peak bond stress (MPa)
so (mm)
Nie (2018) Batch 1
F-700×300-N 41.00 3.68 0.0478 F-700×300-P 44.10 3.83 0.0498 F-800×280-N 41.00 3.68 0.0478 F-800×280-P 44.10 3.83 0.0498
Nie (2018) Batch 2
F-600×220-N 40.30 3.64 0.0474 F-700×200-N 39.60 3.61 0.0469 F-600×280-N
42.00 3.73 0.0484 F-700×260-N
Page 423
392
Table 5.5 Test and predicted loads
Specimen Test result (kN)
3D FE models (kN) 2D FE models (kN)
Prediction Prediction/test Prediction Prediction/test
CB-Rec 340 335.6 0.99 333.5 0.98 CB-T 510 513.0 1.01 510.5 1.00
O-700×300-N 182 189.2 1.04 165.2 0.91 O-800×280-N 181 189.5 1.05 158.9 0.88 O-600×220-N 316 310.9 0.98 243.8 0.77 O-700×200-N 300 300.5 1.00 236.5 0.79 F-700×300-N 207 215.2 1.04 216.8 1.05 F-800×280-N 219 219.9 1.00 221.5 1.01 F-600×280-N 260 258.8 1.00 279.7 1.08 F-700×260-N 270 269.2 1.00 296 1.10 F-600×220-N 407 385.0 0.95 407.4 1.00 F-700×200-N 410 386.5 0.94 406.4 0.99 F-700×300-p 228 284.3 1.25 277.1 1.22 F-800×280-P 215 262.8 1.22 249.6 1.16 Average = 1.03 0.99
STD = 0.09 0.12
Page 424
393
CHAPTER 6
3D SHELL MODELS FOR RC T-BEAMS
6.1 INTRODUCTION
The proposed 3D FE model presented in chapter 5 is, referred to hereafter
as a 3D solid model, given that 3D solid elements are used to represent concrete.
As revealed in the previous chapter, this model has performed well in modelling
the load-displacement curve and crack patterns of RC T-section beams
(T-beams). The 3D solid model, however, is not economical of computational
time as the number of elements is greater than that of the 2D FE model
proposed by Nie (2018), although it does provide more accurate predictions. To
reduce the FE modelling computational time of the dynamic explicit method,
reducing element numbers is a good method when the loading time duration and
the incremental time step length are fixed. Thus, a 3D shell FE model (referred
to as 3D shell model), in which the 3D solid elements are replaced by 3D S4R
shell elements for modeling concrete, was proposed for modelling T-section
beams (T-beams). Compared to the 2D FE model, the 3D shell model can
simulate the non-uniform deformation in the transverse direction of a T-beam.
Compared with the 3D solid model, the construction of a 3D shell model is
easier. Thus, several 3D shell models are proposed based on different
considerations/assumptions and compared in this chapter, with the aim of
Page 425
394
finding the most suitable 3D shell models which can be a substitute of 3D solid
model for modelling T-beams.
6.2 PROPOSED 3D SHELL MODELS
The 3D shell models are proposed based on the 3D solid model presented
in chapter 5. The major difference between 3D shell models and 3D solid
models is the type of concrete element as mentioned earlier. In the construction
of a 3D shell model, concrete of the beam web part was represented by 3D S4R
shell elements available in ABAQUS (2012). B31 beam elements were used to
model steel bars. Bond-slip behaviour between steel bars and concrete are
considered using the proposed 3D prism cohesive model. As for the slab
component, if there is only one layer of longitudinal bars, the slab shell layer is
located at the height of the longitudinal bars, as shown in Figure 6.1(a). If two
layers of longitudinal reinforcing bars exist, which is a usual case, the slab is
modelled by one shell layer as shown in Figure 6.1(a) or two shell layers as
shown in Figure 6.1(b). The effect of considering bond-slip between slab
longitudinal steel bars and concrete in the FE model needs to be investigated ,
so slab bars are either modeled by the discrete bar method which models bars
with beam element B31, or the smeared bar method in which an embedded layer
of steel within the concrete layer is defined to simulate the bars. As a result, a
total of seven 3D shell models, including 3 models with one layer of shell for
slab modeling and 4 models with two layers of shell for slab modelling, are
proposed and discussed next, as shown in Table 6.1. In Table 6.1, except for the
Page 426
395
T-3D-Solid model proposed in chapter 5, all remaining models use shell
elements to represent concrete. In the seven models, only T-2D-Shell-1-SB
model uses the smeared bar method to model slab bars, the rest six 3D shell
models use the discrete bar method to model slab bars, with or without
considering bond-slip behavior between bars and concrete. The models not
considering bond-slip behaviour between slab bars and concrete have “NBS” in
their names. For those models with two slab shell layers, if the deformation of
the two layers follows the plain sectional assumption such as an integral shell
layer, they have “R” in their names.
6.2.1 3D Shell Models with a One-shell-layer Slab
The positions of the shell layers of web and slab in the FE models with
one-shell-layer slabs (referred to hereafter as 3D 1-shell models) are shown in
Figure 6.1(a). Making using of the symmetry condition, only a half the T-section
beam is modeled to reduce the calculation time. As a result, the thickness of the
shell layer of beam web is half that of the beam web. In the beam web, the
stirrups or longitudinal bars at the same position (same height) are modelled
with only one bar meshed with the B31 beam elements with the section area and
bond area considered. The bond-slip behaviour of the steel-concrete interface is
modelled using the proposed 3D prism cohesive model detailed in Chapter 5.
For the slab, as there is only one shell layer with a thickness equal to that of the
slab, that layer is assumed located at the middle of the slab. It should be noted
that the slab bars of the T-3D-Shell-1-SB model are smeared in the slab shell, so
Page 427
396
that the area of the steel layer area is equal to the sum of the complete slab bar
area. One smeared steel layer is used to model the slab longitudinal and
transverse bars as they had a same bar spacing but were orthogonal. Bond-slip
behaviour of the slab steel-concrete interface is ignored (i.e. there is no slip
between the bar and surrounding concrete) in the T-3D-Shell-1-SB model.
For the T-3D-Shell-1 and T-3D-Shell-1-NBS models, slab bars are
modelled with beam elements B31. The slab bars sited in the same position are
modelled with only one bar of the same sectional area to their sectional area
sum. For the T-3D-Shell-1 model, the bond-slip behaviour between slab
longitudinal bars and concrete are modelled using the proposed triangular
cohesive model. The T-3D-Shell-1-NBS model does not incorporate bond-slip
behaviour between slab longitudinal bars and concrete (i.e. there is no slip
between the bar and surrounding concrete). For both T-3D-Shell-1-SB and
T-3D-Shell-1-NBS models, the transverse bars are connected directly to the
concrete elements, hence ignoring the bond-slip behaviour of the steel-concrete
interface.
6.2.2 3D Shell Models with a Two-shell-layer Slab
The positions of the shell layers representing the web and slab of a 3D FE
model in the case of a two-shell-layer slab (referred to, below, as a 3D 2-shell
model) are as shown in Figure 6.1(b). The beam web in 3D 2-shell models is
represented in the same way as for 3D 1-shell models. The two shell layers of
Page 428
397
the beam slab are located at the respective heights of the two slab bar layers.
The thickness of each shell layer is half the thickness of the real beam slab. The
offset values for each shell layer were defined the positions of each layers of
slab relative to the position of bars. A discrete bar method was applied to model
both the longitudinal and the transverse slab bars. For both T-3D-Shell-2 and
T-3D-Shell-2-NBS models, no direct interaction condition is applied to the two
slab layers. For the T-3D-Shell-2-R and T-3D-Shell-2-NBS-R models, direct
interaction is applied to the two slab layers using equation restraints in
ABAQUS (2012). To ensure the two slab layers deform in the same way as the
one layer, the following equations are applied to any two nodes, with one node
in each of the two slab layers but located in the same position along both the X
and Z directions:
`
, ,
, ,
, ,
,
,
000
000
x t x b
y t y b
z t z b
t b
t b z t t b
t b x t t b
y yx x dz z d
(6.1)
where ,x t , ,y t , ,z t , tx , ty and tz are the angular and line displacements
of the node of the top layer and ,x b , ,y b , ,z b , bx , by and bz are the
angular and line displacements of the corresponding node of the bottom layer
and t bd is the vertical distance separating the two layers. It should be noted
Page 429
398
that the above restraints are proposed based on the assumption that the
deformation of the nodes in the layer following a plane section.
In all 3D 2-shell models, no bond-slip behaviour is allowed at the interface
between slab transverse bars and concrete. For the T-3D-Shell-2 and
T-3D-Shell-2-R models, the proposed 3D prism cohesive model introduced in
the last chapter will be used for modelling of bond-slip behaviour and applied in
the longitudinal direction. The bond-slip is not considered (i.e. there is no slip
between the bar and surrounding concrete) in the T-3D-Shell-2-NBS and
T-3D-Shell-2-NBS-R models.
6.3 VERIFICATION OF 3D SHELL MODELS
The specimen CB-T described and tested by Nie (2018) was modeled
herein to verify the proposed 3D shell models. To save time, only half of the
specimen was modelled taking advantage of symmetry. The boundary
conditions of a 3D shell model are illustrated in Figure 6.2. Next, the predicted
load-displacement curves and crack patterns using the shell models were
compared with test results and those of the 3D solid model to assess the
accuracy/efficiency of the proposed 3D shell model, based on which the most
suitable 3D shell model will be determined. The effect of bond-slip behaviour
between steel bars and concrete in the longitudinal direction will also be
evaluated and discussed. The comparisons between the best 3D shell model with
2D models in terms of the load-displacement curve will also be conducted. The
Page 430
399
element size was chosen to 20 mm, the loading time duration was 50T0, the
damping coefficient β was 2e-6 according to the convergence studies presented
later.
6.3.1 3D Shell models with One-shell-layer Slab
The predicted load-displacement curves are shown in Figure 6.3. As
indicated in figure, the 3D-Solid model of the T-beam (referred to hereafter as a
T-3D-Solid model) predicts a slightly more accurate load-displacement curve
than those of the 3D 1-shell models. The 3D 1-shell models predict an
appreciably higher post-cracking stiffness and yield load than those of the
T-3D-Solid model, although both the shell model and the solid model produce
quite similar results for post-yielding loading displacement responses. Among
those 3D 1-shell models, the two which ignore the bond-slip behaviour of
longitudinal bars in the slab, predict a slightly higher post-cracking stiffness and
yield load than those predicted by the 3D-Shell-1 model, which incorporate the
modeling of the bond-slip model. Based on the above, it is apparent that the
considering bond-slip behaviour of bars in the longitudinal direction results in a
more accurate predicted load-displacement curve, especially for the
post-cracking stiffness and yield load of the T-beams.
The predicted crack patterns of the 3D 1-shell models are shown in Figure
6.4, from which it is seen that the web crack and slab crack patterns appear and
connected at the web-flange interface. The test crack pattern is indicated by
Page 431
400
red-dashed lines, indicating that the beam web cracks are well predicted by the
3D 1-shell models. For cracks propagated in the slab, however, the prediction of
T-beam 3D-Shell-1 model (referred to hereafter as a T-3D-Shell-1 model) is the
most successful. The other two models predict more densely spaced cracking
near the centre of the specimen slab, this can be attributed to that they ignore the
bond-slip behaviour of the longitudinal bars in slab. Thus, the
T-3D-Shell-1model, which considers the bond-slip behavior of the longitudinal
bars, is the best of the 3D 1-shell models for simulating the behaviour of RC
T-beams.
6.3.2 3D Shell Models with a Two-shell-layer Slab
The load displacement curves predicted by predicted 3D 2-shell models are
shown in Figure 6.5. Again, the T-3D-Solid model also predicts a better/accurate
load-displacement curves than those of the 3D 2-shell models. Similar to the 3D
1-shell model, the 3D 2-shell models without considering the bond-slip behavior
of longitudinal bars in slab (i.e. T-3D-Shell-2-NBS and T-3D-Shell-2-NBS-R)
predict appreciably higher post-cracking stiffnesses and yield loads than those
of the T-3D-Solid model, whereas the 3D 2-shell models (T-3D-Shell-2 and
T-3D-Shell-2-R) which ignore that bond-slip behaviour, predict obviously better
results (closer to test results and the predictions of T-3D-Solid model) in terms
of post-cracking stiffnesses and yield loads, thus demonstrating that the
inclusion of the bond-slip behaviour in the 3D 2-shell models leads to a more
accurate predicted load-displacement curve.
Page 432
401
The predicted cracks patterns using the 3D 2-shell model are shown in
Figure 6.6 with the test observed crack patters marked by red-dashed lines. It
should be noted that in Fig. 6.6., the web, top and bottom slab crack patterns are
placed, in the figure, at the top, middle and bottom respectively to demonstrate a
whole image of the crack pattern. It is obvious that the 3D-Shell-2 and
3D-Shell-2-R models of the T-beam (referred to, below, as T-3D-Shell-2 model
and the T-3D-Shell-2-R modes) predict better crack patterns regarding the
number and positions of cracks than the two models not considering the
bond-slip behavior of the longitudinal bars in the slab. Comparing Figures 6.6(a)
and (c) or (b) and (d), it is seen that the plane sectional restraint condition
applied for the two slab layers has a significant effect on the predicted crack
patter: it causes the cracks on the two shell layers to appear and propagate in the
same position, whereas in the two shell models (3D-Shell-2 and
3D-Shell-2-NBS) without including such restraint of plan section, even the main
crack may appear at obviously different locations. Comparing the predicted
load-displacement responses, it can be seen that the effect of the plan section
restrain is marginal. In summary, applying the plan section restraint leads to a
more accurate predictions, especially for crack patters.
To further investigate the effect of applying the plan sectional restraint,
several models, tabulated in Table 6.2 were studied. The differences between the
models lie in the density of the nodes applied with plan sectional restraints. As
the element size of the T-3D-Shell-2-R model is 20 mm, the restraint distances
Page 433
402
(distance between nodes with the plan section restraint) in both X and Z
directions was chosen as 20 mm in the T-3D-Shell-2-R. For the
T-3D-Shell-2-R-X100 and T-3D-Shell-2-R-X200 models, the restraint distances
in the X direction are 100 and 200 mm respectively but the restraint distance in
the Z direction still remains as 20 mm. Similarly, the T-3D-Shell-2-R-Z100 and
T-3D-Shell-2-R-Z200 models have the restraint distances of 100 and 200 mm in
the Z direction with the restraint distance along the X direction, remaining
unchanged at 20 mm.
The predicted load-displacement curves of the models listed in Table 6.2
are shown in Figure 6.7. As indicated in Figure 6.7, decreasing the density of
plane sectional restraints by increasing the restraint distances along the X or Z
directions (i.e. transverse and longitudinal directions) has a minor effect on the
predicted load-displacement curves. This is to be expected, as the T-3D-Shell-2
model, in which the plane section restrain is applied, predicts a
load-displacement curve, close to that of the T-3D-Shell-2-R model as discussed
above. The predicted crack patterns are shown in Figure 6.8. It is found that
increasing the restraint distance in the Z direction, results in predicted cracks in
the two slab layers, appearing and propagating at different position along the Z
direction. This situation is more obvious with increasing restraint distance in the
Z direction. Decreasing restraint distance along the X direction has much less
effect on predicted crack patterns than a decrease in restraint distance along the
Z direction, Hence, increasing restraint distance along the X direction is the
Page 434
403
more suitable modelling option in term of the accuracy of predicting the
cracking behavior of the T-section beams.
