Você está na página 1de 466

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Design and analysis of complex D-Regions in reinforced concrete structures


Yindeesuk, Sukit
ProQuest Dissertations and Theses; 2009; ProQuest Dissertations & Theses Full Text
pg. n/a
DESIGN AND ANALYSIS OF COMPLEX D-REGIONS IN REINFORCED
CONCRETE STRUCTURES
BY
SUKIT YINDEESUK
B.S., Kasetsart University, 2001
M.S., University of New South Wales, 2004
DISSERTATION
Submitted in partial fulfillment of the requirements
for the degree of Doctor of Philosophy in Civil Engineering
in the Graduate College of the
University of Illinois at Urbana-Champaign, 2009
Urbana, Illinois
Doctoral Committee:
Associate Professor Daniel A. Kuchma, Chair
Professor Amr S. Elnashai
Associate Professor Arif Masud
Assistant Professor Bassem Andrawes
Assistant Professor Paramita Mondal
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3395554
All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.
UMI
__.Dissertation Publishing--...._
UMI 3395554
Copyright 2010 by ProQuest LLC.
All rights reserved. This edition of the work is protected against
unauthorized copying under Title 17, United States Code.
Pro uesr
---
ProQuest LLC
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106-1346
---
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract
STM design provisions, such as those in Appendix A of ACB 18-08, consist of
rules for evaluating the capacity of the load-resisting truss that is idealized to carry the
forces through the D-Region. These code rules were primarily derived from test data on
simple D-Regions such as deep beams and corbels. However, these STM provisions are
taken as being sufficiently general and conservative that they can be used for the design
of all possible D-Regions, including those regions in which a highly statically
indeterminate and complex truss is selected and designed to carry the imposed loadings.
Since STM design provisions are only for capacity assessment and given the wide range
of applicability of the method, members designed using these STM provision may not
necessarily exhibit satisfactory performance under service load levels or during
overloads. More particularly, the limitation of current STM design provisions include: 1)
no methodology for satisfactorily estimating the complete load-displacement response
history, cracking loads, stress and strain states in reinforcement, and failure modes; 2) a
lack of guidance for selecting the shape, dimensions, and stress limits of STM
components; 3) a lack of guidance for proportioning of forces in statically indeterminate
STM designs; and 4) an inability to employ performance-based design concepts.
To overcome these limitations, an experimental and computational program was
conducted. This research aimed to generate the data needed for evaluating and validating
the STM approach, for advancing STM shape selection techniques, and also to develop
more reliable computational models for predicting the response of cracked structural
concrete. In addition, a design and analysis framework that enables for automated
nonlinear FEA design validation is proposed that integrates the current STM design
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
provisions with two additional components: 1) the shape selection technique of STM
based on topology optimization and 2) the new automated nonlinear FEA tool that was
specialized for modeling and analyzing complex D-Regions.
lll
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
To my Mother and Father
lV
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgements
I would like to thank my adviser, Professor Daniel Kuchma, for his sincere and
invaluable attention, suggestion, encouragement and support throughout my Ph.D.
research. I also would like to express my sincere gratitude to Professors Amr Elnashai,
Arif Masud, Bassem Andrawes, and Paramita Mondal, who served as my committee
members and offered invaluable input and support. Additionally, I would like to offer
thanks to Professor John Popovics, who served as my committee during my preliminary
exam and provided useful review of my work and comments which increased the quality
of my research.
I gratefully acknowledge the National Science Foundation for funding this project
through my adviser and Thai Royal Government for awarding me a scholarship for
completing my Ph.D. and conducting this research.
I am also thankful to my colleagues Ken Marley, Christopher Hart, and Jason
Hart for their assistance in the experimental programs, Dr. Heui Hwang Lee, Dr. Sang-
Ho Kim, and Dr. JungWoong Park for useful discussion about nonlinear FEA and
computational techniques. Special thanks to Tim Prunkard and the staff in the Civil
Engineering Machine Shop for their support in fabrication and test setup of all specimens
tested in the research program. I also wish to thank Dr. Shaoying Qi, as my friend, for our
exchanges of knowledge through discussions.
Finally, I would like to express my gratitude to my parents and my brothers for
their love, patience, encouragement and support throughout my Ph.D. studies.
v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table of Contents
List of Figures ..................................................................................................................... ix
List of Tables .................................................................................................................... xxi
Chapter 1 : Introduction ....................................................................................................... 1
1.1 Background .............................................................................................................. 1
1.1.1 Band D Regions ............................................................................................. l
1.1.2 Design ofD-Regions Using STM .................................................................. 2
1.1.3 Safety, Validation, and Limitations of the STM in D-Region Design ........... 3
1.1.4 Analysis Approaches for D-Regions and Their Limitations .......................... 6
1.1.5 Simple and Complex D-Regions: A New Terminology for This Research ... 8
1.2 Objectives ................................................................................................................ 9
1.3 Overview of Dissertation ........................................................................................ 12
Chapter 2: Literature Review ............................................................................................ 14
2.1 History and Development of STM for Design ofD-Regions ................................ 14
2.2 Shape of STM ........................................................................................................ 15
2.3 Calculating the Capacity of a STM Model ............................................................ 21
2.3.l Geometry of STM Components .................................................................. 22
2.3.2 Stress Limits ................................................................................................ 25
2.3.3 Plastic Truss Capacity ................................................................................. 28
2.3.4 Capacity of Non-Plastic Indeterminate Truss .............................................. 29
2.3.5 Serviceability and Ductility Requirements .................................................. 32
2.4 Nonlinear STM Model ........................................................................................... 35
2.5 Nonlinear FEA for Structural Concrete ................................................................. 52
2.5.1 Discrete and Smeared Crack Approaches ................................................... 53
2.5.2 Crack Orientation: Fixed, Rotating, and Delayed Rotating Approach ........ 55
2.5.3 Selected FE Frameworks ............................................................................. 58
2.5.4 FE Modeling of Reinforcement ................................................................... 68
2.5.5 FE Modeling of Bond-Slip Problems .......................................................... 70
2.5.6 Mesh Dependent Problem in Smeared Crack Approach ............................. 74
2.6 Summary ................................................................................................................ 76
Chapter 3: Experimental Research Program ..................................................................... 78
3.1 Objective of the Experimental Test. ...................................................................... 78
3.2 Design of the Test Specimens ............................................................................... 79
3 .3 Material Properties ................................................................................................ 96
3 .3 .1 Concrete ....................................................................................................... 96
3.3.2 Reinforcement ............................................................................................. 97
3.4 Fabrication of Test Specimens .............................................................................. 97
3.5 Experimental Test Set-up .................................................................................... 107
3 .6 Instrumentation .................................................................................................... 107
3.7 Test Procedure ..................................................................................................... 116
3.8 Summary .............................................................................................................. 116
Chapter 4: Experimental Validation of STM in Design of Complex D-Regions ............ 118
4.1 Introduction ......................................................................................................... 118
4.2 D-Regions Designed for Non-Reversed Static Loadings .................................... 119
4.2.1 Load-Deformation Response of Test Specimens ...................................... 119
Vl
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.2.2 Comparison of Measured and Calculated Strengths .................................. 121
4.2.3 Comparison of Modes of Failure ............................................................... 123
4.2.4 Development of Cracking in Test Structures ............................................ 130
4.2.5 Straining and Load Transfer Characteristics in Test Structures ................ 136
4.2.6 Influence of Local Ductility in Test Structures ......................................... 142
4.3 D-Regions Designed for Reversed Static Loadings ............................................ 144
4.3. l Load-Deformation Response of Test Specimens ...................................... 144
4.3.2 Comparison of Measured and Calculated Strengths .................................. 146
4.3.3 Comparison of Modes of Failure ............................................................... 147
4.3.4 Development of Cracking in Test Structures ............................................ 149
4.4 Summary .............................................................................................................. 151
Chapter 5: Shape of STM in Design of Complex D-Regions ......................................... 153
5.1 Introduction ......................................................................................................... 153
5 .2 Application of Topology Optimization for Shape Selection of STM ................. 154
5.2.1 D-Regions Subjected to Non-Reversed Static Loadings ........................... 155
5.2.2 D-Regions Subjected to Reversed Static Loadings ................................... 165
5.3 Experimental Validation of Topology Optimization for Shape Selection of
STM for D-Regions Subjected to Non-Reversed Static Loadings ...................... 172
5 .3 .1 Comparison of Load-Deformation Responses .......................................... 172
5.3.2 Comparison of Modes of Failure ............................................................... 173
5.3.3 Comparison of Crack Development and Propagation ............................... 174
5.3.4 Influence of Orientation of Tension Tie on Structural Responses ............ 176
5.4 Experimental Validation of Topology Optimization for Shape Selection of
STM for D-Regions Subjected to Reversed Static Loadings .............................. 178
5 .4.1 Comparison of Load-Deformation Responses .......................................... 178
5.4.2 Comparison of Modes of Failure ............................................................... 179
5 .4.3 Comparison of Crack Development and Propagation ............................... 181
5.5 Computational Validation of Topology Optimization for Shape Selection of
STM for D-Regions ............................................................................................. 183
5.5.1 Computational Models ofD-Regions by Vector2 ..................................... 183
5.5.2 Calibration of Computational Models ....................................................... 187
5.6 Numerical Results and Structural Performance Evaluation and Comparison ..... 190
5.7 Summary .............................................................................................................. 194
Chapter 6: Nonlinear FEA for Automated Analysis of Complex D-Regions ................. 195
6.1 Embedded Reinforcement Element Including Bond Stress-Slip ......................... 196
6.1.1 Concept and Proposed Formulation ........................................................... 199
6.2 Numerical Examples ............................................................................................ 210
6.2.1 Anchorage Tests ......................................................................................... 211
6.2.2 Two-Spanned Continuous Deep Beam (Foster et al, 1992) ....................... 220
6.2.3 Simply Support Deep Beams with Openings ............................................. 224
6.3 A New Constitutive Model for Reinforced Concrete Structures ......................... 229
6.3.l Important Features ...................................................................................... 229
6.3.2 Proposed Formulations ............................................................................... 230
6.3.3 FE Procedure and Nonlinear Solving Scheme (Element Level) ................ 250
6.3 .4 FE Procedure and Nonlinear Solving Scheme (Structural Level) .............. 254
6.4 Numerical Examples ............................................................................................ 256
vu
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.4.1 Shear Panels ................................................................................................ 257
6.4.2 Simply Support Deep Beams with an Opening .......................................... 265
6.4.3 Simply Support Deep Beams with a Dapped End and an Opening ............ 273
6.4.4 Simply Support Deep Beams with Openings ............................................. 285
6.5 Mesh Dependency Reduction .............................................................................. 293
6.6 Influence of Orientation of Principal Stress in Concrete and Crack Angle on
Structural Strength and Responses of Complex D-Regions ................................ 299
6.6.1 Element Level ............................................................................................ 299
6.6.2 Structural Level ......................................................................................... 306
6.7 Summary .............................................................................................................. 314
Chapter 7: Design and Analysis Framework for Effective Structural Performance
Design of Complex D-Regions ........................................................................................ 316
7.1 Introduction ......................................................................................................... 316
7.2 Proposed Design and Analysis Framework ......................................................... 319
7.3 Application of Proposed Design and Analysis Framework ................................. 321
7.3.1 A Propped Cantilever Deep Beam with an Opening .................................. 321
7 .3 .2 Simply Support Deep Beams with an Opening ......................................... 341
7.3.3 Simply Support Deep Beams with a Dapped End and an Opening ............ 353
7.3.4 Shear Wall with Openings .......................................................................... 373
7.4 Summary .............................................................................................................. 396
Chapter 8: Conclusions and Recommendations for Future Work ................................... 399
8.1 Summary and Conclusions .................................................................................. 399
8.1.1 Experimental Validation of Complex D-Regions Designed by STM ....... 399
8.1.2 Method and Guidelines to Facilitate Improved STM Shape Selection .... .400
8.1.3 Nonlinear FEA for Automated Analysis of Complex D-Regions ............ .401
8.1.4 Design and Analysis Framework for Effective Structural Performance
of Complex D-Regions .............................................................................. 404
8.2 Recommendations for Future Work .................................................................... 405
References ....................................................................................................................... 409
Appendix A Post Processing Tools Based on Augustus for Embedded
Reinforcement Element Including Bond Stress-Slip ................................................. 415
Appendix B Automatic Mesh Generation by Patran ..................................................... .419
Appendix C Analysis Results of Reinforced Concrete Shear Panels ............................. 423
Author's Biography ......................................................................................................... 444
Vlll
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Figures
Figure.1.1 Examples of Disturbed Regions (Taken from Tjhin 2004) ................................ 2
Figure 1.2 General Procedure for the Design ofD-Regions by the STM ........................... 3
Figure 1.3 Examples of STM (Taken from Tjhin 2004) ..................................................... 5
Figure 1.4 Reinforced Concrete Truss (Adapted from Polla, 1992) ................................... 5
Figure 1.5 Idealized Reinforced Concrete Truss for More Complex D-Regions: (a)
Selected STM (One of Dozens of Possibilities); (b) Idealized Reinforced Concrete
Truss for More Complex D-Region ................................................................................ 6
Figure 1.6 The Proposed Design and Analysis Framework .............................................. 10
Figure 2.1 Ritter's Truss Model (Ritter, 1899) .................................................................. 14
Figure 2.2 Shape ofSTM based on Empirical Model (Reproduced from FIP, 1996): (a)
Empirical STM for a Two Span Deep Beam; (b) Empirical STM for Beam-Column
Joints; and ( c) Empirical STM for Post Tension at End Supports ................................ 17
Figure 2.3 Shape of STM based on Crack Pattern (Reproduced from Fu, 2001): (a)
Crack Pattern at a Dapped End; (b) STM Shape No.I; (c) STM Shape No.2; (d)
STM Shape No.3; and (e) STM Shape No.4 ................................................................ 18
Figure 2.4 Shape of STM (a) Stress Trajectory and (b) Load Path Method (Taken from
Schlaich et al, 1987) ..................................................................................................... 19
Figure 2.5 Topology Optimization for Generating Shape of STM (Taken from Liang et
al, 2001 ): (a) Geometry of a Shear Wall with Openings; (b) Optimal Topology; (c)
Shape of STM based on the Optimal Topology; and (d) Dimensions ofSTM ............ 20
Figure 2.6 2D Structural Grid for Generating Shape of STM (Adopted from Ali and
White, 2001): (a) Ground Structural Grids; (b) Structural Grid at the 25th Iteration;
( c) Structural Grid at the 45th Iteration; and ( d) Optimal Structural Grid at the 50
1
h
Iteration ......................................................................................................................... 21
Figure 2.7 Types of Strut and Nodal Zones (Adapted from Tjhin, 2004): (a) Prismatic
Strut; (b) Bottle-Shaped Strut; (c) Fan-Shaped Strut; (d) CCC Node; (e) CCT Node;
(f) CTT Node; and (g) TTT Node ................................................................................ 24
Figure 2.8 Guidelines for Dimensions of Nodal Zones (Taken from ACB 18-05): (a)
One Layer of Steel; and (b) Distributed Steel .............................................................. 25
Figure 2.9 Rigid Perfectly Plastic Stress-Strain Relation (Taken from Tjhin, 2004) ........ 29
Figure 2.10 Non-Plastic Indeterminate STM: (a) Complex Type of D-Regions; (b)
Simple Type of D-Regions; and (c) Parameter Sensitivity of Nonlinear STM ............ 30
Figure 2.11 Indeterminate STM (Hwang and Lee, 1999) ................................................. 31
Figure 2.12 Nonlinear STM (Sundermann and Mutscher, 1991): (a) Continuous Deep
Beam; (b) Shape of STM including "Free Node"; and (c) Predicted Structural
Responses ..................................................................................................................... 3 7
Figure 2.13 Behavior Models for Struts and Ties (Sundermann and Mutscher, 1991):
(a) Reinforced Concrete Tie under Tension; and (b) Fan Type Strut under
Compression ................................................................................................................. 38
Figure 2.14 Shapes of STM for Load Transfer Mechanism (Hwang and Lee, 1999): (a)
Diagonal Mechanism; (b) Horizontal Mechanism; and ( c) Vertical Mechanism ....... .40
Figure 2.15 The Softened Stress-Strain Relationship of Concrete (Zhang and Hsu,
1998) ............................................................................................................................. 41
Figure 2.16 Nonlinear STM (Yun, 2000): (a) Specimen (Simply Supported Beam); (b)
lX
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
STM based on Stress; (c) Refined FE Models for Nodal Zones; and (d) Predicted
Structural Responses ..................................................................................................... 42
Figure 2.17 Nonlinear STM (To et al, 2001): (a) Effective Dimensions ofSTM
Members; (b) STM for a Structural Frame; and (c) Predicted Structural Response ... .45
Figure 2.18 CAST GUI (Tjhin, 20004) ............................................................................ .47
Figure 2.19 Nonlinear STM (Tjhin, 2004): (a) Dimensions and Reinforcement Details;
(b) One Direct Strut STM; (c) Determinate STM; (d) Indeterminate STM; and (e)
Predicted Load-Deflection ............................................................................................ 49
Figure 2.20 Nonlinear STM (Park and Kuchma, 2007): (a) Determinate STM; (b)
Effect of Multiple Compression Field Relationships; and ( c) Performance of the
Model ............................................................................................................................ 51
Figure 2.21 Discrete and Smeared Cracks (Reproduced from Bazant, 1983): (a)
Cracking Process; (b) Stress Distribution along Section A-A; (c) Discrete Crack;
And (d) Smeared Crack ................................................................................................ 53
Figure 2.22 Deformed Mesh in Discrete and Smeared Cracks (Reproduced from Rots
and Blaauwendraad, 1989): (a) Deformed Mesh in Smeared Crack Model; and (b)
Deformed Mesh in Discrete Crack Model.. .................................................................. 54
Figure 2.23 Illustration of Applied Stresses on Element, Average Stresses in Concrete,
and Average Stresses in Reinforcement (Adapted from Hsu, 1998) ........................... 55
Figure 2.24 Average Stresses on Crack Surfaces of Reinforced Concrete-Fixed Crack
Approach (Adapted from Hsu, 1998) ........................................................................... 56
Figure 2.25 Comparison of the FE Results between Fixed and Rotating Crack
Approach (Adapted from Rots and Blaauwendraad, 1989): (a) Fixed Crack
Approach; and (b) Rotating Crack Approach ............................................................... 57
Figure 2.26 Description of Modified Compression Field Theory (Vecchio and Collins,
l 986) ............................................................................................................................. 59
Figure 2.27 The Advancement of Modified Compression Field Theory (Vecchio and
Collins, 1986): (a) Compression Softening Effect; (b) Tension Stiffening Effect;
and (c) Mechanism at Local Crack Level ("Crack Check") ......................................... 60
Figure 2.28 The Deficiency of Modified Compression Field Theory (Vecchio, 2000):
(a) No Slip along Crack and Reorientation of Principal Stress; (b) Substantial Slip
along Crack and Reorientation of Principal Stress; (c) Measured Principal Stress
and Strain Direction; and ( d) Measured v.s. Predicted Principal Stress and Strain
Direction ....................................................................................................................... 61
Figure 2.29 The Condition at Local Crack Level of Disturbed Stress Field Model
(Vecchio, 2000): (a) Free Body Diagram for Relating Local Crack to Continuum
Level; (b) Continuum Level; (c) Local Crack Level; and (d) Local Crack
Mechanism .................................................................................................................... 62
Figure 2.30 The Compatibility Conditions adopted in the Disturbed Stress Field
Model (Vecchio, 2000): (a) Continuum Strains; (b) Slip along Crack Surface; and
(c) Total (Measured) Strains ......................................................................................... 63
Figure 2.31 The Improved Compression Softening and Tension Stiffening Models
adopted for Use in the Disturbed Stress Field Model (Vecchio, 2000): (a)
Adjusting the New Compression Softening Parameter (Cs); and (b) The New Semi-
Empirical for Tension Stiffening Model considering a Bond-Related Parameter
(Bentz, 2005) ................................................................................................................ 64
x
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 2.32 Tension Chord Model (TCM) (Kaufmann and Marti, 1998): (a)
Idealization of Tension Chord Model (TCM); (b) TCM in Y-Direction; (c) TCM in
X-Direction; TCM in General Direction; and (e) Free Body Diagram for Force
Transfer Mechanism from at Crack to between Crack ................................................. 66
Figure 2.33 Constitutive Models adopted in CMM (Foster and Marti, 2003): (a) Tri-
Linear cr-E Relation for Steel; (b) Tension Softening Relation for Concrete; (c)
Biaxial Strength Envelope for Concrete; and (d) Generalized cr-E Relation for
Concrete ........................................................................................................................ 67
Figure 2.34 FE Modeling of Reinforcement: (a) Type of Reinforcement Modeling; and
(b) Mesh Topology Independence of Embedded Reinforcement Elements (Elwi and
Hrudey, 1989) ............................................................................................................... 69
Figure 2.3S Type of Bond-Slip Model: (a) Bond Link (Discrete); and (b) Bond Zone
(Continuous) ................................................................................................................. 71
Figure 2.36 Analytical Local Bond-Slip Relationship for Monotonic Loading (CEB-
FIP Model Code 1990) ................................................................................................. 74
Figure 2.37 Mesh Dependent Problem in Deep Beams (Dodds et al, 198S): (a) Load
Deflection Curves for Deep Beams; and (b) Crack Patterns for Deep Beams ............. 7 S
Figure 2.38 Effectiveness of the Equivalent Strength in Reducing the Mesh Dependent
Problem (Bazant, 1983) ................................................................................................ 76
Figure 3.1 Types and the Overall Dimensions for All Complex D-Regions: (a) The
Simply-Supported Deep Beam with a Rectangular Opening beneath the Point of
Loading; (b) The Propped Cantilever Beam with an Opening; ( c) The Simply-
Supported Beam with a Dap at One End and an Opening adjacent to the Support
at the Other End; and (d) The Simply-Supported Beam with Two Rectangular
Openings at One End .................................................................................................... 80
Figure 3.2 The Four STMs used to Design the Simply-Supported Deep Beam with
an Opening underneath the Loading Point: (a) Model 4a; (b) Model 4b; (c) Model
4c; (d) Model 4d, and (e) Model 4e (the Optimized Truss) ......................................... 82
Figure 3.3 The Three STMs used to Design the Propped Cantilever Beam with an
Opening: (a) Model Sa; (b) Model Sb; (c) Model Sc; and (d) Model Sd (the
Optimized Truss) .......................................................................................................... 83
Figure 3.4 The Four STMs used to Design the Simply-Supported Beam with a Dap at
One End and an Opening adjacent to the Support at the Other End: (a) Model 6a;
(b) Model 6b; ( c) Model 6c; and ( d) Model 6d ............................................................ 84
Figure 3.S The Additional Three STMs used to Design the Simply-Supported Beam
with a Dap at One End and an Opening adjacent to the Support at the Other End:
(a) Model 7a (without Distributed Reinforcement); (b) Model 7b (the Optimized
Truss); and (c) Model 7c (the Optimized Truss with Orthogonal Reinforcement) ...... 86
Figure 3.6 Three STMs used to Design the Simply-Supported Beam with Two
Rectangular Openings at One End: (a) Model 8a; (b) Model 8b (The Optimized
Truss); and ( c) Model 8c (The Conventional Truss) .................................................... 88
Figure 3.7 Internal Forces in the Indeterminate Truss proportioned by Using the
Equations provided by PIP Recommendations 1996 ................................................... 89
Figure 3.8 Summary of Reinforcement Layouts and Details of All 17 Specimens .......... 9S
Figure 3. 9 Concrete Compressive Stress-Strain Relationships ......................................... 96
Figure 3 .10 Steel Reinforcement Stress-Strain Relationships ........................................... 98
Xl
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 3 .11 Reinforcement Cage for a Truss Model 4A to D ......................................... 101
Figure 3.12 Reinforcement Cage for a Truss Model 5A to C ......................................... 103
Figure 3.13 Reinforcement Cage for a Truss Model 7A to C ......................................... 105
Figure 3.14 Reinforcement Cage for a Truss model 8A to C .......................................... 106
Figure 3.15 Experimental Test Set-Up: (a) Specimen 4a to 7c; (b) Specimen 8a
to 8c ............................................................................................................................ 108
Figure 3.16 Examples oflnstrumentation: (a) Reinforcement Strain Gage Locations
in One of the Complex D-Regions, and (b) Krypton LED and Surface Strain
Gages Locations in One of the Complex D-Regions ................................................. 109
Figure 3 .1 7 Details and Locations of the Instrumentations in All Specimens ................ 115
Figure 4.1 Measured Load-Deformation Response of Test Structures ........................... 121
Figure 4.2 Mode of Failure for Specimens 4A to 7 A: (a) Specimen 4A to 4D; (b)
Specimen 5A to 5C; and (c) Specimen 6A to 7A ....................................................... 128
Figure 4.3 Observed Cracking in Specimen 5A to 6D: (a) Specimen 5A to 5C Tested
at P= 200 kN ( 45 kips), 310 kN (70 kips), and 400 kN (90 kips); and (b) Specimen
6A to 6D Tested at P= 222 kN (49.9 kips), 356 kN (80.3 kips), and 534 kN (120
kips) ............................................................................................................................ 133
Figure 4.4 Plots of Measured Strain in Specimen 6A: (a) Concrete Gauges; and (b)
Reinforcement Gauges ............................................................................................... 141
Figure 4.5 Plots of Measured Strains from Concrete Surface Strain Gauges of
Specimen 7B and 7C .................................................................................................. 142
Figure 4.6 Observed Cracking in Specimen 6A (Provided Distributed Reinforcement)
and 7A (No distributed Reinforcement) ..................................................................... 143
Figure 4. 7 Measured Load-Deformation Response of Specimen 8A and C ................... 145
Figure 4.8 Force Applied to the Shear Wall Side of Specimens 8A to SC ..................... 146
Figure 4.9 Mode of Failure for Specimens Meshonly, 8A, and 8C ................................ 148
Figure 4.10 Observed Cracking in Specimen 8A, 8C, and Meshonly ............................. 151
Figure 5.1 Flowchart of PBO Technique for Non-Reversed Loadings ........................... 156
Figure 5.2 An Example of Checker Board Pattern (Li et al, 2001) ................................. 159
Figure 5.3 Design Domain for the Michell Simply Supported Structure ........................ 161
Figure 5.4 Results in the Selected Iteration of Topology Optimization: (a) Topology
at 5th Iteration; (b) Topology at 35th Iteration; (c) Topology at 50th Iteration; and
(d) Topology at 5ih Iteration ..................................................................................... 162
Figure 5.5 Performance Index History of Michell Simply Support Structure ................. 162
Figure 5.6 Optimal Topology and Performance Index History generated from
Performance Based Topology Optimization (PBO) Technique: (a) D-Region in
Fig.3.la; (b) D-Region in Fig.3.lb; and (c) D-Region in Fig.3.lc ............................. 165
Figure 5.7 Shear Wall with Openings subjected to Reversed Static Loadings ............... 166
Figure 5.8 The Conventional Approach for Design of Complex Regions subjected to
Multiple Load Cases (Tjhin and Kuchma, 2002) ....................................................... 167
Figure 5 .9 Flowchart of PBO Technique for Reversed Loadings ................................... 169
Figure 5.10 Optimal Topology, STM, and Performance Index History of the Shear
Wall Side of D-Regions in Fig. 3.ld: (a) Optimal Topology; (b) Performance
Index History; (c) STM generated from CAST for Load Case 1; and (d) STM
generated from CAST for Load Case 2 ...................................................................... 170
Figure 5.11 Measured Load-Deformation Response of the Test Complex Regions
Xll
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(Specimen 6A to 7C), Note: 1) 1 in.= 25.4 mm. 2) 0.2248 kips= 1 kN ................... 173
Figure 5 .12 Modes of Failure for Models 7 A to 7C ........................................................ 174
Figure 5.13 Observed Cracking in Specimen 6A to 7C tested at P=222, 356, and 534
kN (49.9, 80.3, and 120 kips) ..................................................................................... 175
Figure 5.14 Measured Load-Strain in Surface Concrete and Reinforcement around
the Opening of Model 7B (Note: 0.2248 kips= 1 kN) ............................................... 177
Figure 5.15 Measured Load-Strain in Surface Concrete and Reinforcement around the
Opening of Model 7C ................................................................................................. 177
Figure 5.16 Measured Load-Deformation Response of Specimens 8A to C .................. 178
Figure 5 .17 Mode of Failure for Specimens Meshonly, 8A, 8B, and 8C ........................ 180
Figure 5.18 Observed Cracking in Specimen 8A, 8C, and Meshonly ............................. 182
Figure 5.19 FE Models of Specimen 4A to 7C ............................................................... 187
Figure 5.20 Effects of Mesh Refinement on Predicted Load-Deformation Response of
Complex D-Regions (Note: 1) 1 in.= 25.4 mm, 2) 0.2248 kips= 1 kN ) ................. 189
Figure 5 .21 The Overall Accuracy of the Calibrated FEA Models adopted to Predict
the Structural Responses of the Specimens 4A to 7C ................................................ 190
Figure 5.22 Predicted Load-Deformation Responses of Specimen 4A to 4E (Mesh
Size= 25.4 mm (1 in.)), Note: 1) 1 in.= 25.4 mm, 2) 0.2248 kips= 1 kN ................ 191
Figure 5.23 Predicted Load-Deformation Responses of Specimens 5A to SD (Mesh
Size= 25.4 mm (1 in.)), Note: 1) 1 in.= 25.4 mm, 2) 0.2248 kips= 1 kN ................ 192
Figure 5 .24 Structural Performance Comparison of 16 Complex Regions ..................... 192
Figure 6.1 FE Models of a Complex D-Region ............................................................... 198
Figure 6.2 The Mapped and Physical Coordinates of Embedded Reinforcement
including the Free Body Diagram for Bond Stress-Slip Problem .............................. 200
Figure 6.3 The Local and Global Displacement Components of Embedded
Reinforcement and Reinforced Concrete Element ..................................................... 203
Figure 6.4 Constitutive Model of Bond Stress-Slip and Embedded Reinforcement.. ..... 207
Figure 6.5 The Anchored Reinforcing Bar under Monotonic Pull-Out Loading ............ 211
Figure 6.6 The Predicted Stress Distribution for Anchored Reinforcing Bar under
Monotonic Pull-Out Loading: (a) The Left End Stress= 20 ksi; (b) The Left
End Stress = 40 ksi; ( c) The Left End Stress = 60 ksi; and ( d) The Left End
Stress= 70 ksi ............................................................................................................. 213
Figure 6.7 The Anchored Reinforcing Bar under Monotonic Pull-Push Loading
(Casel) ........................................................................................................................ 214
Figure 6.8 The Predicted Stress Distribution for Anchored Reinforcing Bar under
Monotonic Pull-Push Loading (Case 1): (a) End Stresses= IO ksi; and (b) End
Stresses = 24 ksi ......................................................................................................... 215
Figure 6.9 The Anchored Reinforcing Bar under Monotonic Pull-Push Loading
(Case2) ........................................................................................................................ 216
Figure 6.10 The Predicted Stress Distribution for Anchored Reinforcing Bar under
Monotonic Pull-Push Loading (Case 2): (a) End Stresses= 40 ksi; and (b) End
Stresses = 68 ksi ......................................................................................................... 217
Figure 6.11 The Anchored Reinforcing Bar under Monotonic Push-in Loading ............ 218
Figure 6.12 The Predicted Strain Distribution for Anchored Reinforcing Bar under
Monotonic Push-in Loading (Case 1) ......................................................................... 219
Figure 6.13 The Predicted Strain Distribution for Anchored Reinforcing Bar under
Xlll
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Monotonic Push-in Loading (Case 2) ......................................................................... 219
Figure 6.14 Geometry and FE model ofa Double Span Deep Beam No.5 tested by
Foster (1992): (a) Dimensions and Reinforcement Details; and (b) FE Mesh and
Embedded Reinforcement Elements for the Left Span of a Double Span Deep
Beam No.5 .................................................................................................................. 221
Figure 6.15 Plots of Measured and Predicted Load-Displacement Responses of a
Double Span Deep Beam No.5 tested by Foster (1992) ............................................. 223
Figure 6.16 Plots of Measured and Predicted Strain Distribution along the Bottom
Reinforcement of the Left Side of a Double Span Deep Beam No.5 tested by
Foster (1992) .............................................................................................................. 224
Figure 6.17 The Predicted Strain Distribution for Specimen 8B from the Strain Gages
No.13 to 24 at the Load Levels of 1st crack, 50%Pn, and 75%Pn .............................. 226
Figure 6.18 The Predicted Strain Distribution for Specimen 8B from the Strain Gages
No. 31 to 33 at the Load Levels of 1st crack, 50%Pn, and 75%Pn ............................. 227
Figure 6.19 The Predicted Strain Distribution for Specimen 8B from the Strain Gages
No. 34 to 37 at the Load Levels of 1st crack, 50%Pn, and 75%Pn ............................. 227
Figure 6.20 The Predicted Strain Distribution for Specimen 8C from the Strain Gages
No. 40 to 51 at the Load Levels of 1st crack, 50%Pn, and 75%Pn ............................. 228
Figure 6.21 The Predicted Strain Distribution for Specimen 8C from the Strain Gages
No. 58 to 60 at the Load Levels of 1st crack, 50%Pn, and 75%Pn ............................. 228
Figure 6.22 The Predicted Strain Distribution for Specimen 8C from the Strain Gages
No. 61 to 63 at the Load Levels of 1st crack, 50%Pn, and 75%Pn ............................. 229
Figure 6.23 Reinforced Concrete Element: Crack Angle and Principal Stress Angle .... 231
Figure 6.24 Free Body Diagram for Equilibrium Equation ............................................. 233
Figure 6.25 Free Body Diagram for Local Crack Equilibrium Condition ...................... 234
Figure 6.26 Free Body Diagram for Local Crack Equilibrium Condition including
Bond-Slip Mechanism ................................................................................................ 236
Figure 6.27 Schematic Presentation: (a) Tension Chord Model (Foster and Marti,
2003); (b) Free Body Diagram of Principal Concrete Tensile Stress and Concrete
Stress Between Cracks; and (c) The Force Transfers between and at Cracks in
TCM (Adapted from Foster and Marti, 2003) ............................................................ 237
Figure 6.28 Compatibility Condition: (a) Continuum Strain; (b) Crack Slip Strain;
and (c) Total Strain (Taken from Vecchio, 2000 with Modifications) ....................... 241
Figure 6.29 Material Models: (a) Reinforcing Bars; (b) Concrete; (c) Biaxial
Compression Envelope (Taken from Vecchio, 1992); and (d) New Compression
Softening Coefficient (Cs) .......................................................................................... 243
Figure 6.30 Mesh Dependent Effect in D-Regions ......................................................... 246
Figure 6.31 Components of Stress and Motion along Cracks ......................................... 248
Figure 6.32 Flowchart to Solve for the Nonlinear Response of an Element. .................. 253
Figure 6.33 Flowchart to Solve for the Nonlinear Response (Structural Level) ............. 256
Figure 6.34 Shear Panel subjected to General Loading ................................................... 257
Figure 6.35 Predicted Shear and Shear Strain of PVl 9 ................................................... 258
Figure 6.36 Predicted Shear and Longitudinal Strain of PV19 ....................................... 258
Figure 6.37 Predicted Shear and Transverse Strain of PVl 9 .......................................... 259
Figure 6.38 Predicted Principal Compressive Stress and Strain of PV19 ....................... 259
Figure 6.39 Predicted Shear and Principal Stress Orientation of PVl 9 .......................... 260
XIV
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 6.40 Predicted Shear and Principal Strain Orientation of PV 19 .......................... 261
Figure 6.41 Predicted Shear Strength of the Selected Panels in PV and PB Series
using the Proposed FE Formulation in the Element Level. ........................................ 264
Figure 6.42 Predicted First Shear Crack of the Selected Panels in PV and PB Series
using the Proposed FE Formulation in the Element Level.. ....................................... 264
Figure 6.43 FE Models of Specimen 4A, B, and C: (a) FE Model for Specimen 4A;
(b) FE Model for Specimen 4B; and (c) FE Model for Specimen 4C ........................ 267
Figure 6.44 Predicted Load Displacement for Specimens 4A, B, and C ........................ 268
Figure 6.45 Predicted Failure Mode of Specimen 4A: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern .............................. 269
Figure 6.46 Predicted Failure Mode of Specimen 4B: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern .............................. 270
Figure 6.47 Predicted Failure Mode of Specimen 4C: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern .............................. 271
Figure 6.48 Concrete Strain Prediction in Specimen 4A. ................................................ 272
Figure 6.49 Reinforcement Strain Prediction in Specimen 4A. ...................................... 273
Figure 6.50 FE Models of Specimen 6A, C, and D: (a) FE Model for Specimen 6A;
(b) FE Model for Specimen 6C; and (c) FE Model for Specimen 60 ........................ 277
Figure 6.51 Predicted Load Displacement for Specimens 6A, C, and D ........................ 278
Figure 6.52 Predicted Failure Mode of Specimen 6A: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern .............................. 279
Figure 6.53 Predicted Failure Mode of Specimen 6C: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern .............................. 280
Figure 6.54 Predicted Failure Mode of Specimen 6D: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern .............................. 281
Figure 6.55 Concrete and Reinforcement Strain Prediction in Specimen 6A ................. 282
Figure 6.56 Concrete and Reinforcement Strain Prediction in Specimen 6C ................. 283
Figure 6.57 Concrete and Reinforcement Strain Prediction in Specimen 60 ................. 284
Figure 6.58 FE Models of Specimen 8A, B, and C: (a) FE Model for Specimen 8A;
(b) FE Model for Specimen 8B; and (c) FE Model for Specimen 8C ........................ 286
Figure 6.59 Predicted Load Displacement for Specimens 8A, B, and C ........................ 288
Figure 6.60 Predicted Failure Mode of Specimen 8A: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern .............................. 289
Figure 6.61 Predicted Failure Mode of Specimen 8B: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern .............................. 290
Figure 6.62 Predicted Failure Mode of Specimen 8C: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern .............................. 291
Figure 6.63 Concrete Strain Prediction in Specimen 8B ................................................. 292
Figure 6.64 Concrete Strain Prediction in Specimen 8C ................................................. 293
Figure 6.65 FE models with the Three Different Mesh Refinement Levels of the
Specimens generated to Use in VT2 and the Proposed FE Formulation: (a) Mesh
Size = 8x Max Agg; (b) Mesh Size = 1 Ox Max Agg; and ( c) Mesh Size = 12x Max
Agg ............................................................................................................................. 295
Figure 6.66 The Predicted Load Displacement of Specimen 6C using Three Different
Mesh Refinement Levels analyzed by Using VT2 and the Proposed FE
Formulation ................................................................................................................ 296
xv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 6.67 Comparison of Concrete Strain Prediction between the Proposed
Formulation and VT2 using Different Mesh Refinement in Specimen 6C ................ 297
Figure 6.68 Comparison of Reinforcement Strain Prediction between the Proposed
Formulation and VT2 using Different Mesh Refinement in Specimen 6C ................ 298
Figure 6.69 Influence of the Deviation of the Crack Angle from the Principal Stress
Direction in Concrete on the Structural Responses in the Element Level,
Case 1 ('t': crx:cry=l :0:-0.39) ........................................................................................ 303
Figure 6.70 Influence of the Deviation of the Crack Angle from the Principal Stress
Direction in Concrete on the Structural Responses in the Element Level, Case 2
('t': crx:cry=l :-0.39:-0.39) .............................................................................................. 304
Figure 6.71 Influence of the Deviation of the Crack Angle from the Principal Stress
Direction in Concrete on the Structural Responses in the Element Level, Case 3
('t': crx:cry=l :-0.69:-0.69) .............................................................................................. 305
Figure 6.72 Influence of the Deviation of the Crack Angle from the Principal Stress
Direction in Concrete on the Interaction Diagram of a Shear Panel. ......................... 306
Figure 6.73 Effect of Mesh Size and Difference between Crack and Principal Stress
Angles on Capacity of Specimens 8A, 8B, and 8C .................................................... 308
Figure 6.74 The Crack Pattern just before the Failure of the Specimens 8A to 8C ........ 310
Figure 6.75 FE Models of Three Simply Supported Deep Beams .................................. 311
Figure 6.76 Force Transfer Mechanism in Specimens 8A to 8C .................................... 312
Figure 6.77 Effect of the Difference between Crack and Principal Stress Angles on
the Capacity of Simply Supported Deep Beams ........................................................ 313
Figure 7.1 The Proposed Design and Analysis Framework ............................................ 318
Figure 7.2 Complex D-Regions in a Typical Building .................................................... 321
Figure 7 .3 Member and Loads for the Propped Cantilever Deep Beam with an
Opening ...................................................................................................................... 323
Figure 7.4 Strut-and-Tie Model and Forces: (a) Overview; and (b) Details around the
Opening ...................................................................................................................... 326
Figure 7.5 Force Distribution in Struts and Ties ............................................................. 327
Figure 7.6 The Effective Widths selected for the Struts and Ties and the Utilization
Rates (Fdemand!Fcapacity) in STM of the Deep Beam ..................................................... 327
Figure 7. 7 The Stress-Strain Relationship of "Key" Struts and Ties for Plastic Truss
Analysis ...................................................................................................................... 329
Figure 7.8 The Effective Widths selected for the Struts and Ties and Their
Utilization Rates (Fdemand!Fcapacity) in STM of the Deep Beam (Plastic Truss
Analysis) ..................................................................................................................... 329
Figure 7.9 The Stress-Strain Relationship of "key" Struts and Ties for Nonlinear
Elastic Truss Analysis ................................................................................................ 332
Figure 7 .10 The Effective Widths selected for the Struts and Ties and the Utilization
Rates (Fctemand/Fcapacity) in STM of the Deep Beam (Nonlinear Elastic Truss
Analysis) .................................................................................................................... 332
Figure 7.11 FE Mesh of a Reinforced Concrete Deep Beam constructing with 1355
Quadrilateral Elements and 226 Truss Elements ........................................................ 333
Figure 7.12 The Predicted Load and Displacement of the Deep Beam from
Nonlinear Analysis Techniques: Plastic Analysis, Nonlinear Elastic Analysis,
xvi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and Nonlinear FEA ..................................................................................................... 334
Figure 7 .13 Concrete Compressive Vital Sign in the Deep Beam: Ratio of
Compressive Stress to Compressive Stress Capacity ................................................. 335
Figure 7.14 Combined Deformed Shape and Crack Pattern at the Failure State of the
Deep Beam ................................................................................................................. 335
Figure 7.15 Stress (at Crack Locations) in Reinforcement of the Deep Beam at the
Failure State ................................................................................................................ 336
Figure 7.16 Reinforcement Details of the Scale-Model .................................................. 337
Figure 7.17 Crack Pattern of the Scale-Model at the Failure State ................................. 338
Figure 7.18 Shear Compression Failure above and below the Opening of the Scale-
Model .......................................................................................................................... 338
Figure 7 .19 Members and Loads of a Simply Supported Deep Beam with a
Rectangular Opening .................................................................................................. 342
Figure 7.20 The Optimal Topology for a Simply Supported Deep Beam with a
Rectangular Opening: (a) The Optimal Topology; (b) The Performance Index; and
(c) STM model from CAST ........................................................................................ 345
Figure 7.21 STM Model from CAST for a Simply Supported Deep Beam with a
Rectangular Opening: (a) The Force Distribution in Struts and Ties; (b) The
Effective Widths selected for the Struts and Ties and the Utilization Rates
(Fdemand!Fcapacity) in STM of the deep beam ................................................................. 346
Figure 7.22 Reinforcement Details and FE model of a Simply Supported Deep Beam
with a Rectangular Opening: (a) Reinforcement Details; and (b) FE Model.. ........... 348
Figure 7.23 The Predicted Load and Displacement of the Deep Beam from Nonlinear
FEA by Vector2 .......................................................................................................... 350
Figure 7.24 Concrete Compressive Vital Sign in the Deep Beam: Ratio of
Compressive Stress to Compressive Stress Capacity ................................................. 3 51
Figure 7.25 Combined Deformed Shape and Crack Pattern at the Failure State of the
Deep Beam ................................................................................................................. 351
Figure 7.26 Stress (at Crack Locations) in Reinforcement of the Deep Beam at the
Failure State ................................................................................................................ 352
Figure 7.27 The Structural Performance Evaluation of Designed Simply Supported
Deep Beams with an Opening .................................................................................... 352
Figure 7.28 Members and Loads of a Simply Supported Deep Beam with a Dapped
End and an Opening ................................................................................................... 3 54
Figure 7.29 The Optimal Topology and STM for a Simply Supported Deep Beam
with a Dapped End and a Rectangular Opening: (a) The Optimal Topology; (b)
The Performance Index; (c) STM Model (Optimal); and (d) STM Model (More
Practical) ..................................................................................................................... 356
Figure 7.30 STM Model (Optimal) from CAST for a Simply Supported Deep Beam
with a Dapped End and a Rectangular Opening: (a) The Force Distribution in
Struts and Ties; (b) The Effective Widths selected for the Struts and Ties and the
Utilization Rates (Fdemand/Fcapacity) in STM of the Deep beam .................................... 358
Figure 7.31 STM Model (More Practical) from CAST for a Simply Supported Deep
Beam with an Dapped End and a Rectangular Opening: (a) The Force Distribution
in Struts and Ties; (b) The Effective Widths selected for the Struts and Ties and
the Utilization Rates (Fdemand/Fcapacity) in STM of the Deep beam .............................. 359
xvu
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 7.32 Optimal Reinforcement Details and FE model of a Simply Supported
Deep Beam with a Dapped End and a Rectangular Opening: (a) Reinforcement
Details; and (b) FE Model .......................................................................................... 364
Figure 7.33 More Practical Reinforcement Details and FE model of a Simply
Supported Deep Beam with a Dapped End and a Rectangular Opening: (a)
Reinforcement Details; and (b) FE Model ................................................................. 365
Figure 7.34 Concrete Compressive Vital Sign in the Deep Beam (Optimal
Reinforcement): Ratio of Compressive Stress to Compressive Stress Capacity ........ 366
Figure 7.35 Combined Deformed Shape and Crack Pattern at the Failure State of the
Deep Beam ................................................................................................................. 366
Figure 7.36 Stress (at Crack Locations) in Reinforcement of the Deep Beam at the
Failure State ................................................................................................................ 367
Figure 7.37 Concrete Compressive Vital Sign in the Deep Beam (More Practical
Reinforcement): Ratio of Compressive Stress to Compressive Stress Capacity ........ 367
Figure 7.38 Combined Deformed Shape and Crack Pattern at the Failure State of the
Deep Beam (More Practical Reinforcement) ............................................................. 368
Figure 7.39 Stress (at Crack Locations) in Reinforcement of the Deep Beam (More
Practical Reinforcement) at the Failure State ............................................................. 368
Figure 7.40 The Predicted Load and Displacement of the Deep Beams by Using
the Proposed Nonlinear FEA ...................................................................................... 369
Figure 7.41 Reinforcement Details of Specimen 7B ....................................................... 3 70
Figure 7.42 Reinforcement Details of Specimen 7C ....................................................... 3 70
Figure 7.43 Crack Pattern of the Specimen 7B at the Failure State ................................ 371
Figure 7.44 Crack Pattern of the Specimen 7C at the Failure State ................................ 371
Figure 7.45 The Structural Performance Evaluation of Designed Simply Supported
Deep Beams with a Dapped End and an Opening ...................................................... 3 73
Figure 7.46 Lateral Loads transferred from the Upper Storey to the Shear Wall ........... 375
Figure 7.47 Members and Loads of a Shear Wall with Openings (No Consideration of
the Applied Gravity Load and Bending Moment) ...................................................... 375
Figure 7.48 Optimal Topology, STM, and Performance Index History of the Shear
Wall Side ofD-Regions in Fig. 3.ld: (a) Optimal Topology; (b) Performance
Index History; (c) STM generated from CAST for Load acting on the Left Side;
and (d) STM generated from CAST for Load acting on the Right Side .................... 378
Figure 7.49 The Conventional Approach for Design of Complex Regions subjected to
Multiple Load Cases (Tjhin and Kuchma, 2002): (a) STM generated from CAST
for Load acting on the Left Side; and (b) STM generated from CAST for Load
acting on the Right Side .............................................................................................. 379
Figure 7.50 STM Model (Optimal) from CAST for a Shear Wall with Rectangular
Openings: (a) The Force Distribution in Struts and Ties; (b) The Effective Widths
selected for the Struts and Ties and the Utilization Rates (Fctemanct!Fcapacity) in STM
of the Shear Wall ........................................................................................................ 381
Figure 7.51 STM Model (Conventional Approach) from CAST for a Shear Wall with
Rectangular Openings: (a) The Force Distribution in Struts and Ties; (b) The
Effective Widths selected for the Struts and Ties and the Utilization Rates
(FctemanctlFcapacity) in STM of the Shear Wall ................................................................ 382
Figure 7.52 Optimal Reinforcement Details and FE Model of a Shear Wall with
xviii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Rectangular Openings (7B): (a) Reinforcement Details; and (b) FE Model.. ............ 388
Figure 7.53 Orthogonal Reinforcement Details and FE Model of a Shear Wall with
Rectangular Openings (7C): (a) Reinforcement Details; and (b) FE Model.. ............ 389
Figure 7.54 Concrete Compressive Vital Sign in the Shear Wall (Optimal
Reinforcement)-The Right Side of the Specimen 8B: Ratio of Compressive
Stress to Compressive Stress Capacity ....................................................................... 390
Figure 7.55 Combined Deformed Shape and Crack Pattern at the Failure State of the
Shear Wall (Optimal Reinforcement)-The Right Side of the Specimen 8B ............... 390
Figure 7.56 Stress (at Crack Locations) in Reinforcement at the Failure State of the
Shear Wall (Optimal Reinforcement)-The Right Side of the Specimen 8B ............... 391
Figure 7.57 Concrete Compressive Vital Sign in the Shear Wall (Orthogonal
Reinforcement)-the Right Side of Specimen 8C: Ratio of Compressive Stress
to Compressive Stress Capacity ................................................................................. 391
Figure 7.58 Combined Deformed Shape and Crack Pattern at the Failure State of the
Shear Wall (Orthogonal Reinforcement)-The Right Side of the Specimen 8C
the Deep Beam (More Practical Reinforcement) ....................................................... 392
Figure 7.59 Stress (at Crack Locations) in Reinforcement at the Failure State of the
Shear Wall (Orthogonal Reinforcement)-The Right Side of the Specimen 8C ......... 392
Figure 7.60 The Predicted Load and Displacement of the Shear Walls by Using
the Proposed Nonlinear FEA ...................................................................................... 393
Figure 7.61 Reinforcement Details of Specimen 8B ....................................................... 393
Figure 7.62 Reinforcement Details of Specimen 8C ....................................................... 394
Figure 7.63 Crack Pattern at the Failure State of Shear Wall with Openings: (a) Right
Side of Specimen 8B; and (b) Right Side of Specimen 8C ........................................ 394
Figure 7.64 The Structural Performance Evaluation of Designed Shear Walls with
Openings ..................................................................................................................... 3 96
Figure A I The Screen of Augustus Displaying the Analysis Details and Parameters .... 415
Figure A2 The Screen of Augustus Displaying the Mesh Topological Data, Boundary
Conditions and Load Cases of the Model.. ................................................................. 416
Figure A3 The Screen of Augustus Displaying the Parameters Associated to the
Damage Conditions and Failure State of the Structures ............................................ .416
Figure A4 The Screen of Augustus Displaying the Global Responses of the Structures 417
Figure AS The Screen of Augustus Displaying the Local Responses of the Structures . 417
Figure A6 The Screen of Augustus Displaying the Embedded Reinforcement Element 418
Figure Bl The Screen of Patran Displaying the Overall Features for Generating the
Geometry, Mesh, Loading and Boundary Condition of FE Models .......................... 420
Figure B2 The Screen of Patran Displaying the Features for Generating the Geometry
of the Structure ........................................................................................................... 421
Figure B3 The Screen of Patran Displaying the Features for Generating the Mesh of
the Structure ................................................................................................................ 421
Figure B4 The Screen of Patran Displaying the Features for Assigning the Element
Types to the Existing Mesh of the Structure ............................................................. .422
Figure C 1 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PVI) .......................................... .424
Figure C2 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV 4) ........................................... 425
XIX
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure C3 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV 6) .......................................... .426
Figure C4 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV7) .......................................... .427
Figure CS Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV8) .......................................... .428
Figure C6 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV9) ........................................... 429
Figure C7 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PVl 0) ......................................... 430
Figure C8 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV 11) ......................................... 431
Figure C9 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV12) ........................................ .432
Figure CI 0 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV14) ......................................... 433
Figure C 11 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV16) ......................................... 434
Figure C12 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PVl 8) ........................................ .435
Figure C 13 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV20) ......................................... 436
Figure C14 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV2 l) ........................................ .43 7
Figure C 15 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV22) ......................................... 43 8
Figure C 16 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV24) ......................................... 4 3 9
Figure Cl 7 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV25) ......................................... 440
Figure C 18 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV26) ......................................... 441
Figure C 19 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV27) ........................................ .442
Figure C20 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV28) ......................................... 443
xx
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Tables
Table 2.1 Parameters in the Local Bond-Slip Relationship of CEB-FEP Model Code
1990 for Deformed Bars ................................................................................................ 73
Table 3 .1 Summary of Failure (Ultimate State) Conditions and Nominal Design
Loads of All 17 Truss Models by STM Models and Appendix A of ACl3 l 8-08 ......... 90
Table 3.2 Summary of Concrete Strength Properties ........................................................ 97
Table 3 .3 Summary of Steel Reinforcement Properties .................................................... 99
Table 4.1 Comparison of Calculated Nominal Capacities and Measured Strengths ....... 122
Table 4.2 Comparison of Calculated Nominal Capacities and Measured Strengths ....... 147
Table 5.1 Details of Nonlinear FE Analysis Results of 16 Complex Regions ................ 193
Table 6.1 Material Properties Used in Nonlinear FE Analysis of Deep Beam No.5
(Foster, 1992) ............................................................................................................... 222
Table 7.1 Summary of Material Properties ..................................................................... 324
Table 7.2 Capacity and Failure Mode Comparison of Designs, Analysis, and
Experiment. .................................................................................................................. 3 3 9
Table 7.3 Capacity and Failure Mode Comparison of Designs, Analysis, and
Experiment. .................................................................................................................. 3 72
Table 7.4 Capacity and Failure Mode Comparison of Designs, Analysis, and
Experiment. .................................................................................................................. 3 9 5
xxi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter!
Introduction
D-Regions are those parts of a structure where there is a complex distribution of
stress and strain. The "Strut-and-Tie Model (STM)" is a truss that is idealized to be
contained within the D-Region and that has been designed to support the flow of forces
through this region. The STM design concept, often referred to in the literature as simply
the STM, is based on the lower bound plasticity theorem. Thereby, the calculated
capacity of an STM design can be obtained from a plastic truss analysis; this assumes that
the designed D-Region is sufficiently ductile that a truss mechanism can form. Series of
experimental programs have been conducted to develop and evaluate provisions for
design using the STM concept for strength design. The results from these tests also
illustrate deficiencies in the performance of D-Regions under service loads and ultimate
design loads in some situations. The overall objective of this study is to improve the
current approach for the design and analysis of D-Regions by proposing a more complete
design and analysis framework.
1.1 Background
1.1.1 B and D Regions
A structure, as illustrated in Fig.1.1, can be divided into two types ofregions.
Beam or "B-Regions" are those portions of a structure in which the assumption that plane
section remaining plane is valid such that there is a linear distribution of strain over the
depth or width of the member. These regions may be designed by sectional theories and
code provisions for shear, moment, axial-load, and torsion. Disturbed or "D-Regions" are
St.Venant regions, and are those parts of a structure in which there is a highly nonlinear
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
strain and displacement field due to a geometrical and/or static discontinuity. Examples
of these D-Regions include regions near point loads, corners, and openings as shown in
Fig.1.1.
h, D B
t ....... t
T h2 h
2
h2
-LB i
-,--- +
ii, DLL
h,18 7.,
I h,
______ L
D --r
11111
1
B
1l1!11
(a)
Fig.1.1 Examples of Disturbed Regions (Taken from Tjhin 2004)
1.1.2 Design of D-Regions Using STM
The Strut-and-Tie Model (STM) has been accepted as a simple and rational
approach for the design of D-Regions. As outlined in Fig.1.2, the first step in the STM
design procedure is to divide the structure into B- and D-Regions by considering that the
region of influence of geometrical and static discontinuities. In accordance to the
suggestions of Saint Venant, these are considered to extend a longitudinal distance equal
to about the depth of the member. Next, the engineer is free to select the shape (or
topology) of the load-resisting truss that will be designed to carry the load through this D-
Region. The STM model is composed of compression members or "struts'', tension
members or "ties", and joints or "nodes" as illustrated in Fig.1.3.
2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Determine disturbed
regions
i
Sketch an internal load-resisting truss for each
D-Region that is compatible with the applied
loads and support conditions
i
Determination of forces in all components of this load-
resisting truss of STM components
i
Design each STM component to have
adequate strength.
i
Provide any needed reinforcement to
provide adequate strength and ductility
Fig.1.2 General Procedure for the Design of D-Regions by the STM
The demand on each STM component may be calculated by a simple truss
analysis. If the selected truss is statically indeterminate, then the geometry and stiffness
of each member will be required to determine these demands.
1.1.3 Safety, Validation, and Limitations of the STM in D-Region Design
The safety of the STM design approach is dependent on whether or not the stress
limits in codes of practice and the plasticity assumption are applicable to the D-Region
being designed for the selected shape of strut-and-tie model. For a design in which a
statically determinate truss is selected, the designer may be confident in the safety of a
design if it is justifiably considered that the capacity of the member will not be decreased
by the "filling in" of the empty regions of the idealized truss, such as by filling in the
spaces between the dashed lines in Fig. 1.4 to form a deep beam. The strut-and-tie
3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
approach and associated code provisions have been most thoroughly validated for simple
types of structures, such as deep beams tested by Rogowsky and MacGregor (1983 and
1986) and corbels tested by Mattock et al (1976) and Yong and Balaguru (1994), but
there has been only limited validation for more complex D-Regions ( Maxwell and Breen
, 2000 and Chen et al, 2002).
It is not feasible to fully validate the STM approach and associated code
provisions through the conduct of an experimental research program because the STM
may be used for the design of all regions in structural concrete and since the designer
may choose the shape of the load-resisting truss. For the design of complex D-Regions,
the designer must exercise greater care in the selection of an appropriate idealized truss
and in his or her determination of the applicability of STM code provisions, the plasticity
assumption, and the performance of the structure under service loads and during over
loads. To illustrate a potential concern with a blind application of the STM, consider the
idealized load resisting truss for a <lapped beam with an opening. The selected strut-and-
tie model shown in Fig. 1.Sa is one of more than 30 admissible designs that was
considered for this region. The idealized and designed load resisting truss is shown in
Fig. 1.Sb. While the truss is perfectly admissible, it is reasonable to expect that a crack
could propagate from the inside comer of the <lapped end and lead to failure of the
structure before its nominal capacity, as calculated using a plastic truss assumption, is
realized. Even when following available guidance provided by Schlaich and Schafer
(1991 ), the designer must recognize that the idealized load-resisting truss is just that, an
idealization. In practice, the flow of the forces will be controlled by the full dimensions
of the structure, and this distribution will change due to cracking and other factors that
4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
affect the stiffness characteristics of the continuum of the concrete structure within which
lies the idealized load-resisting truss. The complexity and the characteristics of the flow
of the forces in complex D-Regions and the examination of the safety and applicability of
the STM philosophy and associated ACI code provisions for design of more complex
regions require the more detailed investigation based on the series of experimental
programs.
----strut
-Tie
Node
Fig.1.3 Examples of STM (Taken from Tjhin 2004)
Fig.1.4 Reinforced Concrete Truss (Adapted from Polla, 1992)
5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
(b)
Fig.1.5 Idealized Reinforced Concrete Truss for More Complex D-Regions: (a)
Selected STM (One of Dozens of Possibilities); (b) Idealized Reinforced Concrete
Truss for More Complex D-Region.
1.1.4 Analysis Approaches for D-Regions and Their Limitations
Section 11.8.2 in ACB 18-08 states that
"Deep beams shall be designed using either nonlinear analysis as permitted in
10. 7 .1, or Appendix A." Section 10. 7.1 allows the use of nonlinear stress fields when
proportioning deep beams.
For the ultimate limit state check, section 5.3.2 of CEB-FIP90 and section 6.3.l of
FIP ( 1996) both stated that
6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
"The ULS check should be carried out according to the theory of plasticity (PT)
or an appropriate non-linear method."
In according with the code of practices referenced above, designers are thus
permitted to design D-Regions, such as, deep beams by using either linear, nonlinear, or
plastic analysis. However, while the use of either a nonlinear STM approach or nonlinear
FEA can improve the design and performance of D-Regions, the maturity of these
methods is not sufficient for them to be used by design engineers without considerable
guidance. The currently shortcomings of these methods include:
1) Stress limits for members in codes of practice do not fully account for the
influence of cracking, reinforcement placement, and the state of straining. Consequently,
these stress limits are not sufficiently accurate and often too conservative to use within
nonlinear truss analysis methods for calculating the structural responses required in limit
states design.
2) Existing nonlinear strut-and-tie methods are not applicable to a wide-range of
structures as available constitutive models do not capture the behavior of all struts and
ties and codes of practice do not provide sufficient guidance for assessing effective
member dimensions.
3) Performance based approach cannot yet be applied to D-Regions designed and
analyzed by STM provisions in codes-of-practice due to the limitation of the STM in
obtaining the displacement history, local damages characteristics, and the failure mode of
D-Regions, while the non-linear truss analysis methods developed by researchers are
generally limited to very specific types of STM D-Region designs.
7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4) A nonlinear STM for the analysis of complex D-Regions is still overly complex
for use in practice. It requires careful selection of the STM shape, dimensions,
constitutive models, and stress limits, particularly when completing a complex and
indeterminate STM design.
5) The use of existing nonlinear PEA methods typically requires substantial effort
and time to construct the structural geometry of the FE model, and further requires that
the user is able to make appropriate selections for material models and can correctly
interpret the results of analyses.
6) The result of the prediction from the existing nonlinear PEA is not unique due to
the variation and the uncertainty of the model parameters. In addition, most nonlinear
PEA tools for structural concrete have been designed for users to select from different
models for behavior, constitutive relationships, failure theories, bond properties, mesh
size, etc. This is too flexible for designers who are seeking a unique and best-possible
prediction of the structural response.
1.1.5 Simple and Complex D-Regions: A New Terminology for This Research
For the purpose of this study, the author uses the following classification for D-
Regions; "Simple" and "Complex". Simple D-Regions will be taken as those portions of
structures whose behavior can be "sufficiently" analyzed by using simple nonlinear STM.
These types ofD-Regions include most simple deep beams, corbels, stocky shear walls,
and beam-column connections. The structural behaviors of complex D-Regions are, in
contrast, considered to be "insufficiently" or "impractically" analyzed by a simple
nonlinear STM.
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.2 Objectives
The overall objective of this research study is to improve the current approach for
the design and analysis of complex D-Regions. The research study will include the
examination of the safety and applicability of the STM philosophy and associated ACI
code provisions for the design of more complex regions of structures in which the
idealized internal load-resisting truss has many members. This evaluation will be made
through experimental research programs of the more complex regions of structures. In
addition, an expanded design and analysis framework will be proposed that aims to
ensure good structural performance under service loads and during overloads. This design
and analysis framework is composed of two major parts: 1) Methods and guidelines to
facilitate improved STM shape selection and 2) Nonlinear FEA for automated analysis of
complex D-Regions. The flowchart for this design and analysis framework is illustrated
in Fig.1.6 below.
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Geometry, Loading, and
s. ..
, :_,:,:;, ,, , ,' , : "':',,:


Simple D-Regions

Complex D-Regions


+Yes

N ... Adjust or Refine

Finish

Fig.1.6 The Proposed Design and Analysis Framework
The colored components shown in Fig.1.6 are the three new parts of this
expanded design and analysis framework in which the current scope of the research is to
focus on only the shape selection technique based on topology optimization and the
nonlinear FEA. Those two components are most necessary in design and analysis of
complex D-Regions.
The methods and guidelines to facilitate the STM shape selection are required to
ensure that an appropriate STM shape is adopted for the STM design of D-Regions and
thereby to help ensure good structural performance, including under service load levels.
This is also to overcome the complexity of the STM design since there is no single
solution due to the flexibility in STM shape selection. Since the current research and
development of the techniques for STM shape selection have never been focused on the
practical guidelines leading to the good structural performance design of D-Regions, the
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
proposed research in this part will thus focus on reviewing a very effect technique
(Topology Optimization) for STM shape selection and producing the guidelines to obtain
the STM shapes leading to the good structural performance design of D-Regions.
In topology optimization, D-Regions with the completed descriptions of
geometry, loading, and boundary conditions are discretized into small FE domains by
using either manual or automatic mesh generation available in most of the pre-processing
tools, e.g. Patran, for high quality meshes. The discretized structures are used as the
initial domain where the optimal topology can be obtained by gradually removing the
underutilize elements from the initial domain. The optimal topology is then adopted to
characterize the appropriate STM shape used in the STM design provisions.
After designing D-Regions using STM design provisions in codes of practice, the
validation of complex STM designs is essential to ensure adequate strength and
satisfactory performance at service load level and during over loads. Validation of a
design becomes more important as the complexity of a STM design increases. Fig. 1.4
and 1.Sb provide two examples of idealized internal trusses. Fig. 1.4 presents a deep
beam for which a simple, three-member, statically determinate truss is designed to
support the applied load. The structure is simple, and validation of this design is
unnecessary. In contrast, Fig. 1.Sb presents one of many possible STM designs for a
complex D-Region. For this valid design, concern by the designer is justified, a crack
could develop at the inside comer of the dapped-end and lead to unsatisfactory
performance under service loads and/or failure before the calculated nominal (plastic
truss) capacity is reached. For complex strut-and-tie designs, an engineer's intuition
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
about the structural performance and capacity is likely to be insufficient, and the author
suggests that validation is required.
In the proposed design and analysis framework, for complex D-Regions involving
complexity of geometry, loading, and boundary conditions, an automated nonlinear FEA
model validation methodology is proposed and adopted.
Since the existing nonlinear FEA analysis technique is still unpractical and time
consuming in constructing the model of complex D-Regions and is deficient and
uncertain for predicting the nonlinear structural responses of complex D-Regions as
described in section 1.1.4, this research included the development of improved
computational models that are sufficiently accurate, practical, and reliable in predicting
the nonlinear structural responses of complex D-Regions.
Through the contributions in these areas, the improved and expanded design and
analysis framework for complex D-Regions presented in Fig.1.6 can be realized. This
contribution is proposed to be made through the completion of the following tasks.
Review and propose guidelines for selecting STM shapes for complex D-Regions.
Propose a new nonlinear FE formulation specialized for automated analysis of
complex D-Regions in concrete structures.
Propose and implement the design and analysis framework included the nonlinear
FEA for complex D-Regions.
1.3 Overview of Dissertation
This dissertation consists of 8 chapters. Chapter 2 provides a literature review of
the basis of STM design provision, nonlinear STM approaches developed by
researchers, and nonlinear FEA methods. Chapter 3 describes the experimental program
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
used to validate the subsequent research activities. Chapter 4 presents the results of the
experimental program of complex D-Regions from Chapter 3. Chapter 5 and 6 describe
and present the research activities on assessing the shapes of STM in the design of
complex D-Regions and developing improved FEA for automated analysis of complex
D-Regions, respectively. In each of these chapters, the scope of the research activities
will be presented, analysis results and examples will be provided to illustrate and
validate the research activities. Chapter 7 presents the concepts and application of the
proposed design and analysis framework for effective structural performance design of
complex D-Regions. Lastly, the conclusions, expected contributions to the field, and the
recommendation for future research are presented in Chapter 8.
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 2
Literature Review
This chapter will begin with a summary of the history and development of the
STM design provisions for D-Regions. Next, an overview of nonlinear STM formulations
is presented. Finally, a review is made of nonlinear FEA methods for structural concrete
including the general theory, framework, material models, and mesh dependency.
2.1 History and Development of STM for Design of D-Regions
The Strut-and-Tie Model originated from the 45-degree parallel-chord truss
model for the flow of forces in a simply-supported beam, as introduced by Ritter (1899)
and Morsch (1909). In this parallel-chord truss model, shown in Fig.2.1, the applied loads
are balance by diagonal compressive forces, stirrups lift up the diagonal compression, the
top chord forces are supported by compression in the concrete and bottom reinforcement
serves as the tension chord in the truss.
.
.. '
..
e
<- >
Fig.2.1 Ritter's Truss Model (Ritter, 1899)
Subsequently, a truss model which allows for variable angles of inclination of
diagonal compression was developed by Lampert and Thurlimann ( 1968 and 1969) that
was based on plasticity theory. Schlaich et al (1987, 1991) extended the truss model
approach for the design of D-Regions in structural concrete and referred to this as the
strut-and-tie model (STM) approach. Research on the STM has been conducted by
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
several researchers including Mitchell and Collins (1980), Marti (1985, 1986), Rogowsky
and MacGregor (1983, 1986), and Nielsen et al (1978). Over the past few decades,
provisions for design by the STM have been introduced into codes-of-practice including:
as Appendix A "Strut-and-Tie Model" in Appendix A of the American Concrete
Institute in 2002 (ACB 18-02); these provisions remain practically unchanged
through ACB 18-08;
in the Canadian Standards Association in 1984 (CSA-M84);
in the Comite Euro-International du Beton (1993), CEB-FIP Model Code 1990;
FIP Commission 3 (1999), Practical Design of Structural Concrete, Federation
internationale de la Precontrainte;
Australian Standard in 1994, AS3600(1994 ); these provisions remain practically
unchanged through the draft of AS3600(2001)
While these codes and guidelines all provide rules for assessing the strength of
struts, ties, and nodes, they differ substantially in stress limit values, means in assessing
the influence of geometry and strain on capacity, and in minimum reinforcement
requirements. In addition, they also differ in the level of guidance provided to designers.
For example, the FIP guidelines provide suggest truss model shapes for some common
types of D-Regions and a relationship for calculating the distribution of design forces in
statically indeterminate deep beams. By contrast, the ACI and CSA provisions provide no
such guidance.
2.2 Shape of STM
A linear STM simply refers to a linear elastic truss being used for calculating the
forces in the members of the internal load-resisting truss or STM. Designers are free to
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
select the shape of the truss that will be designed for supporting the flow of forces
through a D-Region to its supports. This freedom is provided by the basis of the STM in
the lower-bound theory of plasticity that assumes that the structures are sufficiently
ductile to support the force flow in the manner selected by the designer. STM regions are
typically quite stocky and not very ductile. As observed by researchers, Kuchma et al
(2007), poor selections of the shape of an STM can have significant affects on structural
performance (Fig.1.4). Thus, it is worthwhile to provide designers with guidance in
selecting the shape of a STM. Currently, there are several techniques used to select STM
shape in both simple and complex D-Regions including (1) previous experience, (2)
empirical models, (3) cracked patterns, (4) linear stress trajectory and (5) optimization
techniques as summarized below.
STM shapes selected based on previous experience are generally depended on the
expertise of the designers and the tradition of practice. This approach is thus suitable for
the simple type of D-Regions experimentally proven to be structurally sufficient for all
limited states.
Empirical models guided in the code of practices and literatures are also useful for
constructing the shape of STM. These types of models are, for example the STM for
simply supported deep beam, frame comers, and end regions guided in FIP
Recommendation (1996) as illustrated in Fig.2.2. Most of them are validated by series of
experimental investigation to achieve the acceptable structural performance.
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I I I I I I I 0.4ql
I I I I I
I I J--J, r
',\// \ Ui
I '' // o
o t 0.16ql '\!f' 0.16ql \
-r ri :ilk i! _
10.sq1 o.sq111 o.sq1 o.4q1t
0.7T
(a)
(b)
(c)
--l>i a
.
T T
-Ja i

I
rl
t
Fig.2.2 Shape of STM based on Empirical Model (Reproduced from FIP, 1996): (a)
Empirical STM for a Two Span Deep Beam; (b) Empirical STM for Beam-Column
Joints; and (c) Empirical STM for Post Tension at End Supports.
For design of D-Regions similar to the existing structures whose cracked pattern
data are available, STM shape can be possibly drawn based on the cracked pattern where
the struts are arranged to be parallel between cracked regions (Schlaich and Schafer
1991).
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
F ,
(<l)
,
,
(b)
(e)
s ------- -
A./' J' F
0 /
Tl
I ,'
I ; I
,'C
1
EG1
(Cl
Fig.2.3 Shape of STM based on Crack Pattern (Reproduced from Fu, 2001): (a)
Crack Pattern at a Dapped End; (b) STM Shape No.1; (c) STM Shape No.2; (d)
STM Shape No.3; and (e) STM Shape No.4
Marti (1985) and Schlaich et al (1987) proposed to use principal stress trajectories
and the load path method to obtain the STM shape. In both techniques, the designers are
required to perform a linear elastic FEA on the uncracked homogenous concrete to obtain
the directions of principal stress flow.
l.:JJ.. .l!O. x ltil
1
4 ..
1
, ..
1
I . t. ..
1
rP
d
~ ~ ~ ~ ~ _ l _ J
-- strut
-11e
a) Uniform Load on a Deep Beam
Fig.2.4 (continued on next page)
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
.,...!!._,
J '
[11LP2p I
I- I
e
--
b) Eccentric Concentrated Load on Long Member
Fig.2.4 Shape of STM (a) Stress Trajectory and (b) Load Path Method
(Taken from Schlaich et al, 1987)
Next, the strut-and-tie model is formed by orientating the struts and ties to the
principal stress flow. However, it is often difficult or not practice to obtain suitable STM
shapes for either simple or complex D-Regions. This is largely due to the impracticality
of providing inclined reinforcement, short ties, and curved struts. Consequently, STM
shapes are often obtained only using the general guidance of principal stress flows.
Liang et al (2000) adopted the topology optimization technique or so called
"performance based optimization: PBO" of 2D continuum plan stress structures for
obtaining STM shapes. PBO technique was developed based on an evolutionary
structural optimization (ESO) technique proposed by Xie and Steven (1993) in which
small numbers of underutilized elements are removed in each iteration of an analysis
until the remaining design domain evolves to an optimal solution. Many numerical
examples of determining STM shapes selection by the PBO technique have been
illustrated for both simple and complex D-Regions including deep beams supporting
bottom point loads, deep beam with openings, and corbels.
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
.t,n n n
o,_t,n D r:
D ll
(c)
Fig.2.5 Topology Optimization for Generating Shape of STM (Taken from Liang et
al, 2001 ): (a) Geometry of a Shear Wall with Openings; (b) Optimal Topology; (c)
Shape of STM based on the Optimal Topology; and (d) Dimensions of STM
In these examples, the optimal STM shape was selected based on the
performance index (Pl) history e.g. the STM shape at the maximum PI was the optimal
STM shape used in STM design process. Although the STM shape obtained from this
technique is unique and apparently useful, research has not been conducted to investigate
if members designed using this PBO shape will exhibit superior performance to members
designed using other shapes.
Another optimization approach adopted in STM shape selection was based on
topology optimization of a structural grid as now presented.
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Park and Yun (2004) adopted the minimum strain energy approach of2D and 3D
structural grids for obtaining the optimal STM shape to be used in the design and analysis
of D-Regions (Fig2.6). Rather than using 2D plane stress elements in the design domain,
the dense 2D or 3D structural grid was adopted to describe the design domain and this
grid is referred to as the so called "ground structure".
(a)
(b)
(c)
(d)
Fig.2.6 2D Structural Grid for Generating Shape of STM (Adopted from Ali and
White, 2001): (a) Ground Structural Grids; (b) Structural Grid at the 25th Iteration;
(c) Structural Grid at the 45th Iteration; and (d) Optimal Structural Grid at the soth
Iteration.
By using the same concept as ESO, small numbers of underutilized grid members
are removed from the ground structure until the optimal grid structure is obtained.
Although the concept and computational task are simple, the topology optimization of
structural grid is depended on the selection of grid size and refinement.
2.3 Calculating the Capacity of a STM Model
The capacity of a STM model can be directly obtained when any failures
(ultimate states) of STM members, for examples, strut crushing, tie yielding, or nodal
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
zone failure, occur. The failure of each STM member is due to the excessive stress or
force demand over the STM component capacity. In ACI318-08, the ultimate state design
of STM components can be given as
(2-1)
where Fu is the factored force acting in a STM component, Fn is the nominal capacity of
a STM component, and is the strength reduction factor specified in code of practices.
The nominal capacity of a STM component can be calculated from Eqn (2-2) where the
stress limit (fcu) provided by the code of practices or literatures is multiplied by the
effective dimension (Aeff) obtained from the geometry of STM components.
Fu = fc'UAef f
(2-2)
2.3.1 Geometry of STM Components
Theoretically, a STM is composed of three components e.g. strut, tie, and nodal
zone in which those components can be specifically varied depended on the assumed
stress and strain characteristics, cracking, and local reinforcing details.
ACI318-02, and its essentially identical provisions in ACI318-05 and ACI318-08,
identifies two types of struts, one that is prismatic (Fig.2.7a) and one that is bottle-shaped
(Fig.2.7b). Prismatic strut occur in D-Regions where the structural boundary surrounding
the strut does not allow for bulging of the strut. Examples of prismatic struts are columns,
the compression zone in beams, and the uniform field of diagonal compression in the web
of beams. Bottle-shaped struts occur in regions where the ends of the strut are narrower
than the strut at mid-length. An example is a strut that flows from the point ofloading to
support in a deep beam that supports a point load at mid-span. The other code provisions
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
such as CEB-FIP(l 990) and FIP(l 996) provide the designer with additional strut shapes
including the fan shape strut (Fig.2.7c) as presented by Marti (1985).
For ties, the types of ties introduced in code of practices (ACI, CEB-FIP, and
FIP) are reinforcing ties, concrete tie, and prestressed tie. Tension ties are usually placed
in the orientation of the principal tensile stress in the D-Regions.
Codes of practice generally define nodal zones into one of several types
depending on the combination of STM members that meet at the zone e.g. CCC-node
(Fig.2.7d), CCT-node (Fig.2.7e), CTT-node (Fig.2.7f), or TTT-node (Fig.2.7g). The
CCC node is the nodal zone subjected to the strut acting on all three faces of the nodal
zone. The CCT-node is adopted for the nodal zone subjected to two struts and one tie
acting on all three faces. When one strut and two ties are acting on all three faces of nodal
zone, the nodal zone is defined as CTT-node. If only three ties acting at the nodal zone,
TTT-node is adopted. Except four types of the nodal zones, there are two additional types
of the nodal zones found in codes of practice and literatures e.g. the hydrostatic nodal
zone and the extended nodal zone.
The hydrostatic nodal zone has the loads that are in the alignment of the axes of
the struts and ties and act perpendicularly on the faces of the nodes. For the CCC
hydrostatic case, the stresses on the face of the nodal zone are equal in all three struts and
leads to the same ratios of the lengths of the sides of the nodal zone as the ratios of the
three strut forces. Appendix A of ACB 18-08 also presents a hydrostatic nodal zone of the
CCT type. A CCT nodal zone subjected to hydrostatic state needs to extend the tie
through the node to be anchored by a plate on the far side of the node, as illustrated in
Fig. 2.8a and b for a bar and group of bars, respectively. In addition, when considering
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the critical section for development of tie reinforcement in a CCT nodal type, ACI3 l 8-08
allows designers to take into account the extended nodal zone (Fig.2.8) which is the
portion of a member bounded by the intersection between the effective dimension of strut
and tie.
,.<'.
"'y
(a)
(d)
(f)
(e)
(g)
Fig.2.7 Types of Strut and Nodal Zones (Adapted from Tjhin, 2004): (a) Prismatic
Strut; (b) Bottle-Shaped Strut; (c) Fan-Shaped Strut; (d) CCC Node; (e) CCT Node;
(t) CTT Node; and (g) TTT Node.
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
f
I
I'
(a)
i
w,
I
1
(b)
' Critical section for
development of
tic reiriforcement
Fig.2.8 Guidelines for Dimensions of Nodal Zones (Taken from ACI318-05): (a) One
Layer of Steel; and (b) Distributed Steel.
2.3.2 Stress Limits
"Stress limits" are used to determine the capacity of STM components and
generally given in terms of cylinder strength of concrete (f c) multiplied by efficiency
factors (u ). Efficiency factors (u) are theoretically adopted in the stress limit calculation
due to the imperfection of the concrete material which is not perfectly plastic and soften
under the transverse tensile stress. In general, efficiency factors (u) are based on
empirical formula obtained from either experiments or analytical approaches such as
nonlinear elasticity or plasticity. In this section, only the stress limits obtained from
experiment and nonlinear elasticity are discussed. The stress limits based on plasticity
theory will be reviewed in the subsequent section.
In codes of practices e.g. ACI3 I 8-02, -05, -08, the permissible stress limits in
STM members depend on member type, concrete strength, provided reinforcement, and
state of cracking of concrete in the D-Regions. The general form of the stress limit of
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
struts and nodal zones defined in these ACB 18 codes are constant and given in Eqn (2-
3).
fcu = 0.85Psf C
(2-3)
in which/cu is stress limit, Ps is coefficient depended on type of strut or nodal zone, and
f c is compressive strength of the concrete. For a prismatic strut, Psis suggested to be 1.0.
The values of 0. 75 and 0.65 are for bottle-shaped strut with and without sufficiently
distributed reinforcement, respectively. For struts in tension members, Psis suggested to
be 0.40 and for other cases, Psis to be 0.60.
In the literatures, researchers have proposed other forms of relationships for
evaluating stress limit. Schlaich et al. ( 1987) proposed several constant stress limits (vf c)
for concrete struts based on type of strut, crack width, and reinforcement such as 0.85f c
( v = 0.85) for struts in a field of undisturbed and uniaxial state of compressive stress,
0.68f c for struts having tensile straining or reinforcement passing perpendicularly to the
strut axis, and 0.5lfc for struts crossing skew cracking or reinforcement. MacGregor
(1988) proposed constant stress limits (0.25f c and 0.45 f c) for severely cracked webs of
slender beams with strut angle of 30 and 45 degrees, respectively. The stress limit for arc
and fan type strut was proposed to be 0.50fc by Alshegeir (1992).
The constant stress limits and efficiency factors shown above in both appendix A
of ACB 18-08 and literatures are gross approximations that are mainly empirically
derived. These constant stress limits are unable to capture the effect of structure geometry
(e.g. shear to depth ratio) and distributed reinforcement on strut capacity. In addition,
stress limits based on crack patterns are difficult to adopt in the design process.
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The more refined version of efficiency factors can be found from the literatures
and codes of practices. The efficiency factors based on one parameter are, for examples,
v = 0.5+1.25/ .J7: proposed by Bergmeister et al. ( 1991 ), v = 3 .2/ R proposed by
I
Exner (1979), and v = 0.36 + 4.8/ f c proposed by Grob and Thurlimann (1976). The one
parameter form of the efficiency factor adopted in the code of practice e.g. AS3600
I
(2001) is v = 0.8 - f c/200 which was directly taken from the formulae proposed by
Nielsen et al. (1978). For the multi-parameters form of stress limits, the formulae
proposed by Warwick and Foster (1993) was derived based on FE models and given as
v= 1.25- f ~ 1500-0. 72ald+O. l 8(ald)
2
for aid< 2.0 and v=0.53- f ~ 1500 for aid> 2.0.
Although these efficiency factors using one or multi-parameters are effective in capturing
the effects of several parameters on the capacity of STM members; the accuracy of the
formulae is limited to a specific range of the input parameters due to the curve fitting
techniques.
The most refined and accurate efficiency factors are theoretically proposed based
on the compression softening models, For examples, the efficiency factors in the forms of
v = 1/(0.8+170E
1
) and v = 0.91 1 + ka&i) proposed by Collins and Mitchell (1986) and
Belarbi and Hsu (1991), respectively. The parameter E1 and ka are the transverse tensile
strain and a coefficient taken into account the type of loading, respectively. Although the
efficiency factors based on the compression softening model are most accurate, the
formulae are not simple to adopt in the design practice since the nonlinear analysis is
required to obtain the transverse tensile strain. In addition, the existing compression
softening models were derived from the curve fitting technique of data from the series of
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the shear panel tests where their behaviors are substantially different from specific types
of D-Regions, for examples, no condensed stress and arc-action effect in the shear panel
while the deep beam does. Thus, a direct use of the existing efficiency factors based on
compression softening models might be inappropriate.
2.3.3 Plastic Truss Capacity
As previously mentioned, theoretically, STM is developed based on the lower
bound of plasticity theorem for design of D-Regions. In general, when D-Regions have
sufficient ductility to allow the force redistribute after concrete cracked and flow closely
to the idealized truss structures, the assumption of the plastic truss analysis together with
the rigid-perfectly plastic stress-strain relations of each STM component may be applied
to calculate the plastic truss capacity.
In general, the rigid-perfectly plastic stress-strain relation is the assumed stress-
strain state that the member can sustain a very large strain after the stress demand reaches
the stress capacity of the member. Thus, the rigid-perfectly plastic stress-strain relation
can be represented by a plot of a constant stress (or effective strength) consistent with the
limit analysis. Nielsen (1967), the first researcher who introduced the rigid-perfectly
plastic stress-strain relation for STM components, proposed the effective strength to
account for the softening behavior in the concrete after reaching its peak stress, f c, and
peak strain, i:;'c. Fig.2.9 illustrates the example of the rigid-perfectly plastic stress-strain
relation for an uniaxial compression member. In addition, using both limit analysis
concept and the effective strength can lead to the ability in solving the statically
indeterminate STM as presented by Muttoni et al (1997).
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Elastic Perfectly Plastic
Stress-Strain Curve
Rigid Perfectly Plastic
Stress-Strain Curve
~ ~ ~ ~ ~ - - - C - . - . . . _ J
0
Typical Concrete Stress-Strain Curve
/in Uniaxial Compression
Fig.2.9 Rigid Perfectly Plastic Stress-Strain Relation
(Taken from Tjhin, 2004).
2.3.4 Capacity of Non-Plastic Indeterminate Truss
There are some situations that the statically indeterminate STM is needed for
design and analysis either complex type of D-Regions (Fig 2.lOa), subjected to complex
geometry, loading, and boundary conditions, or simple type of D-Regions (Fig 2.lOb)
where the refined STM is required to assess the tension demand in the web
reinforcement. In general, it is still very difficult and not practical to adopt indeterminate
STM in design procedure since there is still lack of research works and evidences on how
to assign the proper values of the relative stiffness in solving the indeterminate STM. By
assigning different values of the relative stiffness in each STM member, the level of the
force transfers in the individual STM members will be different and this thus results in
high level of uncertainty in capacity and structural response assessment of D-Regions
(Fig 2.lOc).
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
............ - - - ~ - - . ,._...