6.3.3 Efficiency and Accuracy of the 3D Shell Models
The predicted results reveal that the 3D-Shell-1 and 3D-Shell-2-R models
to be the best 3D 1-shell model and 3D 2-shell model respectively for modelling
a T-beam. As indicated in Figure 6.9, the two shell models predict close
load-displacement curves, both are close to the prediction of the 3D-solids
model. The comparisons of predicted crack patterns shown in Figure 6.4(b) and
6.6(c), indicate that both models satisfactorily predict crack patterns in beam
web and slab. However, the T-3D-Shell-2-R model is less economical of
computational time than the T-3D-Shell-1 model and, even less economical than
the T-3D-Solid model, as demonstrated in Table 6.3 which lists the
computational time. It should be noted that the same computer was used for all
computational times given in Table 6.3. The computational cost for the
T-3D-Shell-1 model was about 64.5% of that for the T-3D-Solid model, whereas
the computational cost for the T-3D-Shell-2-R model was more than twice
(about 2.22 times) that of the T-3D-Solid model. Of further interest is that the
decreasing plane sectional restraint density between the two slabs can reduce the
computational time cost when using 3D 2-shell models, as also illustrated in
Table 6.3. Decreasing restraint density in the X direction, is more effective in
reducing computational time than decreasing along the Z direction.
Page 435
404
To compare the performance of 3D-shell model and 2D FE model, the
load-displacement curve predicted by the T-3D-Shell-1 model is compared with
the 2D FE model of Nie (2018) in Figure 6.10. The T-3D-Shell-1 model, similar
to the T-3D-Solid model presented in chapter 5, is better than the 2D FE model
in predicting post-cracking stiffness as this model is able to capture non-uniform
deformation in the transverse direction.
Thus, in terms of accuracy, modeling convenience (i.e. ease of building the
FE model) and computational efficiency (assessed by computational time), the
T-3D-Shell-1 model can be considered a good substitute for the T-3D-Solid
model, and the T-3D-Shell-2-R with an appropriate restraint distance (e.g.
T-3D-Shell-2-R-X200), although more complicated in building FE model
compared with the T-3D-Shell-1 model, can also be a suitable choice, especially
when the locations (along the beam height) of the bars in slab needs to be
considered.
6.4 PARAMETRIC STUDIES ON THE EFFECTS OF SOME
KEY FACTORS
Of the 3D shell models, the T-3D-Shell-1 model has been found the most
suitable 3D FE model substitute for the T-beam. According to Chen et al. (2015)
and the results presented in the previous chapter (Chapter 5), accurate choices of
element size, loading time duration and damping coefficient β of Rayleigh
damping, are important for dynamic FE modelling if reliable results are to be
Page 436
405
obtained. Thus a parametric study was carried out in a similar way as described
in the previous chapter, to investigate the effects of element size, loading time
duration and the damping coefficient β on the numerical results predicted by the
T-3D-Shell-1 model, with the objective of finding most suitable values for them.
As in the previous chapter, in the parametric study, when one parameter is
changed, the rest parameters will be kept unchanged.
6.4.1 Element Size
A convergence study for different element sizes was carried and the
predicted load-displacement curves are shown in Figure 6.11, in which
T-3D-Shell-1-ELE40 and T-3D-Shell-1-ELE20 stand for FE models meshed
with element sizes 40 mm and 20 mm respectively, and T-3D-Shell-1-ELE4050
meshed mostly with 50 mm elements in the Z and X directions and 40 mm in
the Y direction. As indicated in Figure 6.11, the three models predict similar
load-displacement curves. As the FE model with 20 mm elements provides
predictions almost as accurate as the model with the 10 mm elements, the
former, i.e. FE model meshed element size of 20 mm, was chosen for the
T-3D-Shell-1 model, for modelling T-beams in this study.
6.4.2 Loading Duration
The natural period T0 of the T-3D-Shell-1model was first predicted using
ABAQUS (2012). Loading time spans of 12.5T0 , 25T0, 50T0, 100T0 and 200T0
Page 437
406
were then applied to the T-3D-Shell-1 model. The predicted load-displacement
curves are shown in Figure 6.12. It is evident that load time spans of 12.5 T0 and
25 T0 are not long enough for the T-3D-Shell-1 model to reach stability. When
the load time span is increased to 50 T0, the predicted load-displacement curve
is much closer to those with longer long loading time. The predicted kinetic
energy-displacement curve plotted in a logarithmic form is shown in Figure
6.13(a), which shows that even though kinetic energy increases as load time
duration decreases, kinetic energy is within a reasonable range except at the
beginning of loading (as indicated in Figure 6.13(b)) and without any
unexpected sudden increase. Sudden increases of kinetic energy, featured by the
local peaks of the kinetic energy, caused by concrete cracking are efficiently
damped out with the increase of displacement. Given this, a load time span
equal to 50 T0 was chosen for the T-3D-Shell-1 model.
6.4.3 Damping Coefficient β
As stated in chapter 5, only the stiffness-proportional damping is needed in
Rayleigh damping, so only the coefficient β value needs to be determined for FE
modelling using a dynamic explicit method. The T-3D-Shell-1 model with a
coefficient β value ranging from 0 to 5e-5 was studied. The predicted
load-displacement curves predicted with various values of β are shown in Figure
6.14. As illustrated in Figure 6.14, when β is equal to, or below, 1e-7, the
predicted load-displacement curve deviates significantly from the test response
before it reaches the yield state. When β is larger than 1e-5, the T-3D-Shell-1
Page 438
407
model predicts higher cracking and yield loads, due to the over-shooting effect
on load arising from viscous force associated with the velocity. Thus, the
suggested value for β is between 5e-7 and 1e-5. Predicted kinetic
energy-displacement curves in logarithmic form are shown in Figure 6.15. It is
obvious that when β is not larger than 1e-7, the predicted kinetic energy shows a
sudden increase after concrete cracking and cannot not damped down as
displacement increases. When β is larger than 1e-5, the kinetic energy is also
slightly higher than those corresponding to a β value ranging from 5e-7 to 1e-5,
which is still reasonable. To further narrow the range for β, the T-3D-Shell-1
model with β equal to 1e-6, 2e-6 and 4e-6 was studied. It was found that the
predicted load-displacement curves for the three damping ratio were quite close.
The predicted kinetic energy-displacement curves were also within a reasonable
range. A value for β of 2e-6 was chosen because the T-3D-Shell-1 model
predicts a more stable result than with this value of β as the predicted curves are
sufficiently stable when compared with those with β equal to 4e-6. It is of
interest to note that local peaks of kinetic energy at the two significant moments
corresponding to concrete cracking and steel bars yielding, regardless of the
specific values of β.
To study the effect of insufficient β on predicted crack patterns, the
predicted crack patterns of the T-3D-Shell-1 model with β equal to 1e-8 at
different displacements (marked as moments “a”, “b”, “c” and “d” in Figure
6.14) are presented in Figure 6.17, in addition those of T-3D-Shell-1 model with
Page 439
408
β equal to 2e-6 for comparison. The displacements at moments “a” to “d” are as
indicated in Figure 6.14. Moments “a” and “b” are occurring immediately
before and after appearance of a new crack (at the displacement of 3.28 mm, as
predicted by the model with β equal to 2e-6). Moments “c” and “d” are those at
the lowest and highest points (at the displacement of 5.2 mm and 43.28 mm) of
the load-displacement curve, respectively. As shown in Figure 6.17(a), at the
moment “a”, three flexural cracks propagate in the beam web and slab. When
reaching the displacement of moment “b”, for the model with insufficient β ,
many unexpected large strains and associated cracks arise on the web and slab,
due to local vibration caused by crack initiation and propagation and which has
not been effectively damped out due the low value of damping associated with
the very low value of β. When displacement increases to that of moment “c”,
additional large strains continue to emerge until, finally, at displacement of
moment “d”, the predicted crack pattern is spread over the FE model of
insufficient β , which is totally unreal/unreliable. Thus, β must be large enough
(e.g. >1e-7 for the studied case) to damp out dynamic effects associated with
local vibration caused by local cracking, so as to obtain reliable predicted crack
patterns. However, it should not too large (e.g. <5e-5 for the studied case) to
avoid viscous force leading to the over-shooting (overestimation) of load.
6.5 3D SHELL MODELS FOR A T-BEAM WITH A WEB
OPENING
The above sections have shown that compared with 3D solid model. The
Page 440
409
most suitable 3D 1-shell models (i.e. the 3D-Shell-1 model) well predicts the
cracking and load-displacement behavior of a T-beam, and reduces
computational time as well. Next, the performance of the 3D-Shell-1model will
be further verified by modelling of T-beams weakened by web openings. The
O-700-300-N specimen tested by Nie (2018) was chosen for the analysis. A
convergence study related to element size was first conducted. As shown in
Figure 6.18, even when the element size is 20 mm, the predictions of the
O-700-300-N specimen still cannot converge, suggesting that the element size
needs to be smaller. The 3D-Shell-1 model of O-700-300-N based on 10 mm
elements, however, costs more computational time than that the 3D Solid model
meshed with 20 mm elements. Given that the 3D Solid model is more accurate
than the 3D-Shell-1 model, the latter proves not to be a satisfactory 3D Solid
model substitute when modelling a T-beam with web opening weakening. This
is partially because when the slab is modelled with only one shell layer, that
layer lacks the ability to simulate cracks propagating in slab originated from the
right bottom corner of the opening in the web. As shown in Figure 6.19, the
cracks emerging from the right corner of the opening near the slab are critical
and extend to the bottom of the beam. As shell elements was used instead of
solid element to simulation slab, the progressive propagation of these cracks
within the slab could be effectively simulated by the shell elements.
Theoretically, this problem can be solved by modeling the slab using muti-layer
shells; it can be envisaged that this will lead to complicated restraint conditions
between the different shell layers which will greatly increase the computational
Page 441
410
cost as discussed above, making this modeling approach much less appealing.
6.6 CONCLUDING SUMMARY
The 3D Solid model proposed in Chapter 5 proved to be both reliable and
accurate when modelling a T-section beam (T-beam), but is not economical in
computational time compared with the 2D FE model proposed by Nie (2018).
Thus, representation of the beam with shell elements was alternatively proposed,
by which the number of elements can be greatly reduced. Several shell element
models have been built and used for simulation studies. A T-beam tested by Nie
(2018) was modeled using these shell element models and the 3D T-beam solid
model presented in chapter 5 was used as the control model for comparison
purpose. The following conclusions are drawn based on the numerical results,
comparisons and discussion presented in this chapter:
(1) The 3D-Solid model for T-beams predicts the most accurate results in terms
of the load-displacement curve and crack patterns.
(2) Those 3D 1-shell and 2-shell models incorporating the bond-slip relationship
between slab longitudinal bars and concrete, predict better
load-displacement curves and crack patterns than those which do not
consider the bond slip behaviour.
(3) In the 3D 2-shell model, the plane section restraints applied between the two
shells representing slab have only slight effect on the predicted
Page 442
411
load-displacement curves. The crack patterns, however, are better predicted
when plan section restraints are applied. Decreasing the density of restraints
in the X direction has less effect on the predicted crack patterns than
decrease of restraint density in the Z direction.
(4) Both 3D-Shell-1 and 3D-Shell-2-R models work well as models for the
T-beam. As the 3D-Shell-1 model uses less computational time than the
3D-Shell-2-R mode, the former is the better substitute for 3D solid model in
modeling T-beam.
(5) A 20 mm element size, a time span of 50T0 and a damping coefficient β
equal to 2e-6, are suggested for the FE 3D-Shell-1 model of the T-beam.
(6) An element size of 20 mm is not small enough for the 3D-Shell-1 model to
obtain convergence when it used to model a T-beam weakened by web
openings. Thus, the 3D-Shell-1 model is not a good substitute for the 3D
solid model when modelling such a beam. A 3D shell model with muti-layer
shell slab might be a solution to the convergence problem, but it is not an
efficient method as it not only requires more computational time but gives
less accurate predictions than the 3D solid model.
6.7 REFERENCES
ABAQUS (2012). ABAQUS Analysis User's Manual (Version 6.12), Dassault
Systems SIMULIA Corporation, Providence, Rhode Island, USA.
Page 443
412
Chen, G.M., Teng, J.G., Chen, J.F. and Xiao, Q.G. (2015). “Finite element
modeling of debonding failures in FRP-strengthened RC beams: A dynamic
approach”, Computers and Structures, 158, 167–183.
Nie, X.F. (2018). Behavior and Modeling OF RC Beams with an
FRP-Strengthened Web Opening, Doctoral degree thesis: The Hong Kong
Polytechnic University.
Page 444
413
(a) model with a one-shell-layer slab
(b) model with a two-shell-layer slab
Figure 6.1 Shell layer positions of 3D shell models
Figure 6.2 Boundary conditions of 3D 1-shell models
x
y
x
y
z
y
Page 445
414
Figure 6.3 Predicted load-displacement curves of 3D FE 1-shell models versus
test result and that of the 3D solid model
(a) T-3D-Shell-1-SB
(b) T-3D-Shell-1
0 10 20 30 40 50 600
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test result T-3D-Solid T-3D-Shell-1-SB T-3D-Shell-1 T-3D-Shell-1-NBS
Element size=20 mm Loading duration=50T0 Damping coefficient β=2e-6
Page 446
415
(c) T-3D-Shell-1-NBS
Figure 6.4 Predicted crack patterns of 3D 1-shell models
Figure 6.5 Predicted load-displacement curves of 3D 2-shell models versus test
result and that of 3D solid model
0 10 20 30 40 50 600
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test result T-3D-Solid T-3D-Shell-2 T-3D-Shell-2-NBS T-3D-Shell-2-R T-3D-Shell-2-NBS-R
Element size=20 mm Loading duration=50T0 Damping coefficient β=2e-6
Page 447
416
(a) T-3D-Shell-2
(b) T-3D-Shell-2-NBS
(c) T-3D-Shell-2-R
(d) T-3D-Shell-2-NBS-R
Figure 6.6 Predicted crack patterns of 3D 2-shell models
Page 448
417
(a) Different restraint density along X direction
(b) Different restraint density along Z direction
Figure 6.7 The effect of plane sectional restraint density between the two slab
shell layers on the predicted load-displacement curves
0 10 20 30 40 50 600
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test result T-3D-Shell-2-R T-3D-Shell-2-R-X100 T-3D-Shell-2-R-X200
0 10 20 30 40 50 600
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test result T-3D-Shell-2-R T-3D-Shell-2-R-Z100 T-3D-Shell-2-R-Z200
Element size=20 mm Loading duration=50T0 Damping coefficient β=2e-6
Element size=20 mm Loading duration=50T0 Damping coefficient β=2e-6
Page 449
418
(a) T-3D-Shell-2-R-X100
(b) T-3D-Shell-2-R-X200
(c) T-3D-Shell-2-R-Z100
Page 450
419
(d) T-3D-Shell-Dy-2-R-Z200
Figure 6.8 The effect of plane sectional restraint density between the two slab
shell layers on the predicted crack patterns
Figure 6.9 Comparison between predicted and test load-displacement curves.