t
I
t I
' .
I




I
l
I
(b)

t



I
I
I
' I
I
I
t_...__._.
200
-STM
3
-STM
111"'---lr------l--------+---------J-----I 4
-<>-STM
0 2 4 6 8 10
Local Axis 2 Displacement {mm)
(c)
12
Fig.2.10 Non-Plastic Indeterminate STM: (a) Complex Type of D-Regions; (b)
Simple Type of D-Regions; and (c) Parameter Sensitivity of Nonlinear STM.
Some researchers and code of practices provide guidelines in solving the statically
indeterminate STM; however, those guidelines are generally limited to some specific
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
types of D-Regions. For example, Hwang and Lee ( 1999, 2000) considered the problem
of determining the distribution of forces in statically indeterminate trusses (Fig.2.11).
..!.!
(.
; ~ ,
I i ''' v ... i \' ' F,tan9
I: + t -2-
\ l F, -DsinB
\.L-
....
)_ F,tan9
2
Section 2-2
(Horizontal forces not shown)
v ..
r---Q
- - - -
, .....
\ , ..... <r F. \
\ . \
__!) \ ~ ,..,. F,cot9
\ -Dcos9 2
'<l- F. cot 9
2
Section 1-1
(Vertical forces not shown)
Legend
- - - Compression strut
--Tension tie
Coordinate.
d'i:t t 11 r
~ h
Fig.2.11 Indeterminate STM (Hwang and Lee, 1999).
It was proposed to solve for these forces by assuming that the relative stiffness
between diagonal strut, horizontal, and vertical ties were constant and obtained from the
relationship proposed by Schafer (1996) and Jennewein and Schafer (1992). The constant
relative stiffness assumption ignores the reality that the true behavior and load transfer
mechanism throughout limit states are fully nonlinear, particularly post cracking states
and thus that the relative stiffness should be not constant but rather a function of the
stress and strain or damage in the model.
FIP (1996) also provides some guidelines in solving specific types of
indeterminate STM. For example, a two span deep beam as illustrated in Fig.2.2a. is
modeled as indeterminate STM where the provision suggested the equation of the forces
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in the tie above the internal support and in the internal spans that can be used to solve for
the remaining forces in the indeterminate STM.
2.3.5 Serviceability and Ductility Requirements
While STM is developed based on the lower bound plasticity theorem for design
D-Regions at the ultimate state and experimentally validated for its conservatism of
structural strength design, the structures designed by STM might be unsatisfied in the
structural performance under service states since the current STM does not sufficiently
consider the serviceability requirement, for examples, crack control, deflection limits, and
vibration.
Initially Schlaich et al (1987) proposed that the serviceability design for D-
Regions will be automatically satisfied by adjusting the shape of STM such that the
orientation of each STM component in the principal tensile or compressive direction
according to the linearly elastic analysis of the structure and using the STM shape for
ultimate state design. Although the proposed serviceability design criteria are
conceptually simple, the shapes of STM adjusted in this way often lead to the unpractical
reinforcement patterns. In addition, principal stress direction generally provides only
unclear shapes of STM that are unpractical to use in obtaining the refined STM. To be
more practical and systematic, Tjhin (2004) proposed the serviceability design criteria by
performing the nonlinear analysis of STM for quantifying service-load response. The
procedure was proposed to begin with the standard STM design procedure for strength
and then load-deformation analysis was performed to obtain the structural responses
under service state which were used to directly check with the serviceability criteria
defined in the codes of practice. Although the proposed procedure is very systematic and
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
simple, there are several deficiencies in the nonlinear analysis STM technique, for
examples, lacks of guidelines to select the appropriate stress-strain relations, effective
dimensions, and stress limits in each STM component. These analysis parameters greatly
affect the predicted structural responses and thus result in unacceptable level of
uncertainty when adopting in the design process.
For codes of practices, ACl3 l 8-08 briefly addresses the serviceability design for
D-Regions in deep beams, mainly by controlling of deflection and cracking. For deep
beams, the code allows for the estimate the deflection by using elastic analysis of STM or
nonlinear analysis when nonlinear stress distribution and effects of cracking are
considered. The crack control required in ACl318-08 is to calculate the crack width in the
ties by considering the reinforcing ties distributed within prism concrete. CIB-FIP(l 990)
and FIP(1996) also address the serviceability design ofD-Regions by using STM
oriented in a linear elastic stress field. The design also focuses on crack control and
limiting deflections. For crack control, the code requires the designer to check the crack
width of ties subjected to the maximum force where the effective area of tie is considered
based on CIB-FIP (1990). The limited deflection of D-Regions is verified by using linear
STM with consideration of tension stiffening effect in ties.
For the ductility requirement for STM, theoretically, the lower bound plasticity
theorem assumes the existence of the very large strain limits in each STM component
while the concrete material permits only limited strain or plastic deformations. Thus,
after cracked, the structures must be designed to provide the sufficient local ductility to
allow the redistribution of the internal stress to the equilibrium state idealized as the
internal trusses (STM). Marti (1985) suggested to provide some amounts of the
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
distributed reinforcement to insure the required redistribution of the internal forces in the
crack states. Schlaich et al (1987) proposed to select the internal structural system (STM)
in the way that the deformation limit (capacity of rotation) is not exceeded in any
material points before the assumed state of stress is achieved.
Some codes of practices imply the ductility requirement in design of D-Regions
using STM, for examples, FIP (1996) states the ductility requirement indirectly in the
crack control section 7.5.5 and 7.5.6.l that
Section7.5.5
"In every region where under SLS conditions the tensile strength of concrete may
be exceeded, a minimum amount of reinforcement should be provided to ensure a
predictable behavior of the member"
Section 7.5.6.l
"A D-Region may be considered with a strut-and-tie model for the verification
of cracking at the SLS. The model should be oriented by the stress fields determined from
a linear-elastic analysis"
These requirements are thus similar to the guidelines proposed by marti (1985)
and Schlaich et al (1987), respectively.
ACl318-08 also addresses indirectly the local ductility requirement in the
distributed reinforcement of the bottle shape strut in section A.3.3 that
"The reinforcement required by A.3.3 is related to the tension force in the
concrete due to the spreading of the strut"
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This is thus to provide the sufficient local ductility limit in the region of the bottle
shape strut where the high local ductility demand is expected to occur due to the induced
bursting tension force of the spreading flow of forces.
2.4 Nonlinear STM Model
The distribution of forces within a steel truss and its ultimate capacity can be
predicted with reasonable accuracy if the dimensions and material properties of each
member of the truss are known. The same is true for a reinforced concrete truss. While
the STM is only an idealized truss within a continuum of concrete, researchers have
nevertheless explored the potential for characterizing the geometry and material
properties of struts and ties sufficiently accurately so as to be able to predict the load-
deformation response of STM designs. The developed methods enable a level of
performance-based design to be completed that can examine the anticipated condition of
a member under service load levels as well as the behavior under an overload. The
development of suitably accurate models is particularly important for highly statically
indeterminate STM designs as the underlying assumption of the STM approach that each
member, including compression members, has a linear-elastic perfectly-plastic response
is not necessarily accurate. This is thus required tremendous efforts on the research works
to improve the current analysis techniques of nonlinear STM.
More than a decade, several attempts have been made by researchers to develop
the appropriate nonlinear STM models and frameworks for predicting either shear
strength or nonlinear structural responses of D-Regions. Some of the researchers are
Sundermann and Mutscher (1991), Hwang and Lee (1999, 2000, 2001), Yun (2000), To
et al (2000, 2001), Tjhin and Kuchma (2004), and Park and Kuchma (2007).
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Sundermann and Mutscher (1991) proposed an analytical method for nonlinear
STM analysis developed by extending the conventional STM to include the nonlinear
stress-strain relations. The method enabled the design engineer to be able to obtain the
ultimate load, the support reactions of indeterminate structures, key stresses and strains as
well as the overall load-deflection response. The method laid out the concept of a
practical tool to analyze a structure nonlinearly without the use of sophisticated nonlinear
FEA. The method was formulated based on the following steps. The shape of STM was
obtained by considering the flow of forces or so called "Load Path Method". The more
refined shape of the STM was also possible to obtain by performing linear FEA. The
STM shape adopted in the method was composed of struts, ties, and nodal zones
regularly with a new intuitive nodal component called "Free nodes".
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Free Nodes
,,



\=.=I
"==\

1
.. 2.oom 1. 2.oom.1
(a)
Deflection [mm]
B/A
4,0 --------1,--..:---t---i-
i
a.s----------
3,0 f..---+----+---+--r--+-1--+-
ZS


1,0
0,5 r---+.,,,,,,e:..-+---+---l--+
o,o ....... ....__....._...J..,.__..\---L..__.L---L---1-'---......
0,0 1,0 2.0 3,0 4,0 5,0
' '
(b)
Load [MN/m]
(c)
10

Fig.2.12 Nonlinear STM (Sundermann and Mutscher, 1991): (a) Continuous Deep
Beam; (b) Shape of STM including "Free Node"; and (c) Predicted Structural
Responses.
Free nodes (Fig.2.12b) were the movable nodal zones or the movable node
linking the STM members. The location of free nodes was allowed to move to any
locations in D-Regions causing the minimum stationary potential energy of the structure
during the loading steps. After locating the nodes, the effective width of STM members
and stress limits were then calculated.
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(b)
Ec1
, ...
x
fc
1
=-a
1
fem
E'c1 = O'e: Ee0
f,;
1
=0'
8
Ec0
Fig.2.13 Behavior Models for Struts and Ties (Sundermann and Mutscher, 1991):
(a) Reinforced Concrete Tie under Tension; and (b) Fan Type Strut under
Compression.
The stress-strain relationship of struts and ties adopted in this method can be
varied depended on types of struts (e.g. fan or bottle type strut) and ties (e.g. in a partially
prestressed ties) as determined by users. The stress-strain relationship (Fig.2.13) of struts
adopted a modified rule in the MC 90 and the DIN I 056, respectively. With this rule, the
stress-strain relationship of struts was obtained based on 3 factors e.g. the actual strength
(af), the tangent modulus at the origin (aE), and the strain reached at ultimate stress (m:).
The method also claimed that the time dependent behavior was possible to model such as
creep in concrete by introducing the creep coefficient or debonding in cracked ties. An
example of analysis of a continuous span deep beam subjected to continuous loading
using the method was given in Fig.2.12. From Fig.2.12, the predicted load-deformation
response of the structure agreed quite well with experimental results in the post cracking
state up to the capacity of the structure. The capacity of the structure was; however,
slightly underestimated due to the pre-yielding of the ties. In addition, the internal
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
support reaction obtained from the method was quite underestimated and this would be
due to the premature yielding of ties.
Although this is a simple method with predicted that correlated well with
experimental test results, the analysis example was completed by several unreasonable
assumption such as no consideration of different types of struts in stress-strain
relationship e.g. a =1.0 for all struts, pre-cracking in concrete covered a set of ties, and
over estimation of the level of reinforced concrete tie located above the internal support.
These assumptions would result in inaccuracy of the model when applied to experimental
test data beyond that from which it was derived.
Hwang and Lee (1999, 2000, and 2001) proposed the softened strut-and-tie model
for determining the shear strengths of interior and exterior beam-column joints for
seismic resistance. The model was developed based on the strut-and-tie concept and
derived to satisfy equilibrium, compatibility, and the constitutive laws of cracked
reinforced concrete. The STM adopted in the method was selected to resist the
earthquake forces induced in the joints e.g. the vertical and horizontal shear forces, the
tension forces in the reinforcement, and the compression force equilibrating the tension
force. The shapes of STM (Fig.2.14) adopted in this method were those with one
diagonal strut, two minor struts with horizontal ties, and two minor struts with vertical
ties.
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
~
I
jd
~ ~ ~ ~
J--a
(a)
Diagonal strut
Flat strut
. - - Horizontal tie
Nodal zone
Vertical tie
Steep strut
Fig.2.14 Shapes of STM for Load Transfer Mechanism (Hwang and Lee, 1999): (a)
Diagonal Mechanism; (b) Horizontal Mechanism; and (c) Vertical Mechanism.
To analyze the model, the combination of the different types of STM was required
to capture the load mechanism at the failure of the joints and thus the structural system
became indeterminate. To solve the indeterminate system, the relative stiffness between
diagonal struts, horizontal, and vertical ties was assumed to be constant and obtained
from the relationship proposed by Schafer (1996) and Jennewein and Schafer (1992).
This relationship was obtained from the linear analysis for a specific type of beam-
column joints. The relative stiffness was then used to solve the indeterminate structure to
obtain the internal STM forces and internal stresses which were later checked with the
softened stress-strain relationship (Fig.2.15) of concrete proposed by Zhang and Hsu
(1998).
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
- - ; ~ l ---_ . _ - - ~
reinforced
I concrete
Fig.2.15 The Softened Stress-Strain Relationship of Concrete (Zhang and Hsu,
1998).
The proposed model was subsequently verified by comparing the predicted shear
strength of 63 tested joints and resulted in a good agreement of predicted shear strength.
Although the predicted shear strength obtained from the method correlated well with the
shear strength of the tested joint, there are some concerns with this method: (1) the
relative stiffness between diagonal strut, horizontal, and vertical ties obtained from the
linear analysis and assumed to be constant throughout limit states is improper and
theoretically incorrect since the actual joint behavior and load transfer mechanism
throughout limit states are fully nonlinear, particularly for post-cracking states and thus
the relative stiffness should be not constant and needs to change depending on the state of
stress and strain in the joint, and (2) the high tension demand occurring in the joint is
considered to cause bond degradation between the reinforcement and concrete. The bond
degradation significantly affects the behavior and shear strength of the joint and thus
should be included in the model.
Yun (2000) proposed a nonlinear strut-and-tie approach to design and analyze D-
Regions. The approach started by analyzing plain concrete nonlinearly using the finite
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
element analysis and displaying the principal stress flows. The shape of STM was then
sketched using the principal stress flow. The effective stress in every strut and tie
member was obtained from the principal stress ratios of the nonlinear FEA of plain
concrete in corresponding D-Regions considering the degree of confinement in the strut-
tie members and the difference between the alignment of strut members and the principal
compressive direction. The effective area of each strut-tie member was iteratively
selected until the level of stress demand obtained from linearly elastic STM so as to
equate to the effective stress. After obtaining the strut-tie effective dimension, the truss
model was analyzed considering nonlinear in which the equivalent uniaxial stress-strain
relationship of 2D plain concrete was adopted for struts.
N I I. .I ~ I .
I '
(a) (b)
90
60
/
30 . _.. "' -e-Curreont, Type I STM
-o-Currnt, Type 11 STM
_..,_Current, fYP ill $TM
- - Conven11onal. Type 1 STM
-Test
o - - - ~ - ~ -
o 0.0003 0.0006 0.0009 0.0012 0.0015
Longitudinal Strain (1n./1n )
(c) (d)
Fig.2.16 Nonlinear STM (Yun, 2000): (a) Specimen (Simply Supported Beam); (b)
STM based on Stress; (c) Refined FE Models for Nodal Zones; and (d) Predicted
Structural Responses.
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subsequently the nonlinear FEA was adopted to check the failure mechanism of
critical nodal zones. If there was no failure mechanism occurring in the critical nodal
zones, the amount of required reinforcements was calculated and placed within the
effective dimensions of strut-tie model. The results obtained from Yun's nonlinear STM
approach show a good agreement with the experimental results for both shear strength
and load deformation response prediction of the structures.
Although Yun's approach provides an intuitive method for nonlinear STM
analysis and presented good correlations with experimental data, the approach is not
practical and contains theoretical drawbacks. To obtain the effective area of each strut-tie
member, the nonlinear FEA of the plain concrete regions corresponding to each strut-tie
member is required. This is improper in the practical sense since if one already has
nonlinear FEA model of the structure, only a run of the nonlinear FEA is required to
validate the structural performance and a nonlinear FEA would generally be expected to
be more accurate than the STM. Another theoretical drawback is that obtaining a stress
limit from biaxial strength envelope of plain concrete is unable to capture the effect of
distributed reinforcement provided in concrete regions. In addition, since the
reinforcement in ties will be provided after completing all steps, it is thus impossible to
include tension stiffening or bond slip in this nonlinear STM approach. In checking the
local failure of the nodal zones, Yun adopted the refined nonlinear FEA of nodal zone
regions (Fig.2.16c) by assuming the boundary of the nodal zone regions applied by
traction obtained from stress in STM members. In the actual structure, no nodal zone
exists and thus using a refine model of FEA for modeling non-exist structural
components is improper. In addition, applying the traction forces obtained from STM
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
components to the boundary of nodal zone may not satisfy the actual boundary condition
of the true structural regions and thus affects the obtained results.
To et al (2000, 2001) proposed a monotonic nonlinear strut-and-tie computer
model for both B-and D-Regions and then applied this method to obtain the monotonic
load-displacement enveloped curve of reinforced concrete knee joints under reversed
cyclic loadings. The model seemed to be formulated just to validate the experimental
results and is not intended for use as a practical design tool. The envelope curve obtained
from the approach showed a good agreement with the experimental results. The paper
also provided guidelines in determining dimensions, stress-strain relationships, and stress
limits of strut-and-tie members for B-Regions; however, only stress limits of strut were
given for case of D-Regions. The struts and ties in B-Regions were dimensioned by using
sectional analysis of the beam and column in these B-Regions (Fig.2.17a). The plane
section assumption was adopted in calculating the compression and tension regions in the
section which are then later used to calculate the dimension of strut and tie, respectively.
The stress limit of the longitudinal strut in B-Region was assumed to be obtained from
the maximum fiber compressive stress in the section.
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
500
300
!

Push
j'::... _
(b)
100
0
-100
Joint opening direction
-300
-700
-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 02
Vertical displacement (m)
(c)
Fig.2.17 Nonlinear STM {To et al, 2001): (a) Effective Dimensions of STM
Members; (b) STM for a Structural Frame; and (c) Predicted Structural Response.
The stiffness of concrete strut was adopted as a constant and equal to that
suggested by a relationship in Eurocode 2. The stress limit of the diagonal strut in B-
Region was assumed to be 0.85fc for an upper bound solution. For D- Regions, only the
stress limits in concrete struts were provided considering the cracking and level of
transverse tensile strain induced by the reinforcement. To assign the stress limits of the
concrete strut, the observed cracked concrete and measured tensile strain level in the
re bar were required and this model was thus only for validation of an experimental
program, not for design. In addition, the shape of strut-and-tie adopted in the analysis was
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
also alternated depended on the level of cracking in beam-column joint and direction of
the applied load during the experiment. It was found that some STM models adopted in
the paper were indeterminate STMs (Fig.2.17b); however, no detail was given on how to
assign the relative stiffness of every individual strut-and-tie members.
Tjhin and Kuchma (2004) innovatively developed a computer aided strut-and-tie
(CAST) to provide designers with friendly graphic user interfaces (GUI) and utilizing
tools for transparent and practical design processes that can significantly overcome the
time-consuming and unnecessary calculation. The examples of GUI of CAST are
illustrated in Fig.2.18. Since CAST was also developed to integrate with the nonlinear
STM analysis, the extension of capabilities within the CAST tool can be used for
exploring suitable material models and geometric definitions for development of the
suitable models.
One of the extension and application of the CAST tool was presented by Tjhin
(2004) for a design procedure considering both strength and serviceability requirements
of D-Regions by using nonlinear strut-and-tie models. The feasibility of this approach
was presented using the experimental results from a deep beam test. The approach used
in this paper was formulated based on several assumptions. The approach adopted the
same shape of STM for the entire loading stages. Four
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig.2.18 CAST GUI (Tjhin, 2004)
types of failures modes were considered in the model i.e. yielding of ties, crushing of
struts or nodal zones, and diagonal splitting of struts. The stress-strain relationships and
effective widths of struts and ties could be characterized by the user. Every STM member
was assumed to have uniformly distributed stress across an effective width.
From the above assumptions, the nonlinear STM (as illustrated in Fig.2.19) was
adopted to generate the load-deformation response of a simply supported deep beam with
an applied pointed load at the center of the span tested by Lee (1982). The shear span-to-
depth ratio of the deep beam tested was about 1.60. Three different types of STM shapes
were adopted in the nonlinear analysis including a determinate STM, an indeterminate
STM, and one direct STM model. The stress-strain relationship used in the analysis
adopted the parabola shape with reduction of 0.85 (stress limit) at the peak of the
compressive strength for struts and the tri-linear stress-strain relationship including the
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
tension stiffening effect for ties. The effective width used in strut was selected such that
the stress at the collapse load was equal to 0.85f c.
From the analysis results, a different load deformation response was predicted for
each of the three STM shapes, particularly on post cracking stiffness and strain history in
longitudinal steel. The results were compared with the experiment to compare and
contrast the appropriateness of the approaches. Subsequently, the paper provided design
guidelines for strength and serviceability based on the STM. The guidelines started with
the general process of STM for strength design and later the nonlinear STM was adopted
to determine the response quantities at service load level and check the relevant response
quantities against the serviceability criteria.
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
' .

' -m- 16_,
;t 1 woo-
"
(a)

2000
!llOO
,,;;: HIOO
;;: 140i)
I
ll'OO
100()
:;:,.
-
-
-

21)/j
'
0
Cl 2
;q
.
'
f


lllM,,,.. ........
!2116
TOO mm
(c)
"
II 10
mm
$0.t
,,.
1 Ill 111
(d) (e)

Fig.2.19 Nonlinear STM (Tjhin, 2004): (a) Dimensions and Reinforcement Details;
(b) One Direct Strut STM; ( c) Determinate STM; ( d) Indeterminate STM; and ( e)
Predicted Load-Deflection
Although the paper provided intuitive and practical concepts in design and
analysis of D-Regions using nonlinear STM, there are several concerns that may limit the
application. The stress limit adopted for the bottle shape strut in the deep beam example
over-estimated (0.85f c) the stress capacity relative to that permitted in ACl3 l 8-08 of
0.75Psfc. The effective width obtained by using a stress limit of 0.85fc may under
estimate the effective dimension of struts and thus affect the stiffness prediction of the
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
structure. In design process, there is no data of the designed specimens to compare with
the nonlinear analysis, so selecting the most suitable STM model through comparisons
with existing data (e.g. experimental results) is not possible. Also, the stress-strain
relationship for bottle-shaped struts is insufficient and does not adequately capture the
effects of cracking, transverse tensile strain, and distributed reinforcement.
Park and Kuchma (2007) extended the work by Hwang and Lee (1999, 2000,
2001) by incorporating a new nonlinear solving scheme based on adjusting the secant
stiffness of STM components, eliminating the use of indeterminate STM by adopting a
proposed determinate STM shape, and investigating the effect of multiple compression
softening relationships on the performance of models.
Although the proposed model is simple, practical, and the results are sufficient
accuracy in shear strength prediction of deep beams, there are some drawbacks due to its
assumptions and formulations. Firstly, since the model is formulated based on
determinate STM where the internal forces in STM members can be simply obtained
from only equilibrium, the model is not capable of capturing the redistribution of the
stress and changing of forces in each STM member (relative stiffness) correctly,
particularly for orthogonally reinforced deep beams.
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
'
Mean 1.41
v1+--sl
4
COV=0.18
Fa
~ F
f
/ ~ 'F, ty; I
,l
d 11
z I 2
, ~ ~
h, Fv
t.i.
/ /
1 ~ 1
I / /
F1 h F d ~
\,.?cl. T
.. ' 4

(a)
IJ
0 1 2 3
aid
0.9
fc' (bl)
0.8 0 2 4 6 8 10
~ 0.7
' I ..
4 ~
06
..
~
: 0.5
!;;
'.l
c
~ 0.4
I
..
0.:3
"""
I -
;t 2
0.2
fc = 13 8MPa (2001 psi)-// ,

1
0.1
t<. = ne lllPa (10675 psi)_..;

0 0
0 2 4 6 8 10
0 20 40 60
E1(10"
3
)
r<(MPa)
( b) ( c)
Fig.2.20 Nonlinear STM (Park and Kuchma, 2007): (a) Determinate STM; (b) Effect
of Multiple Compression Field Relationships; and ( c) Performance of the Model.
In another word, the degradation of stiffness in each D-Region location is
different and depends on material, stress and strain states, and detailing. Using
determinate STM to pre-define the portion of the force in each STM member is
unsuitable. Secondly, changing secant stiffness in each iteration should affect the relative
stiffness of STM members and thus ratio of forces; however, in the proposed model it
does not. Lastly, the paper does not address the procedures or guidelines to calculate the
effective area in tension ties members which might affect the calculation of strains and
thus the predicted shear strength. In addition, the effective area of tension ties is also
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
important when the nonlinear characteristics in the web reinforcement represented as ties
are required.
2.5 Nonlinear FEA for Structural Concrete
While non-linear STM models may be able to provide an efficient and sufficiently
accurate approach for the performance evaluation of selected STM models, they are
ultimately limited by the assumptions made in the selection of material models and the
dimensions of STM components. The more complete means of evaluating analytically the
performance of a STM designed structure is to conduct non-linear continuum Finite
Element Analysis (FEA). As the complexity of the STM design increases, so does the
need for validation by leading analytical methods. A very brief summary of the history of
development of non-linear FEA of structural concrete is presented below so as to
illustrate to the potential, limitations, and challenges for the use of FEA for STM design
evaluation.
When Ngu and Scordelis (1967) firstly adopted the nonlinear FEA for structural
concrete, they inspired a new field of endeavor to which hundreds of researchers have
devoted their careers. Due to the rich history of FEA development, only selected key
developments will be presented in this literature review. In addition, since the author will
study and propose a nonlinear FEA methodology based on a compression field approach,
only the literatures relevant to nonlinear elastic FEA framework will be discussed. A
more complete summary on the development of nonlinear FEA approaches for structural
concrete are available from the state-of-the-art by ASCE-ACI (1993) and Finite Element
Analysis of Reinforced concrete structures, a special publication (SP-237) by ASCE-ACI
committee 447(2006).
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.5.1 Discrete and Smeared Crack Approaches
Concrete is a brittle material that cracks when tensile stress exceeds its tensile
capacity. In nonlinear FEA, two general approaches for modeling cracking in concrete
are discrete crack modeling and smeared crack modeling.
11
i
.
lo--
(a)
-a-
-
(c)
A
repreoento!iw
volume
Section A-A
(b)
(d)
Fig.2.21 Discrete and Smeared Cracks (Reproduced from Bazant, 1983): (a)
Cracking Process; (b) Stress Distribution along Section A-A; (c) Discrete Crack;
and ( d) Smeared Crack.
When discrete crack modeling is adopted, as shown in Fig.2.21c, a geometrical
discontinuity simulated by the use of interface elements or mesh reconstruction is
typically required (Carter et al 2000). This way of representing a crack; although realistic,
causes some difficulty in computation, particularly when no cracking surface is pre-
defined in the models. Ngu and Scordelis (1967) and Nilson (1968) adopted the discrete
cracking modeling approach. Fig.2.22b presents an example of deformed mesh of a
structure with discrete cracks.
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
For smeared cracks, as shown in Fig.2.2ld, the cracked material is represented as
a continuum with reduced stiffness properties (Rots and Blaauwendraad, 1989). This
approach is more convenient for computational implementation and it was believed to be
more realistic for simulating the micro-cracking process (Fig.2.2la) causing by the
inhomogeneity of concrete (Bazant and Cedolin, 1983 ). Most researchers have adopted
the smeared crack approach including Bazant and Cedolin (1983), Vecchio (1989),
Cervenka (1970), Foster et al (1992), Kaufmann and Marti (1998), and Pang and Hsu
(1995). Fig.2.22a presents an example of a deformed mesh of a structure with smeared
cracks.
(a)
(b)
Fig.2.22 Deformed Mesh in Discrete and Smeared Cracks (Reproduced from Rots
and Blaauwendraad, 1989): (a) Deformed Mesh in Smeared Crack Model; and (b)
Deformed Mesh in Discrete Crack Model.
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.5.2 Crack Orientation: Fixed, Rotating, and Delayed Rotating Approach
Different assumptions have been made in the FEA of concrete to describe the
continued formulation of cracking; those being fixed, rotating, and delayed rotating.
~ . 2
-4S'
I '" I
+
-
flJ a -
c
(0,JJJ a +
Reinforced concrete Concrete component Reinforcement Component
Fig.2.23 Illustration of Applied Stresses on Element, Average Stresses in Concrete,
and Average Stresses in Reinforcement (Adapted from Hsu, 1998)
In fixed crack approaches, once cracking has been calculated to occur in an
element, the direction or orientation of the crack is fixed throughout the remaining steps
in the analysis. For example, as illustrated in Fig.2.23, the crack orientation is fixed
equally to the principal stress direction ( a
2
) of the applied stress at the time that cracking
occurs while the average principal stress direction (a) of concrete can rotate. Pang and
Hsu (1996), Darwin and Pecknold (1976), and Okamura and Maekawa (1991) all took
this approach. Since the crack direction was fixed while the principal concrete stress is
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
able to rotate, a shear stress along a crack is theoretically induced on the crack surface
(Fig.2.24) and several developers include this shear stress in forming the stiffness matrix
of reinforced concrete element.
1
Fig.2.24 Average Stresses on Crack Surfaces of Reinforced Concrete-Fixed Crack
Approach (Adapted from Hsu, 1998)
By contrast, rotating crack approach assumes that the crack angles are coincident
with the principal stress direction (a) of concrete in post cracking. This thus eliminates
the need to calculate the shear stress induced along the crack surface. The researchers
adopt this approach are Cervenka (1970), Foster et al (1992), Kaufmann and Marti
(1998), and Pang and Hsu (1995), etc. The example of FE results for both fully rotating
and fixed crack model is illustrated in Fig.2.25 below.
It should be noted that a special type of rotating crack approach entitled " the
modified compression field theory (MCFT)" proposed by Vecchio and Collins (1986)
was formulated to allow the existence of the shear stress along the crack surface which is
used to check the ability of the cracked concrete in transferring the average principal
tensile stress across cracks. By considering the shear along the crack surface in MCFT,
the improvement in shear strength prediction of reinforced concrete plane stress
structures can be obtained over the regular rotating crack approach.
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Deformed mesh Crack pattern
(a)
(b)
Principal tensile stress trajectory
""'',
..
: .... :: :: ..
:.::::::;;'.!!HH;;::
'
: .. ::::::;'1]11:::::::
:: : : ;; ;: ': H :: :: : .. : :
:: :: ;; ;; ; I fl 1: :: :: .
: \. :: :: : :
: :: :11 .;;I I:::::: :: ::
::::t: }ftl:!:::::::::
: : : ;; in;;; 1; ; ; ; ; ; ....
" '. :ll::: :: : : :: : ::
! : : : : : ; : : : ; .
Fig.2.25 Comparison of the FE Results between Fixed and Rotating Crack
Approach (Adapted from Rots and Blaauwendraad, 1989): (a) Fixed Crack
Approach; and (b) Rotating Crack Approach.
A delayed rotating crack approach was proposed by Vecchio (2000) to merge the
fully rotating crack model and a fixed crack model. The approach allows full
consideration of crack shear stress and slip along a crack and eliminates the coincidence
requirement between the direction of cracking and principal strains. Vecchio's Disturbed
Stress Field Model (DSFM) is more fully described in the next section.
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.5.3 Selected FE Frameworks
In this section, selected FE frameworks that are relevant to this research will be
briefly reviewed: 1) MCFT (Vecchio and Collins, 1986); 2) DSFM (Vecchio, 2000); and
3) CMM (Kaufmann and Marti, 1998 and Foster and Marti, 2003).
Modified Compression Field Theory (MCFT)
The Modified Compression Field Theory (MCFT) was proposed by Vecchio and
Collins in 1986 to be an extension of the Compression-Field Theory (CFT). The MCFT
and CFT are both smeared rotating-angle crack approaches where the cracked concrete is
treated as a new material having its own average stress-strain material properties. The
equilibrium equations and compatibility condition are formulated based on average stress
and strain (Fig.2.26). The advancements of the MCFT over the CFT is consideration of
the tension stiffening effect which is the average tension contribution of cracked concrete
as shown in Fig.2.27b. The compression softening model shown in Fig.2.27a was
proposed to capture the stiffness and strength degradation in concrete due to transverse
tensile straining, and innovatively introducing the crack check (Fig.2.27c) to limit the
level of tensile stress crossing cracks.
Although the MCFT has been recognized in its reliability and accuracy for
predicting the responses of cracked concrete elements and structures, there are some
deficiencies that affect the accuracy of the theory in some situations as follows: (1) The
MCFT assumes coincidence between the direction of average
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(I)
c;
.2
-
'6
c;
0
(.)
c;
~
...
- en
c;
co
...
-
Cl)
,
(I)
(I)
Q,)
...
a;
Reinforcement
, . , ~ '
(; y
rt
fSlt-=
(' )(
Cracked
Concrete
0 =(Jc
~
I:
2
fc1
v
Reinforced
Concrete
Ultimate
Yield /
\,.....-.--
/
Cracking
)'
Fig.2.26 Description of Modified Compression Field Theory
(Vecchio and Collins, 1986)
principal stress and strain which is not true when substantial changes in crack and stress
orientation and slip occur as shown in Fig.2.28c and d; (2) The MCFT does not explicitly
consider shear stress and slip along the crack in the theory and this can significantly
affect the predicted deformation (Fig.2.28b); (3) The MCFT adopts the excessive level of
compression softening model which results in under-estimation of the shear strength
when there are no significant level of cracking, slip, and stress reorientation (Fig.2.28a);
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(4) The predictions of the MCFT are mesh-dependent which substantially affects the
predicted strength, particularly when cracking is dominant.
,
(!),'
\
\
1
er
f -----
Cl 1+/200E
1

\ \
CD
(a)
(b)
(c)
1.4
1.2
1.0
j 0.1
-::::
.... : 0.1
...
1.0
'".
:: ..
0.8 .... ,
.
O.f Q,S 1.0 t,1 1.4 1.CI
f
0.8 .
jj
..
- 0.4 : ... 1....... __ _
. .. . .
0.2
00
8 8 10 12 14
f I (X 10.
3
)
G)
f fsY
t fsycr
I \
\
\
18
Fig.2.27 The Advancement of Modified Compression Field Theory (Vecchio and
Collins, 1986): (a) Compression Softening Effect; (b) Tension Stiffening Effect; and
(c) Mechanism at Local Crack Level ("Crack Check").
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10r--------------
-;;
vi 500
"'
l:!

i
40
f
8
30

0
z
if
20
...
0
z
0
!i
z
::J

2 3 4 5 & 7 8
SHEAR STRAIN, y
(a)
6 =90"- VI

10 20 30 40 !!00
INCLINATION OF PRINCIPAL COMPRESSIVE STRAIN, VI
(c)
MCFT
ii' ).
! 1.6 /
. ,,,, ....
:i 1.2 /../

i 0.8
Experiment
ili 0.4
Panel PB20

SHEAR STRAIN, y
(b)
z
I
65
Panel PV19
lfi 60
E
.....;55
s::
u,,
/
z-
ii! Q) 50
II.
IL
0
I
I
z
45
0

Stress
(Experiment)
:::;
400
1.0 2.0 3.0 u 4.0
!
SHEAR STRESS, v (MPa)
(d)
Fig.2.28 The Deficiency of Modified Compression Field Theory (Vecchio, 2000): (a)
No Slip along Crack and Reorientation of Principal Stress; (b) Substantial Slip
along Crack and Reorientation of Principal Stress; (c) Measured Principal Stress
and Strain Direction; and (d) Measured v.s. Predicted Principal Stress and Strain
Direction.
Disturbed Stress Field Model (DSFM)
To improve the accuracy of the MCFT, in 2000 Vecchio proposed a new
approach called the "Disturbed Stress Field Model (DSFM)". The theory was formulated
based on a smeared delayed-rotating crack model which combines the advantages of fully
fixed and fully rotating approaches. The DSFM eliminates the deficiency of MCFT by
introducing the explicit model to consider crack shear slip in for both local crack checks
(Fig.2.29) and element compatibility conditions (Fig.2.30), decoupling the principal
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
stress and principal strain directions, and revising the compression softening (Fig.2.3la)
and tension stiffening models (Fig.2.31b).
y
2 1
Average
~ ; = = i
fs,
x
't'xyk Section B
n
y
Ve. = L pi(_{,,,, - /,)cos 0n, sin On,
j;J
At crack
't'xyk Section C
(cf
0
_ Yci +Yeo
5
- l.8w-
0

8
+ (0.234w-
0

707
-0.20) fee
( d)
Fig.2.29 The Condition at Local Crack Level of Disturbed Stress Field Model
(Vecchio, 2000): (a) Free Body Diagram for Relating Local Crack to Continuum
Level; (b) Continuum Level; (c) Local Crack Level; and (d) Local Crack
Mechanism.
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
q y
L Y,12
r
e.,,-1,
1
l
Ee
x
1 1--...j 1-ecx
()
y

as
y/2
'Y..;12
"Y - -
s-
s
YI

= --ys/2 sin(26)
s en-
= -y,,/2 sin(20)
x
r..;; 5.ls
(b)
y
= "Ys cos(26)
r
'Y/2
Ey-ly

[Es] =
e;
2
1

l
[E] = [Ee] + (
5
]
x
x C.rl-
1 u. e: ...
(c)
Fig.2.30 The Compatibility Conditions adopted in the Disturbed Stress Field Model
(Vecchio, 2000): (a) Continuum Strains; (b) Slip along Crack Surface; and (c) Total
(Measured) Strains
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.2
1.0

J
0.8
N

' . "' "
' DD Dg

-0.6
'
..... ....._ C D dJ
........ 0.55
---a @'
1.0

;o.a
1
u 0.6
s
N

0.4
0.2
00
-------
PV23
+ PV24
PV25
" PV27
D PV28
O Ultimate load stage
1 2

&1
) )
I
----- 1982 Veechio-Cofino
----
1987 Colno-Mitched
1967 Tamai et al
o.o.i-------....----.-----1
i 1760
1::
1000
I 750
500
J?
l5 250
!
0.0 2.0 4.0 8.0 8.0
Principal Tenll Stniln ., (x 10;
CoefHcient. 3.C M ,,,.,,
,,, ..
_.; ; Houoton Panel
Toronto Panel ; " ; ; ;
.,,"' Elements
Elements "8 Toronto Shell
; ; ,,, "' Elements
...
;
100 200 300 400 500
M parameter= cone area/bar clrcum (mm)
--...; ..........................
---
-----, ......
Cs=1.0
4 5
.. , . ., ..
' .. . .
;4
' .
'
'
.
univetlllyotTOllllllO
Pane!Eilmentl
70mmlhl:l<
50 mm spacing
6mmd-rba1>
(b)
..
"" .
, fl "
UnwersRy Of Toronto
Shell Elements
285mmlhlck
72 mm spacing
20mmdiameterball
\
6
.,,.. . : .. ...
' ... \I ,_
.. . . . .
Ul'IYefllty DI HOUSIOll
Panel Elements
178mm!hick

20mmdia.-ba1>
Fig.2.31 The Improved Compression Softening and Tension Stiffening Models
adopted for Use in the Disturbed Stress Field Model (Vecchio, 2000): (a) Adjusting
the New Compression Softening Parameter (Cs); and (b) The New Semi-Empirical
for Tension Stiffening Model considering a Bond-Related Parameter (Bentz, 2005).
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Although the DSFM can substantially improve the analysis result for cases where
the MCFT was not successful, there are some drawbacks of the theory inherent to its
assumption and formulation: (1) the procedure to calculate the local shear stress along
cracks based on the incremental strain components of reinforcing bars at crack in the
principal concrete stress direction does not fully capture the distribution of stresses in
reinforcement due to the effects of cracking and bond; and 2) the relationship for tension
stiffening (Fig.2.31b) was based on a model proposed by Bentz (1999) which obtained
from new curve fitting parameters for an attempt to capture the characteristics of bond
between concrete and reinforcement. Although the model was reported to show some
improvements in predicting tensile concrete contribution between cracks (Bentz, 2005),
the model cannot really predict and characterize the bond slip between concrete and
reinforcement between cracks, particularly when the capability of bond stress transfer of
orthogonal reinforcing bars in each direction is different, In addition, since the model
was based on the average tensile stress and strain in cracked concrete, the tension
softening effect which presents the micro-cracking process in concrete was implicitly
included.
Cracked Membrane Model (CMM)
In 1998, Kaufmann and Marti proposed a new model called the "Cracked
Membrane Model (CMM)" for cracked, orthogonally reinforced concrete panels
subjected to in-plane stress. The model was formulated based on rotating crack approach,
the Compression Field Theory, and the Tension Chord Model (TCM) (Kaufmann and
Marti, 1998). By employing the concept of the TCM (Fig.2.32), crack spacings and
tensile stresses between the cracks are determined from equilibrium equations and bond
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
stress-slip relationship. In addition, the model was formulated and presented in the way
that either fixed or rotating crack FE
'J...,-fct
( c)
( d)
r Srm/2 ~ t
crack
Srm/2
r-
-- Nc/2
_ .......
( e)
Fig.2.32 Tension Chord Model {TCM) (Kaufmann and Marti, 1998): (a) Idealization
of Tension Chord Model (TCM); (b) TCM in Y-Direction; (c) TCM in X-Direction;
(d) TCM in General Direction; and (e) Free Body Diagram for Force Transfer
Mechanism from at Crack to between Crack.
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
approach is possible to apply. Subsequently, in 2003, Foster and Marti extended the
model for implementing in FE framework. The extension of CMM by Foster and Marti
was emphasized on the rotating crack approach where the cracks are stress free and able
to rotate. In the FE version, the concrete material model (Fig.2.33d) in compression was
generalized for both softening and strengthening effects e.g. for transverse tensile strain
or bi-axial compression stress (Fig.2.33c). The concrete material model in tension was
proposed innovatively to decouple the tension softening (micro-cracking process) and
tension stiffening related to bond stress-slip. This important feature makes the model
rational in the tensile contribution of cracked concrete. In addition, the tension softening
model of concrete was included the fracture mechanic concept (Fig.2.33b) to reduce the
mesh dependent problem (Petersson, 1981 ).
Oc1
/ct -
O.tfct
(a)
-Ozc/fcp
(0.6, 125)
(l.25,0.6)
! j
I WEd
/J= 1.......,
1.0 -<:J1c/fcp
0,(, P1cp ecp 11EcP
( c) ( d)
Fig.2.33 Constitutive Models adopted in CMM (Foster and Marti, 2003): (a) Tri-
Linear cr-E Relation for Steel; (b) Tension Softening Relation for Concrete; (c)
Biaxial Strength Envelope for Concrete; and ( d) Generalized cr-E Relation for
Concrete.
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Although the features of the model and series of the numerical examples show the
accuracy and validity of the model, there are some deficiencies of the model as follows:
(1) the approach assumes that the directions of principal stress and strain are coincident
which is the same shortcoming as in the MCFT; and 2) no consideration of crack shear
stress or slip at crack by any means which should cause the model to overestimate the
strength or underestimate the deformation at the failure.
2.5.4 FE Modeling of Reinforcement
In FE modeling of reinforced concrete structures, there are several ways to model the
reinforcement (Fig.2.34a) by adopting the followings: smeared reinforcement, truss or
discrete element, and embedded element. When the smeared reinforcement is adopted in
the model, the reinforcement is assumed to be uniformly distributed within specific
regions e.g. element domain. This approach is thus suitable for modeling the distributed
reinforcement where the pattern and configuration of the reinforcement are not important
and thus not included as the main parameters in the model. A disadvantage of the
approach is that it is unable to consider bond-slip between the smeared reinforcement and
concrete. Vecchio and Collins (1986), Vecchio (2000), Kaufmann and Marti (1998), and
Foster and Marti (2003) have all adopted this smeared crack modeling approach.
Truss or discrete elements are used by adding 1 D lower or higher order elements
between the existing nodes of concrete elements. This approach is thus appropriate for
modeling of main reinforcing bars where the orientation and configuration of the bars are
important. This approach also allows consideration of the bond-slip effect by using
special discrete elements or adding springs at the discrete
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Smeared reinforcement
L
L
o Concrete element node
Truss element
(a)
Reinforcing
la ers
L
( b)
Embeded element
t+-- - _L _ _.....
Fig.2.34 FE Modeling of Reinforcement: (a) Type of Reinforcement Modeling; and
(b) Mesh Topology Independence of Embedded Reinforcement Elements (Elwi and
Hrudey, 1989).
element nodes. The difficulty in using this approach is that the topology of the mesh
requires some adjustment to match with the configuration of the reinforcing bars. This, it
is impractical for modeling complex or congested reinforcement. Some of researchers
who use this approach in their FE models are Vecchio (1990, 2000), Lee and Kuchma
(2007), Foster et al (1992), and others.
To overcome the limitation of the discrete element when using with the complex
reinforcement pattern, a special discrete element type was developed called the
"embedded element". This type of element allows users to place it within the element
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
domain or at the element boundary in any orientation or configuration which is thus
independent to the topology of the mesh (Fig.2.34b ). Researchers who have adopt this
approach are Phillipps and Zienkiewicz (1976), Chang (1989), Ranjbaran (1990), and
others. In addition, several researchers demonstrate the possibility to include bond-slip in
the embedded element by using different approaches i.e. Elwi and Hrudey (1989),
Schafer (1975), and Kwak and Fillppou (1995).
2.5.5 FE Modeling of Bond-Slip Problems
Except for nonlinear material behavior, cracking, strengthening and softening, the
bond-slip problem is one of the important behavioral aspects in reinforced concrete
structures that researchers have been interested in since trying to explain how reinforcing
steel and concrete interact. Based on a literature review, there are two major
classifications of bond slip modeling: (1) discrete and (2) continuous. The first bond-slip
model was proposed by Ngo and Scordelis in 1976 and classified as discrete bond-slip
type (Fig.2.35a). The model simulated the bond-slip phenomenon by using the bond-link
element which has no physical dimension to connect between concrete and reinforcement
nodes. Some researchers adopts this approach are Kwak and Fillipou (1995), and
Okamura and Maekawa (1991).
The second approach for bond-slip was proposed by de Groot et al ( 1981) by
using a special element called "the bond-zone element", which is one of the continuous
bond-slip types (Fig.2.35b ). This element simulated the bond-slip by using contact
element and special properties of the bond zone.
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
/------- Concrete
~ , - ~ - . . . _ _ _ - _ - _ - - . - . , - _ . - - Reinforcement
/
,,--- Concrete
(a)
I I 11 - . _ ( '. 6: ~ - : ~ : . ~ :
; t t t : concrete . o
.................. . " .. .-. .....
t t tt
( b)
Fig.2.35 Type of Bond-Slip Model: (a) Bond Link (Discrete); and (b) Bond Zone
(Continuous).
The contact element provides full and continuous compatibility between reinforcing
steel and concrete which are available for both linear and higher order fields. This
important feature makes the bond zone result a more accurate prediction of interaction
between reinforcement and concrete than the bond-link element as pointed out by Keuser
and Mehlhorn (1987). However, several researchers still adopt the bond-link element in
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
their analysis due to the computational simplicity. Additional researchers who have
adopted this model are Keuser and Mehlhorn (1987) and Elwi and Hrudey (1989).
Although there have been several works conducted to study and improve the
bond-slip modeling, the complexity of its phenomenon makes the existing models not
quite reliable, particularly the cost of computation due to increasing number of degrees of
freedom from taking the double nodes( Kwak and Fillipou, 1995). To reduce the cost of
computation due to double nodes, some researchers, such as Kwak and Fillipou (1995),
proposed to adopt the static condensation technique to reduce degrees of freedom to be
solved in this procedure.
In addition to the bond-slip modeling, another ingredient to model the bond-slip
problem is the local bond-slip relationships. The local bond-slip relationship is adopted to
link the slip to the average bond stress. The slip is defined as the relative movement
between reinforcement and concrete. In this proposal, the local bond slip relationship of
CEB-FIP model code 1990(CEB-FIP, 1993) is reviewed in details given below.
In CEB-FIP model code 1990, the local bond slip relationship was given as
for 0 < s < s1
T = Lmczx
T = Tf f OT S3 < S
The parameters in the relationship can be obtained from the table 2.1 for deformed bars.
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 2.1. Parameters in the local bond-slip relationship of CEB-FIP model code
1990 for deformed bars.
--
Unconfined concrete Confined concrete
Good All other Good All other
bond conditions bond conditions bond conditions bond conditions
mmorMPa in. or psi mmorMPa in. or psi mmorMPa in. or psi mmorMPa in. or psi
SJ 0.6 0.024 0.6 0.024 1.0 0.039 1.0 0.039
s2 0.6 0.024 0.6 0.024 3.0 0.118 3.0 0.118
1.0 0.039 2.5 0.098
clear rib clear rib clear rib clear rib
S3
spacing spacing spacing spacing
a 0.4 0.4 0.4 0.4 0.4 0.4 0.4
-
max
24'1/\,
1:J
0.15Tmax 0.15 Z"max 0.15 'rmax 0.15Tmax 0.40rmax 0.40rmax 0.401inax 0.401inax
In addition, the plot of the local bond-slip relationship can be illustrated in Fig.2.36.
In general, the local bond-slip relationship depends on a considerable number of
influencing parameters e.g. bar roughness, concrete strength, state of stress, boundary
condition, and concrete cover. However, the CEB-FIP model code 1990 provides a much
simpler form for the ease in practical design and analysis. The more complex local bond-
slip relationship can be obtained from the literatures e.g. Eligehausen et al (1983), El
Maaddawy et al.(2005), Foster (1992), den Uijl, (1998), Bigaj (1999), Hayashi and
Kokusho (1985), and Yankelevsky (1985).
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Tmax.
I
I
I
I I
- r = Tmax (sis,)"
I I
I 1
I I
r, ----l--....l---------------
1 I
Slip. s
Fig.2.36 Analytical Local Bond-Slip Relationship for Monotonic Loading
(CEB-FIP Model Code 1990)
2.5.6 Mesh Dependent Problem in Smeared Crack Approach
The dependency of a members predicted response on the selected mesh density
remains a critical issue for providing accurate predictions of response of concrete
structures. As reviewed early, the smeared crack approach simulates the cracking process
by means of the strain localization in displacement field or a reduced stiffness. As
pointed out by Bazant and Cedolin ( 1979, 1982, and 1983), the smeared crack approach
is mesh dependent and this results in the divergence of the solution, as the finite element
mesh is refined, due to the infinite increase of apparent tensile stresses. This thus allows
cracks to propagate even at really small load levels. Dodds et al (1985) also demonstrated
the mesh dependent effect in modeling reinforced concrete beams by using smeared crack
approach. Their examples show that the mesh refinement affects the predicted responses
of reinforced concrete beams, particularly for deep beams where the failures are often
governed by cracking.
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
l I\
--
(m (rn
',__I .. 1-r
/ '
r I
/ //
/,'
/
050 /' l
/: IP/?
" / [ 1T
.. 120'1 //) i. // I . 1s
. b--- l
I / i-15"-.-1
80 "i/
l Thru Def,\
J 't - 10 Elements Thru Dq;jh

.. - 15 Elements Thru Depth


I - 20 Elements lhru
----'-"----J...----
C. ,
0.U
'<
Deflection, inches
(a)
(b)
Fig.2.37 Mesh Dependent Problem in Deep Beams (Dodds et al, 1985): (a) Load
Deflection Curves for Deep Beams; and (b) Crack Patterns for Deep Beams.
Several researchers have attempted to solve the mesh dependency problem including
Bazant and Cedolin ( 1979), Cervenka ( 1994 ), Foster and Marti (2003 ), and Choi and
Kwak (1990). Bazant and Cedolin (1979) proposed three different techniques for
reducing the mesh dependence in smeared crack modelingapproaches: (1) energy
variation due to crack advance; (2) equivalent strength criterion; and (3) fitting
asymptotic series to nodal displacements. Fig 2.38 illustrates the effectiveness of the
equivalent strength criterion in reducing the mesh dependence.
75
,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-,
Fig.2.38 Effectiveness of the Equivalent Strength in Reducing the Mesh Dependent
Problem (Bazant, 1983)
Cervenka (1994) and Foster and Marti (2003) adopt an approach to reduce the
mesh dependency by adjusting the shape of strain softening curve of concrete based on
fracture energy and element length scale such as crack band or element width
(Fig.2.33b ). Another intuitive technique to reduce mesh dependency was proposed by
Choi and Kwak (1990) who adjusts the slope of softening curve by assuming a function
to describe the distribution of the micro-cracking process within the element domain.
Based on all the techniques, the technique used by Cervenka (1994) and Foster and Marti
(2003) seems most efficient and the simplest for the nonlinear FEA of concrete.
2.6 Summary
This literature review began with a summary of the history and development of
the Strut-and-Tie Model followed by STM design provisions that have been provided in
codes-of-practice. Next, approaches for solving for the capacity of statically determinate
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and indeterminate trusses were presented, including the plastic truss model. While the
STM is described as a complete and reliable methodology that is solidly based on the
lower-bound theory of plasticity, it was pointed out that experiments have revealed that
STM designed structures may perform poorly under service load levels if an appropriate
shape of STM is not selected. It was also illustrated that members may not achieve their
calculated capacity if the structure is not sufficiently ductile to support the load in the
manner idealized by the designer; the STM method assumes that the behavior of strut-
and-ties is linear-elastic perfectly-plastic, which is particularly not true for struts and
where for large truss models in a complex continuum.
Section 2.3 presented a summary of guidance and methods for selecting suitable
shapes of strut and tie models. Two approaches were then presented for examining all
aspects of the performance of structures design by the STM. Section 2.4 presented the
nonlinear STM method which shows promise as an approach that can be developed and
calibrated for typical and relatively simple STM regions, such as deep beams, corbels,
and joints. Section 2.5 presented a summary of non-linear FEA methods for structural
concrete, which are a more reliable method for assessing the performance of all STM
designed regions.
This literature review has presented the needed background material for
presenting the research activities which are: the experimental validation of STM in design
of complex D-Regions (Chapter 3 and 4); STM shape and dimension assessment
(Chapter 5); automated non-linear FEA analysis (Chapter 6); and the proposed design and
analysis framework for effective structural performance design of complex D-Regions
(Chapter 7).
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 3
Experimental Research Program
This chapter provides a description of an experimental research program in which
17 complex D-Regions were designed and load tested to failure as part of the author's
research and a graduate level course on the mechanics of structural concrete ( CEE561) at
the University of Illinois at Urbana-Champaign. These tests were conducted during the
fall semesters of 2004, 2005, 2006, 2007, and 2008. The chapter presents the objectives
of the experimental tests in section 3 .1, the design of the test specimens in section 3 .2, the
material properties in section 3.3, the fabrication oftest specimens in section 3.4, the
experimental test specimens in section 3.5, the instrumentation in section 3.6, and the test
procedure in section 3.7. This chapter presents the testing program. Chapter 4 presents
the experimental test results.
3.1 Objective of the Experimental Test
Since most previous experimental research has focused on quite rather simpler D-
Regions, such as deep beams, and designed by few members of STM, more extensive
experimental data was needed on the response of complex D-Regions where the influence
of the shape of the STM are considered as main test parameters. The main objectives of
the experimental tests are thus to provide experimental data for: 1) investigating the
applicability of the STM approach and the associated ACI design code provision in
Appendix A for the design of complex D-Regions; 2) studying the influences of STM
shapes on structural responses and performance under service and over load of complex
D-Regions; and 3) calibrating and validating new nonlinear finite element models that are
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
to be used specialized for predicting the response of complex D-Regions designed in
accordance with the STM.
3.2 Design of the Test Specimens
A total of 17 test specimens of four different types of complex D-Regions were
designed based on the strut-and-tie approach and the associated code provisions in
Appendix A of ACB 18-05. The overall dimensions for all complex D-Regions are given
in Fig.3.1 for: a) the simply-supported deep beam with a rectangular opening beneath the
point of loading; b) the propped cantilever beam with an opening; c) the simply-
supported beam with a dap at one end and an opening adjacent to the support at the other
end; and d) the simply-supported beam with two rectangular openings at one end.
All four complex D-Regions in Fig.3.1 were designed by the STM with different
multiple truss shapes from "conventional" to "permissible but perhaps unwise" since it
was reasonable to expect that different engineers would adopt different and widely
varying STMs to use in their designs. Fig.3.2-3.6 presents the shapes of STMS that were
used for each of the designs.
Fig.3.2 shows the four STMs used to design the simply-supported deep beam with an
opening underneath the loading point. These shapes of STM were adopted from the
models suggested by Reineck (2002) in ACI SP-208. Model 4a as illustrated in Fig. 3.2a,
STM shape consists of a three-pin truss that transfers the load along inclined compressive
struts to a lower truss. The inclined compressive struts were equilibrated by a tension tie
above the opening. Model 4b as illustrated in Fig. 3.2b consists of a statically
indeterminate truss above the opening where vertical tension ties are placed to resist the
tension force induced across the direct struts. The top truss is adopted to transfer the force
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
from the point of loading flowing along two separate load paths to the inclined
compression struts to the lower truss. The truss model 4c in Fig. 3.2c is a statically
indeterminate truss having a top truss to transfer the load from the loading point to the
lower truss.
p
8
N

0
I()
f"-400-i

M
'-c::r---------l:::l""'J_
100H200
1
---1000---+--tl100
(a}
p
1
I()
_L
D -!-
so I j_
150 150
(c)
p

406-+ 279-j 279
(b)
LO
N
.-
.-
1400
(d)
p
900
Fig.3.1 Types and the Overall Dimensions for All Complex D-Regions: (a) The
Simply-Supported Deep Beam with a Rectangular Opening beneath the Point of
Loading; (b) The Propped Cantilever Beam with an Opening; (c) The Simply-
Supported Beam with a Dap at One End and an Opening adjacent to the Support at
the Other End; and (d) The Simply-Supported Beam with Two Rectangular
Openings at One End.
In addition, the secondary resisting truss mechanism can form in this truss model if the
horizontal tie above the opening reaches its capacity, then any additional force will be
transferred by the cantilever trusses extending over the opening. In Fig. 3.2d, the shape of
model 4d was selected such that the force from the loading point is transferred to the
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
support directly without having any lower truss system. The bottom tension tie was
provided to equate the force equilibrium of the truss members above the supports. Based
on topology optimization (more details of the technique will be given in Chapter S),
model 4e, as illustrated in Fig. 3.2e, composes of three compression struts and one
tension tie to carry the load from the top part of the deep beam to the bottom part below
the opening where three tension ties are diagonally oriented to equilibrate the forces
meeting at the nodal zones.
The STM shapes used to design the complex D-Regions in Fig. 3.lb are
illustrated in Fig. 3.3. These shapes were selected based upon intuition. Model Sa as
illustrated in Fig. 3.3a has two loading paths to transfer the load from the loading point
by a direct strut to the simply support and to transfer another portion of the load around
the opening by top and bottom trusses to the continuous support. In Fig. 3.3b, model Sb
transfers a portion of the load to the simple support by an indirect strut and another
portion of the load to the continuous support by diagonal struts and ties around the
opening. Fig. 3.3c illustrates model Sc the load transfers to the simple support by an
indeterminate STM composed of both direct and indirect STM while another portion of
the load around the opening transfers to the continuous support along a flatter truss
system than in model Sa. Based on topology optimization (more details of the technique
will be given in Chapter S), model Sd combines the two systems of truss: 1) three
compression struts and one straight tension tie are used to transfer the load to the pinned
support where a diagonal tie is required to help lifting the force up to satisfy the force
equilibrium; and 2) a truss system that distributes forces around the opening to the
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
cantilever support and that is optimized to be composed of compression struts and
diagonal ties as illustrated in Fig. 3.3d.
(a} (b)
(c} (d}
Fig.3.2 The Four STMs used to Design the Simply-Supported Deep Beam with an
Opening underneath the Loading Point: (a) Model 4a; (b) Model 4b; (c) Model 4c;
(d) Model 4d, and (e) Model 4e (the Optimized Truss).
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
...
o',
...
tl] .......
.....
(c)
I
I
I
I
I
I
I
(b)
(d)
Fig.3.3 The Three STMs used to Design the Propped Cantilever Beam with an
Opening: (a) Model Sa; (b) Model Sb; (c) Model Sc; and (d) Model Sd (the
Optimized Truss).
This part of the optimized truss is unable to automatically satisfy the equilibrium and thus
the alternative truss system (Fig. 3.3d) is adopted by using a straight strut and tie at the
top and bottom of the deep beam to transmit the compressive and tensile stress at the
cantilever side, respectively. In addition, both diagonal struts and ties are placed to
equilibrate the forces distributed around the opening.
The four truss systems used to design the complex D-Regions in Fig. 3.lc are
illustrated in Fig. 3.4. These four were selected from 31 proposed designs to capture a
wide range in STM shapes. In Model 6a as illustrated in Fig. 3.4a, the force transfers
force from the top loading along three struts, where at the <lapped end a diagonal tie is
used to lift this force up and transfers it back to the left support by a prismatic strut. Two
other truss systems are used to carry a portion of the load over the opening to the right
support. In Fig. 3.4b, model 6b is designed as if it were a beam with a depth equal to the
top half of the member since by only using a truss system on the top half of the member
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
to transfer the load. The load transfers to the left support by a rectangular truss while
another portion of the load is transferred by another truss system to the right support.
Model 6c as illustrated in Fig. 3.4c is designed to have load transfer systems over the full
member depth. One truss system is used to transfer the load to the left support while
another transfers the load around the opening to the right support. In Fig. 3.4d, model 6d
having five diagonal struts are used to transfer the load to the lower truss systems in the
bottom half of the member. Three struts transfer the load to the left support while the load
around the opening is transferred by two struts to the right support.
(b)
(c) (d)
Fig.3.4 The Four STMs used to Design the Simply-Supported Beam with a Dap at
One End and an Opening adjacent to the Support at the Other End: (a) Model 6a;
(b) Model 6b; (c) Model 6c; and (d) Model 6d.
Fig.3.5 shows three additional STM shapes that are used to design the
complex D-Regions in Fig. 3.lc. Model 7a (Fig. 3.Sa) is a similar truss system as Model
6a; however, no distributed reinforcement is provided. This test was conducted to
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
investigate the effects of the local ductility on the structural responses and mode of
failures of the complex D-Regions designed by using STM. For Model 7b (Fig. 3.5b), the
shape of STM is obtained based on the topology optimization technique (more details of
the technique will be given in Chapter 5). Model 7b composes of two truss systems to
transfer the load to the dapped end and around the opening to the pinned support. The
load is transferred to the dapped end by two compression struts and a straight and
diagonal tension tie. The diagonal tie is placed to life the force from the bottom of the
deep beam up to the dapped end where the set of compression struts are provided to
satisfy force equilibrium. Two compression struts are formed to distribute the force
around the opening while the two diagonal ties placing above and below the opening to
help to resist the induced bursting tension force for stability and equilibrium of the
system. Since the diagonal tension tie in the optimized STM shape (Fig. 3.5b) might be
impractical in the field, the adjusted STM shape based on the optimized one is thus
adopted in model 7c as illustrated in Fig.3.5c. The adjusted shape is intented to place the
tension tie orthogonally adjacent to the opening. In addition, the force is selected to flow
above the opening by the diagonal struts and the orthogonal ties while only longitudinal
tension tie is placed under the opening. This is to investigate how the behavior in this
region is different from the optimized shape that is presented in Fig. 3.5b.
Fig.3.6 shows the three STMs used to design the complex D-Regions in Fig. 3.ld.
The truss models adopted in design of the complex D-Regions are selected to transfer the
load separately to each side of the supports. One side of the specimens without openings
is designed as a half of simply supported deep beam, while the another side with
openings the load is designed to transfer as a cantilever shear wall. To design the shear
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
wall side, the reversed static loads are needed to be considered. The completed details of
the reversed static load design by STM will be given in Chapter 5. And it should be noted
that the STM shape for the design of the shear wall side was included in the experimental
program only for the most complicated shape.
(a)
(c)
..+...N2
.. .... .. I I ...... N40
N15
I I
I I
I I
I
(b)
N 3
Fig.3.5 The Additional Three STMs used to Design the Simply-Supported Beam
with a Dap at One End and an Opening adjacent to the Support at the Other End:
(a) Model 7a (without Distributed Reinforcement); (b) Model 7b (the Optimized
Truss); and (c) Model 7c (the Optimized Truss with Orthogonal Reinforcement).
In model 8a (Fig. 3.6a), one directed strut that transfer the load directly to the
support is adopted in the deep beam side. A tension tie at the bottom of the deep beam is
used to equilibrate the force above the support. In the shear wall side, only the distributed
reinforcement is provided on each surface of the specimen. This is for a comparison with
other STM shapes adopted to design the shear wall side in the other models. Model 8b as
illustrated in Fig. 3.6b transfers the force in the deep beam side by using a determinate
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
truss model composed of a vertical tension tie and indirect struts. A bottom tension tie is
also provided to equilibrate the forces at the support and at the truss node where a vertical
tension tie and inclined strut meet. In the shear wall side, topology optimization is used to
generate the shape of STM. Compression struts are used to transfer the loads around the
top opening to the concrete region adjacent to the lower opening. The vertical and incline
tension ties are used to complete the force equilibrium of the region adjacent to the top
opening. In Fig. 3.6c, Model 8c transfers the forces in the deep beam side by using an
indeterminate STM composed of a vertical tie, indirect struts, and a direct strut. The
internal forces in the indeterminate truss are proportioned by using the equations
provided by FIP Recommendations 1996 as illustrated in Fig. 3.7. In the shear wall side,
the force transfers around the top and below openings by prismatic and bottom shaped
struts. The orthogonal tension ties aligned in horizontal and vertical direction are used to
equilibrate the force induced from the struts in the regions adjacent to the openings.
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Deep Beam Side
,
,
..
,,
,,
..
,.
,.
,.
..
,.
...
...
,,.
,
.....
,
N 9 N S."''
------
N
9
,
I
,
,
(a)
Deep Beam Side
I
,
I
,
,
,
N
I
,
N
,
,
,
,
,
,
I
,
,
I
I ' ,
(b)
Deep Beam Side
,
,
,
,
,
N 6
,
,
,,.., , ...
, , '
, .. ,
,,. ,
.. ,
,, ,. ,
.. ,
,
,
,
,' ,.',
, ,'
, , ,
4
N
7
6 N f1.,." N ei,
.. 'T'_... ---------------
9
(c)
Shear Wall Side
D
Mesh only
D
Shear Wall Side
Shear Wall Side
Fig.3.6 Three STMs used to Design the Simply-Supported Beam with Two
Rectangular Openings at One End: (a) Model Sa; (b) Model Sb (The Optimized
Truss); and (c) Model Sc (The Conventional Truss).
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
F=2(F1 +F2)
+
, 1 ~ .
a
.. ,
2F1/F=(2a/jd-1 )/3 for jd/2 <a < 2jd
Fig.3. 7 Internal Forces in the Indeterminate Truss proportioned by Using the
Equations provided by FIP Recommendations 1996.
All STM shapes for these complex D-Regions were then designed by the STM
approach using the "CAST-Computer Aided Strut-and-Tie" program to satisfy the
provisions in Appendix A of ACB 18-05. CAST allows designer to select and readily
adjust the geometry of the load resisting truss. It provides for linear truss analysis and
evaluated the stresses in struts, ties, and nodal regions. It was the intent to design each of
these structures for the maximum possible load for each of the selected STM shapes.
With this approach, the capacity was expected to be limited by the structural dimension
and stress limits for strut, ties, and nodes in the ACB 18 provisions and not by the
provided amount ofreinforcement. Table 3.1 summarizes the failure (ultimate state)
co_nditions and nominal designed loads of all 17 truss models suggested by STM models
and the associated ACI design code provisions in Appendix A. Fig.3.8 shows
reinforcement layouts and details of all 17 specimens.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3.1 Summary of Failure (Ultimate State) Conditions and Nominal Design Loads of
All 17 Truss Models by STM Models and Appendix A of ACB 18-08.
Specimen Pn,aci Failure Members
(kN) End nodes Member Type
4A 260 S3 39 Bottle Shaped Strut
S4 39 Bottle Shaped Strut
4B 176 6 12 Tie
4C 188 11 12 Bottle Shaped Strut
4D 207 9 6 Bottle Shaped Strut
4E 330 S3 39 Bottle Shaped Strut
S4 39 Bottle Shaped Strut
SA 238 79 87 Tie
SB 488 20 32 Bottle Shaped Strut
SC 2SO 7 8 Tie
SD S3S 20 32 Bottle Shaped Strut
6A 6Sl 10 12 Bottle Shaped Strut
22 16 Bottle Shaped Strut
23 26 Bottle Shaped Strut
6B 624 3 s Tie
6C 860 16 3 Tie
6D S63 23 28 Tie
7A 6Sl 16 22 Prismatic Strut
7B 1160 21 2S Bottle Shaped Strut
2 40 Bottle Shaped Strut
7C 730 2 32 Bottle Shaped Strut
2S 34 Tie
8A SI 37 39 Tie
8B 740 36 3S Prismatic Strut
8C 737 36 3S Prismatic Strut
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
p
4A 48
4C 40
p
4E
Fig.3.8 (continued on next page)
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3#3
5A
2#4
2#3
D
3#4
5C
50
Fig.3.8 (continued on next page)
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
#3
4x2 2x2#
4#4
4
2#4
2x2#
6A (with meshing) 4#3
I
4x2#3 3x2#3 4x2#3
II
4#4
f#4
I
j
;;CJ

68 4#3
r
4x2#3 2x2#3 4x2#3 4x2#3
4#4
I
'J.--#4
I
4#4
l
II
-
r-
CJ13
\_4tl4
I
/#4
2x2#3 2x2#3
I
6C
4#3
I
2x2#3
2#3 1#3
0#3
II


2x2#3

r
3x2#3 j_#4
I l
60
Fig.3.8 (continued on next page)
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
4x2 2x2#
4#4
2#4
2x2#
?A (without meshing)
4#4
4#4
D
4
78
x2#
4x2#
IJ
3
4#3
7C
Fig.3.8 (continued on next page)
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I /
2#4

I
8A
2#4
88

,..--
1 v1.U">

1x1#4
".I' ,/ T
1x1#3
,,
1x1
"-.. 8#4 v
7
'

1 v1'""
I'

'''
I
I
4#4

I

I
- '\. 12#4
BC
Fig.3.8 Summary of Reinforcement Layouts and Details of All 17 Specimens.
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.3 Material Properties
Material tests were performed to measure the mechanical properties of the
concrete and the steel reinforcement. A summary of these properties is provided in this
section.
3.3.1 Concrete
All specimens in the experimental programs were casted by using self-
consolidating concrete (SCC) with less than 7% of the coarse aggregate above Yi" in
diameter. The use of SCC provided an extremely fluid and workable mix that ensured
proper placement. Fig. 3.9 shows the plots of concrete compressive stress-strain
responses for all types of D-Regions. Table 3.2 shows the concrete compressive strengths
finally achieved at time of test. These strength values are adopted in all subsequent
analyses and modeling by finite element analysis.
45 ~ - - - - - - - - - - - - - - - - - ~ ~ ~ - - - - - - - - - - - - - - - - - - - - - ~ - - - - - - - - - - - - ~ - - - - -
40
35
cu 30
0..
:ii: 25
-
ti)
ti) 20
~
us 15
10
5
-6Ato60
--- 7A to 7C
- 5A to50
- 4A to 4E
-8Ato8C
0 - - ~ ~ ~ ~ ~ . , - - ~ ~ ~ ~ - - . ~ ~ ~ ~ ~ - . - ~ ~ ~ ~ ~ ~
0 0.001 0.002
Strain(mm/m)
0.003
Fig.3.9 Concrete Compressive Stress-Strain Relationships
96
0.004
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3.2 Summary of Concrete Strength Properties.
Specimen Strength (Mpa) Strain *(mm/mm) Modulus (Mpa)
4A 41.8S 0.0027 29660
4B 41.8S 0.0027 29660
4C 41.8S 0.0027 29660
4D 41.8S 0.0027 29660
4E 41.8S 0.0027 29660
SA 40.SS 0.0028 28964
SB 40.SS 0.0028 28964
SC 40.SS 0.0028 28964
SD 40.SS 0.0028 28964
6A 40.4 0.0028 28897
6B 40.4 0.0028 28897
6C 40.4 0.0028 28897
6D 40.4 0.0028 28897
7A 40.7 0.0031 29000
7B 40.7 0.0031 29000
7C 40.7 0.0031 29000
8A 41.8S 0.0027 31000
8B 41.8S 0.0027 31000
8C 41.8S 0.0027 31000
(*: Strain Value at Peak Stress)
3.3.2 Reinforcement
This section provides the material properties of all of the deformed bar
reinforcement used in each experimental program. Fig. 3.10 shows the plots of rebar
stress-strain responses. Table 3.3 summarizes the material properties of all rebar adopted
in all subsequent analyses and modeling by finite element analysis.
3.4 Fabrication of Test Specimens
The fabrication of test specimens in each testing program is described in this
section. The general fabrication procedure of all specimens were as follows: 1) The wood
framework is cut and assembled according to the geometry dimension of specimen; 2)
The bottom mesh is installed; 3) The main reinforcement steel with attached strain
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
gauges is placed into the formwork at the location per drawings; and 4) The top mesh is
lastly placed.

800
700
- 600
cu
a..
== 500
-
en
en 400
!
......
(/) 300
200
100
/
:/
_,.--
_,. .. ---
__ .,,.
,,""' --
________________ .., __
___ .. -
, ---- --
---..,..,,.
,'/
,-/
,-:/
,,
-
-
-#4 BA to BC
----#4 5A to 7C
---
- -#3 5A to BC
------ #3 4A to 4E
-#4 4A to 4E
0
0 0.05 0.1 0.15
Strain( mmf m)
Fig.3.10 Steel Reinforcement Stress-Strain Relationships
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3.3 Summary of Steel Reinforcement Properties.
Specimens 4A-4E 5A-5D
Bars #3 #4 WWR #3 #4 WWR
Nominal Area (mm2) 71 129 0.27* 71 129 0.27*
Yield Stress, Fy (l.\.1p a) 512 435 703.45 491 348 703.45
Young lVlo d ul us, Es (l.\.1p a) 192481 191630 211000 207065 201145 211000
Strain Hardening, Esh(mm/mm) 0.0071 0.0068 0.0018 0.0041 0.012 0.0018
Hardening Modulus, Esh (l\fp a) 3990 3136 0 5266.4 3061.2 0
Tensile Strength, Fu(l.Vrpa) 825 702 703.45 747 511 703.45
Specimens 6A-7C SA-SC
Bars #3 #4 WWR #3 #4 WWR
Nominal Area (mm2) 71 129 0.27* 71 129 0.27*
Yield Stress, Fy (l.Vrpa) 491 348 703.45 491 458 703.45
Young lVlo d ul us, Es (l.\.1p a) 207065 201145 211000 207065 203556 211000
Strain Hardening, esh(mm/mm) 0.0041 0.012 0.0018 00041 0.0062 0.0018
Hardening l.Vlodulus, Esh (l\fpa) 5266.4 3061.2 0 5266.4 5749.6 0
Tensile Strength, Fu(l.Vrpa) 747 511 703.45 747 705 703.45
(*:Value in Percent,%)
Figs. 3.lla to d illustrate the reinforcement cage for truss models 4a to d,
respectively. A welded wire mesh with a reinforcement ratio similar to that required in
ACB 18-08 for deep beam was provided in nearly all specimens so as to provide a
minimum level of ductility. Construction tolerances were not exceptionally tight, as the
welded wire mesh was large enough that it made it difficult to tie stirrups and bars in the
exact location intended in the CAST model. It should be noted that, at the time of the
experiment, the specimen designed using topology optimization (model 4e) was not
available since it was planed to conduct the computational prediction of the structural
responses.
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(b) Specimen 4a
(b) Specimen 4b
Fig.3.11 (continued on next page)
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(c) Specimen 4c
( d) Specimen 4d
Fig.3.11 Reinforcement Cage for a Truss Model 4A to D
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figs. 3.12a to c illustrate the reinforcement cages for truss models Sa to c,
respectively. To ensure the proper anchorage and behavior at the boundary of the
cantilever side of the specimens, the formworks were built with an extra part of the
concrete that extends from the fix ended side as originally designed in CAST model
further down below the specimen as L-shape. A welded wire mesh was also provided in
all specimens in an effort to ensure a more ductile response. Similar to model 4e, the
specimen designed using topology optimization (model Sd) was not available since it was
planed to conduct the computational prediction of the structural responses.
(a) Specimen Sa
Fig.3.12 (continued on next page)
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(b) Specimen Sb
(c) Specimen Sc
Fig.3.12 Reinforcement Cage for a Truss Model SA to C
Figs. 3.13a to c illustrate the reinforcement cages for truss models 7a to 7c,
respectively. In these fabrications, except specimen 7a, all specimens are fabricated with
the additional weld wire mesh placed just below the surface on both sides of the
specimen to ensure the ductility as required per ACB 18-08.
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a) Specimen 7a
(b) Specimen 7b
Fig.3.13 (continued on next page)
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(c) Specimen 7c
Fig.3.13 Reinforcement Cage for a Truss Model 7A to C
Figs. 3.14a to c illustrate the reinforcement cage for truss models 8a to 8c,
respectively. To ensure the sufficient anchorage of all reinforcements, the horizontal and
inclined reinforcements were anchored to the steel end plates, and the vertical
reinforcements were bent to u-shape at one end and 90 degree hook at another end.
(a) Specimen 8a
Fig.3.14 (continued on next page)
105
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(b) Specimen 8b
(c) Specimen 8c
Fig.3.14 Reinforcement Cage for a Truss Model 8A to C
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.5 Experimental Test Set-up
The experimental test set-up for D-Regions in Figs. 3.la, b, and c was completed
by placing the specimen on the top of a steel beam (Fig. 3.15a) inside the 600 kips (2700
kN) uniaxial servo-controlled hydraulic frame. Except for specimens of D-Region in Fig.
3.lb, all specimens were simply supported by the use of pinned and free roller supports,
and were loaded on the top face at a single location illustrated in Figs.3.la and c. To
provide the fully fixed restraint at the boundary of specimens in Fig.3.lb, the portion of
concrete "L" shape was cast with a steel plate embedded at the bottom, and the high
strength bolts were used to fasten the plate rigidly to the reaction steel beam. In all
specimens, the applied load and reactions were distributed on the reaction beam surface
with the use of 25 mm (1 in) thick by 150 mm ( 6 in.) square steel plates.
For the specimens ofD-Regions in Fig. 3.ld, the test-setup (Fig. 3.15b) utilized the
steel supports bearing on the strong floor and a twenty-eight foot tall, L-shaped reaction
strong wall in Newmark Structural Engineering Laboratory. The load was applied via an
overhead-mounted hand driven hydraulic jack with a 1335 kN (300 kips) capacity.
Similarly, the 25 mm (1 in) thick by 150 mm (6 in) square steel plates were used to
distribute the forces at the loading point and the supports.
3.6 Instrumentation
A combination of traditional and advanced instrumentations was used to collect
the test data for all complex D-Regions. This consisted of load cells, displacement
transducers, electrical resistance strain gages on reinforcement and concrete surface, and
the Krypton K600 Dynamic Measurement Machine (DMM). The K600 camera is able to
107
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
accurately record in 3-Dimensional space the x,y,z coordinates of a large number of Light
Emitting Diodes (LEDs) that are mounted to the surface of the test structure.
(a)
(b)
Fig. 3.15 Experimental Test Set-Up: (a) Specimen 4a to 7c; (b) Specimen Sa
to Sc
108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
(b)
Fig.3.16 Examples of Instrumentation: (a) Reinforcement Strain Gage Locations in
One of the Complex D-Regions, and (b) Krypton LED and Surface Strain Gages
Locations in One of the Complex D-Regions
Figs. 3.16a and b provide an example of the instrumentation used in these tests.
Fig. 3.16a illustrates where reinforcement strain gages were attached to measure strain
development in key ties e.g. ties crossing bottle-shaped struts, ties passing though nodal
zone regions above the supports, ties above the opening, and in the locations of the
expected high tensile straining. The Krypton LED targets and surface strain gages, as
illustrated in Fig. 3.16b, were attached to the surface of the complex D-Regions to
measure the movement of the specific points in D-Regions and the concrete strain
history, respectively. In addition, the dense LED targets were attached to the bottle-
109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
shaped strut above the support in both the longitudinal and transverse locations to obtain
the strain distribution used to characterize the load transfer mechanism of the regions.
tests.
Fig. 3.17 provides the details and locations of the instrumentation used in these
LEDs Location
p
1
1
35 (/
~
Concrete Gauges
Reinforcement Gauges(4A) Reinforcement Gauges (48)
Reinforcement Gauges(4C) Reinforcement Gauges (40)
Fig.3.17 (continued on next page)
110
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.


CG26
LED Mark/
CGiO S

G71a ..... CG24
CG22
Concre1e Surface_/
train Gauge ... CG23
Reinforcement
Surface Strai
/uge
SG71d
-100 0 100 200 300 400 600 600 700 600 900 1000 1100 1200 1300 1400 1500 160 1700 1800 1900
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
6A
CG10

,
'
3>3

G1a
CG9 (
2
CG4
c 8

SG4


a
s 4b

,12
CG1
1b
I CG11
SG7a

-100 0 100 200 300 400 600 600 700 800 900 1000 1100 1200 1300 1400 1500 160 1700 1800 1900
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
68
Fig.3.17 (continued on next page)
111
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-300
-100 o 100 200 300 400 500 eoo 700 eoo 900 1000 1100 1200 1300 1400 1500 160 1700 1800 1900
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
6C
CG12
c ~
~
s 5a

.CG,
1

CG8
~

CG} \
SG10


1
SG6a
SG5ti 12b
CG6
CG9

,
c ~
c
-1 oo o 100 200 300 400 500 eoo 700 800 900 1000 1100 1200 1300 1400 1500 160 1700 1800 1900
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
60
Fig.3.17 (continued on next page)
112
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CG5
CG4
,
CG6
SG3
SG4
SG1
G6
-100 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 160 1700 1600 1900
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
7A
CG5 CG6
~
CG10
I
-100 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 160 1700 1800 1900
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
78
Fig.3.17 (continued on next page)
113
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CG5 CG6
CG4
4fl> SG10.
SG8.1
SG8.2
SG10.2
/cG7
SG10.3

-100 0 100 200 300 400 500 EKlO 700 EKlO 90010001100120013001400150016017001EKJO 1900
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
7C

'!112
?.


'CG14

' .

.17 .

CG4
CG13
I

. "'.

'
CG6
4'CG1

LEDs Marker and Surface Concrete Strain Gauges for Specimens 8A to 8C
Fig.3.17 (continued on next page)
114
a1s
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1000
SG5 SG6 SG? SG8 SG9
1000
-----..
--
x
60
SG5 > SG57

59
--. ...... I
-
-
> SG56
SG61

SG51 ''
SG41
>
SG55
nSG40 SG42

'-'"'""'
vv-.v
...,.., .. ,
'-'V"tU '-'\J' CV
Fig.3.17 Details and Locations of the Instrumentations in All Specimens
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.7 Test Procedure
The test procedure started by placing the specimen in the position underneath the
reaction frame or the reaction wall. Next, the pinned and roller supports were placed and
adjusted underneath the specimen, and the loading plate was placed on the top of the
specimen. Subsequently, the hydraulic jack was lowered into the position.
The specimen was loaded in displacement control so as to obtain a more stable
post-peak response can be obtained. The first loading step was recorded at the time of
first crack occurred, and then subsequent loadings were taken in fixed loading steps (i.e.
such as in 15 or 20 kips increments) or when there audible or visible clue for potentially
significant damage. Cracks were marked and then pictures taken in each loading step and
when crucial structural responses were observed. In each specimen, the test was
continued until the failure of the specimen occurred. In this context, failure was defined
as a very significant reduction in load carrying capacity, perhaps close to 50% of
maximum. Sometimes the loading was stopped prior to such a large reduction in capacity
to facilitate the removal of the test specimen from the test setup.
3.8 Summary
In this chapter, a full description of the 17 experiments on the four different
complex D-Regions was presented. All specimens were designed with the conventional
and very complex truss models to investigate the applicability of the STM approach and
the associated ACI design code provision in Appendix A for the design of complex D-
Regions, to study the influences of STM shapes on structural responses and performance
under service and overloads of complex D-Regions, and to calibrate and validate the new
nonlinear finite element models that is specialized for predicting the response of D
116
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Regions designed by the STM. Both traditional and advanced instruments were installed
both internally and externally to measure and record the structural responses of these D-
Region test specimens. These instrumentations included load cells, electrical resistance
strain gauges, and the Krypton K600 coordinate measurement machine. The specimens
were loaded under displacement controlled until failure occurred. Cracks were marked
and photos were taken during the testing process in every loading step and when the
significant structural responses were occurred. The collected data was used to validate the
applicability of the STM approach and the Appendix A of ACI design code provision in
chapter 4, assess the influences of shapes of STM on the structural responses in chapter 5,
and validate the proposed nonlinear FEA model for structural concrete D-Regions in
chapter 6.
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter4
Experimental Validation of STM in Design of Complex D-Regions
This chapter presents the results of the experimental program that was described in
Chapter 3 and the experimental validation of STM and the Appendix A of ACI 318-08
code provisions in design of complex D-Regions. An introduction to the need for
experimental validation of STM in given in section 4.1, the experimental validation of D-
Regions designed for non-reversed static loadings in given in section 4.2, the
experimental validation of D-Regions designed for reversed static loadings in given in
section 4.3, and a summary of the experimental results and key observations of the
experimental validation of STM is provided given in section 4.4.
4.1 Introduction
The strut-and-tie model (STM) has been originally proposed based on the
concept of truss analogy that was presented by Ritter (1899) and Morsch (1909). The
model was extented to the design of D-Regions, where the strain distribution across
section is highly nonlinear, by Schlaich et al (1987, 1991). With the use of the STM
approach, a load-resisting truss is idealized to carry forces through the D-Regions to its
supports. The designer is free to choose the shape of STM with only limited guidance and
constraints. The safety of STM method relies on the local ductility in the D-Regions
being sufficient to allow the load to be transferred in the manner selected by the designer.
Several code provisions for the design D-Regions by a STM approach are
available. This included Appendix A of ACI 318-08 provisions, the Canadian Standards
Association, the AASHTO LRFD Bridge, and the CEB-FIP Guidelines. These provisions
also assume that the structure is sufficiently ductile as required by STM method. There
118
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
have been few experimental validations of D-Regions designed by the code provisions of
STM and typically, only simple D-Regions with a few truss members, such as deep
beams, have been properly validated.
The results from the tests that were described in Chapter 3 are now used to
validate of STM method for it's safety and applicability for the design of more complex
D-Regions.
4.2 D-Regions Designed for Non-Reversed Static Loadings
In this section, the experimental results of the D-Regions designed for non-
reversed static loadings will be presented and adopted for the validation of the STM
philosophy and associated ACI Code provisions. Based on chapter 3, the D-Regions
designed for non-reversed static loadings are specimens 4A to 7 A.
4.2.1 Load-Deformation Response of Test Specimens
The measured load-deformation responses of test specimens are presented in
Fig. 4.1. The responses of the different D-Regions are plotted separately. The responses
ofD-Regions in Figs. 3.la, 3.lb, and 3.lc are illustrated in Figs.4.la, 4.lb, and 4.lc,
respectively. The load applied to the specimens is measured by the load cell either
available with the hydraulic testing machine or seating on the top of the specimens. The
measured deformation presented in the figure is relative to the supports of the specimens,
e.g., the movement and slip between the bearing plate and structures were able to be
neglected by direct measurement of specimen deformation. The plots of the responses of
a member that contained only welded-wire mesh (WWR) are also included for
comparison.
119
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
300
-
z
c,200
a.
100

0 1 2 3 4 5 6
Displacement (mm)
(a)
500
,''
.; ............
...
..
...
,
...
,
'
,
..
.
'
'
'
, .
' --.. I ,
,
400
z300
.ii:
-
a.
200
100
-----/------
,,....-----
- ...
wwr

0 1 2 3 4 5
Displacement (mm)
(b)
6
Fig.4.1 (continued on next page)
120
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1400
1200
1000
z
800
.ll:
ii:'
600
400
200
0
0 2 4 6 8 10 12
Displacement (mm)
(c)
Fig.4.1 Measured Load-Deformation Response of Test Structures.
From Figs. 4.la to c, it is observed that the shape of STM or pattern of tie
reinforcement had a variable level of influence on the post-cracking stiffness of the
specimens up to the service load levels, on the capacity of the specimens, and on the
ductility of the specimens for the overloads. Insignificant effects on the post-cracking
stiffness were observed for the D-Regions in Fig. 3.la; but significant effects on the post-
cracking stiffness were observed for D-Regions in Figs. 3.lb and c. The capacity ofD-
Regions in Fig. 3.la was not considerably different while the capacity of D-Regions in
Figs.3.1 b and c was greatly affected by the shape of STM. The observed ductility of D-
Regions in Fig. 3.lc was most affected by the shape of STM; however, D-Regions in
Figs.3.la and b had quite similar level of ductility.
4.2.2 Comparison of Measured and Calculated Strengths
The comparison of measured and calculated strengths of specimens 4A to 7 A is
presented in Table 4.1. The measured strength of the specimens is obtained at the peak
point of the measured load-deformation responses illustrated in Fig.4.1. The calculated
strengths of the specimens were provided in Table 3.1.
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In the comparison, the calculated capacity of STM was chosen such that it was
not limited by the calculated capacity of the nodal zones to provide a more critical
assessment of the STM and also since all nodal zones within a continuum are not fully
designed and the dimensions of the nodal zones are sometimes unclear due to the limited
guidance from the code provisions.
Table 4.1 Comparison of Calculated Nominal Capacities and Measured Strengths
Specimen PrestlkN) Pn ac1lkN) PresJPn acl
4a 325 260 1.25
4b 334 176 1.89
4c 312 188 1.65
4d 356 207 1.72
5a 418 238 1.76
5b 551 488 1.13
5c 418 250 1.67
6a 1120 651 1.72
6b 814 624 1.30
6c 1292 860 1.50
6d 863 563 1.53
7a 258 651 0.39
Notes: Prest= measured specimen strength; Pn,aci =calculated model strength.
Based on the Table 4.1, the mean value of the ratio of the measured strength to
the nominal design capacity by STM is 1.54 with a coefficient of variation of 15%. This
thus suggests that the strut-and-tie provisions in Appendix A of ACB 18-08 is
conservative in terms of the capacity as evaluated in this experimental program.
It should be noted that the conservatism of the STM was obtained from the mean
value that was calculated based on only the specimens provided with the distributed
reinforcement, e.g. not included the specimen 7 A, since the minimum distributed
reinforcement is a requirement addressed ACB 18-08. The significance of the minimum
distributed reinforcement is; however, observed and provided with the comprehensive
details in section 4.2.6.
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.2.3 Comparison of Modes of Failure
A comparison of modes of failure is presented in Fig. 4.2. The failure modes of
the different D-Regions are illustrated separately. In Fig.4.2a, the failure modes of D-
Regions in Fig.3.la that were designed with different four truss models are presented. In
these specimens, the similar failure modes were observed. Specimen 4A reached its
capacity by crushing of the concrete above the opening. The crack patterns of the
crushing regions are confirmation that the strut action was developed to transfer the load
to the lower part of the specimen. The failure mode suggested by the STM design of the
model was also similar, which was by crushing of the bottle shaped strut spanned
between nodes N39 and N53 as described in Table 3.1. The failure mode of specimen 4B
by crushing of the compression strut was also observed in the experiment. The more
redistribution of the internal stress was observed in this specimen through the secondary
failure mode which occurred due to the cantilever-like action after the concrete regions
above the opening that acted like the struts were completely crushed. The STM model of
the specimen; however, was unable to capture the same failure mode since the model
predicted the failure mode by yielding of the tension tie that spanned between nodes N6
and N12. The same failure mode by crushing of the concrete regions above the opening
was observed in Specimen 4C. In this experiment, a greater redistribution of the stress
after the first failure mode was particularly observed due to the ability to resist tension
demand of the tension tie that spanned between N21 and N22. This particular truss shape
was intentionally selected to resist the secondary failure by the cantilever action. The
prediction of the failure mode suggested by STM in this model was slightly inaccurate
since the model suggested the failure mode by crushing of the strut spanned between N 11
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and Nl2. In specimen 4D, it was observed that the failure mode occurred by crushing of
the strut above the opening and followed by the secondary failure mechanism by the
sliding shear failure underneath the opening. The secondary failure mode was not quite
surprising since insufficient reinforcement was provided in the area underneath the
opening to prevent this sliding failure. The STM design for this model predicted the
failure mode would be by crushing of the strut that spanned between nodes N6 and N9
which is inaccurate for the location of the observed failure.
The failure modes of D-Regions in Fig. 3.lb are illustrated in Fig.4.2b. Specimen
SA was observed the failure mode by crushing of the concrete regions above and
underneath the openings. No secondary failure mechanism was observed in this specimen
since there is a significant drop of the load-deformation response as illustrated in Fig.
4.lb. The STM design for this specimen predicted that the failure mode of the specimen
would be by yielding of the tension tie that spanned between nodes N79 and the support,
and no significant level of the compressive stress demand in the struts above and
underneath the opening was captured in the model. The failure mode of specimen SB was
observed by crushing of the concrete region above the opening and immediately followed
by the slip along the crack underneath the opening. The slippage along the crack
underneath the opening was definitely influenced by the orientation of the reinforcement
since insufficient reinforcement was provided in the location that cracks propagated
through. The failure mode suggested by the STM model of the specimen was also unable
to correctly capture the crushing location of the strut although the failure mode was
suggested by the crushing of the strut that spanned between N20 and the support.
Specimen SC was experimentally observed to have a very similar failure mode as in
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
specimen SA in that the struts above and underneath the opening were crushed and this
was followed by slippage along the cracks. The failure mode was also unable to be
captured correctly by the STM design which indicated that yielding of the tension tie that
spanned between nodes N7 and N8 was predicted to limit the capacity of this structure.
125
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Crushing (4A and C)
Cantilever act n
(48)
(4A)
(4C)
Crushing (40)
Crushing (48)
(a)
Ca tilever action ( 48)
(48)
(40)
Fig.4.2 (continued on next page)
126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Crushing ~ < : } : { ; ~ ~
(SA, B, and C)
(5C)
Crushing
(SA and C)
rack( SB)
(b)
(SB)
Fig.4.2 (continued on next page)
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Split . . . ( 6b)
. -