0 10 20 30 40 50 600
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test result T-3D-Solid T-3D-Shell-1 T-3D-Shell-2-R
Element size=20 mm Loading duration=50T0 Damping coefficient β=2e-6
Page 451
420
Figure 6.10 Comparison of predicted load-displacement curves between the 2D
and 3D-Shell-1 models
Figure 6.11 Effect of element size on the predicted load-displacement curves
0 5 10 15 20 25 30 35 400
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
CB-T-Test CB-T-2D CB-T-3D-Shell-1-BS
0 10 20 30 40 50 600
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test T-3D-Shell-1-ELE4050 T-3D-Shell-1-ELE20 T-3D-Shell-1-ELE10
Loading duration=50T0 Damping coefficient β=2e-6
Page 452
421
Figure 6.12 Effect of loading time on the predicted T-3D-Shell-1 model
load-displacement curves
(a) Curves of kinetic energy versus displacement
0 10 20 30 40 50 600
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test T-3D-Shell-1-12.5T0
T-3D-Shell-1-25T0
T-3D-Shell-1-50T0
T-3D-Shell-1-100T0
T-3D-Shell-1-200T0
0 10 20 30 40 50 6010-8
10-6
10-4
10-2
100
102
104
106
Displacement (mm)
Kin
etic
Ene
rgy
(kN
. mm
)
3D-Shell-1-BS-12.5T 3D-Shell-1-BS-25T 3D-Shell-1-BS-50T 3D-Shell-1-BS-100T 3D-Shell-1-BS-200T
Element size=20 mm Damping coefficient β=2e-6
Page 453
422
(b) Curves of kinetic energy to internal energy ratio versus displacement
Figure 6.13 Effect of loading time on the predicted kinetic energy
Figure 6.14 The effect of β value on the predicted load-displacement curves
0 10 20 30 40 50 6010-5
10-4
10-3
10-2
10-1
100
101
102
103
Displacement (mm)
Kin
etic
Ene
rgy/
Inte
rnal
Ene
rgy
(%)
3D-Shell-1-BS-12.5T 3D-Shell-1-BS-25T 3D-Shell-1-BS-50T 3D-Shell-1-BS-100T 3D-Shell-1-BS-200T
0 10 20 30 40 50 600
100
200
300
400
500
600
Displacement (mm)
Loa
d (k
N)
Test 3D-Shell-1-BS-No damping 3D-Shell-1-BS-Beta=1e-8 3D-Shell-1-BS-Beta=5e-8 3D-Shell-1-BS-Beta=1e-7 3D-Shell-1-BS-Beta=5e-7 3D-Shell-1-BS-Beta=1e-6 3D-Shell-1-BS-Beta=5e-6 3D-Shell-1-BS-Beta=1e-5 3D-Shell-1-BS-Beta=5e-5
a b c d
Element size=20 mm
Loading duration =50T0
1%
Page 454
423
(a) Curves of kinetic energy versus displacement
(b) Curves of kinetic energy to internal energy ratio versus displacement
Figure 6.15 The effect of β value on predicted kinetic energy
0 10 20 30 40 50 6010-8
10-6
10-4
10-2
100
102
104
106
Displacement (mm)
Kin
etic
Ene
rgy
(kN
. mm
)
3D-Shell-1-BS-No damping 3D-Shell-1-BS-Beta=1e-8 3D-Shell-1-BS-Beta=5e-8 3D-Shell-1-BS-Beta=1e-7 3D-Shell-1-BS-Beta=5e-7 3D-Shell-1-BS-Beta=1e-6 3D-Shell-1-BS-Beta=5e-6 3D-Shell-1-BS-Beta=1e-5 3D-Shell-1-BS-Beta=5e-5
0 10 20 30 40 50 6010-6
10-4
10-2
100
102
104
Displacement (mm)
Kin
etic
Ene
rgy/
Inte
rnal
Ene
rgy
(%)
3D-Shell-1-BS-No damping 3D-Shell-1-BS-Beta=1e-8 3D-Shell-1-BS-Beta=5e-8 3D-Shell-1-BS-Beta=1e-7 3D-Shell-1-BS-Beta=5e-7 3D-Shell-1-BS-Beta=1e-6 3D-Shell-1-BS-Beta=5e-6 3D-Shell-1-BS-Beta=1e-5 3D-Shell-1-BS-Beta=5e-5
1%
Page 455
424
Figure 6.16 The effect of β value on load-displacement curves and kinetic energy
curves
(a)Moment a
0 10 20 30 40 50 600
200
400
600
Loa
d (k
N)
Displacement (mm)
0 10 20 30 40 50 600
4000
8000
12000
Kin
etic
Ene
rgy
(kN
. mm
)
0 10 20 30 40 50 600
4000
8000
12000
0 10 20 30 40 50 600
4000
8000
12000
Test 3D-Shell-1-BS=1e-6 3D-Shell-1-BS=2e-6 3D-Shell-1-BS=4e-6
Element size=20 mm Loading duration=50T0
=1e-8
=2e-6
Page 456
425
(b) Moment b
(c) Moment c
=1e-8
=2e-6
=2e-6
=1e-8
Page 457
426
(d) Moment d
Figure 6.17 Predicted crack patterns with β =1e-8 and 2e-6 at critical moments
0 10 20 30 40 50 60 70 80
0
50
100
150
200
250
300
350
Displacement (mm)
Loa
d (k
N)
Test O-700-300-N-Shell-1-ELE4050 O-700-300-N-Shell-1-ELE20 O-700-300-N-Shell-1-ELE10
Loading duration=50T0
Damping coefficient β=2e-6
=1e-8
=2e-6
Page 458
427
Figure 6.18 Convergence study on element size for modeling O-700-300-N
specimen
Figure 6.19 Predicted crack pattern of 3D shell-1 model for specimen
O-700-300-N
Page 459
428
Table 6.1 Model names and details
Model name Concrete element
type
Layer number
Bond-slip between slab bars and
concrete Other issues
3D-Solid Solid / √
/
3D-Shell-1-SB
Shell
1 Smeared bars
3D-Shell-1 √ 3D-Shell-1-NBS ×
3D-Shell-2
2
√ 3D-Shell-2-NBS ×
3D-Shell-2-R √ The deformation of two slab layers are restrained 3D-Shell-2-NBS-R ×
Table 6.2 Model names and restraint frequency
Model name Plane sectional Restraint
distance along X direction (mm)
Plane sectional Restraint distance along Z direction (mm)
T-3D-Shell-2-R 20 20 T-3D-Shell-2-R-X100 100 20 T-3D-Shell-2-R-X200 200 20 T-3D-Shell-2-R-Z100 20 100 T-3D-Shell-2-R-Z200 20 200
Table 6.3 Total time cost by FE models
Model Total time T-3D-Shell-1 15:07:02
T-3D-Shell-2-R 51:53:57 T-3D-Solid 23:26:40
T-3D-Shell-2-R-X100 17:22:04 T-3D-Shell-2-R-X200 16:17:23 T-3D-Shell-2-R-Z100 30:28:47 T-3D-Shell-2-R-Z200 27:9:14
Page 460
429
CHAPTER 7
EFFECTIVE SLAB WIDTH OF REINFORCED
CONCRETE FRAMES WITH SLAB UNDER
TENSION
7.1 INTRODUCTION
As previously indicated in Chapter 2, the longitudinal bars within the
effective slab width contribute greatly to the negative bending strength of a
T-section beam (T-beam). However, currently, a uniform method, to determine
the effective width of a slab under tension, especially for an RC structure with
cast-in-place slabs does not exist. Existing research in this area is
experimentally based (Ehsani and Wigh 1982, Durrani and Zerbe 1987, T
Pantazopotrlou et al. 1988, French 1991, Li 1994, Jiang et al. 1994, Wu et al.
2002, Wang et al. 2009, Zhen et al. 2009, Yang 2010, Sun et al. 2010, Qi et al.
2010, He 2010). Table 2.1 is a summary of the suggested values of effective
slab widths given by these researchers, based on their experimental results. It is
obvious that the effective slab width beff is mainly related to beam width bw,
beam height hb, effective span of beam l0, slab thickness t, and clear distance
between beams sn. Most researchers determine the effective slab width when the
storey drift ratio is equal to 1/50. Some researchers, however, have used
Page 461
430
equations, based on their simulation data, to determine the effective slab widths.
Ning et al. (2016) studied the effective slab width of an RC frame both
experimentally and by means of FE simulation. Ning et al. (2016) conducted
parametric studies on the axial compression ratio, concrete strength,
reinforcement ratio of slabs, thickness of slabs and dimension of the transverse
beams. An equation involving only the main beam width and height to predict
the effective slab width with a 95% guaranteed accuracy was then proposed.
Many researchers have studied the effective slab width of a composite
structure under positive loading (under which the slab is under compression). A
few, such as Nie and Tao (2012), also studied the effective slab width of
composite structures under negative loading (under which the slab is under
compression). By conducting parametric studies through FE simulation on
column dimensions, steel beam height, RC slab width and thickness, transverse
beam, and yield stress of the longitudinal reinforcement, Nie and Tao (2012)
found that the main factors which influence the negative effective flange width,
included the column dimensions, the steel beam height, the flange width of the
transverse beam, and the yield stress of the longitudinal reinforcement. The slab
width was found to be less influential.
In Table 7.1, results of two papers (Ning et al., 2016 and Nie and Tao,
2012) giving the effective slab width under negative bending, based on FE
simulation results are compared. Ning et al. (2016) studied the effective slab
width of an RC frame and Nie and Tao (2012) studied the effective slab width
Page 462
431
of a composite structure at ultimate state under both positive and negative
bending. Ning et al. (2016) did not study the effect of beam length on beff, which
is a parameter considered in all design codes except EC8. Nie and Tao (2012)
found the beam length was less influential. In addition, Ning et al. (2016) did
not study the effect of slab width on beff, a parameter also considered by all
design codes except EC8. The equations used by Nie and Tao (2012) and Ning
et al. (2016) for calculating the effective slab width under negative bending use
the stress in slab bars to calculate the effective width. There is a slight
difference in that Nie and Tao (2012), in their calculations, used yield stress
instead of maximum stress in slab bars, which was used in Ning et al. (2016)'s
equation.
By comparing the five different codes (GB 50010-2010, ACI318-05 &
FEMA-356, EC8, NZS-3101:2006) as indicated in Table 2.2, it is apparent that
the effective slab width of an interior joint is wider than that of an exterior joint
under the same conditions. Only Eurocode 8 takes column width into account
when determining the effective slab width. The application of this code is the
simplest. In contrast, the equation suggested in NZS-3101:2006 is the most
complicated of all. The equations suggested in the Chinese code GB
50010-2010, ACI318-05 and FEMA-356 are similar. The difference in these
equations in the codes for effective slab widths further strengthen the need to
clarify which are the key factors affecting effective slab width of a RC joint.
In this chapter, a parametric study on the effective slab width for RC
Page 463
432
structures with slab under tension is conducted via FE modelling to find the key
factors affecting the effective slab width. Simple equations for slab effective
width at both exterior and interior joints are then proposed, based on the results
of the parametric study. It is to be noted that the modelling relates only to weak
beam and strong column joints which fail due to beam flexural failures.
7.2 PROPOSED SIMPLIFIED 3D FE MODELS
7.2.1 FE Model Details and Boundary Conditions
The 3D shell models, with only one shell slab layer, as proposed in chapter
6, were used for the parameter study. In the models, to guarantee all the slab
longitudinal bars are under tension when the beam enters the yielding stage, the
thickness of this layer must be small relative to the depth of the main beam.
Slabs under negative bending were the cases studied and described in this
chapter. The control model and the others simulated are listed in Table 7.2. As
shown, the main parameters studied are the main beam dimensions of beam
width, height and length, the beam reinforcement ratio, the slab width and
thickness, the slab longitudinal bar distance, the column width and transverse
beam dimensions and stirrup spacing. The effect of bond-slip between concrete
and bars, yield stress of bars, and the concrete strength on the predicted
effective slab width was also studied. The frame of the control model are shown
in Figure 7.1. The control models are shown in Figure 7.2. Only one half of a
specimen was modelled and a concentrated load was applied to simulate the
Page 464
433
situation (the distribution of moment) of an RC structure under lateral force. A
frame with following details was chosen to reflect the commonly accepted
parameters in practice, as detailed next. The main beam span is 4000 mm and
the height and width are 400 mm and 200 mm respectively. The width of the
slab flange at one side is 3000 mm. The column width and height are both 400
mm. Concrete with cylinder strength of 40 MPa was assumed for slab, beam
and column. Deformed bars with a yield stress of 400 MPa and plain bars with a
yield stress of 300 MPa respectively constitute the main beam and slab
longitudinal bars. The stirrups in the main beam consist of 8 mm plain bars with
a yield stress of 300 MPa at a 100 mm spacing. The slab bars are reinforced
with two layers 8 mm plain bars at a 200 mm spacing. The height and width of
the transverse beams are 300 mm and 150 mm. Two deformed bars comprise
the upper and lower longitudinal bars. The stirrups are also 8 mm plain bars
with a yield stress of 300 MPa at a 100 mm spacing.
The boundary conditions of the control models are illustrated in Figure 7.2.
To save computational time, only half of a specimen was modelled. The exterior
edge of the slab and the transverse beam are symmetrically restrained in the X
direction as the joints belong to the RC multi-bay frame in Figure 7.1. The
column was modelled by restraining all the degrees of freedom of element
nodes within the column section, similar to Nie and Tao (2012), as the column
is assumed to be elastic and sufficiently strong to take the loading. For an
exterior joint, only one beam under negative loading was modelled. For an
Page 465
434
interior joint, two beams, under respective positive and negative loading, were
modelled.
7.2.2 Ultimate State in Simulation
As suggested in the Chinese deign code (GB50010-2010), a 1/50 drift ratio
is acceptable as representing the ultimate state for most RC Frames. In this
chapter, drift ratio is defined by dividing applied displacement by the length of
beam, which is 4000 mm for control model and a drift ratio of 1/50 was
assumed as the ultimate state during the FE simulations. If the peak load is
reached, however, before that ultimate state assumption, that peak load is used
for the calculation of effective slab width.
7.2.3 Equations for Effective Slab Width
Nie and Tao (2012) and Ning et al. (2016) calculated the effective slab
width effb of the FE models, using the predicted stress of the slab bars at the
same section. However, when the flange is wide, the plane section assumption
may be violated. Thus, in this chapter, the effective slab width has been
back-calculated using the peak moment obtained for the critical section. With
this peak moment value, the effective slab width can be calculated based on the
plane section assumption. The equations used in Teng et al. (2002) were
adopted herein. The part accounting for FRP contribution was removed.
Page 466
435
10 ( )cu sieff w bi
c fr i
f Ak bx b bd
(7.1)
11
( ) ( ) ( ) ( )2
nfcu si
peak 2 eff w bi si si siic fr i
hf Ah h hM k bx - k x b b - A - d2 d 2 2
(7.2)
In the above equation, the slab bars are treated as a layer of steel with a
thickness of si
fr i
Ad
mm. The concrete tension stress is not included as it's
negligible. The compressive zone height x and the effective slab width effb
can be obtained using Equations 7.1 and 7.2.
7.3 PARAMETRIC STUDIES
The exterior joint and interior joint models were used in the following
parametric study and the summarized results are shown in Tables 7.3 and 7.4.
The calculated effective flange width is based on Equations 7.1 and 7.2. It
should be noted that in the parametric studies presented next, when one
parameter is changing, the rest parameters (conditions) are kept the same as
those in the reference beam.
7.3.1 Effect of Stress-strain Models of Steel
Elastic-perfect plastic and nonlinear stress-strain curves related to the
hardening range shown in Figure 7.3 were used in two models to investigate the
steel stress-strain effect on the predicted effective slab width. As indicated in
Page 467
436
Tables 7.3 and 7.4, the steel stress-strain model effect is is small. To simplify
the parametric study, the elastic-perfect plastic model was used in the remaining
analyses.