: ( 6d)

( 6a)
(7a)
(6a) (6b) (6d)
(c)
Fig.4.2 Mode of Failure for Specimens 4A to 7 A: (a) Specimen 4A to 4D; (b)
Specimen SA to SC; and (c) Specimen 6A to 7A.
Fig. 4.2c illustrates the observed failure mode of the D-Region in Fig. 3.lc. Four
distinct failure modes were observed for the four specimens designed using different truss
shapes. The capacity of the specimen 6A was reached when the diagonal strut 29-30
began to crush. Subsequently, the load-carrying capacity of this strut was exhausted and
thus this leads to a significant drop in stiffness and the load shifted to being carried by the
128
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
truss that spanned above the opening. This truss was unable to carry this additional
burden and quickly failed as illustrated in Fig.4.2c. The STM design for specimen 6A
indicates that the capacity is controlled by struts connecting nodes 10-12, 22-16, and 23-
26 as presented in Table 3 .1. This thus failed to capture the correct location of the failure
mode by crushing of the strut 29-30. Specimen 6B reached its capacity when the cracking
extended up from the base of the horizontal tie that is above the opening from nodes 3 to
5, and subsequently the concrete strut from nodes 2 to 4 crushed and split as also
illustrated in Fig. 4.2c. The STM model used for designing this specimen indicates that
the capacity is limited by yielding of the tension tie that connects node 3 to 5. The
capacity of Specimen 6C was reached when the strut from node 20-24 crushed, split, and
subsequently extended into a compression shear crack. The STM model for specimen 6C
is unable to predict the accurate location and the type of the failure since yielding of the
tension tie connecting nodes 16 and 3 was predicted to govern the failure mechanism.
Crushing of the strut between nodes 34 and 36 was observed as the failure mode of the
specimen 6D as illustrated in Fig. 4.2c. The STM design capacity for specimen 6D was
controlled by the tie connecting nodes 23 and 28 which is not the mode of failure
observed from the experiment.
Thus, based on the caparison between the failure modes predicted from STM
models and those observed from the experiment, the ability of STM to predict the failure
modes is poor. This is not surprising because, the trusses are statically indeterminate, and
the relative stiffness of the struts and ties are not considered and difficult to determine
due to the limited guidance from the code provisions. Some of the models may also not
have exhibited sufficient ductility from the plasticity assumption of STM to be satisfied.
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In addition, in STM, only the approximate discretization of the force transfer
inside the continuum as a truss structure is used in the design to estimate the nominal
capacity and failure mode of STM with adequate margin of safety. The ability to predict
the failure mode of STM thus depends on how well the designed STM (the truss shape,
STM member orientation, provided reinforcement, and the designed magnitude of
demand) could mimic the load transfer characteristics inside the cracked D-Regions. The
direction of the diagonal compression force and the assumed angle for strut might be
different due to the nonlinearity of the material caused from concrete, crack, provided
reinforcement, and due to the demand from the flow of force.
To provide in mode details on when and why the STM would provide a reasonably
estimate of failure mode and capacity, the completed study on the development of
cracking (Section 4.2.4) and the load transfer characteristics (Section 4.2.5) inside D-
Regions is necessary.
4.2.4 Development of Cracking in Test Structures
The development of cracking in the test structures is presented in this section.
Understanding how the cracking develops and propagates in D-Regions designed by
STM is important because the STM is only a procedure for calculating capacity and also
given the flexibility afforded to the designer in shape selection of STM. Thus, there is the
legitimate concern that the STM permits structures to be designed that may have
sufficient capacity but that could experience very significant and undesired cracking
under service load levels.
With the use of high-resolution photographs taken during the experimental
programs, the development of cracking was recorded by crack markings. Figs 4.3a and b
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
present the state of measured cracking at three different load levels of D-Regions in
Figs.3.lb and c, respectively. In some of the specimens, the cracking state of the
specimen reinforced with only welded-wire reinforcement (WWR) is also provided to
compare the extent of cracking with other specimens that had tie reinforcement as
required by their particular STM design.
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
p
Specimen 5A (1/2Pn=119 kN, 3/4Pn=178 kN,Pn=238 kN)
p
200 kN
Specimen 5 B(1/2Pn=244 kN, 3/4Pn=366 kN,Pn=488 kN)
p
400 ~
310 kN .. /
....
200 kN
Specimen 5C (1/2Pn=125 kN, 3/4Pn=187 kN,Pn=250 kN)
(a)
Fig. 4.3 (continued on next page)
132
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
!P

......... - -7-;:::..--r I TT _______ _
"'-(' , 534 kN , ,,.! ' , .......... '.\ . 1
\ . / .'--t"" ll 222kNi
\356 kN

i534 kN .-----
_J . I , f t ! r-- . 3 . 56 . / r 1 i
(' . .,_ ': : </
-' .! \ . \ \ ...... ' l \
. \ \ / _) '
\ I l ; ; ( \ I I l \ lh l \ }j 1.LL_

(a) Specimen 6a ('/,P, = 326 kN, 'l,P, = 488 kN, P, = 651 kN) (d) Specimen 6d ('t,P, = 245 kN, '/,P, = 368 kN, P, = 490 kN)
534 kN
i \"-. .. \-.. .. '."' ... 356 kN
'j \ '\ '.
' \ kN \ '
-.-
-'",-:\
(b) Specimen 6b ('/,P, = 312 kN, '/,P, = 468 kN. P, = 624 kN)
IP
.....:t.:..
356lkN.- .... /
: J ("

" ' 534 kN
\'
i
.. ' ' \
: 1. > '
. . .
,;;,.
(c) Specimen 6c ('i,P, = 430 kN, 'i,P, = 645 kN, P, = 860 kN)
(b)
191 kN ...- \.._
/ookN \
j120 kN \ ... __ _
(e) Mesh Specimen
Fig. 4.3 Observed Cracking in Specimen SA to 6D: (a) Specimen SA to SC tested at
P= 200 kN (45 kips), 310 kN (70 kips), and 400 kN (90 kips); and (b) Specimen 6A to
6D tested at P= 222 kN (49.9 kips), 3S6 kN (80.3 kips), and S34 kN (120 kips).
The cracking state of the specimen SA to SC was illustrated in Fig.4.3a. Three
different load levels: 200 kN ( 4S kips), 310 kN (70 kips), and 400 kN (90 kips), are
presented for comparing the crack development and propagation in the specimens. For
specimen SA, at the load level of200 kN (4S kips), a crack was observed to have
developed in the web region within the shear span. By increasing the load to the level of
310 kN( 70 kips), new cracks had formed at the comer of the opening and the bottom of
the specimen at mid-span (akin to flexural cracks). The existing crack in the web region
133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
propagated upward to the point of loading and downward to the support. After applying
the load up to the level of 400 kN( 90 kips), new cracks formed in the concrete region
above the opening and propagated upward to the point of loading, at the bottom of the
specimen (flexural cracks), and in the cantilever region due to the negative bending
moment (flexural cracks). No crack propagation from the existing cracks at the comer of
the opening was observed and this was due to the presence of the orthogonal
reinforcement that surrounded the opening. For specimen SB, at the level of 200 kN (4S
kips), several flexural cracks formed at the bottom side of the specimen. A few cracks
also developed at the comer of the opening due to the high tension demand at this
location. At the load level of 310 kN (70 kips), more flexural cracks developed at the
bottom of the specimen and the existing flexural cracks propagated upward. New web
shear cracks were observed at the web regions within the shear span of the specimen. In
addition, cracks at the comer of the opening propagated upward to the loading point. At
the load level of 400 kN ( 90 kips), new cracks formed above and underneath the
opening. The crack above the opening also propagated upward to the loading point in the
horizontal orientation to the top surface of the specimen. This crack potentially caused
the splitting mechanism on the concrete adjacent to the loading point as observed at the
higher load levels. For specimen SC, at the load level of200 kN (4S kips), flexural cracks
formed at the bottom of the specimen. A few cracks developed at the comer of the
opening. By increasing the applied load up to the level of 310 kN (70 kips), the crack at
comers propagated away from the opening and a new web shear crack was observed in
the web region adjacent to the cantilever side of the specimen. No flexural crack at the
cantilever side was observed. In addition, the inclined cracks also formed underneath the
134
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
opening at this loading step due to the high tension demand and a lack of provided
reinforcement. At the load level of 400 kN (90 kips), a new web shear crack formed in
the shear span of the pinned support side and propagated downward towards the support.
The existing inclined cracks underneath the opening propagated downward to the bottom
of the specimen and potentially caused the splitting and sliding along the crack surfaces
at the higher applied loads. By comparing the cracking pattern at the same loading level
of specimen 5B with the crack pattern observed in specimen 5A and C, insignificant
levels of crack development were observed in specimen 5B; in addition, the stiffness of
specimen 5B is much greater than those specimens (Fig.4.lb). This is not surprising since
the orientation of the reinforcement above and underneath the opening of specimen 5B
was adjusted based on the principal tensile stress of the elastic analysis and the fewer
quantity of the reinforcement was provided, this thus allows more stress redistribution
after cracks formed and leads to the structure with higher stiffness.
Fig.4.3b illustrates the cracking state of the specimen 6A to 6D. The significant
cracks in the WWR specimen occurred at the load of 98 kN (22 kips); there were only a
few large and capacity-limiting cracks at the ultimate capacity of 191 kN ( 43 kips). By
contrast, at a slightly higher load in specimen 6a through 6d, very limited cracking was
observed. Cracks that developed initially in specimen 6a though 6d generally did not
lengthen as the load approached failure. As the load increased, the internal stress
distribution occurred and this leaded to the development of cracks in the new location. In
addition, by considering the level of the crack propagation and development in the
regions of the specimens, specimen 6C was observed to have the best structural
performance under service loads with very limited crack width and propagation. This
135
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
distinct structural performance could be also illustrated by the very high stiffness of the
specimen, particularly, the post-cracking stiffness. Specimen 6B was; however, observed
to have the poorest structural performance since the specimen has more significant cracks
at service load levels along the bottom of the specimen due to no tie to control the
development of cracks at the base of the structure.
From the above study on the observed crack development and propagation, it was
clearly shown that the presence and the pattern of the tie reinforcement could potentially
affect the development of the first crack and the propagation of the damage due to cracks
in D-Regions. In addition, the proper orientation and pattern of the tie reinforcement
could be significant in controlling the development and propagation of cracks which
leads to the less and controllable damage and higher structural performance.
4.2.5 Straining and Load Transfer Characteristics in Test Structures
In this section, straining and load transfer characteristics in tested specimens are
presented based on the measured strains on the reinforcement and the surface of the
concrete in specific locations. The strain history in the strain gauges is very useful for
studying how the force transfer mechanism in D-Regions is influenced by the selected
truss models. In particular, it is important to assess if the internal force inside D-Regions
is transferred in the same way as the assumed truss model?
The study was focused on only the selected D-Regions in Fig. 3.lc that is most
complex and has the sufficient experimental data from the instruments. Figs. 4.4 and 4.5
illustrated the plots of strains measured by the strain gauges attached to the reinforcement
and the surface of concrete of the selected specimens: 6A, 7B, and 7C at the selected
locations illustrated in Fig. 3.17. For the specimen 6A, the locations of the strain gauges
136
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
were intentionally selected at the center of all STM members. For specimens 7B and 7C,
those locations were similar for the concrete surface gauges.
In Fig.4.4, the measured strains at the locations of key STM members of
specimen 6A are presented. Considering the STM model of the specimen 6A illustrated
in Fig. 3.Sa, strain gauges no.25 and 24 measured the longitudinal and transverse
straining in the bottom shaped strut connecting nodes N2 and N25. Those strain gauges
measured tensile straining (expansion) without any compressive strains (contraction).
Thus, the measured strains were correlated well with the calculated stress demand in the
STM model since very low compressive stress demand (nearly zero stress demand) was
predicted to transfer in this strut. In addition, this can explain by considering the crack
patterns of the specimen in Fig.4.3b in which cracks propagated upward the location near
the strain gauges and thus the compressive stress representing the bottle shaped strut was
reoriented in the different direction. The strain gauges no.22 and 23 measured the
longitudinal and transverse straining in the bottle-shaped strut (between nodes N2 and
N23). The measured strains from those strain gauges are in compression (contraction)
and tension (expansion), respectively. This correlates well to the STM model that
assumed that there was a bottle-shaped strut in this location. In addition, the slope of the
load-strain curve for the strut, which could be considered as axial stiffness of the strut,
shows a very stiff response. This also correlated well with the very few crack
development observed in the region as illustrated in Fig. 4.3b. The strain gauges no. 29
and 28 measured the longitudinal and transverse straining in the prismatic strut (between
nodes N22 and N16) reported both contraction and compressive strains. The measured
strain thus correlated well with the STM model that assumed the prismatic strut
137
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
developed in the regions. It should noted that the contraction strain occurred in the strain
gauge no. 28 when no applied force to the structure was due to the existing shrinkage
strain of the harden concrete (Fig.4.3b). The axial stiffness or the slope of the load-strain
response of this prismatic strut was also reported to be distinctively most stiff and this is
due to no transverse tensile strain that softens the compression strut as in the case of
bottle shaped struts. In addition, the strain gages (Fig. 4.4b) attached to the reinforcement
that represents the tension ties were all in tensile strain (expansion). This thus confirms
the validity of the STM model assumed for the load transfer mechanism of the specimen
6A.
In Fig.4.5, the measured strains from the concrete surface strain gauges at the selected
locations of specimen 7B and C are presented. At a very small applied load level (before
first cracking), all strain gauges at the same locations of both specimens were measured
almost the same level of the strain. After fist cracking, the stress redistribution occurred
and the internal truss mechanism formed and transferred the load like the assumed STM
models of both specimens. At this stage, strain gauges at the same locations of both
specimens began to capture different level of strains. This is not surprising since the level
of strains now depends on the nonlinear characteristics of STM members.
Strain gauge (CG2) in specimen 7B reported the higher compressive strain level
than in specimen 7C while the transverse strain (CGI) near the CG2 gauge of both
specimens reported slightly values in the strain level. This is because a portion of the load
in specimen 7C was transferred to the support by the upper load paths as illustrated in
Fig. 3.5c (Struts connecting nodes N2 and N27, N26 and N27, and N21 and N26) while a
portion of the load in specimen 7B was more likely to be transferred by the strut
138
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
connecting nodes N2 and N25 as illustrated in Fig. 3.5b. In addition, as illustrated in
Fig.4.5, the load transfer by the upper path in the specimen 7C was also evident by the
level of strain measured in the gauges CG3-7C and CG5-7C which was distinctively
higher than the strain level measured by strain gauges CG3-7B and CG5-7B,
respectively.
Thus, based on the measured strain data in Figs. 4.4 and 4.5, the load transfer
characteristics ofD-Regions is initially similar to an elastic structure of 2D continuum
just before the cracks occurs. After cracks, whether D-Regions can transfer the load to the
supports by the internal truss mechanism as envisaged by the designer or not is
significantly influenced by the selected STM shapes, the provided reinforcement in
tension ties and distributed reinforcement, and the magnitude of the stress intensity of the
design. These factors are significant in controlling the crack development and
propagation inside the D-Regions to properly form in the patterns that forms the
"equilibrating" and "stable" internal truss mechanism to transfer the forces to the
supports. The proper reinforcement provided in the D-Regions could effectively control
the crack development and propagation that allows the same internal truss mechanism as
the designed STM to form. If the magnitude of the demand is not too high from the
calculated demand in the designed STM, at least the load resistance equal to the designed
STM could be expected from the internal truss that forms inside the D-Regions. When
the improper reinforcement (either tie or distributed reinforcement) is provided, the crack
could potentially propagate in the orientation that is totally different from the orientation
of the designed struts and then causes the reorientation of the internal stress to form the
new internal truss mechanism that can be totally different from the designed truss. The
139
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
new internal truss mechanism thus leads to the different failure mechanism that might be
different from the failure mode and capacity suggested by the firstly designed STM. If
the provided reinforcement is not sufficient to allow the new truss mechanism form, the
redistribution of the stress inside the D-Regions will occur again to find the new
equilibrium and stable state of the new internal truss mechanism which crucially relies on
the local ductility provided in the structure by the use of distributed reinforcement. The
insufficient local ductility could lead to the failure of the D-Regions due to the unstable
crack propagation which causes from the reorientation and redistribution of the internal
truss mechanism.
140
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
........ ----------.-- 1200
CG29 CG25 CG23
CG22
----CG23
CG24
-CG25
-CG28
- CG29
CG28

I
\ \
CG22 ..

'\
': \
II
:.

! ' 1-!f
\ ! 1
: ' .v
:i/
'/ '
.- -- --yl .
I
:. .......-
'I
:1
: I
'.I /CG24
I
\i
1000
800
r
0
600

........
400
200
......... ................ ........
-0.006 -0.004
0 0.0005
-0.002 Strain(r'?imlmm) 0.002
(a)
/
---- SG5a
-SG6b
- -SG71a
SG71b
-SG11a
0.001 0.002
(b)
0.004
1000
800
r
600 g
a.

400
200
Fig.4.4 Plots of Measured Strain in Specimen 6A: (a) Concrete Gauges; and (b)
Reinforcement Gauges
141
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CG2-7C
CG1-7C
1200
1000
800
z

600
0
....
400
200

-600 -500 -400 -300 -200
Strain(oo)
-100 0 100 200
Fig.4.5 Plots of Measured Strains from Concrete Surface Strain Gauges of Specimen
7B and 7C
4.2.6 Influence of Local Ductility in Test Structures
The influence of local ductility in test structures on the structural response,
crack development and propagation characteristics, and failure mode is presented in this
section. Based on the experimental results of specimen 6A and 7 A, Fig. 4.6 illustrates
the comparison of the crack development and propagation between specimens 6A and 7 A
at three different loading levels. By comparing the figure with the propagation of cracks
of specimen 6A in Fig. 4.3b, it is shown that the well distributed crack pattern which
developed in specimen 6A led to a greater stress redistribution at the higher loading steps
when the distributed reinforcement was provided. When distributed reinforcement was
not provided however, specimen 7A (Fig. 4.2c), a crack was able to propagate and this
led to a brittle mode of failure at the very low loading step (Fig. 4.lc). In addition, the
142
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
capacity of specimen 7A was observed to be only 39% of the nominal design load
calculated based on STM of the Appendix A of ACB 18-08 provisions and thus results in
the unsafe design as presented in Table 4.1. However, when the distributed reinforcement
was provided in the design (specimen 6A), the same STM shape and reinforcement
resulted in the conservative design where the measured capacity of the specimen was
much higher than the nominal design load based on STM.
Specimen 6a (1/Pn = 326 kN,
3
/
4
Pn = 488 kN, P" = 651 kN)
With Distributed Reinforcement
245 k ~ ~ /!
,/ l . . ... '
/rt ~ .. 2oo kN
......,..,..._./ _ _;
g t
.
.
. :
\ ____
1 ;
:
, I
'-----------------\
155 k,N t ';
I
L
________ __ _ ______ L ___ _t ____ i ___ _
Specimen 7a (1/
2
Pn = 326 kN,
3
/
4
pn = 488 kN, Pn = 651 kN)
(No Distributed Reinforcement)
Fig. 4.6 Observed Cracking in Specimen 6A (Provided Distributed Reinforcement)
and 7 A (No Distributed Reinforcement)
Thus, to achieve the conservative design of D-Regions by using STM, local
143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ductility and cracking control is needed; this can be provided by the use of the distributed
reinforcement.
4.3 D-Regions Designed for Reversed Static Loadings
In this section, the experimental results of the D-Regions designed for reversed
static loadings will be presented and adopted for the evaluating and validation of the
STM philosophy and associated ACI Code provisions. The D-Regions designed for
reversed static loadings are the specimens 8A to 8C, as presented in chapter 3. In this
section, only specimens 8A and 8C which were designed by the conventional STM shape
and orthogonally reinforced will be presented while the specimen 88 designed by the
STM shape selected from the topology optimization will be discussed in Chapter 5.
4.3.1 Load-Deformation Response of Test Specimens
The measured load-deformation responses of test specimens are presented in
Fig. 4.7. Since the experimental program was designed to mimic the load and boundary
conditions of the cantilever shear wall using the half side of the simply supported
structure, only the responses of the structure measured in the shear wall side will be
presented in this section. In addition, the load applied to the specimens measured by the
load cell seating on the top of the specimens was used to calculate the load acting directly
on the shear wall by a simple equilibrium as illustrated in Fig.4.8 and the measured
deformation presented in the figure is relative to the supports of the specimens, e.g., the
movement and slip between the bearing plate and structures are neglected from the
results.
144
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
450
400
350
z 300

-
't:I 250
ta
.3 200
150
100
1.
50
0
0
---
---- '
,. '
"/ ---- -
Meshonly
- -BA
-BC
....... -
2 4
Displacement (mm)
6
Fig. 4.7 Measured Load-Deformation Response of Specimen 8A and C.
8
From the Fig. 4.7, similar to D-Regions designed for non-reversed static loading,
it is also observed that the shape of STM and the pattern of tie reinforcement had a
variable level of influence on the post-cracking stiffness of the specimens up to the
service load levels, on the capacity of the specimens, and on the ductility of the
specimens during overloads. Both Specimen 8A and 8C were observed to have their post
cracking stiffness and capacity higher than the specimen in which only distributed
reinforcement as a minimum reinforcement.
After first crack, the post-cracking stiffness of specimen 8A was observed to be
much lower than the stiffness of specimen 8C. The capacity of those D-Regions was also
greatly different since specimen 8A was observed to reach its capacity just after the few
cracks observed and its capacity was much lower than the capacity of specimen 8C. This
145
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
is because specimen 8A was provided only distributed reinforcement in the shear wall
side while specimen 8C had been designed and provided the tension ties to resist the
tension demand up to much higher load levels. However, the observed ductility of
specimen 8C was significantly lower than the ductility of specimen 8A since the initial
failure of specimen 8A occurred due to yielding of the distributed reinforcement crossing
the crack and followed by fracturing of those reinforcement while 8C caused by
immediately crushing of the strut adjacent to the lower opening and thus the structure
completely collapsed in the more brittle manner.
LC)
N
...--
...--
Pn
M-- ---------------1400. ...............________ - i 9 ~ 0 ~ 0 ---I
250 265
t=175mm
300
13
0.395Pn 0.605Pn= Force Applied to the Shear Wall
Fig. 4.S Force Applied to the Shear Wall Side of Specimens SA to SC
4.3.2 Comparison of Measured and Calculated Strengths
The comparison of measured and calculated strengths of the specimen 8A and
8C is presented in Table 4.2. The measured strength of the specimens is obtained at the
peak point of the measured load-deformation responses that is illustrated in Fig.4.7. The
calculated strengths of the specimens were provided in Table 3.1.
146
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As shown in Table 4.2, the value of the ratio of the measured strength to the
nominal design capacity by STM is greater than unity. Thus, for reverse loading design
of the D-Region, the conservatism of the strut-and-tie provisions in Appendix A of
ACB 18-08 is also obtained in this experimental research program.
However, It is important to note that the complete validation of STM in design
of the more complex D-Regions subjected to reverse loadings or seismic actions is still
required the more extensive experimental programs using the instruments that utilize
more realistic seismic actions; for example, NEES@UIUC MUST-SIM Facility, and this
is not the goal of the current research program.
Table 4.2 Comparison of Calculated Nominal Capacities and Measured Strengths
Specimen PTes1(kN) Pn ac1(kN) PTestf P n acl
SA 299 50 5.9S
SC 755 737 1.02
Notes: Prest= measured specimen strength; Pn,aci =calculated model strength.
4.3.3 Comparison of Modes of Failure
The comparison of modes of failure between specimen 8A and 8C is presented in
Fig. 4.9. The specimen that was provided with only the minimum distributed
reinforcement is also included in the figure. The failure mode of both specimen
"Meshonly" and 8A occurred due to the sliding shear failure along the surfaces of the
crack that developed in the concrete regions between the two openings and the shear
failure of the concrete strut adjacent to the lower opening. The sliding shear failure was
observed to occur just after the fracture of the distributed reinforcement crossing the
crack. For the "Meshonly" specimen, after the upper part of the specimen slid along the
147
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
crack surface, the shear demand was induced so intensively at the concrete underneath
the lower opening and shapely caused the collapse of the structure.
(Meshonly)
(8C)
e (M honly and 8A)
C ack/slide (88 and C)
........__
rack/slide (8A)
Crush (88 and C)
honly)
(8A)
Fig.4.9 Mode of Failure for Specimens Meshonly, SA, and SC
148
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The same sliding failure mode was also observed in specimen SA; however, with the
existence of the bottom reinforcement in the specimen, the sliding mechanism was
retarded due to the contribution of the dowel action of the bottom reinforcement.
Although the STM model of the specimen SA was incapable of capturing the sliding
shear failure due to the limitation of STM technique, the STM model provides the useful
information to the designers; for example, yielding of the distributed reinforcement which
can be interpreted as a potential factor for the sliding mechanism along the crack in the
specimen. The failure mode of specimen SC was observed to occur by crushing of the
concrete strut adjacent to the lower opening. The failure mode was quite brittle and no
warning sign was detected. The failure mode suggested by STM of the model was also
similar by crushing of the prismatic strut spanned between nodes N35 and N36 as
predicted in Table 3.1.
4.3.4 Development of Cracking in Test Structures
The development of cracking in test structures is presented in this section. Figs
4.lOa, b, and c illustrates the state of measured cracking in specimens SA, C, and
meshonly, respectively, at five different load levels: 155 kN (35 kips), 222 kN (50 kips),
311 kN (70 kips), 533 kN (120 kips), and 711 kN (160 kips). For specimen SA, at the
load level of 155 kN (35 kips), the first crack development at the comer of the lower
opening was observed. Upon increasing the load to the level of 222 kN ( 50 kips), a new
crack formed at the comer of the upper opening and propagated downward in the
direction of the support. The exiting crack propagated from the lower opening upward to
the region underneath the upper opening. At this loading, a longitudinal crack along the
strut adjacent to the opening was initially observed. After applying the load up to the
149
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
level about 311 kN ( 70 kips), the existing crack propagated deeply to the region just
underneath the upper opening. The crack width of this large and capacity-limiting crack
was observed to be much wider due to yielding of the distributed reinforcement and
finally leads to the failure of the structure by sliding along the crack surface.
In Fig. 4.lOb presents the crack development and propagation in specimen 8C
was illustrated. At the load level of 222 kN ( 50 kips), the first crack occurred at the
comer of the lower opening. By increase the load to the level of 311 kN (70 kips), the
existing crack propagated from the comer of the lower opening upward. A new crack
formed at another comer of the lower opening and propagated downward to the support.
At the load level of 533 kN (120 kips), the diagonal cracks initially formed at the regions
between the openings, underneath the upper opening, and above the lower opening. The
longitudinal cracks along the strut adjacent to the lower opening were initiated. After
applying the load to the level of711 kN (160 kips), several diagonal cracks additionally
formed adjacent to the openings and propagated across the openings. The longitudinal
cracks of the strut adjacent to the lower wall were widened and were considered to reduce
the strut strength and this finally leads to the failure mechanism of the specimen.
By considering the level of the crack propagation and development in the
regions of the specimens, specimen 8C was thus observed to have much better structural
performance under service loads and higher cracking load than specimen 8A since fewer
large and capacity-limiting cracks were developed which allows more stress
redistribution in the structure and leads to the better structure performance.
150
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
p
(a) Specimen 8A (1/2Pn= 25 kN,3/4Pn= 37.5 kN, Pn=50 kN)
p
533
(b) Specimen BC (1/2Pn= 369 kN,3/4Pn= 553 kN, Pn= 737 kN)
p
(C) Meshonly
Fig. 4.10 Observed Cracking in Specimen 8A, 8C, and Meshonly.
4.4 Summary
In this chapter, a full description of the experimental validation of D-Regions
designed by the STM approach and the associated ACI design code provisions in
Appendix A was presented. The validation was undertaken by using the experimental
results from the experimental programs that was described in Chapter 3. Several
significant issues of STM in design of D-Regions were covered; for examples, the
151
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
structural performance of D-Regions under service loads and over loads, the
conservatism of STM in design of D-Regions with several truss members and
indeterminate load-resisting truss structures, the local ductility in D-Regions, the
characteristics of cracking and load transfer mechanism inside D-Regions. Both the
reverse and non-reverse load designs of D-Regions were also included in the
experimental programs to evaluate and validate the application of the STM.
Based on the experimental validation, the results indicate that: 1) the strength
calculated by the ACI STM Code provisions yielded conservative estimates of capacity
for D-Regions even when intentionally questionable truss shapes were selected in the
design; 2) the distributed reinforcement must be provided to ensure the sufficient local
ductility of D-Regions that prevents the premature and undesirable failure modes as
required in the plasticity theorem; 3) the load-deformation response under service levels
can be significantly influenced by the selected shape of the load-resisting truss; 4) A poor
selection for the shape of a STM can lead to unacceptable levels of cracks and damage
under service loads; and 5) the load transfer mechanism of D-Regions can be
significantly influenced by the selected shape of the load-resisting truss.
In the next chapter, the shape selection technique entitled "topology optimization"
will be introduced and discussed. The structural performance of D-Regions designed
based on the technique will be experimentally and computationally investigated to
provide the in-depth comprehension and details on how to select the STM shape leading
to the significant improvement of the structural performance of D-Regions, particularly,
under service loads.
152
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 5
Shape of STM in Design of Complex D-Regions
This chapter presents the experimental and computational validation of the STM
shape selection technique entitled "Topology Optimization". The introduction to the need
of the validation process of shape selection technique is presented in section 5.1. The
theory and application of topology optimization for design of D-Regions subjected to
non-reversed and reverse static loadings are presented in section 5.2. Experimental
validation of topology optimization for non-reversed and reversed static loadings is
discussed in section 5.3 and 5.4, respectively. Subsequently, the results from the
computational calibration and validation are presented in section 5.5. The structural
performance evaluation and comparisons are then given in section 5.6 and the summary
of both experimental and computational validation of topology optimization are given in
section 5.7.
5.1 Introduction
After the boundaries of the D-Region, support conditions and the applied loadings
have been determined, the next step is to select the general shape of the STM to use in
design. The flexibility in freely selecting the STM shape in the design process can create
difficulties for designers, particularly when encountering D-Regions with very complex
geometries, loadings, and boundary conditions. An inadequate topology, or shape, for the
STM could result in ineffective structural performance during service states, for
examples, uncontrollable crack width and propagation, low cracking load, and low
capacity at the ultimate states (Kuchrna, Yindeesuk et al, 2008 and Tyler Ley et al, 2007).
153
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As presented in Chapter 2, there are several techniques used to select STM shape.
These include previous experience, empirical models, crack patterns, linear or nonlinear
stress trajectory, and topology optimization entitled "Performance Based Topology
Optimization (PBO)". Amongst the above STM shape selection techniques, PBO
technique is the most comprehensive. It can be readily incorporated in the STM design
procedure with the use of computer software in both design and analysis ofD-Regions
e.g. Tjhin and Kuchma (2002).
Although several numerical examples of STM shape selection by PBO, as
demonstrated by Liang et al (2002), were successfully illustrated including those for deep
beams with openings, deep beams with bottom pointed loads, corbels, and stocky shear
walls, there is no further experimental or analytical verification of the structural
performance in both service and ultimate states on those D-Regions. To confirm the
validity of the optimization technique that may be adopted in the STM design procedure
or framework to generate STM shape for effective structural performance design of D-
Regions, a series of both experimental and computational investigations were conducted
as reported in this chapter.
5.2 Application of Topology Optimization for Shape Selection of STM
In this section, the theory and application of topology optimization for shape
selection of STM are presented. The section will firstly present the optimization
technique for D-Regions subjected to non-reversed static loadings; this is followed by a
presentation of the technique for reversed static loadings.
154
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.2.1 D-Regions Subjected to Non-Reversed Static Loadings
An introduction to topology optimization was provided in the literature review in
Chapter 2. In this current section, the theory is more fully described and the author's
MATLAB implementation of selected topology optimization approach is presented.
In topology optimization methodology, as stated in Eq (5-1), the approach is to
seek the most efficient material distribution over a design domain that achieves the
required objective function (e.g. minimizing structural weight and satisfying design
constraints such as stress limits). Several topology optimization techniques have been
developed in recent years including the homogenization method (Hassani and Hinton
1999; Bendsoe and Sigmund 2003), the evolutionary structural optimization-ESQ
approach (Xie and Steven 1997), rule-based optimization (Seireg and Rodriquez 1997),
and soft-kill optimization (Mattheck 1998).
n
Minimize W = Lwi(t) (5-1)
l=l
Subject to Ci C* ( i=J, ... ,n)
where Wis the total weight of a structure, wi is the weight of the i
1
h element, tis the
thickness of elements, C is the value of the design parameters in the i
1
h element, C* is the
prescribed limit (e.g. maximum stress) of C, and n is the total number of elements in the
discrete design domain.
The performance-based topology optimization method (PBO) proposed by Liang
et al (2002) is based on evolutionary structural optimization (ESO). The PBO concept is
simple and robust and involves the gradual removal of underutilized elements from the
design domain. Once several iterations have been achieved, the optimal topology design
will be obtained. In PBO, to monitor when the optimal topology was achieved, the
155
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
performance index (Pl) approach was introduced. A simplified flow chart for the PBO
technique is presented in Fig.5.1.
The formulation of performance index (Pl) and the criteria used to decide which
elements to remove depend on the constraints and element removal parameters used in
the PBO technique. There are several possible constraints and element removal
parameters adopted in PBO such as mean compliance or strain energy density of an
element, Von Mises stress in an element, and virtual strain energy of an element. In this
research, mean compliance and element removal parameters based on mean compliance
will be adopted due to its simplicity and reliability.
Start
Model the initial structure
Perform FEA
Evaluate the Performance Index
Consider the underutilized elements
Eliminate smal number elements
No
Save the model and obtain the
most optimum at highest PI
End
Fig.5.1 Flowchart of PBO Technique for Non-Reversed Loadings
156
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
PBO for Structures with Mean Compliance Constraints
The topology optimization problem for a continuum structure subjected to the
single mean compliance constraint can be stated as follow:
n
Minimize W = L w, (t) (5-2)
1=1
Subject to C :'.S C *
where Wis the total weight of a structure, Wi is the weight of the i
1
h element, tis the
thickness of elements C is the constrained strain energy, C* is the prescribed limit of C,
and n is the total number of elements in the design.
The element removal criteria are determined by performing the sensitivity
analysis on the mean compliance with respect to element removal as demonstrated by
Liang and Steven (2002). The element strain energy is a measure of element contribution
to the overall stiffness of a structure, and is denoted as
(5-3)
in which {ue} is the displacement vector of the e
1
h element and [ke] is the stiffness matrix
of the e
1
h element. The strain energy density of an element is defined as ve =Ice I !we ,
where we is weight of an element.
To obtain the maximui stiffness and minimum weight design, the elements with
the lowest strain energy should be eliminated from the structure. Since the strain energy
of elements is approximately evaluated by neglecting the higher order terms in the
sensitivity analysis, only a small number of elements with the lowest strain energy are
allowed to be removed from the structure in each iteration. The element elimination ratio
(EER), or the maximum percentage of the underutilized elements to be removed in each
157
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
iteration, of 1-2 % can be used to obtain an accurate solution at an acceptable
computational cost.
The performance index used in mean compliance constraint is adopted that use
energy-based performance indices proposed by Liang and Steven (2002) for evaluating
the performance of structures and to obtain the optimal topology.
By adopting the scaling design concept, PI is defined as
PI = CoWo
C;W;
where Co and C; are the strain energy of the structure at the initial iteration and ;th
(5-4)
iteration respectively, and W
0
and W; represent the total weight of the structure at the first
and ;th iteration, respectively.
To prevent numerical instability, such as checker board pattern shown in Fig. 5.2,
a technique using the average values of element strain energy in the neighboring elements
proposed by Youn and Park (1997) is adopted in this proposal. The step-by-step of the
technique is described as follows:
158
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a) Solution with checkerboard pattern"
(b) Solution without checkerboard
Fig. 5.2 An Example of Checker Board Pattern (Li et al, 2001)
Calculate the nodal strain energy (Eqn.5-5) of an element based on
averaging the strain energy of neighboring elements
} M
nd = M re
(5-5)
where Snd is the nodal strain energy , M is the total number of elements
that connect to the node, and Ye is strain energy (Eqn.5-3) of the e th
element.
Calculate the average strain energy (Eqn.5-6) of the element using the
corresponding nodal strain energy (Eqn.5-5).
qe
1 Q
Q


(5-6)
where Se is the average strain energy of the element and Q is the total
number of nodes in the element.
159
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Implementation of PBO Techniques
In this research, MATLAB was adopted as a programming language to implement
the PBO technique. The implementation was made using linearly elastic FEA using CST
elements. The MATLAB code was written to allow using the input node and element
data from Patran, a general pre-processing tool for FE that provides automatic mesh
generation. A brief description of the automatic mesh generation available in Patran will
be given in Appendix B. After inputting nodal positions and element identifiers, material,
loading, boundary condition data, and PBO parameters, the MATLAB code performs
topology optimization iteratively until converging to the reduced structure. During this
optimization process, the code generates a simple text file including the node, element,
and filled-voided element data. This file can be used to display the evolving topology
using FEPLOT, which is a general post processing tool for FEA that can import a simple
format text file from MATLAB. The program generates a PI history file at the end of
optimization loop to allow designers to choose the most optimal topology (the one with
highest PI).
Numerical Examples for Topology Optimization
In this section, numerical examples will be used to illustrate the efficiency of the
code and validity of the employed PBO techniques in using topology optimization for
selecting STM shape for the design of D-Regions.
160
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Michell Simply Supported Structure
A continuum domain with boundary condition and a bottom point loading at
midspan, as illustrated in Fig.5.3, was used for topology optimization of this Michell
simply supported structure. The domain was discretized into 4800 CST elements. The
element removal ratio (ERR) of 1 % was adopted throughout the optimization process.
In this example, the PBO technique with mean compliant constraints was adopted
for topology optimization. Figs. 5.4a to d illustrate the evolving topology of the Michell
simply supported structure. The optimal topology is given in Fig.5.4d. The performance
index (Pl) (Fig. 5.5) corresponding to the optimal structure was 1.46. The material used
in the optimal structure was thus more effective than the initial structure.
Fig.5.3 Design Domain for the Michell Simply Supported Structure
161
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a) (b)
(c) (d)
Fig.5.4 Results in the Selected Iteration of Topology Optimization: (a) Topology at
5th Iteration; (b) Topology at 35th Iteration; (c) Topology at soh Iteration; and (d)
Topology at 57th Iteration.
1.6
1.4
1.2
><
Cl>
'O
.5 0.8
I-Pl History I
c:
0.6
0.4
0.2
0
0 20 40 60 80 100
Iteration No.
Fig.5.5 Performance Index History of Michell Simply Support Structure
D-Regions in Fig.3.la to c were designed by STMs whose shapes were selected
from the optimal topology. The results of the optimal topology for the selected complex
162
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
regions, illustrated in Fig.5.6, represent the idealized truss or shape of STM that transfers
the load from the loading point to the support. By following the optimal topology, STM
of the selected complex regions can be obtained as illustrated in Figs. 3.2e, 3.3d, 3.Sb,
and 3.Sc which correspond to models 4E, SD, 7B, and 7C, respectively. All STMs were
constructed by using the program CAST (Computer Aided Strut-and-Tie). The STM
shape as illustrated in Fig.3.2e is composed of two compression struts and one tension tie
to carry the load from the top part of the deep beam to the bottom part below the opening
where two tension ties are diagonally oriented to equilibrate the forces meeting at the
nodal zones. The optimal STM shape for the propped cantilever deep beam with opening
combines the two systems of truss: 1) three compression struts and one straight tension
tie are to transfer the load to the pin support where a diagonal tie is required to help lift
the force up to satisfy force equilibrium 2) the truss system, distributing forces around the
opening to the cantilever support, is optimized to be composed of compression struts and
diagonal ties as illustrated in Fig.5.6b. This part of the optimized truss is unable to
automatically satisfy the equilibrium and thus the alternative truss system (Fig.3.3d) is
adopted by using a straight strut and tie at the top and bottom of the deep beam to
transmit the compressive and tensile stress at the cantilever side, respectively. In addition,
both diagonal struts and ties are placed to equilibrate the forces distributed around the
opening. The optimal STM shape, illustrated in Fig.5.6c, is composed of two truss
systems to transfer the load to the <lapped end and around the opening to the pinned
support. The portion of the load transferred to the <lapped end is by two compression
struts and each of straight and diagonal tension tie. The diagonal tie is placed to lift the
force from the bottom of the deep beam up to the <lapped end where the set of
163
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
compression struts are formed to satisfy force equilibrium. Unlike in the previous case,
the optimal topology of the force around the opening will automatically satisfy the
equilibrium. Two compression struts are formed to distribute the force around the
opening while the two diagonal ties placing above and below opening helps to resist the
induced bursting tension force for stability and equilibrium of the system. Since the
diagonal tension tie in the optimized STM shape might be unpractical in the field, the
adjusted STM shape based on the optimized one is thus adopted in Fig. 3.Sc. The
adjusted shape uses tension tie orthogonally adjacent the opening. In addition, the force is
selected to flow above the opening by the diagonal struts and the orthogonal ties while
only a longitudinal tension tie is placed under the opening. This was done to investigate
how the behavior in this D-Region designed by using the orthogonal ties is different from
the D-Region designed by using the optimized shape of ties.
164
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.8
Performance Index History
1.6
1.4
1.2
ii::
0.8
l-4PBOI
0.6
0.4
0.2
o
o 20 40 60 80
(a)
Iteration No
1.6
Performance Index HI story
1.4

1.2
ii::0.8 1-sPBOI
0.6
0.4
0.2
o
0 20 40 60 80
(b)
Iteration No
1.6
Performance Index History
1.4

1.2
li:0.8
0.6
1-rnl
0.4
0.2
o
(c)
o 20 40 60 80
Iteration No
Fig.5.6 - Optimal Topology and Performance Index History generated from
Performance Based Topology Optimization (PBO) Technique: (a) D-Region in
Fig.3.la; (b) D-Region in Fig.3.lb; and (c) D-Region in Fig.3.lc.
5.2.2 D-Regions Subjected to Reversed Static Loadings
The conventional design approach, as used by Tjhin and Kuchma (2002), for
100
100
100
application of the STM for the design of D-Regions subjected to reversed static loadings
(Fig.5.7) is as follows: 1) Determine the geometry, boundary condition, and loadings of
D-Regions; 2) Select the different STM shapes for different load cases, for examples,
reversed static loadings on left and right of the shear wall in Fig. 5.8; 3) Design D-
165
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Regions by using the designated STM shape for each load case; and 4) Superimpose the
most dominated reinforcement from all load cases in the final design.
0.30H
0.1 SH
1.165H
0/IH
0.275H
0.30H
H=965 mm
thickness=1 mm
0.275H
0.1H
Fig. 5.7 Shear Wall with Openings subjected to Reversed Static Loadings.
Based on the 2nd step of the conventional approach, the designers need to
perform the STM shape selection processes for each load cases. This can greatly increase
the difficulty in using the STM for the design of D-Regions, particularly, for very
complex geometry D-Regions subjected to very complex loadings and boundary
conditions. As stated in the work by Liang (2007), the single objective and multi
constraints topology optimization can be used to obtain the unique optimal STM shape
for multi load cases design of D-Regions. This thus significantly reduces the complexity
and cost of the design process.
166
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
'
'
'
'
'
D
'
'
Conventional approach
'
'
'
'
'
'
'
'
'
D
I
Load C ~ a s e
Fig.5.8 The Conventional Approach for Design of Complex Regions subjected to
Multiple Load Cases (Tjhin and Kuchma, 2002)
The formulation of PBO for STM shape selection for D-Regions subjected to
reversed static loadings or multi load cases can be achieved by extending the single
objective function and single constraint PBO (Eqn. 5.2) to the single objective function
and multiple constraints PBO as proposed by Liang (2007).
The single objective function and multiple constraints topology optimization
problem can be stated as:
n
W = LWe(t)
Minimize
1
=1 (5-7)
Subject to Cj ::=; Cj * (j=l, ... ,m)
where Wis the total weight of a structure, We is the weight of the e
1
h element, tis the
thickness of elements, Cj is the j1h constrained strain energy, Cj* is the prescribed limit of
Cj, m is the total number of constraints, and n is the total number of elements in the
design.
To solve Eqn (5-7), Liang (2007) proposed a new two-level control scheme and
this has been included in the PBO algorithm. In the two-level control scheme, the first
167
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
level control is in the element removal step in which the element removal task is
proposed to begin by calculating the strain energy densities of elements in the design
domain. Subsequently, to decide the elements to be removed in each iteration, a logical
AND condition is employed to account the effects of the multi constraints and load cases.
In the Logical AND condition, an element is eliminated from the design domain only if
its strain energy density is the lowest for all load cases (Liang, 20007). This condition
thus leads to the stiffest structure with respect to the worst loading condition. For the
second level control, the performance index for monitoring the optimal solution of the
topology is proposed to be calculated from the most critical load cases. By using this two
level control scheme, the optimal topology results in the overall stiffness performance
and minimum weight design with respect to all load cases.
By adding the new two-level control scheme in the PBO algorithm, a simplified
flow chart for the PBO technique is now obtained as shown in Fig.5.9. The flow chart
was implemented by extending the MATLAB program for non-reversed static loadings to
reversed static loadings by incorporating the two level control scheme which is straight-
forward and simple for the implementation.
168
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Start
Model the initial structure
Perform FEA
For all load cases
Evaluate the Performance Index (Pli)
For all load cases (i=1 to N)
Consider the underutilized elements
that are critical in all load cases
Eliminate a small number elements
Save the model and obtain the
most optimum at highest Pli
End
Fig.5.9 Flowchart of PBO Technique for Reversed Loadings
The Optimal Shape of STM and Design Using Computer-Aided Strut-tie (CAST)
for Reversed Static Loadings
Due to the simplicity of the comparison with the experimental program, the
shear wall in Fig.5.8 subjected to the reversed static loadings was rotated clockwise in the
configuration similar to the D-Regions used in the testing program as was shown in Fig.
3.ld. Subsequently, the shear wall side of the D-Region in Fig. 3.ld was designed by
STMs whose shapes were selected from the optimal topology generated from the flow
chart of the single objective function and multi constraints topology optimization in Fig.
5.9.
169
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.