7.3.2 Effect of Bond Slip Behaviour
The bond slip effect between concrete and bars was investigated. The
bond-slip behaviour between steel bars and concrete surface is represented by
COH3D8 cohesive elements and CEB-FIP(1993) bond-slip model was used.
The proposed 3D prism cohesive model was used in simulation to connect steel
bars and concrete. The elements of steel bars for models E-BS-N and I-BS-N,
were directly connected to the concrete nodes. For models E-C and I-C,
bond-slip behaviour was assumed for longitudinal bars in the beam and slab and
stirrups in the beams. As shown in Tables 7.3 and 7.4, the bond slip effect on
the predicted effective slab width of interior joint is obvious. The inclusion of
the bond-slip behaviour also leads to more accurate predictions as demonstrated
in chapter 6. Hence bond slip effect was considered in the remaining analyses.
7.3.3 Effect of Yield Stress of Steel Bars
The strength effect of steel bars was investigated by changing yield stress
of longitudinal bars in both main beam and slab bar. Beam bars of 300-600 MPa
yield stress and slab bars of 200-400 MPa yield stress were assumed in the
simulations. As shown in Tables 7.3 and 7.4, for both interior and exterior joints,
Page 468
437
increasing the bar yield stress of main beam longitudinal bars and slab bars
leads to less predicted effective slab width of both exterior and interior joints.
The effect of slab bar yield stress is more obvious than that of main beam
longitudinal bars. The effect of the yield stress of main beam longitudinal bars
is insignificant for exterior joints.
7.3.4 Effect of Concrete Strength
To investigate the concrete strength effect on predicted effective slab width,
concrete of a cylinder strength of 20, 30, 40, 50, 60 MPa was assumed. As
shown in Tables 7.3 and 7.4, an increase in concrete strength results in a greater
effective slab width, except for model E-CS-60, exterior joint with concrete
strength equal to 60MP. It is possible that increased concrete strength leads to
increased beam shear strength and thus the shear deformation along the
transverse direction is weakened. Thus the deformation of the critical section is
more uniform. The effect of concrete strength is more obvious for interior joints
than for exterior joints.
7.3.5 Effect of Beam Length, Width and Height
The effects of beam heights of 300 mm to 800 mm, beam widths of 100
mm to 300 mm, half beam lengths of 1000 mm to 3000 mm were all examined.
The study on beam length effect was only considered for exterior joints as the
results of exterior joints have indicated that the effect of the beam length is
Page 469
438
insignificant, which is probably because the effective slab width is more
strongly affected by the critical section properties and boundary conditions. As
shown in Tables 7.3 and 7.4,the effect of beam height and width is more
significant for interior joints than for exterior joints. Increasing the beam height/
width leads to wider predicted effective slab width of interior joints.
7.3.6 Effect of Beam Reinforcement Ratio
In this study, the longitudinal reinforcement ratio of main beam is changed
through changing the diameter of the bars. The cohesive element area is also
modified accordingly. The area of the upper bars is always equal to that of the
lower bars. As shown in Tables 7.3 and 7.4, the beam longitudinal
reinforcement ratio of one side is from 0.5% to 1.5%. The effect of beam
longitudinal reinforcement is insignificant for both exterior and interior joints.
7.3.7 Effect of Column Width
Only column width has been taken into consideration in this study, due to
limitations of the proposed models. The column section was assumed to be
square. Column widths of 300 to 600 mm were included in the analyses. As
shown in Tables 7.3 and 7.4, increasing column width has almost no effect for
both exterior and interior joints. A possible reason is that, the proposed models
assume the column to be elastic and sufficiently strong
7.3.8 Effect of Slab Size.
Page 470
439
The effects of flange widths of 1600 mm, 2000 mm, 3000 mm and
thickness of 80 mm, 100 mm 120 mm, 140 mm were studied and the effect of
flange thickness was found to be small. A possible reason is that the flange
thickness is relatively small compared to the beam height. The change of flange
thickness slightly changed the position of slab layer in the model. Increasing the
flange thickness slightly increases the effective slab width in exterior joints. As
shown in Tables 7.3, an increase in the flange width leads to a almost the same
predicted effective slab width of exterior joints, which indicates that the
effective slab width will keep constant if the flange width is wider than it. Thus
the effect of flange width was not included into the parametric study for interior
joints.
7.3.9 Effect of Slab Reinforcement Ratio
The slab reinforcement ratio effect was investigated by changing the slab
bar spacing as, in practice, changing slab bar spacing is more usual than
changing the diameter. As indicated in Tables 7.3 and 7.4, slab bar spacing has
a great effect on the predicted effective slab width. A decrease in spacing leads
to a decrease in effective slab width. This may be because increasing bar density
restricts the crack propagation in the transverse direction. Besides, the tensile
stress of slab concrete was not considered in the calculation when using
Equations 7.1 and 7.2.
7.3.10 Effect of Transverse Beam
Page 471
440
To study the effect of a transverse beam on predicted effective slab width,
the transverse beam height, width and also stirrups related to beam torsional
stiffness, were investigated. As indicated in Tables 7.3 and 7.4, transverse beam
width only, has an obvious effect on predicted effective slab width at exterior
joints. Increasing the transverse beam width increases predicted effective slab
width at exterior joints. For interior joints, the effect is marginal.
7.4 DISCUSSIONS
According to the FE modelling results, the key factors influencing the
effective slab width at exterior joints includes concrete strength, slab bar yield
stress, spacing and transverse beam width. For interior joints, the factors of
concrete strength, beam height, width, longitudinal bar yield stress, slab bar
yield stress and spacing are of great effect. As the column is assumed to be
elastic, its effects are insignificant. The effect of slab thickness is very limited,
even though it is included in many proposed equations for effective slab width
(French 1991, Li 1994, Jiang et al. 1994, Wu et al. 2002, Wang et al. 2009,
Yang 2010, Qi et al. 2010, He 2010, GB 50010-2010, ACI318-05 & FEMA-356,
EC8, NZS-3101:2006).
The predicted effective slab width at interior joints is greater than that for
exterior joints. This is confirmed by equations proposed in the literature
(Durrani and Zerbe 1990, Zhen et al. 2009, Sun et al. 2010, Qi et al. 2010) for
exterior and interior joints. The reason relates to this fact is that an interior joint
Page 472
441
slab under hogging moment is provided with additional restraint by the slab
subject to a sagging moment.
To generate a simple equation to calculate effective width, two equations
(7.3) and (7.4) were proposed for exterior joints and interior joints respectively.
Only the above key factors are included in the proposed equations. As shown in
Figure 7.4 and 7.5, approximately linear relationship exists between each of
these key factors and the effective flange width, with the only exception being
the flange width for exterior joints.
'
6000( 0.0011 0.783)
(0.0033 0.2755)(0.0013 0.1482)(0.002 0.1167)
eff w e sbys e cs e fbs e tbw
e sbys sy
e cs c
e fbs fr
e tbw tb
b bf
fd
w
(7.3)
where e sbys , e cs , e fbs , e tbw account for the effects of slab bar yield
stress syf , concrete strength 'cf , flange bar spacing frd and transverse beam
width tbw of exterior joints.
'
6000( 0.0021 1.4089)
(0.0087 0.4025)(0.0004 0.6245)(0.0013 0.5231)( 0.0003 0.8884)
(0.0025 0.2672)
eff w i sbys i cs i bh i bw i bbys i fbs
i sbys sy
i cs c
i bh b
i bw b
i bbys by
i fbs fb
b bf
fhw
fd
(7.4)
Page 473
442
where i srs , i cs , i bh , i bw , i fw , i frp account for the effects of slab bar
strength syf , concrete strength 'cf , beam height bh , beam width bw , beam bar
yield stress byf and flange bar spacing fbd at interior joints. The coefficients
accounting for the effects of key factors were calculated independently and their
inter-effects were not considered. This is similar to Nie and Tao (2012), who
generated a equation for calculating effective slab with of composite frame
system by using three coefficients 0 , tr and r to represent the effects
resulting from different key factors. In the equation for 0 , the contribution
from column dimension and steel beam height to 0 , were also independently
considered. Coefficients tr and r , which respectively presents the effect of
transverse beam and bar yield stress, were also independently considered.
As the column is assumed to be elastic, to accurately predict the effective
slab width, a weakening coefficient i c and e c for interior and exterior
joints should be included. i c and e c equal to 0.5 are suggested for interior
and exterior joints respectively.
7.5 CONCLUDING SUMMARY
Many researchers (Ehsani and Wigh 1982, T Pantazopotrlou et al. 1988,
Durrani and Zerbe 1990, French 1991, Li 1994, Jiang et al. 1994, Wu et al.
2002, Wang et al. 2009, Zhen et al. 2009, Yang 2010, Qi et al. 2010, He 2010)
have recognised the need to identify the effective slab width of an RC frame and
Page 474
443
numerous experimental studies have been conducted. Parametric studies, based
on the experimental approach, have been carried out but these studies have been
limited to study the effect of a few significant factors, although various different
equations have been proposed in the codes of different countries (GB
50010-2010, ACI318-05 & FEMA-356, EC8, NZS-3101:2006). Thus,
parametric studies covering a wide range factors have been conducted in this
chapter using the 3D 1-shell model proposed in Chapter 6.
Based on the FE modelling results and subsequent discussions presented
above, the following conclusions can be drawn:
(1) The key factors affecting the predicted effective slab width at both exterior
and interior joints include concrete strength, slab bar spacing, yield stress of
steel.
(2) The transverse beam width only has an obvious effect on the effective slab
width for an exterior joint. The effects of beam bar yield stress, beam height
and width are more obvious for an interior joint than for an exterior joint.
(3) The effective slab width for an interior joint is much wider than that for an
exterior joint with the same properties.
(4) Based on numerical results, two simple equations for predicting the effective
slab width were proposed for interior and exterior joints respectively. Only
key factors are included in the equations. A weakening coefficient
Page 475
444
accounting for effect of column should be included to more accurately
predict the slab effective width.
7.6 REFERENCES
ACI 318 (2005). Building Code Requirements for Structural Concrete and
Commentary (ACI 318-05), ACI Committee 318, American Concrete
Institute, Farmington Hills, MI.
Durrani, A.J. and Zerbe, H.E. (1987). “Seismic resistance of R/C exterior
connections with floor slab”, Journal of Structural Engineering, ASCE,
113(8), 1850-1864.
Ehsani, M.R. and Wight, J.K. (1982). Behavior of External Reinforced Concrete
Beam to Column Connections Subjected to Earthquake Type Loading, Univ.
of Michigan, Ann Arbor, Mich.
Eurocode 8 (2004). Design of Structures for Earthquake Resistance – Part 1:
General Rules, Seismic Actions and Rules for Buildings (EN 1998-1: 2004),
CEN, Brussels.
French, C.W. (1991). Effect of Floor Slab on Behavior of Slab-Beam-Column
Connections, Design of Beam-Column Joints for Seismic Resistance.
GB-50011 (2010). Code for Seismic Design of Buildings, Architectural &
Building Press, Beijing, China. (in Chinese)
Page 476
445
He, J. (2010). Study on Influence of Cast-in-situ Slab on RC Frame Structure
Achieving “Strong Column Weak Beam”, Master degree thesis: Huazhong
University of Science and Technology, China. (in Chinese)
Jiang, Y.S., Chen, Z.F., Zhou, X.P. et al. (1994). “Seismic studies of frame
joints with cast-in-place slab”, Journal of Building Structures, 15(6), 11-16.
(in Chinese)
Nie, J. G., and Tao, M. X. (2012),"Slab spatial composite effect in composite
frame systems. I: Effective width for ultimate loading capacity",
Engineering structures, 38, 171-184.
Ning, N., Qu, W., and Ma, Z. J. (2016), "Design recommendations for achieving
“strong column-weak beam” in RC frames", Engineering Structures, 126,
343-352.
NZS-3101 (2006). Concrete Structures Standard, Standards New Zealand,
Wellington, New Zealand.
Qi, C.H. et al. (2010). “Effect of floor slabs on the mechanical properties of
reinforced concrete frame structures”, Journal of Catastrophology, 25(S0),
105-110. (in Chinese)
Sun, Y. (2010). The effect of Slabs on Strong Column-Weak Beam Mechanism
of RC Frame Structures, Master degree thesis: Harbin Institute of
Page 477
446
Technology, China. (in Chinese)
Teng, J. G., Chen, J. F., Smith, S. T., and Lam, L. (2002), FRP: strengthened
RC structures. Frontiers in Physics, 266.
Wang, S.G., Hang, X.L. and Ji, J. (2009). “The effect of slabs on the failure
mode of reinforced concrete frame structures”, Journal of Civil,
Architectural & Environmental Engineering, 31(1), 66-71.
Wu, Y., Lei, J.C., Y, H. et al. (2002). “Discussion on the problem that slab
reinforcement getting involved in the negative moment capacity of beam
end”, Journal of Chongqing Jianzhu University, 24(3), 33-37. (in Chinese)
Yang, Z.L. (2010). Research on “Strong Column Weak Beam” Yield
Mechanism Factors of Reinforced Concrete Frame Structure, Master
degree thesis: Qingdao Technological University, China. (in Chinese)
Zhen, S.J., Jiang, L.X., Zhang, W.P. and Gu, X.L. (2009). “Experimental
research and analysis of effective flange width of beam end section in
cast-in-site concrete frames”, Structural Engineers, 25(2), 134-140. (in
Chinese)
Page 478
447
Figure 7.1 Frame and joints of the FE models
(a) Exterior joint
Symmetric section
Load
Symmetric section
Constraints of all
DOFs
Page 479
448
(b) Interior joint
Figure 7.2 Control FE models of exterior and interior joints
Figure 7.3 Two stress-strain models for steel bars
Load
Load
Load
Symmetric section
Constraints of all
DOFs
Symmetric section
Page 480
449
(a) Effect of slab bar yield stress
(b) Effect of concrete strength
y = -0.0011x + 0.783
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
0 100 200 300 400 500
Eff
ectiv
e sl
ab w
idth
/ Pra
tical
wid
th
Yield strength (MPa)
Slab rebar yield stress
线性 (Slab rebar yield stress)
y = 0.0033x + 0.2755
0.0%
5.0%
10.0%
15.0%
20.0%
25.0%
30.0%
35.0%
40.0%
45.0%
50.0%
0 10 20 30 40 50 60 70
Eff
ectiv
e sl
ab w
idth
/ Pra
tical
wid
th
Concrete strength (MPa)
Concrete strength
线性 (Concrete strength)
Page 481
450
(d) Effect of flange bar spacing
(e) Effect of transverse beam width
Figure 7.4 Key factors effecting the predicted slab width of exterior joints
y = 0.0013x + 0.1482
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
0 50 100 150 200 250 300 350
Eff
ectiv
e sl
ab w
idth
/ Pra
tical
wid
th
Flange rebar distance (mm)
Flange rebar distance
线性 (Flange rebar distance)
y = 0.002x + 0.1167
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
0 50 100 150 200 250
Eff
ectiv
e sl
ab w
idth
/ Pra
tical
wid
th
Transverse beam width (mm)
Transverse beam width
线性 (Transverse beam width)
Page 482
451
(a) Effect of beam bar yield stress
(b) Effect of slab bar yield stress
y = -0.0003x + 0.8884
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
0 100 200 300 400 500 600 700
Eff
ectiv
e fla
nge
wid
th/ P
ratic
al w
idth
Yield strength (MPa)
Beam bar yield stress
线性 (Beam bar yield stress)
y = -0.0021x + 1.4089
0.0%
20.0%
40.0%
60.0%
80.0%
100.0%
120.0%
0 100 200 300 400 500
Eff
ectiv
e fla
nge
wid
th/ P
ratic
al w
idth
Yield strength (MPa)
Slab rebar yield stress
线性 (Slab rebar yield stress)
Page 483
452
(c) Effect of concrete strength
(d) Effect of beam height
y = 0.0087x + 0.4025
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
100.0%
0 10 20 30 40 50 60 70
Eff
ectiv
e fla
nge
wid
th/ P
ratic
al w
idth
Concrete strength (MPa)
Concrete stength
线性 (Concrete stength)
y = 0.0004x + 0.6245
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
100.0%
0 100 200 300 400 500 600 700 800 900
Eff
ectiv
e fla
nge
wid
th/ P
ratic
al w
idth
Beam height (mm)
Beam height
线性 (Beam height)
Page 484
453
(e) Effect of beam width
(f) Effect of flange bar spacing
Figure 7.5 Key factors affecting the predicted slab width of interior joints
y = 0.0013x + 0.5231
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
100.0%
0 50 100 150 200 250 300 350
Eff
ectiv
e fla
nge
wid
th/ P
ratic
al w
idth
Beam width (mm)
Beam width
线性 (Beam width)
y = 0.0025x + 0.2762
0.0%
20.0%
40.0%
60.0%
80.0%
100.0%
120.0%
0 50 100 150 200 250 300 350
Eff
ectiv
e fla
nge
wid
th/ P
ratic
al w
idth
Flange rebar spacing (mm)
Flange rebar spacing
线性 (Flange rebar spacing)
Page 485
454
Table 7.1 Some existing parametric studies on the effective slab width
Time Researchers software Parameters Model Equation used to
calculate effb Proposed effb
2016 Ning et al. Abaqus Axial compression ratio, concrete strength,
reinforcement ratio of slabs, thickness of the slabs and the stiffness of the transverse beams
3D solid element models 1
max
n
bi bi
eff wb
sb b
at 2% drift ratio
6.4eff w bb b h
2012 Nie and Tao MSC.Marc
(2005)
Column dimension, steel beam height, RC slab width and thickness, transverse beam flange
width, and yielding strength of the longitudinal reinforcement
3D nonlinear shell-solid elaborate finite element model
1
sy
n
bi bi
eff w
sb b
f
at 2% drift ratio
eff w fb b b
Note: beff: effective slab width; bw: beam width; hb: beam height; bi : the i slab bar stress; maxb : max stress of slab bars; bs : distance
between slab bars; syf : yield stress of slab bars; : coefficient; fb : the flange width.
Page 486
455
Table 7.2 Parameters investigated in studying slab effective width Parameters Values/considerations
Stress-strain curves Elastic-perfect plastic or nonlinear
stress-strain curve considering hardening range
Bond-slip effect With or without including the
bond-slip relationship Yielding stress of bars fy (MPa) 200 300 400 500 600
Concrete strength fc (MPa) 20 30 40 50 60 Column width (mm) 300 400 500 600
Beam
Length (mm) 2000 3000 4000 5000 6000 Width (mm) 100 150 200 250,300 Height (mm) 300 400 500 600 700 800 Longitudinal
reinforcement ratio (%) 0.5 0.75 1 1.25 1.5 ( through changing
area)
Flange
Width (mm) 2000 2500 3000 Thickness (mm) 80 100 120 140
Longitudinal reinforcement distance
(mm) 100 160 200 240 300
Transverse beam
Height (mm) 200 300 400
Width (mm) 100 150 200
Stirrup spacing (mm) 60 100 200
Page 487
456
Table 7.3 Predicted effective flange width of exterior joints
Model name Parameter Parameter values Sectional strength (Kn·m)
Calculated flange
width of one side
(mm)
Calculated flange width of one side/flange
real width
E-C Stress-strain
curves
Elastic-perfect plastic model
226.5 1283 42.8%
E SSC-N Nonlinear stress-strain
model considering hardening range
233.4 1362 45.4%
E-BS-N
Bond-slip
Tied to concrete node 230.1 1324 44.1%
E-C Bonded to concrete node
through cohesive element
226.5 1283 42.8%
E-BBYS-300 Beam bar yielding
stress
300 201.9 1294 43.1% E-C 400 226.5 1283 42.8%
E-BBYS-500 500 252.0 1282 42.7% E-BBYS-600 600 272.6 1225 40.8% E-SBYS-200 Slab bars
yielding stress
200 212.7 1692 56.4% E-SBYS-300 300 226.5 1283 42.8% E-SBYS-400 400 232.3 1012 33.7%
E-CS-20 Concrete strength (MPa)
20 190.7 953 31.8% E-CS-30 30 212.0 1153 38.4%
E-C 40 226.5 1283 42.8% E-CS-50 50 238.0 1385 46.2% E-CS-60 60 234.6 1330 44.3%
E-BH-300
Beam height (mm)
300 157.6 1313 43.8% E-C 400 226.5 1283 42.8%
E-BH-500 500 298.1 1291 43.0% E-BH-600 600 372.4 1314 43.8% E-BH-700 700 444.2 1315 43.8% E-BH-800 800 517.4 1322 44.1%
Page 488
457
Table 7.3 ( Cont.)
E-BL-1000
Beam length (mm)
1000 224.4 1259 42.0% E-BL-1500 1500 224.2 1258 41.9%
E-C 2000 226.5 1283 42.8% E-BL-2500 2500 225.7 1274 42.5% E-BL-3000 3000 226.9 1287 42.9% E-BW-100
Beam width(mm)
100 202.9 1113 37.1% E-BW-150 150 216.4 1206 40.2%
E-C 200 226.5 1283 42.8% E-BW-250 250 230.3 1300 43.3% E-BW-300 300 232.1 1303 43.4%
E-BTRR-0.5 Beam tension reinforcement
ratio (%)
0.5 171.5 1240 41.3% E-BTRR-0.75 0.75 198.0 1250 41.7%
E-C 1 226.5 1283 42.8% E-BTRR-1.25 1.25 249.5 1253 41.8% E-BTRR-1.5 1.5 278.0 1286 42.9% E-CW-300
Column width (mm)
300 223.5 1249 41.6% E-C 400 226.5 1283 42.8%
E-CW-500 500 231.0 1334 44.5% E-CW-600 600 231.5 1340 44.7% E-FW-1500
Flange width (mm)
1600 222.7 1240 77.50% E-FW-2000 2000 227.3 1292 64.60%
E-C 3000 226.5 1283 42.8% E-FRD-100
Flange bar distance(mm)
100 259.7 836 27.9% E-FRD-160 160 234.1 1096 36.5%
E-C 200 226.5 1283 42.8% E-FRD-240 240 216.5 1404 46.8% E-FRD-300 300 210.3 1652 55.1%
E-FH-80 Flange height
(mm)
80 226.4 1240 41.3% E-C 100 226.5 1283 42.8%
E-FH-120 120 225.1 1312 43.7% E-FH-140 140 225.1 1361 45.4%
E-TBH-200 Transverse beam height
(mm)
200 217.9 1186 39.5% E-C 300 226.5 1283 42.8%
E-TBH-400 400 236.6 1399 46.6% E-TBW-100 Transverse 100 196.1 946 31.5%
Page 489
458
E-C beam width (mm)
150 226.5 1283 42.8% E-TBW-200 200 249.8 1553 51.8%
Table 7.4 Predicted effective flange width of interior joints
Model
name Parameter Parameter values
Sectional
strength
(Kn*m)
Calculated
flange
width of
one side
(mm)
Calculated
flange width
of one
side/flange
real width
I-C
Stress-strain
curves
Elastic-perfect plastic
model 307.0 2274 75.8%
I SSC-N
Nonlinear stress-strain
model considering
hardening range
311.2 2331 77.7%
I-BS-N
Bond-slip
Tied to concrete node 348.3 2860 95.3%
I-C
Bonded to concrete node
through cohesive
element
307.0 2274 75.8%
I-BBYS-300
Beam bar
yielding stress
300 287.9 2362 78.7%
I-C 400 307.0 2274 75.8%
I-BBYS-500 500 325.3 2177 72.6%
I-BBYS-600 600 342.3 2064 68.8%
I-SBYS-200 Slab bars
yielding stress
200 303.8 3347 100.0%
I-SBYS-300 300 307.0 2274 75.8%
I-SBYS-400 400 310.5 1741 58.0%
I-CS-20
Concrete
strength
(MPa)
20 241.7 1683 56.1%
I-CS-30 30 280.4 2047 68.2%
I-C 40 307.0 2274 75.8%
I-CS-50 50 333.8 2525 84.2%
I-CS-60 60 358.2 2756 91.9%
I-BH-300
Beam height
(mm)
300 203.5 2208 73.6%
I-C 400 307.0 2274 75.8%
I-BH-500 500 430.8 2500 83.3%
I-BH-600 600 555.5 2632 87.7%
I-BH-700 700 673.2 2672 89.1%
I-BH-800 800 799.9 2747 91.6%
Page 490
459
Table 7.4(cont.)
I-BW-100
Beam width(mm)
100 254.0 1908 63.6% I-BW-150 150 293.9 2253 75.1%
I-C 200 307.0 2274 75.8% I-BW-250 250 337.5 2570 85.7% I-BW-300 300 355.6 2719 90.6%
I-BTRR-0.5 Beam tension reinforcement
ratio (%)
0.5 260.9 2343 78.1% I-BTRR-0.75 0.75 284.9 2321 77.4%
I-C 1 307.0 2274 75.8% I-BTRR-1.25 1.25 329.3 2230 74.3% I-BTRR-1.5 1.5 359.3 2289 76.3% I-CW-300
Column width (mm)
300 309.5 2301 76.7% I-C 400 307.0 2274 75.8%
I-CW-500 500 308.1 2289 76.3% I-CW-600 600 313.2 2358 78.6% I-FRD-100
Flange bar distance(mm)
100 359.1 1513 50.4% I-FRD-160 160 331.9 2095 69.8%
I-C 200 307.0 2274 75.8% I-FRD-240 240 304.1 2682 89.4% I-FRD-300 300 292.4 3123 100.0%
I-FH-80 Flange height
(mm)
80 303.0 2135 71.2% I-C 100 307.0 2274 75.8%
I-FH-120 120 308.0 2287 76.2% I-FH-140 140 308.2 2290 76.3%
I-TBH-200 Transverse beam height
(mm)
200 324.2 2510 83.7% I-C 300 307.0 2274 75.8%
I-TBH-400 400 310.8 2325 77.5% I-TBW-100 Transverse
beam width (mm)
100 319.8 2449 81.6% I-C 150 307.0 2274 75.8%
I-TBW-200 200 322.8 2490 83.0%
Page 492
461
CHAPTER 8
THREE-DIMENSIONAL FINITE ELEMENT
MODELLING OF RETROFITTED RC
BEAM-COLUMN-SLAB JOINTS
8.1 INTRODUCTION
An experimental study of RC beam-column joints, retrofitted by the
proposed techniques has been presented in Chapters 3 and 4. The 3D FE
analysis of T-beams with a web opening has been presented in Chapter 5. The
FE modelling of RC beam-column joints has previously been conducted by a
few researchers. For example, Li et al.(2009) used DIANA software to model
test specimens and carried out parametric studies. The FE modelling of RC
beam-column joints weakened by openings however, has not been previously
studied. The effect of slits on RC joints and frames has been studied through FE
modelling by Zhang et al. (2011), Wang el al. (2012) and Zhang (2013). The
investigations were based, simply, on FE modelling not substantiated by
experimental data. It is noted that RC Joints with transverse grooves (TGs) had
not as yet, been examined either experimentally or by FE modelling. Thus, to
fill the gap in the existing knowledge on FE analysis of RC beam-column joints
with weakened beams, 3D FE modelling of RC beam-column joints retrofitted
Page 493
462
by the proposed techniques has been investigated by this study and will be
presented in this chapter.
8.2 THE PROPOSED 3D FE MODELS
The proposed FE models for retrofitted RC beam-column joints are based
on the T-beam models using solid elements described in Chapter 5. Similar to
the 3D models in Chapter 5, the element size chosen for the RC joint models
was 20 mm. The concrete was modelled by the 3D solid elements C3D8R in
ABAQUS (2012) and the constitutive equation proposed by Saenz (1964) for
the state of uni-axial compression was adopted. p and Ec were determined
based on test results. If test results were not available, they were set equal to
0.002 and 4730 'cf respectively as in Chapter 5. In terms of tension-softening
curves, the constitutive equation curve proposed by Hordijk (1991) was adopted.
For tension damage, the power low model used in Nie (2018), presented in
Equation 8.1 was adopted.
1 (1 )ntt
cr
wdw
(8.1)
where n is equal to 5 following Chen et al. (2012); is crack opening
displacement; is crack opening displacement at the complete release of
concrete stress.
According to the FE analysis of T-beams with weakening openings and
Page 494
463
FRP strengthening described in Chapter 5, the FRP confinement effect should
not be ignored. Thus, Yu el al.'s (2009a, b) modified Plastic-damage model was
adopted for the concrete of the web chord, which was strengthened by FRP
wrapping.
As concrete in the joint region concrete was reinforced by the steel stirrups,
the stress-strain model for confined concrete proposed by Mander et al. (1988),
as expressed in Equation (8.2), was employed for the concrete in the joint
region.
'
ccc r
f xrfr -1+ x
(8.2)
where 'ccf is compressive strength of confined concrete; cc is the compressive
strain of concrete at ultimate state; c
cc
x
in which c is compressive
concrete strain. The cc and r can be obtained using the following
equations:
'
'1 5 1cccc co
co
ff
(8.3)
c
c sec
ErE E
(8.4)
in which and are unconfined concrete strength and corresponding 'cof co
Page 495
464
strain, respectively; is equal to 0.002 if test data is not available, cE is
equal to '4730 cof and secE is equal to '
cc
cc
f
.
The steel bars were modelled with the B31 beam elements and an
elastic-perfect plastic model was applied to represent their properties if a yield
platform was available, as in Chapter 5. Otherwise, the model proposed by
Ramberg Osgood (1943) was applied. The bond-slip behaviour between steel
bar and concrete was represented by COH3D8 cohesive elements and the
CEB-FIP(1993) bond-slip model was used. The 3D prism cohesive model
proposed in Chapter 5 was used to simulate bond behaviour between steel bars
and concrete.
FRP used to wrap the concrete chord was modelled with the S4R 4-node
shell elements and treated as a linear elastic brittle material with elastic modulus
of 227.380 GPa based on test material properties. Only in the fibre direction was
defined with stiffness. The Poisson's ratio of the FRP was set as 0.001. The FRP
reinforcement fails when it reaches its tensile rupture strength of 2820 MPa,
obtained from coupons test. The simplified bond-slip model Lu et al. (2005)
was adopted to define the bond behavior of FRP and concrete.
The dynamic explicit approach was used to avoid convergence problems,
as for the FE models discussed in Chapters 5 and 6. When building the 3D
model, the densities of materials were set equal to the actual values: 2.5e-9
co
Page 496
465
ton/mm3 for concrete and cohesive elements, 7.25e-9 ton/mm3 for steel
elements and 1.75e-9 ton/mm3 for FRP elements. The stiffness-proportional
damping coefficient β was set equal to 0.000002 based on the parametric study
shown in Chapter 5.