(a)
I
I
,, '
0 ~ ~
,,,,"" D
(c)
1.6
1.4
1.2
0:
0.8
0.6
0.4
0.2
0
0 10 20 30
1terat1o'WN0. so
60 70
(b)
Fig.5.10-0ptimal Topology, STM, and Performance Index History of the Shear
Wall Side of D-Regions in Fig. 3.ld: (a) Optimal Topology; (b) Performance Index
History; (c) STM generated from CAST for Load Case 1; and (d) STM generated
from CAST for Load Case 2.
The optimal truss results in the unique solution that can be adopted to transfer the
reversed loads represented by the left (load case I) and right (load case 2) applied loads
to the shear wall as illustrated in Fig. 5.8. The designer is more convenient to design D-
Regions by simply constructing the STM model of the same shape for all load cases and
subsequently applies the loads associated to the load case; for example, Figs. 5.lOc and d
170
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
illustrate the STM models for load cases 1 and 2 that are subjected to the load applied to
the left and right side of the shear wall, respectively.
As illustrated in Fig.5.lOc, the truss model to transfer the load in case 1 consists
of straight and diagonal tension ties connecting nodes N38 and N46 and nodes N37 to
N45, respectively, to resist the tension demand at the base of the shear wall (the top left
side of the specimen in the testing program). A prismatic strut connecting nodes N46 and
N45 was used to transfer the load from the loading point of the shear wall to the regions
adjacent to the opening where a diagonal tension tie connecting nodes N37 and N45 was
placed to equilibrate the force that flows downward to the base (the bottom left side of
the specimen in the testing program) of the shear wall by the bottle-shaped strut. For load
case 2, a prismatic strut connecting nodes N35 and N36 was used to transfer the applied
load from the loading point (the bottom right of the specimen in the testing program) to
the regions adjacent to the openings where a diagonal tie connecting nodes N39 and N45
was placed to resist the tension demand that occurs when the force flow downward to the
base (the right top of the specimen in the testing program) of the wall by the bottle
shaped strut connecting nodes N37 and N45.
In the actual design process for D-Regions subjected to reversed static loads, the
reinforcement can be obtained by superimposing the required reinforcement from the
most critical load cases. However, for the limited numbers of the specimens used in the
experimental program, only the most complicated case; for example, shear walls
subjected to the load case 2 (applied load to the right side of the shear wall) will be
considered. Thus, only the reinforcement from this load case was used in constructing the
specimens in the experimental program as illustrated in Fig. 3.8.
171
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In the next section, the experimental validation of those D-Regions designed
using the truss shapes based on the optimization scheme will be presented and compared
their structural performance during service and over loads with other D-Regions designed
using different truss shapes.
5.3 Experimental Validation of Topology Optimization for Shape Selection of
STM for D-Regions Subjected to Non-Reversed Static Loadings
In this section, the experimental validation of topology optimization for shape
selection of STM for D-Regions subjected to non-reversed static loadings will be
presented. Due to the limited experimentation, only specimens 7B and C in Fig.3.13 were
constructed and experimentally validated. Specimens 6A to 7 A were designed using
different STM shapes while the specimens 4E and SD were computationally validated by
nonlinear FEA tools which are presented in section 5.5.
5.3.1 Comparison of Load-Deformation Responses
The measured load deformation responses of the complex regions are illustrated
in Fig.5.11. The plot shows the influence of STM shapes on the structural performance
during the service state, the strength, and the observed ductility. From the plot, the
complex region (7B) designed using STM shape generated from PBO technique are
stiffest and has the highest strength. This is due to the provided reinforcement in the
optimal STM shape and orientation that can effectively limit numbers of crack developed
and propagated, retard the degradation of the structural stiffness, and allow the maximum
possible strength. In addition, since the specimens used in the comparison had different
amounts of reinforcement, the comparison between the amount of provided
172
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reinforcement (i.e. ratio of strength to reinforcement volume) in each specimen is thus
necessary and is presented in Section 5.6.
1400
__ ..
1200
1000
-z
800 ..::.::
-
"C
cu
600
0
...J
400
200
0
0 2
-78
- - - 7A
---a-7C
-68

---6A
-+-60
-MeshOnl

4 6 8 10 12
Displacement (mm)
Fig.5.11 Measured Load-Deformation Response of the Test Complex Regions
(Specimen 6A to 7C), Note: 1) 1in.=25.4 mm. 2) 0.2248 kips= 1 kN
5.3.2 Comparison of Modes of Failure
14
Fig. 5.12 described the failure state of the specimens 7B and 7C. The capacity of
specimen 7B was reached when diagonal strut connecting nodes N2 and N40 (Fig.3.5b)
began to crush. Subsequently, the load-carrying capacity of the structure strut was
significantly exhausted and thus leads to the brittle failure of the structure. The failure
mode suggested by STM of the model was also similar by crushing of the same strut as
described in Table 3 .1. The failure mode of specimen 7C was observed to occur due to
the sliding shear mechanism along the existing crack underneath the opening. This failure
mechanism is abrupt and different from the suggested failure mode by STM of the model
since the failure mode was suggested to occur due to simultaneously yielding of the
173
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
tension tie connecting nodes 25 and 34 and crushing of the strut connecting nodes 2 and
32. The significant difference in the predicted failure mode of the specimen from STM
and the observed failure mode from the experiment is; however; not surprising since the
STM model does not account for the nonlinear characteristics in the material and only
compression and tension responses of STM components can be considered in the STM.
(7b) (7c)
Fig. 5.12 -Modes of Failure for Models 7 A to 7C
5.3.3 Comparison of Crack Development and Propagation
The effectiveness of the reinforcement orientated in the direction of optimal STM
shape in limiting the numbers of crack and controlling the crack propagation is illustrated
in Fig.5.13. Fig.5.13 presents the number of crack and crack propagation in the complex
regions designed with the various STM shapes under the three selected loading steps of
222, 356, and 534 kN (49.9, 80.3, and 120 kips). At the loading step of 222 kN (49.9
174
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
kips), while the other specimens encounter with the several cracks, the minimal numbers
of crack were observed in specimen 7B and 7C and at only the locations where the
maximum principal tensile stress occurs e.g. the comer of the opening and the <lapped
end. When the load reached 534 kN, cracks were observed in other specimens to develop
more and propagate closely to the mechanism observed at the ultimate state; however,
specimen 7B and 7C are moderately cracked, well distributed, and no sign of crucial
damages.
,.::-\
IP
~
; ... .;,
\I
222 kN
"'....,
356 kN
'\\
\,
245 kN
.. ~ : o ~ kN
/
'
\
155 k,N
I l
\
__ ,
*-"" ~
(a) Specimen 6a ('/,P, = 326 kN, '/,P, = 488 kN, P
0
= 651 kN) (d) Specimen 7a ('/,P., = 326 kN, '/,P
0
= 488 kN, P, = 651 kN)
534 kN
'\. .... '
. \ \ '\\':<:" .... 356 kN
; \._ . \\ ', I
'! ~ ~ k N \
-,,-
, 534 kN
/; '
. ;356 kN ,
222 kN
\
~ \
(b) Specimen 6b ('/,P, = 312 kN, '/,P, = 468 kN. P, = 624 kN) (e) Specimen 7b ('/,P,, = 580 kN, 'IP,= 870 kN, P, ~ ... 1160kN)
....
' ,'- !)34 kN
\'-'
534 kN.
356 kN';' .
\ 222 kN
'\
:
':
;: :
__ I
', \
(c) Specimen 6c ('/,P,, = 430 kN, '/,P., = 645 kN, P, = 860 kN) (f) Specimen 7c ('/,P
0
=365 kN, 'l.P, = 547.5 kN, P, =730kN)
Fig.5.13 -Observed Cracking in Specimen 6A to 7C tested at P=222, 356, and 534
kN (49.9, 80.3, and 120 kips)
175
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.3.4 Influence of Orientation of Tension Tie on Structural Responses
The author further examines the load transfer characteristics above and below the
opening of the specimen 7B and C (Fig. 5.14 and 5.15) to study the effectiveness of the
optimal and orthogonal orientation of reinforcement. In specimen 7B, the diagonal
reinforcing bars (strain gage #17, 7, 8, and 9) above and below the opening can pick up
and then transfer the tensile forces as effectively as the orthogonal bars in specimen 7C
(gage #10.1, 8.3, 11.1). However, orthogonal bars above the opening in specimen 7C
were unable to effectively control the diagonal cracks (Fig. 5.13t) developed due to the
transverse tensile stress and thus degraded the stiffness of the compression strut (#8c) to
transfer the compressive forces. This thus leads to the redistribution of the compression
stress slightly to the regions above and intensively to the regions below the opening. Due
to this intensive stress and no reinforcing bars under the opening, the shear compression
and slip failure eventually occurred in the specimen 7C. In contrast, the diagonal
reinforcing bars in specimen 7B were observed to effectively control the diagonal cracks
and thus allow the compression strut to pick up more loads (#8c, 9c, and 1 Oc ). This thus
confirmed the effectiveness of the load transferability in the optimal orientation over the
orthogonal orientation. In addition, when the orthogonal bars are unavoidably required
due to the practical issues, the structural behavior should be accurately predicted in order
to effectively and safely design the complex regions throughout limit states.
176
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Strain Measurement for Load Transfer Mechanism in The Right Side of Specimen 7B
1200
j -+-14
--17
-18
;-7
..... 8
-- 9
. -x-8c
--e--9c
-10c
1000
800 -
z

600 :;

0
400 .J
200

-1000 -500
O Strain(ue) 500
1000 1500
Fig.5.14 -Measured Load-Strain in Surface Concrete and Reinforcement around
the Opening of Model 7B (Note: 0.2248 kips= 1 kN)
Strain Measurement for Load Transfer Mechanism in The Right
Side of Specimen 7C
-
--....____ ,_.!

-10.1
-- 11.1
- - - - 8.1
- - -8.2
--- 8.3
-8.4
--Be
-9c
-1oc
-1500 -500 500 1500 2500 3500
Strain (ue)
1200
1000
800
,,.....,

....._,
"O
ro
400.3
200
Fig.5.15 -Measured Load-Strain in Surface Concrete and Reinforcement around
the Opening of Model 7C
177
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.4 Experimental Validation of Topology Optimization for Shape Selection of
STM for D-Regions Subjected to Reversed Static Loadings
In this section, the experimental validation of topology optimization for shape
selection of STM for D-Regions subjected to reversed static loadings will be presented by
considering the experimental results for specimen 8B as were introduced in the section
4.3.
5.4.1 Comparison of Load-Deformation Responses
The measured load-deformation responses of test specimens are presented in
Fig. 5.16.
600
500
400
-
z

-
"C
300
cu
0
...J
200
100
... ----
0
0
...... -- .__..,,,,,.. ..... ,
,- '
'
- Meshonly
- -BA
-88
-BC
....._ __ .....
-...... _
2
- . -- .. -- .. - ...
4
Displacement (mm)
6
Fig. 5.16 Measured Load-Deformation Response of Specimens SA to C.
8
From Fig. 5.16, by comparing the structural responses of specimen 8B with the
responses of specimen 8A, 8C, and "meshonly", specimen 8B was observed to have the
178
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
stiffest post cracking stiffness and the highest capacity. After cracking initiated, the post-
cracking stiffness of specimen 8A and "meshonly" was observed to be much lower than
the stiffness of specimen 8B. The post cracking stiffness of specimen 8C was slightly
different from the stiffness of specimen 8B up to the approximate load level of 200 kN (
25% of the structural capacity). For the higher load level, the stiffness of specimen 8B
was measured to be much higher than that of specimen 8C. This thus indicates more
damages in specimen 8C although the capacity of specimen 8C was observed to be
slightly lower than that of specimen 8B. In addition, both specimens 8B and 8C were
observed to have lower ductility than specimens 8A and the "meshonly" specimen.
5.4.2 Comparison of Modes of Failure
The modes of failure of specimens 8A to 8C are presented in Fig. 5.17. Similar to
specimen 8C, the failure mode of specimen 8B was also observed to be the same brittle
failure mode by crushing of the concrete strut adjacent to the lower opening. This type of
brittle failure modes observed in both specimen 8B and 8C was additionally evident by
the low ductility as was illustrated in Fig. 5.16. The failure mode suggested by STM of
the model was also similar, by crushing of the prismatic strut that spanned between nodes
N35 and N36 as described in Table 3.1.
179
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
e (M honly and 8A)
C ack/slide (88 and C)
rack/slide (8A)
~ ~ ~ ~ Crush (88 and C)
honly)
(Meshonly) (8A)
(88) (8C)
Fig.5.17 Mode of Failure for Specimens Meshonly, 8A, 8B, and 8C
180
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.4.3 Comparison of Crack Development and Propagation
To demonstrate the effectiveness of the reinforcement pattern obtained from the
optimal STM shape in controlling the crack development and propagation, the
comparison of crack development between specimen SB and other test structures is
presented in this section. Figs 5.18a, b, c, and d illustrates the state of measured cracks of
specimens SA, B, C, and "meshonly", respectively, at five different load levels: 155 kN
(35 kips), 222 kN (50 kips), 311 kN (70 kips), 533 kN (120 kips), and 711 kN (160 kips).
At the loading step of 222 kN (50 kips), while several cracks were observed in the other
specimens, no cracking was observed in specimen SB until the load step of 267 kN ( 60
kips). This thus demonstrates that using the optimal reinforcement pattern could lead to
the structure exhibiting a higher cracking load. In addition, this significant observation is
thus to confirm that cracking is strongly influenced by the pattern of selected
reinforcement. At the load step of 311 kN (70 kips), the minimal numbers of crack were
observed in specimen SB at the comer of the opening which is the location where the
maximum principal tensile stress occurs; however, both specimen SA and "meshonly"
were observed to form the mechanism. When the load reached 533 kN (120 kips), cracks
in specimen SB were observed to develop and propagate slightly from the comer of both
openings while the specimen SC was observed to have cracks develop more severely. At
the load level of71 l kN (160 kN), specimen SC was observed to have cracks propagate
so to create the capacity limiting mechanism observed at the ultimate state while just few
cracks in specimen SB were observed to develop and propagate near the corner without
any significant structural damage and degradation.
lSl
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
p
{a) Specimen BA {1/2Pn= 25 kN,3/4Pn= 37.5 kN, Pn=50 kN)
p
(b) Specimen BB (1/2Pn= 370 kN,3/4Pn= 555 kN, Pn= 740 kN)
p
{c) Specimen BC (1/2Pn= 369 kN,3/4Pn= 553 kN, Pn= 737 kN)
p
(d) Meshonly
Fig. 5.lS Observed Cracking in Specimen SA, SC, and Meshonly.
182
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.5 Computational Validation of Topology Optimization for Shape Selection of
STM for D-Regions
In this section, the computational validation of topology optimization for shape
selection of STM for D-Regions will be presented. Since there was no experimental data
for specimens 4E and SD (designed based on the optimal STM shapes selected from
topology optimization), the calibrated computational models of specimens 4E and SD are
adopted to predict the structural responses of those specimens which are subsequently
used to compare with the structural responses of specimens 4A to 4D and SA to SC that
are also predicted from the same calibrated computational models. Section S.5.1 will
provide comprehensive details of computational models of D-Regions produced by the
computational tool" Vector2". The details of pre-processing tool so called "Patran" and
its powerful mesh generation adopted in generating the input files for "Vector2" are
presented in Appendix B. The calibration of computational models will be given in
section 5.S.2.
5.5.1 Computational Models of D-Regions by Vector2
In the computational validation of topology optimization, the computational
models of D-Regions were produced by the commercial software program "Vector2".
Vector2 was developed by Vecchio's Vector Analysis Group and enables nonlinear
analysis based on Modified Compression Field Theory (MCFT) and Disturbed Stress
Field Model (DSFM) as proposed by Vecchio and Collins (1986) and Vecchio (2000),
respectively. The nonlinear models adopted in Vector2 account for the smeared, or
average, stress-strain relationships in cracked concrete and reinforcement. The models
also incorporate an equilibrium check at crack locations to ensure the ability in
183
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
transferring the shear and tensile stresses across the crack that limits the capacity of the
two dimensional continuum of concrete structures. Several material constitutive models
can also be included in the analysis, including compression softening, tension stiffening,
and tension softening models. For discretization of the structure domain, Vector2 allow
for the use of the combinations of low order elements including the constant strain
triangle and four nodes quadrilateral elements for reinforced concrete domain and the
constant strain truss element for the reinforcement domain. The use of truss element to
model reinforcement with complex orientation often found in D-Regions designed by
STM leads to the complexity of modeling process. Thus, the author adopts the mesh
generation capability from the pre-processing tool called "Patran" to generate the finite
element mesh corresponding to the orientation of the reinforcement and allow the
connection of the truss elements to the concrete mesh at the true locations of the
reinforcement pattern. The comprehensive details of the mesh generation used in Patran
are presented in Appendix B.
The comprehensive details of computational models for specimens 4A to 6D are
illustrated in Fig. 5.19. The computational models of all specimens were produced for a
whole specimen due to no plane of symmetry in those specimens. Smeared reinforcement
was adopted in all models to model the distributed reinforcement. The truss element was
adopted to model all main reinforcement in all specimens associated with the details
illustrated in Fig. 3.8. The selection of the most appropriate material models are
completed by the model calibration process. The details of the model calibration are
presented in section 5.5.2. The material properties assigned in all models were adopted
from Table 3.2 and 3.3.
184
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4C 40
4E
Fig. 5.19 (continued on next page)
185
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5A
5C
Fig. 5.19 (continued on next page)
186
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
. '
6A
68
--l
..
-. ' ' '. :
<' , '
a
"
6C
PBO
Fig. 5.19 FE Models of Specimen 4A to 7C.
5.5.2 Calibration of Computational Models
Vector2 belongs to a class of FEA tools for structural concrete in which relatively
limited options are provided to the user for making selection that can dramatically affect
the predictions of the program. The developers of Vector2 consider that the expertise
needed for obtaining accurate predictions should belong to the developers. Nevertheless,
there are three significant choices that the user of Vector2 needs to make that are referred
to as "Model Calibration" in this study. These are: 1) selection of mesh size; 2) selection
187
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of nonlinear FEA framework (MCFT or DSFM); and 3) selection and calibration of
material behavior models.
As well demonstrated in the literature (Dodds et al ( 1984 ), Bazant ( 197 6), and
Bazant and Cedolin (1979)), mesh refinement affects the accuracy of nonlinear FEA of
reinforced concrete adopting smeared (either rotating or fixed) cracked models which are
similar to the MCFT and DSFM. Selecting the appropriate level of mesh refinement can
reduce the problem of mesh dependency recognized in the smeared crack model as
pointed out by several researchers e.g. Dodds et al (1984), Bazant (1997), and Cervenka
and Pukl (1995).
From the calibration process, the mesh refinement resulting in the most accurate
predicted strength, crack pattern, and failure mode is about 2.67 and 12 times the
maximum aggregate size (3/8" or 9.5 mm) for the specimens 4A to 5C and the specimen
6A to 7C, respectively. Fig. 5.20 illustrates an example of the mesh dependent effect in
smeared crack model of nonlinear FE for the specimen 6C. The DSFM (included explicit
cracked shear stress-slip components) with Popovic-NSC ( 1973) for concrete pre-peak
model, Vecchio 1992-A (1993) for compression softening model, Bentz (1999) for
tension stiffening model, bilinear model for tension softening model, and elastic-plastic
with strain hardening in smeared and discrete reinforcement model was finally selected in
the analysis. Fig. 5.21 illustrates the overall accuracy of the calibrated FEA models
adopted to predict the structural responses of the specimens 4A to 7C. The calibrated
FEA resulted in inaccuracy (capacity overestimation) when it was adopted to predict the
specimen with no distributed reinforcement (7 A). This is due to the inaccuracy of
smeared crack model in predicting the structural responses of the structures subjected to
188
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
large crack width where the more accuracy in structural prediction can be obtained by
using the discrete crack model.
Load-displacement comparison for different mesh refinement
1400
1200 Strength (Exp) =1250kN
1000
z 800
-e-6C_8Agg
..:.:::
> 600
-6C_ 10Agg

400
200
0 1 2 3 5
Displacement (mm)
Fig.5.20 -Effects of Mesh Refinement on Predicted Load-Deformation Response of
Complex D-Regions (Note: 1) 1in.=25.4 mm, 2) 0.2248kips=1 kN)
189
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1500 --------------------------
-z
.:iii:::
-
:E' 1000
(J
!
a
"C
500
"C

D..
0
No Distrifed Reinforcement
500
With Distributed Reinforcement
j
Mean (Exp/FEA)=0.92 i
COV=8%
1000 1500
Measured Capacity (kN)
Fig.5.21 - The Overall Accuracy of the Calibrated FEA Models adopted to Predict
the Structural Responses of the Specimens 4A to 7C.
5.6 Numerical Results and Structural Performance Evaluation and Comparison
The predicted structural behaviors of specimens 4E, 5D, and other specimens are
given in Figs. 5.22, and 5.23, and in Table 5.1. Based on Figs. 5.22 and 5.23, specimens
using the STM shape generated by PBO technique (4E and 5D) generally have the
highest cracking load and strength including the maximum stiffness with the minimal
degradation due to the controllable crack width and propagation which can be also
illustrated clearly in Table 5.1 where the estimated crack width was the lowest in most
cases.
The structural performance evaluation of all 16 complex regions are also given
in Fig.5.24 by comparing the level of strength, quantity of the reinforcement, and ratio of
strength to reinforcement quantity. Obviously, strength and ratio of strength to
190
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reinforcement quantity of PBO specimens are generally greatest in all cases although the
quantity of reinforcement of PBO specimens in some cases is not the lowest. This thus
confirms that using topology optimization technique in generating the shape of STM in
design process can result in complex regions with good structural performance
throughout the limit states.
400
---4E
350
----- 4A
300
48
250
-><-4C
-
-;o'-40

-
>
150
100
50
0
0 0.5

1.5 2
Fig.5.22 -Predicted Load-Deformation Responses of Specimen 4A to 4E (Mesh Size
= 25.4 mm (1 in.)}, Note: 1) 1in.=25.4 mm, 2) 0.2248 kips= 1 kN
191
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
700
600
/ ~ 1 / /
,, ,
. ,,'1
--+--50
---- - 5A
500
___ ., __ 5B
50
__ ,,, __ 5C
400
-z
~
-
> 300
200
100
5A 5B
5C
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Displacement( mm)
Fig.S.23 -Predicted Load-Deformation Responses of Specimens SA to SD (Mesh Size
= 2S.4 mm (1 in.)), Note: 1) 1in.=2S.4 mm, 2) 0.2248 kips= 1 kN
1.4
1.2
...
cu
-
1
l!
cu0.8
a.
"C
~ 0 . 6
cu
Strength(kN)*1000
Rebar
Volume(mm/\3)x10/\7
o strength/Rebar
Volume (N/mm/\3)
! 0.4 -+---------1--------lill-----{!!lo-
o
0. 2 --Hllllt-=---fkH
0 -i--..........
4A 48 4C 40 4E 5A 58 5C 50 6A 68 6C 60 78 7A 7C
Specimen No.
Fig.S.24 -Structural Performance Comparison of 16 Complex Regions
192
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 5.1-Details ofNonlinear FE Analysis Results of 16 Complex Regions
Est. Est.
No. Strenath (kN/ki JS)
max max
er er
Mode
width width
of
50% 75% Volume Strength/
Test/ Failure
Pn Pn (x10
6
Volume
Test FEA FEA
(FEA)
(mm) (mm) mm
3
/ft
3
) (N/mm
3
)
4A 32S/73 321/72 1.01 SC 0.07 0.19 1.7S/0.06 0.18
4B 334/7S 306/69 1.09 SC 1.S9 1.16/0.04 0.26
4C 312/70 318/72 0.98 SC 0.46 1.08 2.02/0.07 0.16
40 3S6/80 331/7S 1.07 SC 0.19 0.39 2.19/0.08 0.1S
4E N/A 346/78 N/A SC 0.12 0.2S 1.27/0.04 0.27
BFand
SA 418/93 419/94 1 SC 1.92 3.72 1.3S/O.OS 0.31
SB SS1/124 S41/122 1.02 SC 0.97 1.91 2.01/0.07 0.27
CR
SC 418/93 420/94 1 and BF 1.04 2.38 1.3S/O.OS 0.31
CR
and
SD N/A S74/129 N/A SC 0.86 1.06 1.66/0.06 0.3S
CR
6A 1120/2S2 127S/287 0.88 and BF 0.69 1.48 10.6/0.37 0.11
large
6B 814/183 807/182 1.01 crack 2.6 2.91 7.1/0.25 0.11
SC
6C 1292/290 12S8/283 1.03 and BF 1.07 1.82 11.S/0.41 0.11
CR
60 863/194 923/208 0.93 and BF 0.71 1.6S 6.7/0.24 0.13
CR
and
7B 1300/292 127S/287 1.02 SC 0.4 0.77 8.4/0.30 0.1S
large
7A 2S8/S8 919/207 0.28 crack 2.28 >3 10.6/0.37 0.02
7C 1040/234 127S/287 0.82 CR 0.3 O.S3 9.S/0.34 0.11
Note: 1) SC= Shear Compression Failure
2) CR= Crushing
3) BF= Bursting Failure
4) 1 in.= 25.4 mm
193
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.7 Summary
In this chapter, the author addresses and studies the influence of STM shape on
the structural performance of complex regions based on the experimental and
computational validation of the selected complex regions designed by using STM shapes
obtained from designer's experiences, empirical models, and topology optimization. The
experimental and computational validation confirms that using topology optimization in
selecting STM shape in the design process can produce in the complex regions that are
designed to exhibit good structural performance at both service and ultimate states
because the orientation of the reinforcement in the optimal STM shapes can lead to the
structure exhibiting a higher cracking load and can effectively control the development
and propagation of cracks. In addition, it was observed that cracking is strongly
influenced by the pattern of selected reinforcements. Alternative to placing the
reinforcement in the optimal STM shape, the orthogonal orientation can be sometimes
more practical; however, the structural behavior and performance of the regions should
be assessed based on nonlinear FEA tool for the effective and safe design. In addition, the
distributed meshing must be provided in the D-Regions designed by the STM to provide
the sufficient local ductility that prevents the premature failure due to the unstable crack
propagation as required in some code of practices and by the plasticity theorem.
In the next chapter, a new nonlinear FEA that was specialized for modeling and
analyzing complex D-Regions will be introduced. The theory, formulation, and validation
of the model will be presented in details. The model will lead to the simplicity in
modeling of complex D-Regions and will provide the more complete, accurate, and
reliable nonlinear analysis of the complex D-Regions.
194
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 6
Nonlinear FEA for Automated Analysis of Complex D-Regions
As presented in the new design and analysis framework shown in Fig. 1.5, the
performance of complex D-Regions designed using the STM may be predicted using
Non-Linear Finite Element Analysis, herein referred to as NLFEA If the results of these
analyses indicate that structural performance targets are not satisfied, then the analytically
predicted behavior can be used to inform the design refinement process.
As illustrated in Chapter 2, NLFEA tools for structural concrete can provide
sufficiently accurate means of predicting the structural performance of D-Regions.
Unfortunately, the effort required to use these tools and interpret the results from a
NLFEA often make their use impractical. To overcome this time consuming barrier to the
use ofNLFEA, some researchers (Park et al, 2007) have proposed the adoption of
automated NLFEA. In this approach, embedded reinforcement elements are used rather
than truss elements such as used in many programs including Vector2. This enables the
mesh to be automatically generated independent of the location of reinforcement. The
work begun by Park is extended in this study.
As part of this effort, shortcomings in current state-of-the-art behavioral models
(e.g. MCFT or DSFM), which are: 1) A significant dependency on mesh size and 2) An
ability of the advanced analysis tool "Vector2" implemented MCFT and DSFM models
that can only account the effects of reinforcement by using distributed reinforcement in
each element or truss bars between elements; this both increased the modeling effort for
the use and the dependency of the prediction on mesh size selection, will be overcome
through the development and implementation of new formulations. This includes
195
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
modeling bond-slip along the interface between concrete and reinforcement to improve
the accuracy of predictions and through being able to more fully capture all modes of
failures (Foster et al, 1996). In modeling concrete regions, the existing behavior models
need be improved by replacing assumptions with models. This includes: (i) more fully
describing the tensile response of cracked concrete between and at crack locations, (ii)
accounting for the impact of differences between the direction of principal strain and
stress, (iii) accounting for the impact of differences between the direction of principal
stress and crack orientation, and (iv) considering the effects of dowel action.
This chapter presents the analytical works required for development of an
improved NLFEA methodology for the automated analysis of complex D-Regions.
Section 6.1 presents the concepts and formulations of the embedded reinforcement
element including bond stress-slip. The results to demonstrate the validity of the
proposed formulation are provided in Section 6.2. Section 6.3 presents an improved
constitutive model for reinforced concrete continuums. Section 6.4 presents numerical
examples of this formulation. The mesh dependency reduction is achieved by using the
improved constitutive model as demonstrated in Section 6.5. Section 6.6 provides the
numerical study on the influence of orientation of principal stress in concrete and crack
angle on structural strength and responses of complex D-Regions. Section 6.7 presents a
summary of this component of the study.
6.1 Embedded Reinforcement Element Including Bond Stress-Slip
As described in Chapter 2, relying on the use of truss elements for the modeling
of reinforcement in complex D-Regions is problematic due to mesh topology dependency
as illustrated in Fig. 6.lb. The layout ofreinforcement in complex D-Region, as shown in
196
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. 6.la, can be irregular, particularly around complex aspects of the geometry such as
openings. This impedes using truss elements to well connect to nodes of the neighboring
elements and thus degrades the accuracy of the model. One way to solve this problem is
to use automatic mesh generation as shown in Fig. 6.lc. However, this feature is
uncommon in many NLFEA tools for concrete structures. With the use truss and user-
distributed reinforcement modeling, any changes in the layout of reinforcement would
require the use of a new mesh and this would further complicate the analysis process and
the comparative interpretation of results. An alternative to specialized meshes is to use
embedded reinforcement since it enables reinforcement to be modeled independently of
mesh topology.
- -
"
~ ~ ~ ~ ' ~ ~ ~
5.90J;1.97it0.59
HE
---
(a)
,
I

I

= _:

(b)
~ . . -
--
v
-
v
llii-
IV
F
/
I/
/
!/
Fig. 6.1 (continued on next page)
197
17.94
-
,
/
0.7 0.89
"l
~
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Truss Types
PBO mean comllance Fall06 (Aba Def1rnt1on 2 Truss Types Displacement Factor= O 00
- 100 - 500
- 200
- 300
- 400
(c)
Fig. 6.1 FE Models of a Complex D-Region
Embedded reinforcement modeling approaches have been formulated for the
perfect bond condition such as those by Chang (1989), Ranjbaran (1990) and others.
However, neglecting bond-slip effect can adversely affect the accuracy of analyses as
demonstrated by Foster et al (1996). Several researchers have been working on
formulation of the bond-slip of the embedded reinforcement such as Elwi and Hrudey
(1989), Kwak and Fillippou (1995), Jendele and Cervenka (2006), and others. Most of
the works are formulated based on the bond-link concept which is less accurate than
bond-zone concept (Keuser and Mehlhom, 1987). In addition, available formulations for
accounting for the bond-slip effect have been developed for tangent based analysis
methods, and not for secant based methods as primarily used in the implementation of the
MCFT and DSFM.
198
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Thus, the author proposes to develop a new embedded reinforcement model that
accounts for bond-slip effects in NLFEA formulations that employ fundamental concepts
of the MCFT and DSFM. The formulation is implemented using a displacement based
approach where the displacement fields of embedded reinforcement, reinforced concrete
continuum, and slips are approximated using shape functions and nodal displacements. In
addition, the total secant based approach will be adopted in deriving the element and
structural level stiffness matrices.
6.1.1 Concept and Proposed Formulation
In this proposal, the embedded reinforcement with bond-stress slip formulation is
formulated using the displacement based approach for the embedded reinforcement
technique originally proposed by Ranjbaran (1990) for perfect bond condition and then
extended to including bond-slip effect by using the contact slip concept proposed by
Schafer (1975). The finite element equilibrium equation of the embedded reinforcement
with bond-stress slip is derived using the method of weight residual by the mean of the
Galerkin method.
199
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Mapped Element
ax'
..
A,E,L
Physical Element
y
-r(flu) ax'+dax'
.. .. .. .. ..
dx' ..
Free body diagram
Fig. 6.2 The Mapped and Physical Coordinates of Embedded Reinforcement
including the Free Body Diagram for Bond Stress-Slip Problem.
A free body diagram of the embedded reinforcement under stresses at both ends
and with bond stress acting along the embedded reinforcement is shown for a small
length "dx" in Fig. 6.2. From this, the following equilibrium equation, as given in Eqn 6-
1 below, is obtained.
This leads to the differential equation (DE) or strong form of the bond problem in
Eqn (6-2)
da ,
_x_A

= 0
dx' '
200
(6-1)
(6-2)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
where As and L
0
are area and perimeter of embedded reinforcement, respectively,
and is the bond stress due to slip ( at the interface between the embedded
reinforcement and the concrete.
By adopting the concept of contact problems as demonstrated by Schafer (1975),
the slip can be treated as a new degree of freedom during the formulation process. Rather
than using the virtual displacement of the slip, the author proposes to use the weight-
residual method for a more general treatment within an FE context.
From DE in Eqn. (6.2), the weight residual (weak form) can be determined as
shown in Eqn. (6.3).
fWiRdO = 0 (6-3)
in which R is the residual and obtained in terms of the approximated displacement field
d(i 1
as R = As - L
0
and Wi is a weight function equated to a shape function of
the approximated slip field for the Galerkin approach. For the ID domain of embedded
reinforcement, the weight residual function is formed as:
(6-4)
To obtain Eqn 6-4 in terms of shape functions and nodal displacements of concrete
and embedded reinforcement elements, the compatibility equation or the relative slip
(Eqn 6-5) at the interface between the concrete and embedded reinforcement elements is
used together with the iso-parametric element concept as presented below.
In Fig. 6.2, the relative slip along the x' axis of the physical element can be
expressed as
201
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(6-5)
where u., and uc are the approximated displacement field of the embedded
reinforcement and the reinforced concrete element, respectively. Also [N
6
] and {d} are
the shape function of relative slip and nodal displacement of the reinforced concrete
element and embedded reinforcement.
Generally, the approximated displacement field of embedded reinforcement (constant
strain element) can be given as:
= "N. ( ) = [ _ ic=!.2. ][Ulx's]
US I r USI 2 2
U2x's
(6-6)
in which N;(r) is the ith component shape function and us; is the nodal displacement in
local coordinate ( r) of the embedded reinforcement (Fig. 6.2). To obtain us in terms of
nodal displacements in global coordinates (x-y), coordinate transformation was adopted
as below.
[
Ulx's J = [Ts]
u2x's
usyl
usx2
(6-7)
in which [Ts] can be obtained as [ l C and S are the cosine and sine of the
angle of the embedded reinforcement (8s) as illustrated in Fig. 6.3.
202
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ucy1
01x's
ltCX1 .

y


x
..





.ucy4
Ucx4
..
.usy2
.u2x's
Usx2
..
bondslip etemeQt




.ucy
.<{Jex'

Ucx
. ..