The boundary conditions are shown in Figure 8.1. To save computational
time, only half of a specimen was modelled. Nodes on the section of symmetry
were permitted no displacement in the X direction. Coupling constraints were
applied to the five reference points (referred to as RPs below).
As in experimental test, axial load was applied first and then lateral load ,
in FE modelling, Loads were applied in two steps. In the first step, an axial load
equal to 485 kN was applied to the column top through RP 5. In this step, for
the first test group, the linear displacements of RPs 1-3 were restrained in the X ,
Y and Z directions. For the second test group, Y displacements of the RPs 1 and
3 were not restrained as, during testing, the beam tip loads of the second test
group were released to zero after applying the axial load. In the second test
group, a horizontal load was applied through RP4, until the specimens failed.
During this modelling procedure, the RPs1-3 restraints were kept unchanged for
the first test group. For the second test group, restraints on the vertical
displacements of RPs 1 and 3 were provided. The loading time of the first step
was 0.1s and 50T0 (T0 is the natural period of a 3D model) for the second step.
8.3 PREDICTED RESULTS
Page 497
466
8.3.1 Load-displacement Curves
The predicted curves of lateral loads, beam tip loads, and initial stiffness
values versus lateral drift ratios or displacements are shown in Figures 8.2 to 8.4.
In these figures, the drift ratio values are defined by dividing lateral
displacement by storey height (2400 mm). The peak lateral loads are given in
Table 8.1. The envelope curves are used for comparison herein, as the
specimens were tested under cyclic loading. In Figure 8.2, the 3D models
predict satisfactory results for all specimens except for the two with TGs
(F-G-50-200 and S-G-50-200-100). For the specimens F-G-50-200 and
S-G-50-200-100, the FE model predicts slightly higher post-cracking stiffness
and peak lateral load. The descending portion after peak load is not well
captured either. This can be attributed to the severe slippage of the steel bars in
beam bottom due to the existence of transverse grooves (placed close to or 100
mm away from the beam-column interface) subjected to cyclic loading. As the
3D model was subjected to a monotonic load, the cyclic loading effect was not
simulated. The FE model predicts a slightly lower peak lateral load for
specimen F-S-450-450, but the descending portion is well captured. For
specimens F-S-450-450, S-O-500-180 and S-O-500-180-S-300-300, the FE
models predict slightly lower peak lateral loads.
In Figure 8.3, curves of the left and right beam tip loads versus lateral drift
ratio are presented. The left and right beams are under respective negative and
positive loading. It is seen that the positive beam tip load have all been well
Page 498
467
predicted, whereas, the FE model predicts negative loads higher than test values
for F-G-50-200, S-Control and S-G-50-200. The beam tip loads after peak
points are generally overestimated by the 3D FE models, which might because
that the FE models do not consider the cumulative damage arising from cyclic
loading. The residual deformations of beams, columns and joints are thus
underestimated at a given load. This phenomenon is more obvious for those
specimens did not fail by beam flexural failure.
In Figure 8.4, the first cycles of test load-displacement curves are shown
for to substantiate the proposed models. The FE model predicts a slightly higher
initial stiffness than the test value. The specimen S-450-450 showed a 2mm
initial displacement, which is probably due to ‘mechanical slack’, tiny gaps
existing between testing equipment components.
8.3.2 Crack Patterns
The comparisons between numerical and test crack patterns and test cracks
are shown in Figures 8.5 and 8.6. Test observed crack propagation at the
ultimate state when specimens were subject to push action, are highlighted in
black in Figure 8.5 as a reference to verify the accuracy of the FE models in
predicting crack patterns. As shown in Figure 8.6(a), the cracks in the slab of
the specimen F-Control are well captured. Main cracks in the joints region, are
also correctly predicted. For specimens F-G-50-200 and S-G-50-200-100, the
predicted cracks caused by beam bar bars slipping in the joints region closely
Page 499
468
match with the test results, even though those test cracks were more severe as
cyclic loading was applied. For specimen F-S-450-450, the slab and beam
flexural cracks are also precisely predicted. For specimens F-O-450-150 and
F-O-500-180, the slab cracks caused by the four hinges mechanism (i.e. local
flexural rotation at the two ends of web and flange chords) are also well
captured. Specimen S-Control failed due to hinge formation in bottom column,
which is also predicted by FE model as shown in Figure 8.6(f). The cracks on
the top column, beams and slab are also well predicted. For specimen
S-O-500-180, the cracks in the beam web above the opening are well predicted.
The cracks arising from the anchor within the right hand side slab are also
predicted. For specimen S-O-500-180-S-300-300, the cracks originated from the
ends of the longitudinal slits within the right hand side slab are well captured
with, the same can be said for those cracks caused by four hinges mechanism.
8.4 DISCUSSIONS
8.4.1 Confinement Effect of Joint Steel Stirrups
The comparison between the models with and without a consideration of
the joint stirrups confinement effect is shown in Figure 8.7. As indicated in the
Figure 8.7, the differences between the two models for the first test group are
smaller than for the second, as heavier joint stirrups were provided for the
second test group. The difference became more obvious with increasing
displacement, this is because the joint deformations became more obvious in the
Page 500
469
later range. Stirrups confinement had more obvious effect on the specimens
F-Control, S-Control as the joint deformations were more severe in these
specimens. In general, it can be seen that the inclusion of the confinement effect
for joint concrete results in more accurate predicted peak loads, except for the
two specimens with TGs.
8.4.2 Effect of Opening Shape
A slot-shaped opening has been tested on Specimens S-O-500-180 and
S-O-500-180-S-300-300 and this opening shape led to the four hinges type
deformation in its weakening T-beam as the rectangular one did on the beams of
Specimen F-O-500-180. To get a in-depth comparison between the two types of
openings, a 3D FE model S-O-500-180-R with rectangular-shaped beam
openings was built and studied. The FE mode S-O-500-180-R has the same
properties as the FE model S-O-800-180 except for the opening shape. The
predicted load-displacement curves are presented in Figure 8.8. It is obvious
that the two types of beam opening predicted almost the same
load-displacement curves. The predicted crack patterns by FE model
S-O-500-180-R is shown in Figure 8.9. The comparison between Figures 8.9
and 8.6(h) indicates that the two 3D models also predicted very close crack
patterns, except for some insignificant cracks (e.g. the cracks on the beam web
above the slot-shaped opening). Thus slot-shaped opening is a better option for
the BO technique as it removes less beam web concrete as the rectangular one
Page 501
470
does.
8.4.3 Effect of Longitudinal Slit Length
When applying the SS technique to retrofit RC beam-column joints
F-S-450-450 and S-O-500-180-S-300-300, both longitudinal and transverse slits
were created. The effect of the transverse slits is noticeable as the longitudinal
slab bars crossing the slits were cut. The effects of longitudinal slits on the test
specimens were not clear. Thus two additional 3D FE models F-S-450-200 and
F-S-450-0 were built to study the effect of longitudinal slit length. Similar to FE
model F-S-450-450, both the models F-S-450-200 and F-S-450-0 had 450 mm
long transverse slits, while the model F-450-200 had 200 mm long longitudinal
slits and model F-S-450-0 had no longitudinal slit. The predicted
load-displacement curves of the FE models with longitudinal slit of different
size were presented in Figure 8.10. It is obvious that the length of longitudinal
slits has marginal effect on the predicted load-displacement curves. The model
with shorter longitudinal slits predicted a slight higher peak load. The three FE
models with longitudinal slits different in length also predicted very close crack
patterns as shown in Figure 8.10(a), (b) and Figure 8.6(e). The reason
accounting for above observation might be that the cut transverse slab bars
crossing the longitudinal slits had little effect on the moment capacity of T-beam
along the longitudinal direction. Both transverse and longitudinal slits are
created when applying the SS technique as in a real situation, a seismic load
may come from any direction.
Page 502
471
8.4.4 Prediction Errors
As shown in Table 8.1, it is clear that the predicted lateral loads are more
accurate than the predicted beam tip loads. This is because the predicted peak
lateral loads happened at a 2% drift ratio, but the predicted peak beam tip loads
usually occurred at a later drift ratio as demonstrated in Figures 8.3 and 8.8.
This might be because the cumulative damage due to the cyclic loading is not
simulated in the FE analysis, which can also be the reason why the FE model’s
predictions on the specimens with TGs are not as good as those without TGs.
Also in these cases, the mechanical slack in the testing equipment may play a
role in the difference. The predicted positive beam tip loads are closer to test
results than the negative loads. This may be because some specimens failing by
joint shear failure, column hinge formation or beam bottom bar slippage, so that
these beams did not reach their negative strengths during testing. Besides, the
retrofitted techniques have more effect on beam’s negative strength, as observed
and discussed in Chapter 4. Cyclic loading might further increase this effect. As
the FE models do not simulate the cumulative damage due to cyclic loading,
higher negative loads are predicted, as clearly demonstrated in Figure 8.3.
8.5 CONCLUDING SUMMARY
3D FE models for RC joints retrofitted by the proposed techniques have
proposed and assessed in this Chapter. The following conclusions can be drawn
based on the predicted results and discussions:
Page 503
472
(1) The proposed 3D FE models perform well in predicting load-displacement
curves and crack patterns of retrofitted RC joints. The predictions for
specimens with transverse grooves are not as good as the others because the
beam bottom bars slippage was more noticeable under cyclic loading.
(2) The predicted initial stiffness is slightly greater than in the test value, which
can be attributed to the mechanical slack of test equipment arising from the
small gap between the mechanical components and the corresponding test
equipment.
(3) To obtain more reliable predictions, the joint stirrup confinement effect
should be considered in the 3D FE analysis. The model proposed by Mander
et al. (1988) can be applied. The consideration of the joint stirrup
confinement effect has more effect in the specimens F-Control and
S-Control, which failed by joint shear failure and bottom column hinge
formation, respectively.
(4) The ignorance of cyclic loading effect leads to a slightly higher predicted
negative beam tip loads. The main reason is that cumulative damage caused
by cyclic loading during testing was not simulated in FE analysis.
8.6 REFERENCES
ABAQUS (2012). ABAQUS Analysis User's Manual (Version 6.12), Dassault
Systems SIMULIA Corporation, Providence, Rhode Island, USA.
Page 504
473
CEB-FIP. (1993). CEB-FIP Model Code 90, Thomas Telford, London.
Chen, G.M., Chen, J.F. and Teng, J.G. (2012). “On the finite element modelling
of RC beams shear-strengthened with FRP”, Construction and Building
Materials, 32, 13-26.
Hordijk, D.A. (1991). Local Approach to Fatigue of Concrete, PhD thesis, Delft
University of Technology.
Lu, X.Z., J.G. Teng, Ye, L.P. and Jiang, J.J. (2005). “Bond-slip models for FRP
sheets /plates bonded to concrete”, Engineering Structures, 27(6), 920-937.
Li, B., Tran, C. T. N., and Pan, T. C. (2009)." Experimental and numerical
investigations on the seismic behavior of lightly reinforced concrete
beam-column joints", Journal of structural engineering, 135(9), 1007-1018.
Mander, J. B., Priestley, M. J., and Park, R. (1988). "Theoretical stress-strain
model for confined concrete", Journal of structural engineering, 114(8),
1804-1826.
Mansur, M.A., Tan K.H. and Wei, W. (1999). “Effects of creating an opening in
existing beams”, ACI Structural Journal, 96(6), 899-906.
Nie, X.F. (2018). Behavior and Modeling OF RC Beams with an
FRP-Strengthened Web Opening, Doctoral degree thesis: The Hong Kong
Polytechnic University.
Page 505
474
Ramberg, W., and Osgood, W. R. (1943)," Description of stress-strain curves by
three parameters".
Rots, J. G. (1988). Computational Modeling of Concrete Fracture, Ph.D. thesis,
Delft University of Technology.
Saenz, L. P. (1964). “Discussion of ‘Equation for the stress-strain curve of
concrete’ by P. Desayi and S. Krishan”, ACI Journal, 61(9), 1229-1235.
Wang, X.G., Shan, M.Y., Ge, N. and Shu, Y.P. (2012). “Finite element analysis
of efficiency of slot-cutting around RC frame joint for ‘strong column and
weak beam’”, Journal of Earthquake Engineering and Engineering
Vibration, 32(1), 121-127. (in Chinese)
Yu, T. T. J. G., Teng, J. G., Wong, Y. L., and Dong, S. L. (2010a), "Finite
element modeling of confined concrete-I: Drucker–Prager type plasticity
model", Engineering Structures, 32(3), 665-679.
Yu, T., Teng, J. G., Wong, Y. L., and Dong, S. L. (2010b). "Finite element
modeling of confined concrete-II: Plastic-damage model", Engineering
structures, 32(3), 680-691.
Zhang, J. (2013). Research on Efficiency of Slot-cutting around Frame Joint for
“Strong Column and Weak Beam" under Earthquake Action, Master degree
Thesis: Hunan University, China. (in Chinese)
Page 506
475
Zhang, Y.P., Hao, Z.J., Shan, M.Y. and Ge, N. (2011). “Research on
anti-seismic performance for reinforced concrete frame joint with slot
around”, Building Science, 27(9), 7-11. (in Chinese)
Page 507
476
Figure 8.1 3D FE Model Control specimen
(a) F-Control
0 1 2 3 4 5 60
20
40
60
80
100
120
140
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-Control-Test F-Control-FE
Coupling restraint Coupling restraint
Horizontal Load
Axial Load
RP5
RP4
RP1 RP2 RP3
Coupling restraint
Page 508
477
(b) F-G-50-200
(c) F-S-450-450
0 1 2 30
20
40
60
80
100
120
Lateral Drift(%)
Lat
eral
Loa
d (k
N)
0 24 48 72
Lateral Displacement (mm)
F-G-50-200-Test F-G-50-200-FE
0 1 2 3 4 5 60
20
40
60
80
100
120
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-S-450-450-Test F-S-450-450-FE
Page 509
478
(d) F-O-450-150
(e) F-O-500-180
0 1 2 3 4 5 60
20
40
60
80
100
120
140
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-O-450-150 Test F-O-450-150-FE
0 1 2 3 4 5 60
20
40
60
80
100
120
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-O-500-180 Test F-O-500-180-FE
Page 510
479
(f) S-Control
(g) S-G-50-200-100
0 1 2 3 4 50
20
40
60
80
100
120
140
160
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120
Lateral Displacement (mm)
S-Control-Test S-Control-FE
0 1 2 3 40
20
40
60
80
100
120
140
Lateral Drift(%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 Lateral Displacement (mm)
S-G-50-200-100-Test S-G-50-200-100-FE
Page 511
480
(h) S-O-500-180
(i) S-O-500-180-S-300-300
Figure 8.2 Predicted lateral load versus test result
0 1 2 3 4 5 60
20
40
60
80
100
120
140
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
S-O-500-180-Test S-O-500-180-FE
0 1 2 3 4 5 60
20
40
60
80
100
120
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
S-O-500-180-S-300-300-Test S-O-500-180-S-300-300-FE
Page 512
481
(a) F-Control
(b) F-G-50-200
0 1 2 3 4 5 6-200
-150
-100
-50
0
50
100
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-Control-Test F-Control-FE
0 1 2 3-150
-100
-50
0
50
100
Lateral Drift(%)
Bea
m T
ip L
oad
(kN
)
0 24 48 72
Lateral Displacement (mm)
F-G-50-200-Test F-G-50-200-FE
Page 513
482
(c) F-S-450-450
(d) F-O-450-150
0 1 2 3 4 5 6-150
-100
-50
0
50
100
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-S-450-450-Test F-S-450-450-FE
0 1 2 3 4 5 6-200
-150
-100
-50
0
50
100
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-O-450-150-Test F-O-450-150-FE
Page 514
483
(e) F-O-500-180
(f) S-Control
0 1 2 3 4 5 6-150
-100
-50
0
50
100
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-O-500-180-Test F-O-500-180-FE
0 1 2 3 4 5 6-200
-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
0 24 48 72 96 120 144
Lateral Displacement (mm)
S-Control-Test S-Control-FE
Page 515
484
(g) S-G-50-200-100
(h) S-O-500-180
0 1 2 3-200
-150
-100
-50
0
50
100
150
Lateral Drift(%)
Bea
m T
ip L
oad
(kN
)
0 24 48 72
Lateral Displacement (mm)
S-G-50-100-200-Test S-O-50-100-200-FE
0 1 2 3 4 5 6-200
-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
0 24 48 72 96 120 144
Lateral Displacement (mm)
S-O-500-180-Test S-O-500-180-FE
Page 516
485
(i) S-O-500-180-S-300-300
Figure 8.3 Predicted beam-tip load versus test result
(a) The first test group
0 1 2 3 4 5 6-200
-150
-100
-50
0
50
100
150
Lateral Drift (%)
Bea
m T
ip L
oad
(kN
)
0 24 48 72 96 120 144
Lateral Displacement (mm)
S-O-500-180-S-300-300-Test S-O-500-180-S-300-300-FE
0 5 10 150
50
100
Lateral Displacement (mm)
Lat
eral
Loa
d (k
N)
F-Control-Test F-Control-FE
0 2 4 6 8 10 120
20
40
60
80
Lateral Displacement (mm)
Lat
eral
Loa
d (k
N)
F-G-50-200-Test S-G-50-200-FE
0 5 10 150
20
40
60
80
Lateral Displacement (mm)
Lat
eral
Loa
d (k
N)
F-S-450-450-Test F-S-450-450-FE
0 2 4 6 8 10 120
20
40
60
80
Lateral Displacement (mm)
Lat
eral
Loa
d (k
N)
F-O-450-150Test F-O-450-150-FE
0 2 4 6 8 10 120
20
40
60
80
Lateral Displacement (mm)
Lat
eral
Loa
d (k
N)
F-O-500-180Test F-O-500-180-FE
Page 517
486
(a) The second test group
Figure 8.4 Predicted initial stiffness versus test result
0 1 2 3 4 5 6 70
20
40
60
80
100
Lateral Displacement (mm)
Lat
eral
Loa
d (k
N)
S-Control-Test S-Control-FE
0 1 2 3 4 5 6 70
10
20
30
40
50
60
Lateral Displacement (mm)
Lat
eral
Loa
d (k
N)
S-G-50-200-100-Test S-G-50-200-100-FE
0 1 2 3 4 5 6 70
10
20
30
40
50
60
Lateral Displacement (mm)
Lat
eral
Loa
d (k
N)
S-O-500-180-Test S-O-500-180-FE
0 1 2 3 4 5 6 70
10
20
30
40
50
60
Top-of-column Displacement (mm) L
ater
al L
oad
(kN
)
S-O-500-180-S-300-300-Test S-O-500-180-S-300-300-FE
Page 518
487
(a) Specimen F-Control
Page 519
488
(b) Specimen F-G-50-200
Page 520
489
(c) Specimen F-S-450-450
Page 521
490
(d) Specimen F-O-450-150
Page 522
491
(e) Specimen F-O-500-180
Page 523
492
(f) Specimen S-Control
Page 524
493
(g) Specimen S-G-50-200-100
Page 525
494
(h) Specimen S-O-500-180
Page 526
495
(i) Specimen S-O-500-180-S-300-300
Figure 8.5 Specimen crack patterns under push action
Page 527
496
(a) F-Control
(b) F-G-50-200
Page 528
497
(c) F-S-450-450
(d) F-O-450-150
Page 529
498
(e) F-O-500-180
(f) S-Control
Page 530
499
(g) S-G-50-200-100
(h) S-O-500-180
Page 531
500
(i) S-O-500-180-S-300-300
Figure 8.6 Predicted crack patterns of 3D FE models
(a) F-Control
0 1 2 3 4 5 60
50
100
150
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-Control-Test F-Control-FE F-Control-FE-NJC
Page 532
501
(b) F-G-50-200
(c) F-S-450-450
0 1 2 30
20
40
60
80
100
120
Lateral Drift(%)
Lat
eral
Loa
d (k
N)
0 24 48 72
Lateral Displacement (mm)
F-G-50-200-Test F-G-50-200-FE-NJC F-G-50-200-FE
0 1 2 3 4 5 60
20
40
60
80
100
120
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-S-450-450-Test F-S-450-450-FE F-S-450-450-NJC-FE
Page 533
502
(d) F-O-450-150
(e) F-O-500-180
0 1 2 3 4 5 60
50
100
150
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-O-450-150-Test F-O-450-150-FE F-O-450-150-NJC-FE
0 1 2 3 4 5 60
20
40
60
80
100
120
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-O-500-180-Test F-O-500-180-FE F-O-500-180-NJC-FE
Page 534
503
(f) S-Control
(g) S-G-50-200-100
0 1 2 3 4 50
20
40
60
80
100
120
140
160
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120
Lateral Displacement (mm)
S-Control-Test S-Control-FE S-Control-FE-NJC
0 1 2 3 40
50
100
150
Lateral Drift(%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96
Lateral Displacement (mm)
S-G-50-100-200-Test S-G-50-100-200-FE S-G-50-100-200-FE-NJC
Page 535
504
(h) S-O-500-180
(i) S-O-500-180-S-300-300
Figure 8.7 The effect of joints stirrup confinement
0 1 2 3 4 5 60
50
100
150
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
S-O-500-180-Test S-O-500-180-FE S-O-500-180-NJC
0 1 2 3 4 5 60
20
40
60
80
100
120
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
S-O-500-180-S-300-300-Test S-O-500-180-S-300-300-FE S-O-500-180-S-300-300-FE-NJC
Page 536
505
Figure 8.8 The effect of opening shape
Figure 8.9 Crack pattern predicted by FE model S-O-500-180-R
0 1 2 3 4 5 60
20
40
60
80
100
120
140
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
S-O-500-180-Test S-O-500-180-FE S-O-500-180-R-FE
Page 537
506
Figure 8.10 Effect of longitudinal slit length
(a) F-S-450-200
0 1 2 3 4 5 60
20
40
60
80
100
120
Lateral Drift (%)
Lat
eral
Loa
d (k
N)
0 24 48 72 96 120 144
Lateral Displacement (mm)
F-S-450-450-Test F-S-450-0-FE F-S-450-200-FE F-S-450-450-FE
Page 538
507
(b) F-S-450-0
Figure 8.11 Crack pattern predicted by FE models F-S-450-200 and
F-S-450-0
Page 539
508
Table 8.1 Predicted loads versus test results
Specimen Beam tip load (kN)
Top of column load (kN) Positive Negative
Prediction Test Error (%) Prediction Test Error (%) Prediction Test Error (%)
First test
group
F-Control 87.8 87.2 0.7% -161.1 -156.6 2.9% 133.6 130.2 2.6% F-G-50-200 88.6 87.4 1.4% -109.8 -85.6 28.3% 105.2 96.9 8.6% F-S-450-450 79 83.1 -4.9% -116.8 -109.4 6.8% 104.4 110.9 -5.9% F-O-450-150 86.0 88.5 -2.8% -153.7 -138.3 11.1% 120.5 120.2 0.2% F-O-500-180 86.4 88.2 -2.0% -128.7 -115.7 11.2% 109.6 109.4 0.2%
Second test
group
S-Control 119.4 112.8 5.9% -160.7 -144.4 11.3% 153.1 147 4.1% S-G-50-200-100 119.1 107.3 11.0% -135.6 -111.0 22.2% 125.8 122 3.1%
S-O-500-180 112.9 119.3 -5.4% -132.4 -120.4 10.0% 126.8 135.6 -6.5% S-O-500-180-S-300-300 99 102.6 -3.5% -122.5 -111.7 9.7% 110.4 118.6 -6.9%
Page 540
509
CHAPTER 9
CONCLUSIONS
9.1 INTRODUCTION
This thesis has presented a systematic study into a new seismic retrofit
approach called “beam weakening in combination with FRP (fibre reinforced
polymer) strengthening” (BWFS) method through both experimental study and
numerical investigation. The study has been focused on exploring the effects of
the proposed retrofitted techniques on the seismic performance of RC
beam-column joints via experimental study, and developing advanced
three-dimensional (3D) finite element (FE) models capable of accurately
modelling of T-section beams (T-beam) with a web opening (with or without
FRP strengthening) and the retrofitted 3-D RC joints. The developed 3-D FE
model was also used to study the issue of effective slab width with slab under
tension.
In the first stage of this study, an experimental study had been carried out
to explore the most effective techniques of the BWFS of retrofitting RC
beam-column joints, in the purpose to find a suitable approach for achieving
ductile beam sway mechanisms. Three different beam weakening techniques
were studied by testing 9 full-scale RC joints (under combined vertically
Page 541
510
constant axial loading and horizontally cyclic loading), including: (a) the slab
slit (SS) technique; (b) the beam web opening (BO) technique; (c) the beam
section reduction (SR) technique, in which a deep transverse groove (TG) was
cut on the soffit of the beam near the joint. The latter two techniques had been
carried out in associated strengthening measures (i.e., strengthening the
weakening area in shear using FRP) to satisfy both serviceability and
load-carrying capacity requirements.
The study was then focused on developing of 3D FE solid models (using
solid elements for modelling concrete) capable of accurately modelling of
T-beams with opening, with and without FRP strengthening in terms of the
load-displacement response and cracking behaviour (cracking patterns). The 3D
shell models (using shell elements for modelling concrete) for T-beams has also
been proposed and accessed. An explicit dynamic approach was adopted in
obtaining the solution in which the structural problem is regarded as dynamic
problem and solved using explicit centre-difference-method (CDM). Effects of
several significant parameters, such as loading duration, damping schemes,
computational time and accuracy associated with the explicit dynamic method
have been assessed and discussed in-depth in this thesis.
The final part of the thesis has been concerned with the application of
developed advanced 3D solid FE model and 3D shell FE model. First, the most
suitable 3D shell FE model was used to study the issues of effective slab width
of T-beams in hogging moment zone (and slab in tension). A wide range of key
Page 542
511
parameters influencing effective slab width have been investigated through
parametric studies and new models of effective slab width were proposed for
both interior and exterior joints. Finally, the 3D FE solid model was used for
numerical simulations of the retrofitted RC joints.
9.2 EXPERIMENTAL STUDIES OF RETROFITTED RC
JOINTS
To investigate the effectiveness of the proposed retrofit techniques in
transferring a storey sway failure mechanism to a beam sway mechanism, two
test groups, including a total of 9 specimens, were tested under cyclic loading.
Except for two control specimens, which was designed to violate the SCWB
mechanism (i.e. failing by storey sway mechanism), two specimens retrofitted
by creating 450×150 mm2 and 500×180 mm2 size openings were tested. It was
found that the specimens with the larger opening achieved the beam sway
mechanism. To reduce the opening area, an RC joint retrofitted by slot-shaped
openings was then tested. A 50×200 mm2 TG was made in the soffit of the
beam next to the joint. The first specimen so weakened, adjacent to the
beam-column interface, failed due to slipping of the bars in beam bottom within
the joint region. Another specimen of the same size, weakened by TGs placed
100 mm away from the beam-column interface, was also tested. The failure
mode was similar to the first. A specimen weakened by 450 mm long slab slits
along both the longitudinal and transverse directions was also tested. The steel
bars crossing the slits had all been cut. Finally, a combination of the SS and BO
Page 543
512
techniques was investigated. Based on the test results and discussions presented
in Chapter 4, the following conclusions can be drawn:
(1) The existence of a transverse groove (TG) can lower the moment-capacity of
a T-beam greatly. However, the failure mode was non-ductile because the
main beam bottom longitudinal bars slipped in the joint region, even though
the groove was not placed close to the beam-column interface. The slab
slitting (SS) method can effectively lower the moment capacity of a T-beam
as it removes the contribution of the cut slab longitudinal bars. Meanwhile,
the failure mode is beam end flexural failure. The beam opening (BO) size
should be well designed (e.g. increasing the opening size)to lead to an
obvious four-hinges mechanism if the BO technique is applied as a retrofit
measure. The slab slits can help a specimen, not sufficiently weakened by
the BO technique, to realize the four-hinges mechanism. Slot-shaped
openings have similar effect on the seismic behaviour of a RC beam-column
joint as the rectangular one does.
(2) As the horizontal load recorded by the MTS machine included friction force,
which varied during test and among different specimens, the calculated
column shear force was used for discussions instead. The specimens
retrofitted by the proposed techniques had lower stiffnesses and strengths
than control specimens. Both the BO and SS techniques decrease the
stiffness of T-beams under both negative and positive loading. The TG
technique only has obvious effect on the stiffness and strength of a T-beam
Page 544
513
when it was under negative loading. When the groove is not placed close to
the beam-column interface, the effect of the TG technique on T-beam
stiffness becomes weaker. The positive strength of T-beams is slightly
affected by the existence of BO. The SS technique deceases both negative
and positive strength of T-beams.
(3) When the control specimen has good ductility, the TG technique leads to
very poor ductility. The BO technique keeps the specimen ductility almost
unchanged. The SS technique decreases the ductility of a specimen. When
the control specimen has a poor ductility, the TG technique has a small
effect on specimen ductility but leads to a smaller yield displacement than
that of control specimen. The BO technique slightly increases the specimen
ductility. When the combination of the BO and SS techniques is applied,
specimen ductility can be increased in addition to effectively decreasing the
moment-capacity of T-beams.
(4) The Specimen retrofitted by the SS technique had better energy dissipation
capacity than those by the BO and the TG technique. The specimens
retrofitted the TG technique had best energy dissipation capacity at first but
worse energy dissipation capacity after they reached peak load.
(5) Equivalent viscous damping ratios (EDRs) of specimens basically increased
with increasing drift ratio. The EDRs of those specimens weakened by TGs
were almost the highest before they failed. The EDR of the specimen
Page 545
514
F-S-450-450 was the second highest among the first test group. The BO
technique has slight effect on a specimen EDR. Generally, the EDRs of the
second cycle of each drift ratio are lower than those of the first cycle.
(6) Unlike the BO, SS methods and their combination, the TG method has less
effect on the initial PTPS of specimens. In addition, the PTPS of the
specimens weakened by TGs, reduced the most quickly with increasing drift
ratio, as those specimens had poor ductility and failed soon after peak load
had been reached. The specimen retrofitted by a combination of the BO and
SS techniques has the lowest speed of PPTS descending.
(7) The TG technique shortens plastic hinges length of T-beams as the
deformation of T-beams is concentrated on the groove region.
(8) Except for two control specimens F-Control and S-Control, beams’
deformation contribution to the lateral drift ratio increased with the
increasing lateral drift ratio. Thus the contribution of columns deformation
was decreased with the increasing lateral drift ratio
9.3 THREE-DIMENSIONAL FINITE ELEMENT
MODELLING OF RC T-BEAMS AND JOINTS
9.3.1 3D Solid FE Model of T-beams with a Web Opening.