Reinforced concrete element
.ucy2
.ucy3
Ucx3
..
Fig.6.3 The Local and Global Displacement Components of Embedded
Reinforcement and Reinforced Concrete Element.
Substituting Eqn. (6-7) into (6-6), the approximate displacement field of
embedded reinforcement ( u.) can be obtained as:
[
Usxl I
U = [ N* ] Usyl
s rs U
sx2
Usy2
(6-8)
where [ N* ] was given as [ - C (r-I) - S (r-t) c (r+J) s (r+ll ]
rs 2 2 2 2
The term uc in Eqn (6-5) is the displacement field along the embedded
reinforcement orientation (dotted line in Fig. 6.3) and thus can be expressed in terms of
the shape functions of the reinforced concrete element below.
The displacement field of reinforced concrete element at any point in the element
domain can be expressed as:
203
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ucxl
Ucyl
Ucx2
[ Ucx l = [ N1 0
Ni
0
NJ
0
N
4
OJ
Ucy2
Ucy 0 N
1
0
Nz 0 NJ
0 N
4
Ucx3
Ucy3
Ucx4
Ucy4
where N
1
to N
4
is the shape function in terms of mapped coordinates given as
N; = 0.25(1- ~ ~ )(1- T/T/;) for i = 1 to 4
The term uc can be thus obtained as uc = ucx cos(}+ ucy sin(}= [CS] [ucx]
Ucy
(6-9)
(6-9a).
(6-9b).
Substituting Eqn. (6-8), (6-9), (6-9a), and (6-9b) into Eqn. (6-5) and rearranging the
terms, the shape function [Nb] of the relative slip and nodal displacement { d} of the
reinforced concrete element and embedded reinforcement can be obtained as shown
below.
(6-9c)
To obtain the consistent variable in [Nb] , the equation to link between the
mapped coordinates (s,11) and r was given using linear expression suggested by Elwi and
Hrudey (1989) as:
and 1J = 0.5(1Jb -17a )r + 0.5(1Jb + 1Ja) where (Sa.lla)
and (Sb,llb) are the mapped coordinates of the intersection points between the embedded
reinforcement and the concrete element as illustrated in Fig. 6.2.
204
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As illustrated in Fig. 6.2, the values of (sa,TJa) and (Sb,TJb) correspond to the values
of (Xa,Ya) and (xb,Yb) in the physical element and can be expressed as shown below using
an isoparametric concept.
for i = 1 to 4 and k = a to b (6-10)
By using the nonlinear equation in Eqn ( 6-10), iterative techniques such as the
Newton-Raphson scheme can be used to solve for the values of (Sa,TJa) and (Sb,TJb) when
the values of (xa,Ya) and (Xb,Yb) are available. For example, if (xk,Yk) is in the physical
coordinates, the nonlinear equations can be formed as follows.
(6-1 la)
(6-11 b)
The updating (Sk, TJk) for the n+ I th iteration can be obtained as:
(6-1 lc)
[
f l ~ Jn [ l]n [ ~
where k can be obtained as A-
1
g and A is defined as ~ ~
fi1h g2 o ~ k
ogl Jn
Br/;
og2 or
o'lk
-[Jr which is the negative values of Jacobian matrix at the nth iteration.
After obtaining all approximated displacement fields, the weight residual function
(Eqn 6-4) can be further formulated by integrating the first term by parts once to
eliminate the first derivative of the approximated stress field and invoking the boundary
terms.
l l
P- f Wi,x
1
(A.
1
. Ei x')dx' - f Wi(L
0
r(fiub) )dx' = 0 (6-12)
0 0
205
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[p]
where the boundary term is P = and P.<e
4
xi) and Pccsxi) are the boundary
force vectors of embedded reinforcement and concrete element, respectively. Also the
terms Wi and Wi,x; are the shape function of the relative slip e.g. [Nb f (Eqn 6-9c) and
its first derivative with respect to the xi variable e.g. [Bsrr] (Eqn 6-16a and b),
Bsb
respectively. From Eqn 6-9c, it can be shown that Bs and Bsb are the contribution of the
embedded reinforcement and the bond-slip on the reinforcement element stiffness. To
formulate the FE formulation for secant based approach, the consistent linearized terms
of nonlinear bond stress-slip and stress-strain relationship of embedded reinforcement at
the current state (iteration) ith can be obtained from Fig. 6.4. This leads to Eqn ( 6-13) as
shown below,
(6-13)
(6-l 3a)
(6-13b)
The term e;sx' can be calculated from the first derivative of displacement of
embedded reinforcement as shown below.
. du du
er , __ s __ s dr
sx - dx' - dr dx'
(6-14)
206
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
~ \ u b
S2
,..
S3
Fu
(b)
ts
esh
,..
eu
Fig. 6.4 Constitutive Model of Bond Stress-Slip and Embedded Reinforcement
In Fig. 6.3, the term es is the orientation of the embedded reinforcement relative
to the global x axis. Thus, as suggested by Ranjbaran (1990),
c = coses = ax = ox k and s = sines = Oy = Oy or
ox' or ax ax or ox'
Since cose
2
+sine
2
= 1 then ox'= 1 = (ax
2
+ ay
2
)
s s ' or or or
(6-15)
By substituting Eqn(6-8) and Eqn(6-15) into (6-14), it follows that
u.\'X1 u.u1
i du du d[ -C (r-1)
& , _ __ s __ S _ _L _ ---=--2 ~ ~ ~ ~
sx - dx' - dr J' -
-S <r-1i C (r+l)
S (r;l) ]
1
usyl
=[B,]
u.vyl
2 2
dr
j'
usx2 usx2
usy2 usy2
Hence,
d[ _ C (r-1) _ S (r-1) C (r+l) S (r+l) ]
[
B ] = 2 2 2 2 1.
" dr J
(6-16a)
It should be noted that Eqn (6-16a) is the first derivative of the components of the
shape functions of the relative slip that are contributed from the embedded reinforcement
with respect to the variable x'. Similarly, the term Bsb can be obtained by taking the first
derivative of the components of the shape functions of the relative slip that are
contributed from the reinforced concrete element as shown below.
[
B ] = d[-CN1 -SN1
sb
-CN
2
-SN
2
-CN
3
-SN
3
-CN
4
-SN
4
]
dx'
207
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(6-16b)
Then, to obtain the final form of equilibrium equation of the embedded
reinforcement including bond stress-slip, Eqn (6-5), (6-13a and b), and (6-16a and b) are
substituted in Eqn(6-13), rearranged into matrix forms, and this leads to Eqn(6-17a) as
shown below.
[
F_'] [[Ks]4x4
12x1 - [Ksb]8x4

l + [[Kbss ]4x4 [Kbsc l4xs][us4x1 l


08x8 u Csxl [ Kbcs lsx4 [ Kbcc lsxs u c8xl
(6-17a)
[
[ Kbss + Ks ]4x4 [ Kbsc ]4xs ][us 4x1 l
- [Kbcs+Ksb]
8
x
4
[Kbcc]
8
xs ucsxi
l 1
Where[Ks]4x4 = f [Bsf (As
5
)[B
5
]dx' = f[Bsf (As E,)[B
5
]J* dr,
0
l 1
[Ksb]
8
x
4
= f [B,hf (As E.
1
.)[Bs]dx' = f[Bsbf (As Es)[Bs]J*dr,
0
From Eqn (6-17a), the stiffness matrix representing the contribution of the
embedded reinforcement and bond-slip is non-symmetric due to the term [Ksb] which is
the effect of the bond-slip on the embedded reinforcement due to the deformation of
concrete. In this proposal, to reserve the symmetry of the stiffness matrix, this term is
proposed to be neglected and Eqn (6-17a) is reduced to Eqn (6-17b) as shown below. As
pointed out by Ashraf and Filippou (1999), the influence of the deformation of the
concrete on the slip at the interface between reinforcement and concrete may be
208
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
neglected for the post yielding state of the reinforcement. In addition, when cracks
developed in the regions of concrete surrounding reinforcement, the ability for stress
transfer at the interface (bond) between reinforcement and concrete will be degraded and
the slip thus considerably increases. In this case, it is reasonable to neglect the
deformation of the concrete as well.
[
~ ] = [[Kbss + Ks]
4
x
4
p' [Kbcs]
c 12x1 8x4
[ Kbsc ]4xs ][us 4xi]
[ Kbcc ]
8
x
8
u C
8
x
1
(6-17b)
The validity and the accuracy of the proposed formulation when the term Ksb is
neglected will be demonstrated by using the numerical examples given in section 6.1.2.
The assembly process of the embedded reinforcement including bond stress-slip
in a smeared rotating-angle cracked reinforced concrete element can be subsequently
obtained as shown below.
04x8 ]['iis4x1 J} )
[KcJsxs 'iicsxl e
(6-18a)
[
~
1
] ={[[Kbss+Ks]
4
x
4
P
1
[ Kbcs lsx
4
c s
[Kbsc]4x8 ]['iis4xl ]}
[Kbcc+KcJsxs 'iic
8
x
1
s
(6-18b)
in which e is element number, M is the total element number, Kc is the stiffness matrix of
smeared rotating-angle cracked reinforced concrete element (Vecchio, 1990), and s is the
subscript for structure level.
Static Condensation Technique to Reduce Problem Size
Due to the computational inefficiency from double nodes in the bond-slip
formulation, the static condensation technique is adopted here to reduce the problem size
209
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
as suggested by Kwak and Fillippou (1995). By using the static condensation technique
in Eqn(6-18b), two sets of the equations can be obtained as shown below.
P/ =[Kbss+Ks]us+[Kbsc]uc (6-19a)
~ =[Kbcs]us+[Kbcc+Kc]uc (6-19b)
From Eqn(6-19a), the expression for us =[Kbss+Ksr
1
(P/ -[Kbsc]uc) is substituted
in Eqn(6-19b) and this leads to Eqn(6-19c).
Pc' - [Kbcs]([Kbss + Ksr
1
P/ = (-[Kbcs][Kbss + Ksr
1
[Kbsc] + [Kbcc + Kc])uc (6-19c)
where Peq =P/-[Kbcs]([Kbss+Ksr
1
P/ and Keq= (-[Kbcs][Kbss+Ksr
1
[Kbsc]+[Kbcc+Kc])
When loads are not directly applied to the embedded reinforcement e.g. P/ = o,
this leads to a reduction in the problem size and thereby the solution process as
Pc' = Kequc. Thus, the size of the problem is reduced to the case where no bond stress-slip
is included, and the responses of the embedded reinforcement can be subsequently
obtained by solving the small problem of either Eqns. (6-19a) or (6-19b).
6.2 Numerical Examples
To demonstrate the validity of the proposed formulation for the embedded
reinforcement including bond stress-slip, numerical examples based on the existing
experimental data from the literatures are given in this section. Section 6.2.1 provides the
examples at the anchorage test level. The examples in the structural level are given in
Sections 6.2.2 and 6.2.3.
210
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.2.1 Anchorage Tests
The responses of anchored reinforcing bars under monotonic pull-out loads are
predicted by the proposed formulation. The experimental data for the tests are taken from
Viwathanatepa et al(l 979). In addition, the proposed formulation is compared with the
existing bond-slip models proposed by Viwathanatepa et al (1979) and Yankelevsky
(1985) to demonstrate the accuracy and efficiency of the proposed formulation.
Pull-Out Loading
25in
Fig.6.5 The Anchored Reinforcing Bar under Monotonic Pull-Out Loading
The proposed formulation is used to simulate the response of anchored
reinforcing bars illustrated above under monotonic pull-out loading. The specimen is a #8
bar anchored in a well confined block of 25 in (63.5 cm) width, which corresponds to the
required anchorage length for a #8 bar. The concrete cylinder strength used in the
specimen is 4700 psi (32.4 MPa) and the yield strength of the reinforcing steel is 68 ksi
(467 MPa). The yield strain and the strain hardening modulus are 0.23% and 411 ksi
(2834.50 MPa), respectively. The parameters adopted in the bond stress-slip relationship
(Fig 6.4a) are taken from Yankelevky (1985) as: 'ty=14.7 MPa, 'tu=4.05 MPa, sy=0.62
mm, s2=1.6 mm, and s3=5.9 mm. The concrete regions are discretized into 10 elements
longitudinally and 6 elements transversely. The stress distribution along the embedded
reinforcement is compared with existing experimental data and computational models.
211
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The levels of the applied pull-out loading considered here are to cause the left ended
stress from 20 ksi to 70 ksi.
25

20
-
-
15
-
"'
"'
QI
...
00 10

QI
.....
rJ1
5
0
45
40
35
_30
!25
"'
"'
20
00
- 15
0
\
'
'
'
5
' .. . .
':'
...
'
. : .
. . .
. . .
'
'
---
10
"' - --
.. -
15
Distance(in)
(a)
- Viwathanatepa et al
- - - Yankelevsky
- - - experiment
- Sukit-Yankelevsky
bond-slip
.........
20
- - Viwathanatepa et al
- Yankelevsky
experiment
- Sukit-Yankelevsky
10 .
5 ' ..... ....: . .- . . ..
25
0
0 5 10 15
Distance( in)
(b)
20
Fig. 6.6 (continued on next page)
212
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-
70
60 -
50

<ll
<ll

00
]20
....
00
10
'
- Viwathanatepa et al
Yankelevsky
- experiment
- Sukit-Yankelevsky
'
-
___
.. __ _
0
0 5
80
70 ........
60

.:id
-


lo.
....
00 30
'ii

10
0
0 5
10
10
15
Distance(in)
(c)
15
Distance(in)
(d)
20
- - - experiment
- Sukit-Yankelevsky
bond_slip
20
25
25
Fig. 6.6 The Predicted Stress Distribution for Anchored Reinforcing Bar under
Monotonic Pull-Out Loading: (a) The Left End Stress= 20 ksi; (b) The Left
End Stress= 40 ksi; (c) The Left End Stress= 60 ksi; and (d) The Left End
Stress= 70 ksi.
213
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Comparing the predicted results to the experimental data and other computational
models, the proposed formulation can well predict the experimentally measured stress
distribution. The method seems to slightly overestimate bond stress in cases of low
applied stress; however, the prediction highly agrees with the experimental results at the
higher stress levels even after yielding of the reinforcing bar. In this example, neglecting
the term Ksb in Eqn (6-17a) did not have a significant effect on the accuracy of the
numerical result. Therefore Ksb will be neglected in the proposed formulation
Pull-Push Loading
The propose formulation is further adopted to simulate the response of the
anchored reinforcing bars under monotonic pull-push loading. The material properties are
the same as in pull-out loading except for the yielding strength of the reinforcing steel
which becomes 68 ksi ( 469 MPa).
25in
Fig.6.7 The Anchored Reinforcing Bar under Monotonic Pull-Push Loading (Casel)
214
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
- Sukit-Yankelevsky
bond_slip
10
- Viwathanatepa et al
----- Yankelevsky
-
in
5
~
- Kwak and Filippou
-in
- Experiment
in
f
0
-
rn
5 10
a;
25
Cl)
-5
-
rn
-10
-15
Distance( in)
(a)
30
-Sukit-Yankelevsky
bond slip
- Viwafflanatepa et al
20
Yankelevsky
-
10
-in
- - - Kwak and Filippou
- Experiment
in
!
0
-
rn
5 10
a;
~
.S-10
rn
-20
-30
Distance( in)
(b)
Fig. 6.8 The Predicted Stress Distribution for Anchored Reinforcing Bar under
Monotonic Pull-Push Loading (Case 1): (a) End Stresses= 10 ksi; and (b) End
Stresses = 24 ksi
215
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25in
Fig.6.9 The Anchored Reinforcing Bar under Monotonic Pull-Push Loading (Case2)
The prediction obtained from the proposed formulation as illustrated in Figs. 6.8
and 6.10 results in a slight over estimation of the steel stress in the low level of pull-push
ended stresses; however, the predictions greatly correlate with the experimental data at
the higher ended stress levels. Since the same bond slip relation was adopted throughout
the length of the specimen e.g. no modification of the relation in the outer unconfined
portion of the specimen, the prediction results in slightly overestimate the stress in the
reinforcing bar in the push side even at the high stress level. In addition, by comparing
the accuracy of results obtained from the proposed formulation which is displacement
based formulation to the result obtained from the mixed formulation proposed by Kwak
and Fillippou (1995), the accuracy of the mixed formulation is higher. This is not
surprising; however, the simplicity of the displacement based formulation, the
practicability in the implementation, and the full compatibility of the formulation to the
MCFT based constitutive approach for 2D continuum are overcome by the proposed
formulation.
216
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
::-
I/)
40
30
20
-30
-40
60
40
20
I/)
I/)
Cl)
0
...
-en
'Gi-20
Cl)
-
en -40
-60
-80
5
-Sukit-Yankelevsky bond_slip
-Viwathanatepa et al
------Yankelevsky
- - - Kwak and Fillippou
Experiment
Distance(in)
(a)
Distance( in)
(b)
- Sukit-Yankelevsky
bond slip
------Kwak-and Fillippou
Experiment
Fig.6.10 The Predicted Stress Distribution for Anchored Reinforcing Bar under
Monotonic Pull-Push Loading (Case 2): (a) End Stresses= 40 ksi; and (b) End
Stresses = 68 ksi
217
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Push-In Loading
In this section, the proposed formulation was adopted to predict the strain
distribution along the anchored reinforcing bar under push-in loading. The test specimens
used in this section were from a series of tests by Muller and Eisenbiegler ( 1979). The
specimens, 18 mm diameter deformed bar centrally embedded in a concrete block, were
applied by compressive loads. The material properties are adopted as follows: the
young's modulus of 191695 MPa, the yield strength of 383 MPa, the strain hardening
modulus of 2834 MPa, and the ultimate strength of 575 MPa. The parameters for the
bond stress-slip diagram are from Yankelevky (1985): 1y=21MPa,1u=lO MPa, sy=l
mm, s2=4 mm, and s3=8 mm. The based curves for the bond stress-slip are from
Yankelevky (1985) and CEB-FIP (1990) for easel and case2, respectively.
--" w
CJ1 CJ1
0 0
3
CJ1 !
3
.,J:l.
3 3
0 0
0 0
3 3
3 3
easel
Fig.6.11 The Anchored Reinforcing Bar under Monotonic Push-in Loading
218
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-1.4 ------------ ---------,....------------.
- Sukit-Yankelevsky
-1.2
- - Experiment
-Yankelevsky
- -1
E
-

-c:
c;; -0.6
i.;.
-
en -0.4
-0.2

0 50 100 150
Distance( mm)
Fig.6.12 The Predicted Strain Distribution for Anchored Reinforcing Bar under
Monotonic Push-in Loading (Case 1)
-1.6
- Sukit-CEP-FI P
-1.4
- Experiment
-1.2
-
E
-1
-
E
-Yankelevsky
.5.-o.8
c:
c;;
.:: -0.6
en
-0.4
-0.2
0
0 50 100 150 200 250 300 350
Distance( mm)
Fig.6.13 The Predicted Strain Distribution for Anchored Reinforcing Bar under
Monotonic Push-in Loading (Case 2)
Based on Figs. 6.12 and 6.13, the predictions from the proposed formulation for
both easel and case2 are highly correlated with the experimental data globally, although
219
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the proposed formulation slightly underestimates the strain in the top portion of the bar.
In addition, the proposed formulation can accurately capture the highly nonlinear strain
distribution in the bar tested in case2.
6.2.2 Two-Spanned Continuous Deep Beam (Foster et al, 1992)
In this section, the numerical examples in the structural level are presented to
demonstrate the validity of the formulation. The selected structure is a double span deep
beam No.5 tested by Foster (1992). The details of the deep beam including its FE
meshing are given below.
900
,-<( 250
..
A"""
2Y12
R4@100. -2Y16 300
250
I i
32
'
. .. ,
700

I

I
f ..
I - .
\ I I
I

2Y16
....
R4@100
2Y16 R4
1650
300
Section A-A
i .,.
L

i200
(a)
Fig. 6.14 (continued on next page)
220
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-""'
_.
-...._
....._
-""
,,,_
\ J "'
.......

,,,,,,,.
....
\ \ \ \ I I I
'
...
.......
--
j
...._
l
\ \ \ \
'

,
,
I I I J
\ \
'
1
_..... ._,,,,
'
,
,
I I ......
-""
....... ..._
'
, .,,,,,,
.......
.......
,
,

(b)
Fig.6.14 Geometry and FE model of a Double Span Deep Beam No.5 tested by
Foster (1992): (a) Dimensions and Reinforcement Details; and (b) FE Mesh and
Embedded Reinforcement Elements for the Left Span of a Double Span Deep Beam
No.5
The entire deep beam is modeled using 4 nodes isoparametric elements for
smeared rotating crack and smeared reinforced concrete elements. Fig. 6.14b illustrates
only a half (left span) of the deep beam. The FE formulation for the cracked reinforced
concrete element employs the newly proposed constitutive models for nonlinear FEA of
D-Regions. Section 6.3 provides a full and comprehensive detail of the proposed
constitutive models. The benefit of using embedded reinforcement in the example is
illustrated by using embedded reinforcement elements with bond stress-slip to model top,
bottom, diagonal main and web reinforcement in the beam. The distributed vertical
reinforcement, in this example, adopts the regular smeared reinforcement type to show
221
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the possibility and potential benefit of using both smeared reinforcement and embedded
reinforcement in a model.
The material properties of concrete are: f c (MPa) =26.9, ft (MPa)=l.82,
Ec(MPa)=21520, i::
0
(me )=2.5, and A simple elastic-plastic with strain hardening is
adopted for constitutive model of reinforcing bars with the material properties given in
Table 6.1 below.
Table 6.1-Material Properties Used in Nonlinear FE Analysis of Deep Beam No.5
(Foster, 1992)
Bar Type Es(MPa) Ey (me)
Esh(MPa)
R4 220000 2.5 73000
Y12 200000 2.4 10
Y16 190000 2.4 10
The bond stress-slip model adopted in the analysis employs the local bond stress-
slip model from CEB-FIP model code for monotonic loading and unconfined concrete
type. The good bond conditions are assumed in the model.
The predicted load-displacement responses of the deep beam modeled by using
perfect bond and by considering bond-slip are given in Fig. 6.15. In the perfect bond
condition, the predicted mode of failure is by crushing of concrete near the loading
column and no yielding of the reinforcing bars is predicted. This is different than the
reported experimental data. When including bond stress-slip in the model, the more
accurate result, particularly the mode of failure and displacement at the maximum load,
can be obtained comparing to the case where perfect bond condition is assumed. The
modes of failure obtained from this case are yielding of the bottom reinforcing bars and
222
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
subsequent concrete crushing near the loading column.
700
600
-z
:.. 500
;
c.
400
Cl)
c.
-g 300
0
_J
200
crushing of
concrete
rushing of concrete
-+- Experimental
100
- Su kit-Bon d-Slip(CE B-FI P)
0 2 3
Displacement (mm)
4
-l!l- Sukit-Periect Bond
5 6
Fig.6.15 Plots of Measured and Predicted Load-Displacement Responses of a Double
Span Deep Beam No.5 tested by Foster (1992)
In addition, the interesting result is the predicted strain distribution along the
bottom reinforcing bars. The plots given below present the strain distribution for three
cases: 1) experimental data; 2) FE model from Foster et al 1996 (both perfect bond and
bond-slip models); and 3) the proposed bond-slip formulation. As expected, the model
including bond stress-slip results in better predicted strain distribution along the bar,
particularly around the local mid span of the deep beam where high strain developed up
to 6 millistrains and mainly dominates to the mode of failure of the beam. In addition, the
plot illustrated another important issue which is that since the proposed formulation was
developed based on bond slip, not slip strain as in the model by Foster et al 1996, the
proposed formulation is not sensitive to the selected crack spacing. For example, the
models by Foster et al 1996 experience a significant change in the predicted strain in the
223
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
reinforcing bar when changing the crack spacing from 350 to 360 mm.
14
12
10
-
E
8
-E
E
-
c:
e 6
....
(/)
4
2
0
0 0.5
-+-Experimental(Foster, 1992)
----- FEM Perfect Bond (Faster et al,1996)
__..._FEM bond Cs=360mm, Foster et al, 1996
-*-FEM bond Cs=350 mm (Foster et al,1996)
-sukit-CEB-FIP
-.!P-Sukit-Perfe ctb ond
1.5
Displacement from Outside Edge(m)
Yield strain of bottom bar
2 2.5
Fig.6.16 Plots of Measured and Predicted Strain Distribution along the Bottom
Reinforcement of the Left Side of a Double Span Deep Beam No.5 tested by Foster
(1992)
6.2.3 Simply Support Deep Beams with Openings
In this section, the proposed embedded reinforcement including bond-stress slip
formulation is adopted to predict the strain distribution in the simply support deep beams
with openings of specimens. Only specimens 8B and 8C were intentionally selected for
the comparison with the computational model since the measured strain in specimen 8A
is very low and thus the more useful comparison could be obtained from the specimens
with higher measured strain levels in the reinforcement. The reinforcement details and
the material properties of the specimens are based on Fig. 3.17 and Table 3.2 and 3.3,
respectively. The FE mesh and models of these specimens are illustrated in Fig. 6.42. The
full details of FE models of the deep beams are fully described in the numerical examples
224
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
given in section 6.4.4. In this section, only the detailed comparison of the strain
distribution along the embedded reinforcement will be provided and discussed.
Considering the strain gages attached to the reinforcement in specimen 8B and 8C
as illustrated in Fig.3.17, the predictions of the strain distribution are illustrated in Figs
6.17 to 6.22 for three levels of applied loads corresponding to approximately 1st crack,
50%, and 75% of the capacity (Pn) of the specimen. From Figs 6.13 to 6.22, the strain
distribution measured in the reinforcement of the specimen is nonlinear due to the
interaction of concrete and reinforcement, cracking, and bond characteristics in the
regions. As illustrated clearly in the figures, both perfect bond and bond-slip formulation
result in the predicted strain distribution correlating well with the measured strain
distribution from the experiment; however, the level of the accuracy decreases when the
specimens are under the higher applied loads.
The perfect bond assumption generally leads to the less accuracy in strain
prediction than the bond-slip assumption. The under estimation of the strain level near the
end anchorage of the bars (strain gages: 23-24 and 49-51 in Figs. 6.17 and 6.20,
respectively) and the locations where several cracks occur (strain gages: 18-20 and 45-48
in Figs. 6.17 and 6.20, respectively) is generally obtained using the perfect bond
assumption; however, the improved strain level is obtained by using the bond-slip model.
This thus reveals the significant effect of the slip at the interface between the
reinforcement and concrete that should be included in the analysis of D-Region for the
most accuracy of the prediction of strain induced in the reinforcement and the load
transfer characteristics.
225
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Although the prediction of the strain distribution by using the embedded
reinforcement including bond stress-slip results in the acceptable level of accuracy, the
current bond-slip model is still inaccurate in predicting the strain level at the end
anchorage at the high applied load level; for example, the strain distribution along the
strain gages: 23-24 and 49-51 in Figs. 6.17 and 6.20, respectively. This is due to the
effect of the anchorage details; for examples, the effect of the 90 degree hook or use of a
steel end plate is not accounted for in this model. In addition, the accuracy of the strain
distribution prediction is less in the locations where the interaction of the neighboring
rebars is significant; for example, at the locations between strain gauge No: 31 to 33 and
58 to 60. In those locations, the interaction between the crossing reinforcing bars might
degrade the accuracy of the strain prediction in the embedded reinforcement since the slip
formulation is developed based on only the displacement field of the concrete element
and the reinforcement and thus does not account the effect of the crossing bars.
0.9
0.8
0.7
E'
-eo.6
!.o.s
c

<n
0.3
0.2
0.1
0
13 14 15
!::, 0.75Pn
Oo.5Pn
01st crack
_Exp
=:
16 17 18 19 20 21 22 23 24
Strain Gage No.
Fig.6.17 The Predicted Strain Distribution for Specimen SB from the Strain Gages
No. 13 to 24 at the Load Levels of 1st crack, 50%Pn, and 75%Pn.
226
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-
1
0.9
0.8
0.7
E
eo.6
!.o.5
c
'i! 0.4
<n
0.3
0.2
0.1
0
31 32
Strain Gage No.
6 0.75Pn
0 0.5Pn
D 1st crack
-Exp


33
Fig.6.lS The Predicted Strain Distribution for Specimen SB from the Strain Gages
No. 31 to 33 at the Load Levels of 1st crack, 50%Pn, and 75%Pn.
-
1
0.9
0.8
0.7
E
50.6
E-o.s
c
0.4
<n
0.3
0.2
0.1
0
34 35 36
Strain Gage No.
/:;. 0.75Pn
0 0.5Pn
D 1st crack
-Exp
-Bond-Slip
--- Perfect Bond
37
Fig.6.19 The Predicted Strain Distribution for Specimen SB from the Strain Gages
No. 34 to 37 at the Load Levels of 1st crack, 50%Pn, and 75%Pn.
227
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1
0.9
0.8
- 0.7
E
e o.6
E
- 0.5
c
0.4
<n
0.3
0.2
0.1
0
/.J. 0.75Pn
00.5Pn
D1s1 crack
-Exp ,
-Bond-Slip i
- - - Perfect Bond
i
40 41 42 43 44 45 46 47 48 49 50 51
Strain Gage No.
Fig.6.20 The Predicted Strain Distribution for Specimen SC from the Strain Gages
No. 40 to 51 at the Load Levels of 1st crack, 50%Pn, and 75%Pn.
1.5
1.4
1.3
1.2
1.1
'E 1
e o.9
E 0.8
";' 0.7
0.6
<n o.5
0.4
0.3
0.2
0.1
0
/.J. 0.75Pn
0 0.5Pn
D 1st crack
-Exp
-Bond-Slip
- - - Perfect Bond
60 59 58
Strain Gage No.
Fig.6.21 The Predicted Strain Distribution for Specimen SC from the Strain Gages
No. 5S to 60 at the Load Levels of 1st crack, 50%Pn, and 75%Pn.
228
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.5
1.4
1.3
1.2
1.1
- 1
E
e o.9
E 0.8
0.7
0.6
q) 0.5
0.4
0.3
0.2
0.1
0
A0.75Pn
00.5Pn
01st crack
-Exp
-Bond-Slip
- - - Perfect Bond
!!>-_
--
--
61 62 63
Strain Gage No.
Fig.6.22 The Predicted Strain Distribution for Specimen SC from the Strain Gages
No. 61 to 63 at the Load Levels of 1st crack, 50%Pn, and 75%Pn.
6.3 A New Constitutive Model for Reinforced Concrete Structures
6.3.1 Important Features
In this section, the author will present and discuss the features for the new
nonlinear FE model for reinforced concrete structures that the author proposes. As
reviewed in Chapter 2, the existing well-known computational models (i.e. MCFT,
DSFM, and CMM) for reinforced concrete structural modeling still require significant
improvement to be more suitable for the automated NLFEA framework for complex D-
Regions. Based on the fundamental concepts of those three models, the new FE model
should be formulated and developed to include the following features:
Smeared delayed rotating crack approach
Compression softening
Consideration of local crack phenomenon
229
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Tensile responses of concrete based on tension chord model (between and
the end of cracks)
Fracture mechanics concept for tension softening model (micro cracking
process)
Mesh dependency reduction.
Rational relationships to link between local crack phenomenon and
average element levels.
Decouple tension stiffening from tension softening.
Allow the deviation of principal compressive stress direction from total
(measured) compressive strain direction.
Effects of crack shear slip demand (the induced crack shear and the
deviation of crack angle from the direction of principal compressive stress
of concrete) and capacity (aggregate interlock and dowel action) on the
structural responses at both element and structural levels.
6.3.2 Proposed Formulations
In this section, the author will present and describe the proposed formulation of
the computational models including the above features previously mentioned. In addition,
the model is generalized to be applicable when the crack angle is not coincident with the
principal compressive stress since the crack direction measured from D-Regions tests can
significantly deviate from the principal stress direction and this will affect the shear
demand along the crack surface as presented by Nagle and Kuchma (2007).
230
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The derivation of the formulation starts by firstly considering the free body
diagram of a cracked reinforced concrete element subjected to in-plane stresses as
illustrated in Fig. 6.23.
4
B.
y
' . . '\
' .
. \
'
Fig.6.23 Reinforced Concrete Element: Crack Angle and Principal Stress Angle
In Fig. 6.23, the crack angle (Ser) is allowed to deviate from the principal stress
direction (Sc). By assuming that ~ or the difference between the crack angle and the
principal stress is known, the author can obtain the relationship between the crack angle
and the principal stress direction as in Eqn 6-20.
Ser= Sc+ S (6-20)
The Continuum Level and Local Crack Level
Since the proposed formulation will involve two levels: 1) continuum or element
level and 2) local crack level as adopted generally in both the MCFT and DSFM, a
definition for both levels will be given and explained for ease of formulation
development in the following section.
Consistent with other FE implementations of the MCFT and DSFM, the
continuum or element level is the domain that spans several cracks (Fig. 6. 7). The
231
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
stresses or forces acting on the domain, the material properties, and cracks inside the
domain are assumed to be averaged (or smeared).
The local crack level, as used by Vecchio and Collins ( 1986), is assumed to be on
the surface of the fictitious crack inside the element domain where its capacity for
transferring tensile forces across cracks is assumed to be by a combination of aggregate
interlock, dowel action, and local increases in reinforcement stress. The procedure
proposed to check the ability of the local crack to transfer the forces from the continuum
level is referred to by these authors as the so-called "crack check".
In the MCFT and DSFM, the crack orientation was usually assumed to be in the
direction of the principal compressive stress of the concrete. This accuracy of this
assumption has been questioned by researchers (e.g. Hsu (1998)) for almost two decades
since shear stress component must be disappeared when the fictitious crack is in the
principal stress direction. Since the crack angle is allowed to deviate from the principal
stress direction in the proposed formulation, this questionable assumption is overcome.
Equilibrium Equation
The element or continuum level equilibrium equation of the reinforced concrete
element in Fig. 6.23 can be derived from the free body diagrams (FBDs) of the isolated
regions that are shown in Fig. 6.24. It is interesting to examine how the difference
between the crack angle and the principal compressive stress of concrete (e.g. ~ not
equal to zero) affects equilibrium.
232
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

1----+-l---:
AO
crsx
-
=0
crsx
---
exists
'"Cx1T
crr
(tJ)::2 .. __
crsx_.
i;x1j-
Fig. 6.24 Free Body Diagram for Equilibrium Equation
crcl
:
Fig. 6.24a presents the FBDs when the difference between the angle of cracking
and principal compression is equal to zero. Fig. 6.24b presents the FBDs for the
case when exists. In the rotating crack or delayed rotating crack model, the crack
direction is not fixed but allowed to rotate depending on the state of stress in the material.
The equilibrium equations are simple to form and based on the quantities in the principal
stress direction, e.g. the quantities acting on the cut faces in the principal stress direction.
By using this approach, the equilibrium equations can be derived from the FBDs in Figs.
6.24a and b. and it is found to be able to express these in terms of only Sc as given
below.
ax = ac2 sin z (k + ac1 cosz Be+ Pxfsx
ay = ac
2
cos
2
Be +ac1 sin
2
Be+ pyfsy
Txy = (ac
1
-ac
2
)sin(kcos(k
233
(6-21a)
(6- 21b)
(6-21c)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
For the local crack equilibrium equation and by adopting the same technique as
used in the MCFT, the local crack equilibrium equation can be derived by equating the
internal force components in the continuum level to the internal forces along the average
crack surface at the local cracked level. Since the author considers ~ to be non-zero in
the model, it is also valuable to explore the effect of as described in reference to Fig.
6.25.
'. y
'., Jcrsy
.)'.
' crcl
crsx
____.
(a)
A
Continuum Level
x
Average Crack Level
B y
'
'B
(c)
B
y
crx
x
B
(b)
vc
Mohr's Circle
(d)
Fig.6.25 Free Body Diagram for Local Crack Equilibrium Condition
The internal force components of the continuum level can be obtained by cutting
the reinforced concrete element (Fig. 6.23) along either line A (Fig. 6.25a) or B (Fig.
234
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.25b ). By cutting along line A, which is in the principal concrete stress direction, the
internal force components along the cut face are composed of principal tensile stress of
concrete (crc1) and smeared reinforcing bar stress in x and y direction (PxO'sx and pycrsy).
There is not a shear component since the cut face is in the principal stress direction. By
cutting along line B, which is in the crack angle direction (8cr), the internal force
components along the cut face are composed of the normal stress in the concrete (crci*)
and smeared reinforcing bar stress in the x and y direction (PxO'sx and pycrsy). In addition,
the shear component ( "t'c *) is induced. By using Mohr's Circle for average stresses in
concrete (Nagle and Kuchrna, 2007) or a simple stress vector transformation (in this
research), one can obtain the formulation to calculate the normal and shearing stress
induced on the cut face as described in Eqn (6-22).
* * * * T T
where {ac} =[a cl ac
2
re] , {ac
12
} = [ac
1
ac
2
O]
[

, and = - sin
2

- 2
sin
2


2 B) B)
(6-22)
1

cos
2
sin
2

The internal force components along the average crack surface (at crack location)
can be obtained by cutting the element along line B as illustrated in Fig. 6.25c. The
internal force components induced along the crack surface are composed of the normal
component due to the axial stresses of the smeared reinforcing bar at crack location in x
and y direction and the shearing components due to aggregate interlock and the smeared
dowel stresses that are considered in the proposed formulation. The smeared dowel
stresses are assumed to be induced on the crack surface from the dowel action of the
distributed reinforcement passing through the cracked surface. Rather than considering
235
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the dowel force for each individual reinforcing bar, the author proposes to adopt the
smeared version of the dowel force as the smeared dowel stresses. The formulation for
this smeared dowel stresses will be derived later and as shown in Eqn 6-30.
These internal (resisting) forces along the crack surface are induced internally to
equilibrate the force transfer (demand) at crack. To obtain the equilibrium equation at the
crack, the internal forces in either Figs. 6.25a or b are equated to the internal resisting
forces in Fig. 6.25c. Since using different angle (i.e. 8cr and 8c,) results in the
complexity in forming the equations due to the geometry of cutting element, the author
selected to use Fig. 6.25b and Fig. 6.25c which use the same angle (e.g. 8cr). By
combining both figures, and assuming that Fig. 6.25b is cut in between cracks, and
including the additional mechanism of bond stress-slip based on tension chord model by
Kaufmann and Marti (1998), the FDB in Fig. 6.26 is obtained.
fsycr
Fig. 6.26 Free Body Diagram for Local Crack Equilibrium Condition including
Bond-Slip Mechanism
236
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
It should be noted that crc
1
*and 't'c * in Fig. 6.26 can be obtained from Eqn ( 6-22)
and the principal tensile stress of concrete ( crc
1
) in the equation is calculated from the
combination of tensile stresses of concrete between cracks (crcmxandcrcmy) as illustrated in
Fig. 6.27b. The tensile stresses in the concrete between cracks are derived from the
tension chord model (Fig. 6.27a) which is formulated as a function of bond stress and
average strain in concrete (Foster and Marti, 2003). In addition, stress components due to
aggregate interlock and dowel action are combined as traction along crack surfaces ( 1cr) .

(a) (b)
(C)
Fig. 6.27 Schematic Presentation: (a) Tension Chord Model (Foster and Marti,
2003); (b) Free Body Diagram of Principal Concrete Tensile Stress and Concrete
Stress Between Cracks; and (c) The Force Transfers between and at Cracks in TCM
(Adapted from Foster and Marti, 2003)
237
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Based on Fig. 6.26, the local equilibrium equation can be obtained. This equation
will be used together with the appropriate constitutive models to calculate the slip along
the crack used in the compatibility conditions in the next section.
(6-23a)
It should be noted that the local equilibrium equation given in Eqn (6-23a) is more
refined than the original equation proposed by Vecchio (2000). The original equation was
also derived by considering the stress components along the cutting sections at the
continuum and local crack level; however, only the contribution of the aggregate
interlock was considered. The contribution of both dowel action and the deviation of the
crack angle from the principal stress direction was not included in Vecchio's formulation.
In addition, since no approach was adopted to link the stresses at crack to those between
cracks, the original equation was solved using the ad-hoc assumption of the stresses at
crack location. The stresses at crack location were calculated by assuming that the
incremental strains at the crack location were at maximum in the direction of the
principal stress of concrete. By solving for the value of the incremental strains that
satisfies the tensile stress transfer across the crack, the stresses at crack location were
then obtained from the constitutive model of the reinforcing bars.
In this research, a more complete approach is adopted. By using the tension chord
model, a direct relationship between reinforcing bar stresses at cracks and between cracks
can be derived based on Fig. 6.27c and as given in Eqn (6-23b),
(6-23b)
P,
238
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
where fsim fsh crcmh and Pi are the reinforcing bar stress at crack, between cracks, tensile
stress of concrete between cracks, and reinforcement ratio in the i (e.g. x or y) direction,
respectively.
Fig. 6.27c illustrates how the forces in both concrete and reinforcing bars between
cracks transfer to the reinforcing bar at a crack. By using a simple FBD and considering
one cut section midway between cracks and another at a crack, the force component at
crack (Nscri) in the reinforcing bars must be equal to the summation of the force
components between cracks due to concrete (Ne) and reinforcing bars (Nsi) e.g. Nscri =
Nsi +Ne. Since the force components can be formulated as the product of stress and area,
by considering the reduced concrete area due to the reinforcing bars e.g. Ac= Atotal - As,
the equilibrium equation can be expressed in terms of stresses as fscriAsi = fsAsi
+ O'cmi(Atotal-Asi). Subsequently, by dividing the previous equation by Atotal and
rearranging the terms, Eqn (6-23b) can be easily obtained.
Compatibility Condition
Since, in the proposed model, the author adopts the delayed rotating crack
approach, the compatibility condition is thus simply based on the DSFM (Vecchio, 2000)
with some modifications to include the effect of e In the DSFM, the compatibility
condition of an element is composed of concrete continuum straining and discontinuous
slip straining where the combination of those is called the apparent or measured (total)
straining. The total strains will be defined as
(6-24)
where [Ee] is the concrete continuum strain, and [Es] is the slip strain. It should be noted
that Eqn (6-24) does not account for elastic or plastic strain offsets, and other effects such
239
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
as thermal expansion, mechanical expansion, etc. However, the equation can be extended
to consider those effects using the approach by Vecchio (1992). In addition, the slip strain
[Es] can be calculated by using the relationship below.
Es x=-ysf2sin(28cr)
Es y=ys/2sin(28cr)
y\y=ys cos(28cr)
(6-25a)
(6-25b)
(6-25c)
where Ys is the magnitude of slip strain calculated as the ratio between slip (Os) and crack
spacing (s) e.g. osfs. Fig. 6.28 illustrates the compatibility conditions using Mohr's
circle.
240
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
f:q y
L 'Y,; 12
""'
r


1 1
l
i
c


x
1
I l-c.x
y
( a)continuum strain
----.
Yf2
x
y
( b) crack slip strain
r
y/2
y-ty
1'.q/2
2
1
1

l

x Ext-
1.
1
I E:.r
( c) total strain
Fig.6.28 Compatibility Condition: (a) Continuum Strain; (b) Crack Slip Strain; and
(c) Total Strain (Taken from Vecchio, 2000 with Modifications)
Fig. 6.28c illustrates that the total strain is composed of continuum strain and
crack slip strain as previously discussed. The continuum strain components in x-y
coordinates can be obtained from the concrete principal strains in the Sc direction (Fig.
241
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.28a). Since the local crack surface is in the Ser direction, the crack slip strain
components in x-y coordinates are thus calculated for this Ser (Fig. 6.28b). This is
slightly different from the original formulation adopted in DSFM where Ser was assumed
to equate to Sc.
Constitutive Relationships
In this proposed model, the author adopts the same form of the constitutive
relationship for concrete and reinforcing bars in both tension and compression as
suggested by Foster and Marti (2003) with some modifications. In Fig. 6.29a, the
constitutive relation for reinforcing bars is illustrated as a tri-linear formulation. Fig.
6.29b illustrates the constitutive models of concrete in the general form showing both
compression and tension including softening and strengthening effects. The Thorenfeldt
base-curve is adopted for concrete in compression where the softening (Fig. 6.29d) and
strengthening factor (Fig. 6.29c) are given in Eqn (6.26a-d).
242
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
i;
y
(a)
e
,.. .
"

a.ccr
,..
avecr - i;cr .. . I
a,fct
_ fct
(b)

LS
, 2
0
1
Co!Djlresslon SotenJng Rela!loosbip: no reomntatlon ofprtnclpill stress angle
<JS

(c)
I
0 40
I
I
experiment
--

--cs-0.41 ""9 (Min Error)


(d)
Fig. 6.29 Material Models: (a) Reinforcing Bars; (b) Concrete; (c) Biaxial
Compression Envelope (Taken from Vecchio, 1992); and (d) New Compression
Softening Coefficient (Cs)
The author proposes to slightly modify the parameter (Cs) based on the regression
analysis of the shear panel test data where no crack rotation exists e.g. PV23, 24, 25, 27,
and 28 as illustrated in Fig. 6.29d. The original softening factor proposed by Vecchio
(2000) seems to slightly overestimate the parameter.
243
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
For softening,
fJ = 1
1 +Cs Cd
(6-26a)
Cd = 0.35(- &cl - 0.28)
08
(6-26b)
&c2
cs = 0.41779 from regression analysis (6-26c)
For strengthening under bi - axial compression,
fJ = 1+0.92(- !ct)- 0.76(- !ct )
2
in which i = 1,2
fc fc
(6- 26d)
For the constitutive model of concrete in tension, the author selected to adopt the
tension softening relation based on fracture mechanics concept for simulating the micro
cracking process for crack propagation and controlling the mesh dependency effect as
suggested by Cervenka and Pukl (1992). The tension softening parameters used with tri-
linear relation (e.g. Petersson, 1981) is thus given in Eqn (6-27a to c) below.
a
1
= 1/3
2
az =-a3 +a1
9
l 8EcG I
a3 = 2
51 cJct
(6-27a)
(6- 27b)
(6-27c)
where Ee is the modulus of the concrete, Gris fracture energy of the concrete, fct
is tensile strength of the concrete, and lch is the element length limiters defined as
functions of element dimension and crack orientation (Cenvenka and Pukl, 1992).
The author performed a parametric study to illustrate that the mesh dependent
effect can be more pronounced in D-regions; the results of which are shown in Fig. 6.30.
The structures adopted in this parametric study are composed of three types: a simple
shear panel (Type I), shear panel with discontinuous loading and boundary conditions
244
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(Type II), and shear panel with geometrical discontinuity (Type III). Those structures
were analyzed using the NLFEA frameworks (both MCFT and DSFM) with various
mesh refinement level from very fine to very coarse mesh. From the study, it was found
that using different mesh refinement levels have the least effect on the shear strength
prediction of the simple shear panel (Type I) where no geometrical, loading, and
boundary discontinuity exists. In contrast, when the level of the discontinuity in either
loading or boundary is increased (Type II), the effect of mesh refinement on the predicted
shear strength is more pronounced. In addition, the effect of mesh refinement is most
pronounced when the geometrical discontinuity exists in the structure e.g. Type III. In
this case, the predicted shear strength using a very fine mesh is shown to be 14 times
lower than the predicted shear strength using a very coarse mesh. This is thus why the
element length limiters are required in the proposed model.
For the constitutive model of concrete between cracks under tension ( <rcmx and
<rcmy) which is for consideration of bond-slip effect in the average sense (e.g. tension
stiffening), the author follows the proposed formulation by Foster and Marti (2003) who
derived the formulation based on tension chord models. The formulations are given
below in Eqn (6-28a to d).
245
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
16
14
12
10
-

l;IJ
l;IJ
q:
=
t!6
>
4
2
0
0 5
MeshSensiti.Yity ......