3D FE models for T-beams retrofitted by BO techniques were developed
and verified in Chapter 5. The RC T-beams tested by Nie (2018), with web
Page 546
515
openings, with or without FRP strengthening, were used for assessing the
accuracy of the 3D FE model. To avoid convergence problems, an explicit
dynamic approach which uses explicit centre-difference-method (CDM), was
adopted to obtain the numerical results. In this study, a 3D Prism cohesive
model was proposed for modelling the steel-concrete interface. The effects of a
number of significant parameters affecting the accuracy of the explicit dynamic
approach, including the element size, loading time, damping ratio, calculation
time and calculation precision, were studied. The confinement effect provided
by FRP on the web chord was also investigated. Finally, the predicted results of
the 3D FE solid model were compared with the predicted results of the 2D FE
models proposed by Nie (2018). The following conclusions can be drawn based
on the results and discussions presented in Chapter 5:
(1) The proposed 3D FE models predicted well, the behaviour of the beam
specimens tested by Nie (2018), except for the two beams under positive
loading. Furthermore, the yield/peak loads and test crack patterns of all
specimens were also accurately predicted, though the post-yielding curves of
T-beams with web opening were not well captured. A possible reason is that
the damage in shear is underestimated by the proposed models.
(2) The proposed 3D prism cohesive element modelling method was shown to
be effective and reliable for modelling the steel-concrete interface in the 3D
solid FE model.
Page 547
516
(3) For a quasi-static analysis, the dynamic energy should be a small portion of
the internal energy (normally below 1%), except for a few discrete moments
when local dynamics events occur, such as at the beginning of loading, at
the appearance of significant cracks, and at yielding of the main tensions
bars.
(4) Element size, loading time and stiffness damping coefficient were
determined based on convergence studies. A 20-mm element size, 50T0 (T0
is the FE model natural period) and value of 2e-6 were found
appropriate for the proposed 3D FE models. Only the value of is needed
to define the Rayleigh damping, since the vibrations associated with high
modes are expected to damped out in the adopted explicit dynamic method.
(5) The double precision computation was suggested for numerical simulation
using the 3D solid FE. Single precision computation costs less
computational time, but leads to more accumulation errors and unreal (or
even problematic) crack patterns in the later stage of simulation.
(6) When the confinement effect of the FRP wrapping was simulated using Yu
et al.'s (2009a, b) modified plastic-damage model for FRP-confined concrete,
more accurate results can be obtained, especially for a T-beam with a
smaller opening and a bigger web chord.
(7) The proposed 3D solid FE models are better than the 2D FE model proposed
Page 548
517
by Nie (2018) in the modelling of T-beams, especially for those weakened
by web openings but without FRP shear strengthening. The 2D FE models
usually overestimate the post-cracking stiffness of T-beams, which is better
predicted by the proposed 3D FE models.
(8) A DP+BC model, which is capable of modeling the shear degradation effect
of cracked concrete, was examined and the predictions were compared with
those from the DP model. The effects of three parameters, namely, the
number of critical cracks, the maximum cracking strain and the coefficient
of power law, of the BC model, were investigated through parametric
studies. The results indicated that the DP+BC model predicted the
load-displacement curves and crack patterns similar to those predicted with
the DP model; the DP+BC model predicted slightly better the post-peak
load-displacement response. The DP model is thus recommended for
modelling RC T-beams with a web opening but without FRP strengthening
since it is more convenient to create an FE model using the DP model.
9.3.2 3D Shell FE Models for T-beams
The 3D solid model proposed in Chapter 5 proved to be both reliable and
accurate when modelling a T-beam, but was not as economical as the 2D FE
model of Nie (2018) in terms of computational time. A decrease in the number
of elements proved to be a satisfactory option when the dynamic explicit
method is used. Based on the above considerations, several 3D shell FE models,
Page 549
518
in which concrete was modelled by shell elements (referred as 3D shell models)
were presented in Chapter 6. Different 3D FE models were proposed and
assessed, which included two 3D shell models with slab modelling by only one
shell layer (i.e. 3D 1-shell models) and a few 3D FE models with slab modelled
using two shell layers (i.e. 3D 2-shell models). For both types of shell models,
the effects of the bond-slip behaviour between longitudinal bars and concrete in
slab were investigated. For the 3D 2-shell models, effects of the plane section
restraint applied to the two slab shell layers were investigated. The T-beams
tested by Nie (2018) and the 3D solid model presented in Chapter 5 were used
for comparison purpose in the above assessment. The following conclusions are
drawn based on the numerical results and discussion presented in Chapter 6:
(1) The 3D-Solid models for T-beams are most accurate in terms of the
load-displacement curve and crack patterns.
(2) The 3D 1-shell and 2-shell models incorporating the bond-slip relationship
between slab longitudinal bars and concrete, predict better
load-displacement curves and crack patterns than those not considering the
bond slip behaviour.
(3) In the 3-D 2-shell model, the plane section restraints applied between the
two shells representing slab have only slight effect on the predicted
load-displacement curves. The crack patterns, however, are better predicted
when plan section restraints are applied. Decreasing the density of restraints
Page 550
519
in the X direction has less effect on the predicted crack patterns than
decrease of restraint density in the Z direction, which is the direction of
longitudinal bars.
(4) Both 3D-Shell-1 and 3D-Shell-2-R models (i.e., models with plane section
restraints applied between the two shells) work well as models for the
T-beam. As the 3D-Shell-1 model costs less computational time than the
3D-Shell-2-R mode, the former is the better substitute for 3D solid FE
model in modelling T-beam.
(5) A 20 mm element size, a loading time of 50T0 and a damping coefficient β
of 2e-6, are suggested for the FE 3D-Shell-1 model for modeling T-beams.
(6) An element size of 20 mm is not small enough for the 3D-Shell-1 model to
obtain mesh size convergence when it was used to model a T-beam
weakened by web openings. Thus, the 3D-Shell-1 model is not a good
substitute for the 3D solid FE model for modelling such beams. It not only
requires more computational time but provides less accurate predictions than
the 3D solid FE model.
9.3.3 Effective Slab Width of RC Frames
Using the 3D 1-shell model developed in Chapter 6, parametric studies
were carried out to investigate the effects of a wide range of factors on the
effective slab width of T-section RC beam with slab in tension, with following
Page 551
520
factors investigated: beam length, width, height, bar reinforcement ratio, slab
width, thickness, bar spacing, yield stress of steel bars, transverse beam height,
width stirrup spacing, column width. The effects of the bond-slip relationship
between longitudinal bars and concrete were also investigated. Two types of
stress-strain model for steel bars were studied and described parametrically.
Based on the numerical results and discussions presented in Chapter 7, the
following conclusions can be drawn:
(1) The key factors affecting the predicted effective slab width at both exterior
and interior joints include concrete strength, slab bar spacing, yield stress of
steel.
(2) The transverse beam width only has an obvious effect on the effective slab
width of an exterior joint. The effects of beam bar yield stress, beam height
and width are more obvious for an interior joint than for an exterior joint.
(3) The effective slab width for interior joints is wider than that for interior
joints.
(4) Based on the numerical modelling results, two simple equations for
predicting effective slab width were proposed for interior and exterior joints.
Only key factors were included in the equations. With the objective of
providing more accurate predictions, a weakening coefficient accounting for
column effects was included.
Page 552
521
9.3.4 3D FE Modelling of Retrofitted RC Joints
3D FE models developed for RC joints retrofitted by the proposed
techniques was presented in Chapter 8, which were based on the verified 3D
solid models described in Chapter 5. The 3D FE models for RC joints were
subjected to monotonic loading, rather than the cyclic loading in the tests. The
envelope curves derived from these test results were then used for comparison
purposes. The following conclusions can be drawn based on the predicted
results:
(1) The proposed 3D FE models perform well in predicting load-displacement
curves and crack patterns of the retrofitted RC joints. The predictions for
specimens with transverse grooves are not as good as the others because bars
slippage at beam bottom is noticeable due to cyclic loading, which was not
considered in the 3D FE models.
(2) The predicted initial stiffness are slightly larger than those of the test values
because of the mechanical slack existing in the testing equipments. Such
slack was caused by very small failures of the mechanical components of the
test equipments used to fit together precisely.
(3) The confinement effect of stirrups should be considered in the 3D FE model
to obtain more reliable/accurate predictions. In this study the model
proposed by Mander et al. (1988) was applied.
Page 553
522
(4) The confinement effect of stirrups has a larger effect on the specimens
F-Control and S-Control, in which joint shear failure occurred and bottom
column hinge formed respectively.
(5) The ignorance of the effect of cyclic loading leads to slightly higher negative
beam tip loads. This is mainly because the cumulative damage caused by
cyclic loading during testing is not simulated in FE analyses.
9.4 FURTHER STUDIES
During this study, the effectiveness of the proposed retrofit techniques for
RC frames were investigated through testing RC joints subject to applied cyclic
loading. Due to the limit of time, money and test equipment, only 9 specimens
were tested. To produce comprehensive retrofitting guidelines, more serials of
tests should be conducted. Firstly, due to the limit of test facilities, the applied
axial load was constant. However, in a real situation, the axial load might
increase with the increasing drift ratio, which could decrease the specimen
ductility. Thus, if possible, a serial of test with axial load keeping changing
should be conducted. Secondly, the specimens retrofitted by the TG technique
had poor seismic performance due to the bottom beam bar slippage within the
joint region. Thus, another modification, like filling the groove with a material
with lower compressive strength, should be done for this technique. Finally, to
better verify the effectiveness of the proposed retrofit techniques in realising the
SCWB mechanism, experimental studies on retrofitted RC frames subjected to
Page 554
523
seismic loading should be further explored. These RC frames should be
designed according to previous editions of the seismic design codes. When
subjected to seismic loading, the control specimen should fail by storey sway
mechanism. After applying the proposed retrofit techniques, a beam sway
mechanism should be realised.
The proposed 3D FE models representing retrofitted RC joints considered
monotonic loading only, which needs to be improved to consider the effect of
cyclic loading. The 3D models for the RC Joints retrofitted by the TG technique,
overestimated the peak load, the corresponding displacement at the peak load
and the post-cracking stiffness. This may be due to the incapability to model the
effects of cyclic loading. Thus, a more advanced FE model of retrofitted RC
joints, subjected to cyclic loading, should be developed in the future. The
Software OpenSEES (2009) can be used for this purpose. An appropriate FE
modelling of retrofitted RC frames subject to seismic loading should then be
conducted. With a reliable FE model, parametric studies on the opening sizes,
locations, and various combinations of the three techniques should be done,
with the aim to produce comprehensive retrofitting guidelines. Besides, the
effect of axial load applied on the column top should be investigated with the
verified FE model.
Parametric studies related to effective slab width of RC frames have been
conducted and two simple equations proposed for interior and exterior joints.
The effect of BO or SS on the effective slab width has not be investigated. As
Page 555
524
both the BO and SS techniques might actually be used in practice for the
retrofitting of existing RC frames, the effects of BO and SS on the effective slab
width cannot be ignored. The effect of the SS on the effective slab width is
obvious as slab bars crossing the slits are cut. The BO technique might change
the failure mode from beam flexural failure to four hinges failure. The existence
of web openings has a significant effect on the effective slab width in composite
steel beams and this has been confirmed by Alsarraf and El Din (2016). Thus
the effect of BO on the effective slab width of a RC frame should also be
investigated in the future.
9.5 REFERENCES
ABAQUS (2012). ABAQUS Analysis User's Manual (Version 6.12), Dassault
Systems SIMULIA Corporation, Providence, Rhode Island, USA.
Alsarraf, M. A., and El Din, H. S. (2016). "Effects of Web Openings on the
Effective Slab Widths in Composite Steel Beams", International Journal
of Engineering and Technology, 8(1), 6.
Mander, J. B., Priestley, M. J., and Park, R. (1988). "Theoretical stress-strain
model for confined concrete", Journal of structural engineering, 114(8),
1804-1826.
OpenSees (2009). Open System for Earthquake Engineering Simulation, Pacific
Earthquake Engineering Research Center, University of California at
Page 556
525
Berkeley, http://opensees.berkeley.edu.
Nie, X.F. (2018). Behavior and Modeling OF RC Beams with an
FRP-Strengthened Web Opening, Doctoral degree thesis: The Hong Kong
Polytechnic University.
Yu, T. T. J. G., Teng, J. G., Wong, Y. L., and Dong, S. L. (2010), "Finite
element modeling of confined concrete-I: Drucker–Prager type plasticity
model", Engineering Structures, 32(3), 665-679
Page 558
527
APPENDIX A FRP ANCHOR DESIGN
As shown in Figure A.1, there are three typical adhesive anchor failure
modes, which are concrete cone failure, combined failure and anchor failure
(Kim and Smith 2010).
The anchor resistance force ( uN ) is given as below:
min( , , )u cc cb arN N N N (A-1)
1.5cc 9.68 ' ef cN h f (cone failure ) (A-2)
04.62 ( ' 20 ) cb ef cN d h f MPa (combined failure) (A-3)
09.07 ( ' 20 ) cb ef cN d h f MPa (combined failure ) (A-4)
0.59ar FRP FRP FRPN w t f (anchor failure ) (A-5)
where efh = effective embedment depth of the anchor (mm); 'cf =concrete
cylinder compressive strength (MPa); 0d =diameter of the anchor hole (mm);
FRPw and FRPt = width (mm) and thickness (mm), respectively, of the fiber
sheet used in construction of the FRP anchor; and FRPf = flat coupon tensile
rupture FRP strength (MPa).
The expected failure mode is FRP rupturing before anchor failure. Thus the
design anchor should satisfy the following equation:
FRP uF N (A-6)
Page 559
528
where FRPF is the resistance force of one FRP stripe (N).
0.334 ,5022
2 0
7
82FRP u
t mmw mmG GPa
f MPa
' 35cf MPa
47090 FRP uF wt N
From equation(A-5) and(A-6): FRPw >84.7mm; FRPw =90mm
From equation(A-2) and(A-6): efh > 87.7; 令 efh ≈90mm
From equation(A-4) and(A-6): 0d >18.4mm; 令 0d ≈19mm
The diameter of the anchor holes should not be too big to void small space
between anchor holes. 0d =12 mm is applied into anchor design. The failure
force is then calculated as below:
09.07 9.07 3.14 12 90 30758.2 cb efN d h N
The FRP effective rupture stress is about 60% of u . The design anchor
meets the requirement for the expected failure mode.
The free end of anchor will be fanned out at an angle about 36°
(2arctan(25/80)) and attached to the FRP sheet to avoid the FRP sheet
debonding from the U jacket end.
Page 560
529
Figure A.1 Typical adhesive anchor failure modes (Kim and Smith 2010)
Figure A.2 The detail of FRP sheet anchor
150 450 150450
180
25 5050 2525505025
80
36°
Figure A.3 The layout of FRP Sheet anchors of specimen F-O-450-150
Page 561
530
200
50 100200
50100
5050
80
36°
2525
Figure A.4 The layout of FRP Sheet anchors of specimen F-G-50-200
REFERENCES
Kim SJ, Smith ST. (2010) "Pullout strength models for FRP anchors in
uncracked concrete", Journal of Composites for Construction, 14(4),
406–414.
Koutas, L., and Triantafillou, T. C. (2012), "Use of anchors in shear
strengthening of reinforced concrete T-beams with FRP", Journal of
Composites for Construction, 17(1), 101-107.