Type I Type II
Ref"memeJ( (Li\gg)
--+--TypeI(MCFT)
----TypeI(DSFM)
__._ Typell(MCF D
--Typell(DF S.M)
-*-TypeIII(MCFT)
--.-T eIII(DSFM)
Type ill
20
Fig.6.30 Mesh Dependent Effect in D-Regions
25
In addition, since tension stiffening is derived from the tension chord model and
not treated as the average tensile stress of concrete as usually adopted in MCFT and
DSFM, this thus separates tension stiffening from tension softening which is the
preferred approach of Foster and Marti, 2003.
Tb;A;fct
(J'cmi = ,l=X,y
'bo12
(6-28a)
8m1Esdbi
rb1 = 'hm
2s
(6-28b)
A.= s
I e
S rm10 Sln er
(6-28c)
(6-28d)
where 't"bi, 't"boi is the bond stress and strength in the i direction, Emi is the average
strain in the concrete in the i direction, dbi is the diameter of reinforcing bar in the i
246
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
direction, s is the crack spacing, Ssmio is the crack spacing for uniaxial tension in the i
direction and A-i is the tension stiffening factors in the i direction as originally proposed
by Foster and Marti (2003) with the modification by the author to replace the principal
concrete stress direction (8c) with the crack angle (Ser).
For the constitutive models of the aggregate interlock adopted in the equilibrium
equation along the crack surface (Eqn6-23a), the author selected to use the formulation
by Okamura and Maekawa (1991) as given below.
n ffc
r = ----------
agg O+l (0.31+24w/(a+16))
52
and n = s
(0.5625w
2
)
(6-29)
where "w" and "a" are crack width and maximum aggregate size, respectively.
In Eqn (6-29), the crack shear component contributed by aggregate interlock was
proposed to be in terms of the compressive cylinder strength of the concrete (fc), crack
width (w), maximum aggregate size (a), and magnitude of the slip (8
5
). The formulation
was adopted in the implementation and validation of the DSFM by Vecchio (2000) and
the crack shear slip formulation by Vecchio and Lai (2004), and this results in an
acceptable level of accuracy and the simplicity when used in an FE formulation and
implementation.
The constitutive model of dowel stress is theoretically based on the model
formulated for the individual bar by Walraven (1980) and the author modified it for use
in his smeared reinforcement modeling approach. The modification is presented below.
247
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
w
cosecr
smear r-'
reinforcement
~
sin8cr
Fig. 6.31 Components of Stress and Motion along Cracks
For reinforcing bars orthogonally embedded through a unit area of the inclined
crack surface as shown in Fig. 6.31, the author derived a relationship between the number
(ni) of discrete bars and reinforcing bar ratio (pi) in the i direction as shown in Eqn (7-
30a) below.
3/3
3
EI o
where F . = i s1 s1 s i = x y
d1 2 3 ' '
3 + 6 /JJ1 + 6(/JJ;) + 2(/31 J; )
Ed Emi
4
fJ
= ( I b1 )114 E I . = SI b1
I 4 E I ' IS SI 64
~ SI
r = d . cot(B ) + (Iseri - f,; ) d .
J 1 b1 er
45
b1
~ = 0.2
w+0.2
248
(6-30a)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The dowel smeared stress ( 'tctt) can be calculated from the summation of the
tangential components ( crtctx and crtcty) of dowel stress along the unit area of the crack
surface. By assuming that the tangential components of the dowel stress in each
individual bar are equal, the summation of the tangential components can then be
calculated as the product of the number of bars (ni) and the tangential components as in
Eqn(7-30b) where F di is the dowel force of a discrete reinforcing bar proposed by
Walraven(l 980), Er is the foundation modulus of concrete which is independent of bar
diameter (dbi) and led to the empirical formulation as 34(f c)
05
os-
0

85
, fi is the free length
of bar and proposed to be a function of bar diameter, crack angle, and the difference
between stress at crack and between cracks, and s is the reduction factor due to crack
opening ( w).
Crack Spacing (s)
The crack spacing equation as generally adopted in MCFT and DSFM was
theoretically derived by assuming that the crack angle orients in the principal stress
direction e.g. ~ = 0. Since the proposed formulation allows the deviation of crack angle
from the principal stress direction, the author thus needs to derive the more refined
version of crack spacing equation. Previously, Foster and Marti (2003) adopted tension
chord model approach (TCM) as illustrated Fig. 6.31 b to alternatively derive the original
cracked spacing equation of Vecchio and Collins (1980) where no ~ exists; however, in
this proposal, the author proposed to use the TCM and Fig. 6.31b for deriving a more
refined crack spacing equation considering ~ by setting the equilibrium equation based
on the free body diagram of the triangular element as shown in Eqn (6-3 la) below.
249
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8
2
8
2 {'cm
(J"cm.x COS c + (]'cmy Sln c = J cl
h1cm :<::;fc,
acm.x :-::; A.xfc1 and acmy :-::; A.yfc1
(6-3 la)
(6-3 lb)
(6-3 lc)
(6-3 ld)
Since the summation of the maximum tensile stress of concrete between cracks
must be less than fct as given in Eqn (6-3 la and b) and the maximum tensile stress of
concrete between cracks can be derived based on the tension chord model as in Eqn ( 6-
28c) and (6-31c), the author can thus derive the refined version of crack spacing (s) in
Eqn (6-31d). This refined crack spacing formulation will be adopted in the proposed
model. In addition, as expected, when no ~ exists, Eqn( 6-31 d) becomes the crack
spacing equation proposed by Vecchio and Collins (1986).
6.3.3 FE Procedure and Nonlinear Solving Scheme (Element Level)
The FE Procedure of the proposed FE formulation in the element level will be
described in this section. The nonlinear solution scheme extends from the approach by
Vecchio (2000). This will be presented in a flowchart that is used to implement a small
nonlinear FE code using MATLAB to demonstrate the initial numerical results for this
research.
The procedure is initially described by considering the state of stress at a point in
a reinforced concrete element in matrix form as shown below in Eqn(6-32a) for the
conditions that stress and total strain at a point must be satisfied. It should be noted that
Eqns. (6-32a to c) do not include the elastic or plastic strain offset since the author only
considers only continuum and slip strains.
250
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[a] =[D][]-[aJ
[] =[c]+[s]
[aJ=[Dc][s]
(6-32a)
(6-32b)
(6-32.c)
When implementing the procedure, [D] is obtained from the secant stiffness
matrix of concrete [De]' and reinforcement [Ds]'. Both [De]' and [Ds]' are in the principal
stress direction of concrete and in the orientation of reinforcing bar direction,
respectively, and thus are required to be transformed into the x-y direction by using a
transformation matrix prior to combining into [D]. [De]' can be calculated by using Eqn
(6-33a to d) for uncracked and cracked states. [Ds]' can be simply obtained by using Eqn
(6-35a to b).
(6-317)
(6-3l'J)
(6-3x)
(6-31i)
(6-34)
where E
1
and E2 are the elastic modulus of concrete, and Ee1 and Ee2 are the secant
modulus of concrete (Fig. 6.13b) under tensile and compressive stress, respectively.
251
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In this research, and since the author separates the tension stiffening and tension
softening, to calculate Ee
1
which is "average" tensile contribution of concrete, the author
calculates Eel from Eqn (6-34) as shown below by combining the effects of both tension
stiffening (fe1 cm) and tension softening (fe1
5
) together. fc1 cm is obtained from the local
equilibrium equation of concrete between cracks as given in Eqn (6-31a) and fe
1
5
are
obtained from the tension softening curve.
r
P;E,;
0
0
E.=fs,
SI
&,;

0 0
(6-3:U)
(6-35')
(6-3x)
where E
5
j, fsh and Ei are the average strain, average stress of the reinforcement, and the
total strain component in the i
1
h direction (e.g. x or y), respectively.
Subsequently, [De]' and [D
5
]' are transformed and summed to obtain [D] by using
Eqn(6-36a to c).
[D]
[DJ
[D,);
(6-3fu)
(6-3fh)
(6-3fr)
Based on the proposed concept and formulations described previously, the author
proposes this more comprehensive approach, as illustrated in the flowchart in Fig. 6.32 to
solve for the nonlinear response of a reinforced concrete element.
252
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Determine the prestress vector a
0
Solve for the total slrain vector E=[ex ey ~ y ]
from D
1
(a+ao)
Yes
Fig.6.32 Flowchart to Solve for the Nonlinear Response of an Element
From Fig. 6.32, for a level of applied stresses, the nonlinear analysis for the
element level is initiated by assuming the pre-strain components (can be firstly set to zero
for the first iteration). Then the modulus matrix Dis calculated. For the first iteration, the
D matrix can be calculated from linearly elastic analysis. For any subsequent iteration,
the D matrix can be calculated from the combination of both concrete and reinforcement
parts as given in Eqn 6-36a. Then the pre-stress vector [ cro] can be calculated from Eqn
6-32c and the total strain vector can be simply obtained using Eqn 6-32a. The total strain
vector can be used to check for convergence. When the subsequent iteration is required,
the total strain vector can be used to determine the concrete (Eqn 6-32b) and
reinforcement strains (Eqn 6-35c) components which are then used to update the secant
stiffness for calculating the D matrix. In addition, the equilibrium equation at the local
253
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
crack will be nonlinearly solved for the new value of the slip (Eqn 6-23a) by using any
nonlinear solving scheme such as the Newton-Raphson method. Then the new value of
the slip will be used to update the pre-strain component and the process is repeated until
convergence.
6.3.4 FE Procedure and Nonlinear Solving Scheme (Structural Level)
The FE Procedure of the proposed FE formulation in the structural level will be
described in this section. The proposed formulation in the element level in section 6.3.3
will be extended into the structural level in the manner illustrated in the flow chart in Fig.
6.33. This approach uses the strain offset approach as suggested by Vecchio (1990). The
nonlinear solving scheme in the structural level could be implemented numerically based
on standard nonlinear solving scheme; for examples, the iterative secant stiffness
approach where either load or displacement control is possible in the numerical scheme;
For example, the load control numerical scheme is illustrated in Fig. 6.33.
For the load control scheme, the procedure is initially described by inputting all
structural geometry, material properties, load and boundary conditions. Then, the applied
external load (F) at a specific load level is applied to the structure. Next, the computation
loop over all elements in the structural domain is performed. In each element, the
computation in the element level as illustrated in Fig. 6.33 is adopted. This initiates by
consideration of the prestrain components (Ee and E
5
) from the previous iteration. For the
first iteration, the prestrain components in all elements are initially assumed to be zero
when no initial strain exists. Then the modulus matrix D is calculated. For the first
iteration, the D matrix can be calculated from linearly elastic analysis. For any
subsequent iteration, the D matrix can be calculated from the combination of both
254
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
concrete and reinforcement parts as given in Eqn 6-36a. The completed D matrix is
adopted to calculate the element stiffness matrix [ke] based on the standard FE
formulation given in Eqn 6-37. Then the pre-stress vector [cr
0
] calculated from Eqn 6-32c
is adopted to determine the prestrainjoint force [f
0
] as given in Eqn 6-38. Subsequently,
both element stiffness matrix [ke] and the prestrainjoint force [f
0
] are assembled to the
structural level. The nodal displacement vector [ d] of the structure can be solved from a
simple linear equation in Eqn 6-39. Next the convergence of the solution is checked by
adopting force, displacement, or energy residual approach. Throughout the numerical
examples, the author adopts the energy residual approach to check for the convergence.
If the next iteration is required, the current nodal displacement vector is adopted
to calculate the strain and stress components in the element level. This then leads to the
calculation of the new values of the prestrain component (Es) from the slip (8s) along
crack surface which is solved by using a small Newton Raphson scheme.
(6-37)
where [D] is the element modulus matrix as given in Eqn 6-36a and [B] is the first
derivative of the shape function of the concrete FE.
(6-38)
n
(6-39)
255
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Input structural and material properties
External joint force . F


1
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
"-----------------------r-----------------------
Assemble ke and fo to structural level (Kand Fo)
Done, go to the next load step
*Note: determine slip (8) and slip
strain (8s) from the element level
flowchart
Fig. 6.33 Flowchart to Solve for the Nonlinear Response (Structural Level)
6.4 Numerical Examples
In this section, the author presents the numerical examples to demonstrate the
accuracy and the validity of the proposed FE formulation. Four numerical examples are
given in this section: 1) Shear Panels; 2) Simply Supported Deep Beams with an
Opening; 3) Simply Supported Deep Beams with a Dapped End and an Opening; and 4)
Simply Supported Deep Beams with Openings.
256
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.4.1 Shear Panels
PV 19 (Pure Shear)
890
890
Fig.6.34 Shear Panel subjected to General Loading
PV 19 selected from PV series tested by Vecchio and Collins (1982) is adopted as
the first example in this section. In this analysis, the effect of L18 is not considered. The
shear panel illustrated in Fig. 6.34 is subjected to pure shear ('r:crx:cry=l :0:0). The panel
dimensions are 890 mm x 890 mm x 70mm. The material properties are: fc=19 MPa,
Eo=2.15 mE, fxy=299 MPa, fyy=299 MPa, px=l.785 %, py=0.713%, and Gf=75 Nim.
The analysis results are illustrated in Fig. 6.35-40.
257
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
4
35
3
1.5
05
0
~
a..
:a:
..
"'
!:
"'
ii
"'
.c
"'
/
//
45
4
35
3
2.5
2
1.5
05
/
,
)'
/
/
0 002 0004 0006
shear strain(mmfmm)
0008
-proposed
-exp
-+-MCFT
- DSFM(Hybrid')
- DSFM(Walraven
Stresses)
__.,_ DSFM(5 lag)
0 01
Fig. 6.35 Predicted Shear and Shear Strain of PVl 9
r
/
//
/
;
/
/
__ .PredictecLSllear.stren .an.d Jo.git.Udinal.tensile.strain of PV1 s ~ .

-t-exp
-+-MCFT
-proposed
- DSFM(Hybnd
Vecchio-Lai)
0 012
0 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
0 0.5 1.5 2 2.5
ex(mmfm)
Fig. 6.36 Predicted Shear and Longitudinal Strain of PV19
258
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0.8
06
4
3_5
3 /
.y
I
2.5 I
I
I
2 I
1.5
0 .. 5
0 2
Shear stress and transver.ttJensile.strainof.PY1L ______ ----- --i
-=------=-=-- -+ I

3 4 5 6
ey(mmlm)
-+--exp
-+-MCfT
_,.._ proposed
-osFM
Hybrid(Vecchio-Lai)
7 8 9
Fig. 6.37 Predicted Shear and Transverse Strain of PV19
Prlnc;lpal Compruslve StressS1rllln-PV19
06 08
ec2tecO(mmlmm)
1 2
-proposed
-Exp
-.-MCFT
-ti- DSFM(HYbnd
Vecchio-Lai)
1 4 16
Fig. 6.38 Predicted Principal Compressive Stress and Strain of PV19
259
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
--proposed
-+-Exp
60 -+-MCFT
--OSFM(Hybnd)
-.- OSFM{Waraven stress)
55 -+- OSfM\5 degree lag)
50
4 0 - - ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
0 0.5 1 5 2 2.5 3 3.5 4 45
snear(MPa)
Fig. 6.39 Predicted Shear and Principal Stress Orientation of PV19
Based on an examination of the predictions of the responses of PV19, the
following observations can be made about the effectiveness of the proposed formulation
relative to that of the VecTor2 implementations of the MCFT and DSFM. The proposed
formulation is shown to fairly well predict the nonlinear experimental responses of this
panel, with a slight improvement over the MCFT and DSFM in predicting the capacity
and the deformation at the failure of the panel. However, the proposed formulation was
less successful than the DSFM in predicting the reorientation of principal stress and strain
of the model. This is considered in large part to be because the DSFM was specifically
calibrated using the results of the PV (and PB) series of tests (Vecchio, 2000) whereas the
parameters of the tension chord model, as adopted in solving the slip strain in Eqn 6-23a
of the proposed model, were not calibrated with the data from the Vecchio panel tests.
The calibration of the parameters of the slippage along the crack surface can be found in
260
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Vecchio and Lai (2004). The overall effectiveness of the proposed formulation at
predicting the response of a large number of panel tests is presented in appendix C.
65
_______ P_ .. cte.d.Re.spons.es ..of.PV1S,,J"rincipal.Stra.in.OrtantatiQn
-Proposed
-+-Exp
60
-MCFT
- DSFM(hybnd)
55
_._ DSFM(Waraven Stress)
.,
-'I'- OSFM(5 lag)
-<(
I ,/'
I _____________ ._.

I I
50
45 ............. - ./
/
/.
/
40+----r----,----r-----i----r------.---..-----.----1
0 05 1 5 2 25 3 3.5 4 45
shear(MPa)
Fig. 6.40 Predicted Shear and Principal Strain Orientation of PVl 9
PV and PB Series
In this section, the proposed FE formulation in the element level is adopted to
predict the structural responses of the 3 8 selected shear panels from the PV and PB series
tested by Vecchio and Collins (1982) and Bhide and Collins (1989), respectively. In the
PV series, the shear panels were designed and constructed in the way that the distributed
reinforcement in both x and y direction were fabricated and the main load configurations
of the shear panels are pure shear and combinations of either axial or biaxial compression
and shear. The significant structural responses and failure modes of those panels involve
crushing of the concrete as well as a sudden and explosive failure with a sliding shear.
For the PB series, the shear panels were designed to be reinforced in only one direction
261
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and then mainly loaded in combinations of tension and shear. The important structural
responses and mode of failure of those panels involve cracking and concrete shear failure.
The orientation of the crack direction was substantially occurred in the test series. Based
on the overall structural responses and failure modes, those PV and PB are thus
significantly different in the structural responses and thus are suitable in the
computational validation of the proposed FE formulation.
The accuracy and validity of the proposed formulation in predicting the structural
capacity and cracking loads of all panels are illustrated in Figs.6.41 and 6.42,
respectively. The completed details of the structural predictions in all selected PV shear
panels are given in Appendix C.
In Fig. 6.41, the plot of the ratio of the predicted to the measured shear strength
of both the PV and PB panel series is given. The average ratio of the predicted to the
measured shear strength for PV series is 1.056 with the coefficient of variation(%) of
8.63. For PB series, the average ratio of the predicted to the measures shear strength is
0.956 with the coefficient of variation(%) of 4.73. The average ratio of all panels
including PV and PB panels is 1.007 with the coefficient of variation(%) of 7.21. Based
on these ratio values, the proposed formulation is observed to slightly overestimate the
shear strength of the panels in PV series. A slight underestimation of the shear strength is
obtained in the PB series. In addition, the strength prediction in both shear panels of the
proposed formulation is in the acceptable level with a quite low uncertainty measured by
the coefficient of variation.
For the prediction of the cracking loads, Fig. 6.42 illustrates the ratio of the
predicted to measured cracking strengths. The average ratio of the predicted to the
262
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
measured shear strength for PV series is 0.822 with the coefficient of variation(%) of
16.86. For the PB series, the average ratio of the predicted to the measures shear strength
is 0. 762 with the coefficient of variation (%) of 17. 72. The average ratio of all panels
including PV and PB panels is 0. 787 with the coefficient of variation (%) of 17.2. Based
on these ratio values, the proposed formulation is shown to underestimate of the cracking
loads of the panels in both the PV and PB series. In addition, a considerable uncertainty is
evident in the coefficient of variation. The inaccuracy of the proposed formulation in
predicting cracking loads of the shear panel is not surprising since the formulation adopts
the empirical formulation (Eqn 6-40) in calculating the tensile strength (cracking stress)
of concrete under tension.
fct = 3 3 ~ f; (MP a) (6 - 40)
263
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.4
1.2
c: 1
><
w
-
"go.a
t)
"O

a.
-:J
>0.4
0.2
0


D
0 0

Proposed Model-PV
o Proposed Model-PB
D


D + 0 + +
+ D D
D
PV series Mean=1.056, COV(%)=8.63
PB series Mean=0.956, COV(%)=4.73
All Mean=1.007, COV(%)=7.21
1) Pure Shear (non-marking)
2) Shear and Bi-axial Compression
3) Shear and Bl-axlal tension
4) Shear and axial tension (PB series)
(reinforcement in one direction)
Fig. 6.41 Predicted Shear Strength of the Selected Panels in PV and PB
Series using the Proposed FE Formulation in the Element Level.
1.4
1.2
-a. 1
><
w
-
"O

-
"O

a.
1:"

0.2
0

D +




Proposed Model-PV
o Proposed Model-PB
D
PV series mean=0.822, COV(%)=16.86
PB series mean=0.762, COV(%)=17.72
All mean=0.787, COV(%)=17.183
D

D
D

D
+ D +
2

Note: o
2

1) Pure Shear (non-marking)
2) Shear and Bi-axial Compression
3)Shear and Bi-axial tension
4 )Shear and axial tension (PB series)
(reinforcement in one direction) ;
Fig. 6.42 Predicted First Shear Crack of the Selected Panels in PV and PB
Series using the Proposed FE Formulation in the Element Level.
264
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.4.2 Simply Support Deep Beams with an Opening
In this section, the proposed FE formulation in the structural level is adopted to
predict the structural responses of the 3 selected simply supported deep beams with an
opening (specimen 4A, B, and C) in which the structural geometry and reinforcement
details are illustrated in Figs. 3.la and 3.8, respectively. The details of FE models of the
deep beams are illustrated in Fig. 6.43.
Figs. 6.43a to c illustrate the FE models of specimens 4A, B, and C which were
discretized into the same number of small finite elements by using 556 quadrilateral 4
nodes elements, respectively. To display the FE models and all solutions of the analysis,
the author adopts Augustus which is the post-processing tool for Vector2. By generating
the appropriate output format and importing those output into Augustus, the
comprehensive presentation of the FE models and solutions could be obtained. Addition
information on using Augustus as the post-processing tool is provided in Appendix A.
In all specimens, the main reinforcement was modeled by using the proposed
embedded reinforcement including bond stress-slip. The local bond-slip relation employs
the model from CEB-FIP model code for monotonic loading and unconfined concrete
type. Good bond conditions are assumed in the model. Material properties for concrete
and reinforcement adopted in the FE models are obtained from Tables 3.2 and 3.3,
respectively.
The analysis results of all simply supported deep beams with an opening are
illustrated in Figs. 6.44 to 6.48.
Fig. 6.44 presents a comparison between the predicted and measured load
displacement for specimens 4A to 4C. In the figure, the predicted load displacement
265
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
obtained from the proposed FE models agree well with the measured load displacement.
The responses predicted from the proposed FE models can capture the pre and post-
cracking stiffness of the test specimens quite accuracy, although the models slightly
underestimate the capacity of the specimens.
The predicted mode of failures, crack pattern at failure condition, and the ratio of
the principal compressive stress demand to the compressive capacity of all specimens are
illustrated in Figs. 6.45 to 6.47.
266
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(b)
(c)
Fig. 6.43 FE Models of Specimen 4A, B, and C: (a) FE Model for Specimen
4A; (b) FE Model for Specimen 4B; and (c) FE Model for Specimen 4C.
267
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
350
300
250
-
z
,,_
.::.:: 200
-4A (Exp)
-
'tS
--48 (Exp)

I
0 150
]
...I ------ 4C (Exp)
'
100
\
-4A (FEA)
\
- 48 (FEA)
50
'
4C (FEA)
'
------------
0
0 2 4 6 8
Displacement (mm)
Fig. 6.44 Predicted Load Displacement for Specimens 4A, B, and C
In the overall prediction, the predicted failure condition and crack pattern at
failure mode in all specimens agree well with the experiment as illustrated in Fig. 4.2 in
chapter 4. The proposed FE model predicted the failure modes of all specimens occurred
by crushing of the concrete strut just above the opening. This is evident by the ratio of the
principal compressive stress demand to the compressive capacity which is quite close to
unity. In addition, the severe crack condition in the region above and adjacent to the
opening is also captured by the model at the failure load.
268
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Specimen 4A
fem
Notllf: Looded
-to1r64
-lot)!;i'
-10011
-tl.)(114
(a)
Combined View
I /,','/// '// \ \' ,\',>'\"\._._ __
..\-I- - . I I . / i / i / / \ \ \ \ ' \ ', \ \ . _1-L:
' / / f / / \ \ \ . .I.. _;_._.:....i,. _ _l_
(b)
Fig. 6.45 Predicted Failure Mode of Specimen 4A: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern
269
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
t o o . ~ t o o ~ e
-to007 -to on
-to011 -lo0.26
-tt>DIS -to019
Specimen 48
-toO)l
-to0)7
-to040
-10044
(a)
fem
Wtoo47
-"'"" -l'Q05S
-toU58
Combined View
(b)
W10062
-to066
-lo06!)
llBm.to0.13
emrooao
lilll!ll to 084
11111Eto088
Fig. 6.46 Predicted Failure Mode of Specimen 4B: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern
270
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

W to 004
-to001
-10011
-t<7!01il
---to 0.18
-!0022
-toOIS
-ton:m
Soecimen 4C
fem
Wtoo.32
_ ... ,.
-ti:i 039
-ti:i041
(a)
W1004r
-too.so
-toOS4
-tc.01)7
Combined View
I 'i \
\\\
I
\ \ I \.
I \I\
(b)
',
-toOS'I
-torJ&'S
11m10000
m1110 on
\
\
20.00
Fig. 6.47 Predicted Failure Mode of Specimen 4C: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern
To investigate of the accuracy of the prediction of the local responses of
the specimens, as illustrated in Figs. 6.48 and 6.49, a comparison between the
271
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
predicted and measured strain in concrete and reinforcement is undertaken and
presented. The observed data from all 4 concrete surface gages and 4
reinforcement strain gages are plotted against the predicted values that correspond
to the location of the gages on the specimens. All of the gages show reasonable
agreement with their corresponding model predictions. Differences between the
predicted and actual experimental behavior may be due to several factors
including changes in the direction of principal compressive strain during loading
and position and orientation of the gages relative to cracks and other local
phenomena. These types of local behaviors can greatly affect comparisons
between strain gage data and model predictions.
350 350
300
300
250 250
z
200
z
200
:. :.
't:J 't:J
"'
150
.9 ...I
Fig. 6.48 Concrete Strain Prediction in Specimen 4A
272
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
350
300
250
200
...
150
..J
100
50
0
350
300
250
200
...
150
..J
100
50
0
350
300
250
200
...
150
..J
ST1(Exp)
FEA
100
50
0
0 0.5 1.5
Strain (mm/m)
350
300
250
! 200
...
150
..J
-ST3(Exp) 100
-ST3 FEA
50
0
0 0.5 1.5 0
Strain (mmlm)
0 0.5
I
!
0.5
-ST2(Exp)
-ST2 FEA
1.5
Strain (mmlm)
I
-ST4(Exp) !
-ST4(FEA)i
1.5
Strain (mm/m)
2
Fig. 6.49 Reinforcement Strain Prediction in Specimen 4A
2
2.5
6.4.3 Simply Support Deep Beams with a Dapped End and an Opening
In this section, the proposed FE formulation is adopted to predict the structural
responses of the 3 selected simply supported deep beams with a dapped end and an
opening (specimen 6A, C, and D) in which the structural geometry and reinforcement
details are illustrated in Figs. 3.lc and 3.8, respectively. The details of FE models of the
deep beams are illustrated in Fig. 6.50.
Figs. 6.SOa to c present the FE models of specimens 6A, C, and D. FE models of
those specimens were produced using 198 quadrilateral 4 nodes elements, respectively.
The main reinforcement was modeled by using the proposed embedded reinforcement
including bond stress-slip in all specimens. In these examples, the benefit of using the
embedded reinforcement is distinctively presented. The non orthogonal reinforcement
obtained from STM models of the specimens which causes the difficulty in modeling of
273
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the structures can now be modeled very easily and practically by using embedded
reinforcement. The embedded reinforcement, which is independent to the structural
mesh, requires only the input of the starting and ending points of the end nodes within the
structural domain. Thus, the same structural mesh could be used for all specimens and
only the orientation and the numbers of the embedded reinforcement elements are
required to change in each model. This practicability of the embedded reinforcement
leads to the huge reduction of the time required in the modeling process.
The local bond-slip relation employs the model from CEB-FIP model code for
monotonic loading and unconfined concrete type. The good bond conditions are assumed
in the model. Material properties for concrete and reinforcement are obtained from Table
3.2 and 3.3, respectively.
Figs. 6.51 to 6.57 illustrate the analysis results of all simply supported deep
beams with a <lapped end and an opening. Fig. 6.51 illustrates the plots of the predicted
load displacement for specimens 6A, 6C, and 6D. In the figure, the predicted capacity of
the models correlates well with the observed capacity. The models slightly underestimate
the capacity of both specimen 6A and 6C and slightly overestimate the capacity of
specimen 6D.
The predicted mode of failures, crack pattern at failure condition, and the ratio of
the principal compressive stress demand to the compressive capacity of all specimens are
illustrated in Figs. 6.52 to 6.54. The proposed FE model predicted that the failure
specimens 6A, 6C, and 6D occurred by crushing of the concrete strut just underneath the
loading point and above the support adjacent to the <lapped end. In addition, the severe
crack condition in the region above and adjacent to the <lapped end is also captured well
274
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
by the model at the failure load. The predicted failure condition and crack pattern at
failure mode in most of the specimens except specimen 6D agree well with the
experiment as illustrated in Fig. 4.2c. In the experiment, the failure mode of specimen 6D
occurred due to crushing of the concrete adjacent to the opening; however, no prediction
of concrete crushing in the region could be obtained and thus the ratio of the principal
compressive stress demand to the compressive capacity in the region predicted from the
model is approximately 0.70. This thus indicates that the model underestimates the
demand in this region which might be due to the complexity of the stress redistribution in
the region around the opening, the imperfection of the concrete in the region, and the
influence of the cracks.
As illustrated in Figs. 6.55 and 6.57, the comparison between the predicted and
measured strain in concrete and reinforcement is undertaken and presented. The observed
data from the selected concrete surface gages and reinforcement strain gages in
specimens 6A, 6C, and 6D are plotted against the predicted values that correspond to the
location of the gages on the specimens. Most of the reinforcement gages show reasonable
agreement with their corresponding model predictions except specimen 6D. In this
specimen, less accuracy of the predicted strain values from the model was obtained in the
regions where high complexity of the stress redistribution and crack characteristics exist;
for example, the regions above the opening of specimen 6D. The regions above the
opening were provided only the light distributed reinforcement which causes the region
to behave quite differently from the case where high percentages of reinforcement were
provided. As reported in DSFM paper by Vecchio (2000), the behavior of the lightly
reinforced shear panel involves crack reorientation and slippage along crack surface.
275
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
These cracks that dominate the behavior of shear panel are not distributed, but discrete.
Thus the reorientation and slippage along those cracks cause even more complexity in the
stress and strain prediction. In addition, the crack characteristics of light reinforced panels
are less accurately predicted by using the smeared crack model. Thus, in this region, the
predicted strain in both reinforcement and concrete is quite different from the observed
value. The predicted strain in gage SG9a from the model is overestimated while the strain
in gage CG2 is underestimated. This leads to the inaccuracy of the load transfer
characteristics around the opening such that the less force demand was predicted to be
transferred adjacent to the opening and thus this results in the inaccuracy in the prediction
of the mode of failure of the specimen as previously discussed.
276
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
(C)
Fig. 6.50 FE Models of Specimen 6A, C, and D: (a) FE Model for Specimen
6A; (b) FE Model for Specimen 6C; and (c) FE Model for Specimen 6D.
277
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1200
1000
-
z 800
.::.::
-
i 600
0
...J
400
200
.......................................... ;;;;1
.. .. .................................................................................................................. .
... .................. .
--6C
0
0 1 2 3 4 5
Displacement (mm)
Fig. 6.51 Predicted Load Displacement for Specimens 6A, C, and D
278
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
to .24
-10 0'9
-to034

fJif ! !
_!
/ '
I
/ ;/
/11 / .
I

---
,/
__ / //
_}_L / /
I
/
I < ,___

_'-1.___J -- - -
/
/
T
!
I
I
//
/
/
/
I
I
/
Specimen 6A
fem
-tf.J046
-to 053
-to0.S8
(a)
to oa2
-1<>0.81
-10091
mlllli:tJSIG
Combined View
rrn-
-
___,-
\
\
\
\
l
\
// \
----
1--

J \
I
\
\
' \
\
I
\
!
I \
\\ 1. ,,,,
I _!
I I
*':
I
:
/
I
/
I
I_
' \.
\
I \
\
\
'
I
;-- I
i
\
\ \
!
I
l \
\
-,_.
.......----
fCll'l LoMF$cior = 4 4D
to I 01
-10106
ll!l'lli to 1 11
111111 to 1.1(:.

---
-
\
-
i
\
I
,_
-
\
\
\
\
I
I
I
-
i
I
/ I /
I
1-:-f \
\
I
/
L
i !
/
\
I /
/ i

\
I
I ---
I
I
/ I
I I
}
\
I
,/
/
i
I
/
'
!
i
,_____..___
--
(b)
Fig. 6.52 Predicted Failure Mode of Specimen 6A: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern
279
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Specimen 6C
fem
o m b i ~ ~ J View
(b)
s. tcm LO!tt1 Factor"' 4 60
to 0.95
to 1 ()J
to 1.CM
lo 1 09
Fig. 6.53 Predicted Failure Mode of Specimen 6C: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern
280
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Specimen 60
fem
Combined View
(b)
s; !em 440
to ..o-sr-------
to 096
to 1 01
to 1.os
Fig. 6.54 Predicted Failure Mode of Specimen 6D: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern
281
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0
1200
1000
800 b
"'
0.
600
400
200
1400
..... )1 .... I/
. ' ,
', /
/ .. /
/
/ /
1200
1000
800 r
g
0.
600"'
3
- SG71a
SG71b(Exp)
SG71b (FEA) 400
- SG71a FEA
200
0 0.0005 0.001 0 0015 0.003 0.0035
CG15(Exp)
-CG21(Exp)
CG15(FEA)
CG21 FEA)
1200
1000
800 b
"'
0.

400
200
-------------------! 0
-0 005 -0 004 -0.003 O 001 0
...----------------, 1400
1200
1000
800 r
g
0.
600"'
3
200

0 00002 00008
Fig. 6.55 Concrete and Reinforcement Strain Prediction in Specimen 6A
282
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
f
. cG?(E-;;P)l
-CG6(Exp) I
... CG?(FEA)I

-0001 -0.0008 -0.0006 -0.0004 -0.0002
Strain (mm/mm)
CG23 (Exp) i
-CG19 (Exp) J
i CG23 (FEA) I
I . CG19 (FEA) i
. '
-0.002 -0 0015 -0 001
Strain (mm/mm)
1200
1000 '
z 800
:.

400
200
-0.0005
1200
0
1200
0
0
z
:.
,,
"'
.s
CG21 (Exp)
-CG20(Exp)
CG20 (FEA)
CG21 FEA
1200
1000
800
-0 0005 -0 0004 -0 0003 -0 0002 -0 0001
Strain (mm/mm)
0
1200
1000
800
600
400
200
0
I
SG1Ab (FEA)I

0 0.0005
- '
0.001
Strain (mm/mm)
SG5c(Exp)
!-SG9b (Exp)!
, SG5c (FEA) i
l:. :
0.0015 0.002
0.0005 0.001 0.0015 0.002 0.0025 0.003
Strain (mm/mm)
Fig. 6.56 Concrete and Reinforcement Strain Prediction in Specimen 6C
283
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
z

l!
0
..J
1200
1000
800
z

"Cl 600
.,,
.s
400
200
O
0
1200
1000
800
600
400
200
0
0

.,,
.,,
0
..J
r-
-0.001
.
0.0005
.
.
0.001
1200
1000
..
ii
800 i;
600
400
200
. 0 .--
0
.... , .
.
SG1 (Exp)
-SG6b{Exp)
SG1 {FEA)
SG6b FEA
1=-cci1.3(E;;;;;1
'-CG12 (Exp) I
. CG13 (FEAJ!
i -CG12 (FEA)i
1200 -
1000
0.001
Strain (mm/mm)
00015 0.002 -00012 0.001 -00008 -0.0006 -0.0004 -00002 0
0.002
Strain (mm/mm)
Strain (mm/mm)
.........
-SG6c {Exp) I
-SG3{Exp) I
SG6c (FEA) i
:
0.003 0.004 -0.0005 -0.0004 -0.0003 -0.0002 -0 0001
Strain (mm/mm)
1:..-=sGea (Exp) I
[-SG5b (Exp) I
f SG9a (FEA) f
I: ..: ....: {f'i;t11J
f= cci4-(Exp)
i-CG2(Exp)
, CG4{FEA)
f- CG2 FEA
..... " ........................r --"-.............,................................... - .......,
0.001 0.002 0.003
Strain (mmlmm)
0004
-0002 -0.0015 -0.001
Strain (mm/mm)
-0 0005
1200
1000
800
0
1200
1000
0
Fig. 6.57 Concrete and Reinforcement Strain Prediction in Specimen 6D
284
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.4.4 Simply Support Deep Beams with Openings
In this section, the proposed FE formulation is adopted to predict the structural
responses of the 3 simply supported deep beams with openings (specimen 8A, B, and C)
in which the structural geometry and reinforcement details are illustrated in Figs. 3.ld
and 3.8, respectively. The details of FE models of the deep beams are illustrated in Fig.
6.58.
Figs. 6.58a to c illustrates the FE models of specimens 8A, B, and C. FE models
of those specimens were produced by using 342 quadrilateral 4 nodes elements,
respectively. The main reinforcement was modeled by using the proposed embedded
reinforcement including bond stress-slip in all specimens. The local bond-slip relation
employs the model from CEB-FIP model code for monotonic loading and unconfined
concrete type. The good bond conditions are assumed in the model. Material properties
for concrete and reinforcement are obtained from Table 3.2 and 3.3, respectively.
Figs. 6.59 to 6.62 illustrate the analysis results of all simply supported deep
beams with openings. Fig. 6.59 illustrates the plots of the predicted load displacement for
specimens 8A, 8B, and 8C. In the figure, the models predict the load displacement for all
specimens agree well with the measured load displacement from the experiment. The
initial and post cracking stiffness of all specimens is accurately predicted from the
models. The predicted capacity of the models highly correlates well with the observed
capacity although the models slightly underestimate the capacity of all specimens. In
addition, the prediction of ductility of all specimens also agrees well with the measured
ductility from the experiment.
285
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I
I
!
I
I I
- - - - - - --- - - ~ --+--t--+-+-+-- --+--+ f--+--i---.+ t--I
............. - ......... , .., .. -- ......._... -f-+--+-----1;-+++--+
..... t ...
(a)
I r-
r----
i
(b)
Fig. 6.58 FE Models of Specimen 8A, B, and C: (a) FE Model for Specimen
8A; (b) FE Model for Specimen 8B; and (c) FE Model for Specimen 8C.
286
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The ability of the model to predict the mode of failures, crack pattern at failure
condition, and the ratio of the principal compressive stress demand to the compressive
capacity of all specimens is illustrated in Figs. 6.60 to 6.62. The proposed FE model
predicted quite accurately the observed failure modes of all specimens. Model 8A was
predicted to fail by severely cracking and sliding along the crack plane in the regions
between upper and lower openings. The severe cracks in the regions are predicted to
occur due to the influence of the transverse tensile strain induced from the compressive
struts flowing around the openings to the lower support. Those cracks and transverse
tensile strain soften the concrete regions as illustrated by very high ratio of the principal
compressive stress demand to the compressive capacity (Fig. 6.60a). In addition, the
fracture of distributed reinforcement provided in the regions is also captured by the
model. Model 8B and 8C were predicted to fail by crushing of the concrete strut adjacent
to the lower opening which is just above the support (Fig. 6.60a and 6.61a). In addition,
the crack condition just before the failure of the specimens is also captured by the model.
Model 8B was predicted to have just only few cracks develop and propagate in the
regions adjacent to the openings (Fig. 6.61b) while more severe cracks were predicted to
develop and propagate in Model 8C (Fig. 6.62b ).
287
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
450
400
350
.-..
Z300
.::.::
-
"'C 250
cu
0200
..J
150
100
50
0
0
,,
,.' .
,, .. \
, ~ ~
I - I)
, ~ 1'
I
'
,,.
,'
l
-- l
- .. '
.. . -.. ,
........ -:--: .....
- - - B8 (FEA)
-BC (FEA)
-------- B8 (Exp)
-BC (Exp)
-- BA (FEA)
- -BA (Exp)
2 4
Displacement (mm)
6
Fig. 6.59 Predicted Load Displacement for Specimens SA, B, and C
288
B
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Specimen 8A
fem
(a)
Combined View
1cm Load Fact01"' 0 50
to 068
to 0.69
r-t-t--t-+-t---t-t-+-++-+--if--+--l--1--1---+--l--+--....l....--L -+-- - --........ ---+--i-+-<
i
"
t--+-+--+--!--+--i-+-+---- r--t--t--+--+-+-1--+--+--!--,.---.-,-h-\+--+--+--+--t-+-I
-r,-+--t---+---1--+--t--!-.\ \
r--1'--t-t-t--+-+-+-+--+--+--+--+--+--+--+---1--
- --- -t--t--t--t--+--il--+---+--+--+--1--+-+---i--+- , .. - -"""""
-- - --- -t-t-+--+--+-+--11--+---+--+-+-+--+-+- +----- ..
\
r--
(b)
Dtvplac.emerlt 20 00 x

Fig. 6.60 Predicted Failure Mode of Specimen SA: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern
289
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Specimen 88
fem
(a)
Combined View
to 093
lo IJW
to t 0:::
----
r->-
\
I / / / / \ --- -- 1---:-r--
\ \ \ i i
(b)
M&gniflCut x
Crack\'Vldlh\$ 1 OO!fl!f1,mid,Ulk.,'ll 200mm
Fig. 6.61 Predicted Failure Mode of Specimen SB: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern
290
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
No tile lo.oded
Specimen 8C
fem
" 0 36
-to040
-to0,.4
-to 048
to 052
-to 056
- to 060
-toOS<I
(a)
Combined View
I /
to 088
-10072
-"'"'" ... t<l080
\ \ \
YI&! Signs fem Load factor= 310
to 064
-too&;
111118 to 0 92
11111111 to O 815
\ \ ' \ \

(b)
Fig. 6.62 Predicted Failure Mode of Specimen SC: (a) Ratio of Principal
Compressive Stress Demand to Capacity; and (b) Crack Pattern
The accuracy of the models in predicting the local responses in specimens 8B to
8C is presented in Figs. 6.63 to 6.64. Only the comparison between the predicted and
291
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
measured concrete strains is presented. The completed assessment of the predicted strain
values in the reinforcement is already given in details in section 6.2.3.
The observed data from the selected concrete surface gages in specimens 8B and
8C are plotted against the predicted values that correspond to the location of the gages on
the specimens. Most of the concrete surface gages show reasonable agreement with their
corresponding model predictions. The accuracy in predicting the level of strains in
concrete and reinforcement of the specimens correlate to the accuracy in capturing the
load transfer characteristics, crack propagation, and the failure modes of specimens 8B
and 8C as illustrated in Figs. 6.61 and 6.62.
-CG11 (Exp) I
-CG15 (Exp)
- CG15 (FEA);
--CG11 (FEAr
.......T
-07 -06 -05 -04 -03 -02 -01
Strain(mm/m}
900
800
700
600

500 ...
15"
400
0 -0 4

1-CG14 (Exp) I
l-CG13 (Exp) I
f- CG14 (FEA)I
I
-0.3 -0.2 -0 1
Strain(mm/m)
Fig. 6.63 Concrete Strain Prediction in Specimen SB
292
900
800
700
600
500
0
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-CG11
-CG15 Ex
--CG15 F }
- CG11 FEA}
-2 -1.5 -1 -0.5
Strain (mm/m)
800
700
600
500
z
x:
... ----'
i 1. -- CG14

.s
j -CG13 (Exp} I
I CG14 (FEA)
; I
!
0 -0.3 -0.25 -0 2 -0.15 -0.1 -0 05
Strain (me)
Fig. 6.64 Concrete Strain Prediction in Specimen SC
6.5 Mesh Dependency Reduction
800
In this section, the reduction of the mesh dependency in the smeared crack model
is achieved by the use of tension softening model that is formulated in terms of the
element size (Eqn 6-27c) is presented. As described in chapter 2, mesh dependency
results in the divergence of the solution, as the finite element mesh is refined, due to the
infinite increase of apparent tensile stresses. In addition, the parametric study (Fig. 6.30)
0
also illustrates that the mesh dependency can be most pronounced in complex D-Regions.
Thus, based on these observations, the author performed the nonlinear FEA of the
selected complex D-Regions in various levels of mesh sizes from coarse to fine meshes.
The reduction of the mesh dependency in those D-Regions is presented for both the
structural and local levels.
To demonstrate the achievement in mesh dependency reduction, specimen 6C is
selected in this study. The results from the FE models of the specimen constructed and
analyzed by using VT2 are compared with the results from the models of the specimen
constructed and analyzed by the proposed behavior model. Figs. 6.65a to c illustrates the
FE models with the three different mesh refinement levels of the specimens generated to
293
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
use in VT2 and the proposed FE formulation. The three different mesh sizes are 8, 10,
and 12 times the maximum aggregate size of the mixed concrete used to cast the
specimens. The behavior models adopted in FE models generated in VT2 and proposed
FE are all the same except the formulation of the tension softening model and tension
stiffening model. In VT2, the tension softening model was formulated in terms of crack
spacing while the proposed formulation adopted the formulation included the element
size (Eqn 6-27c). The tension stiffening model adopted in VT2 and the proposed
formulation is selected to be Bentz (2003) model and tension chord model, respectively.
The other behavior models in both FE models are cracked shear stress-slip components
based on Okamura and Maekawa (1991), Popovics (1973) for concrete pre-peak model,
and elastic-plastic with strain hardening in smeared and discrete reinforcement model. In
addition, the perfect bond assumption was used in both FE models.
The prediction of the structural responses obtained from FE models in Fig. 6.65 is
illustrated in Fig. 6.66. All solid lines present the predicted load displacement of
specimen 6C from the proposed formulation while all dash lines present the predictions
from VT2. As expected, the predicted initial stiffness of all FE models are almost
identical. At the load level higher than 600 kN, VT2 predicts a stiffer post cracking
stiffness than the proposed formulation due to the use of different tensile responses of the
concrete (tension softening and tension stiffening) which results in the different level of
the damage propagation from the crack. However, the predictions of the structural
capacity obtained from both models are almost identical in most of the mesh refinement
except at the level of fine mesh. The VT2 model constructed using fine mesh predicts the
structural capacity very lower than the capacities obtained from coarser mesh due to the
294
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
effect of mesh dependency; however, the capacity prediction of the fine mesh model from
the proposed formulation shows only slightly different from the coarse mesh. Thus, the
sensitivity of the prediction to mesh size is significantly reduced.
Vector2 Proposed FE
(a)
~ ~ , _ , , , , . ,
; j t
:
~ I
t I
(b)
(c)
Fig. 6.65 FE models with the Three Different Mesh Refinement Levels of the
Specimens generated to Use in VT2 and the Proposed FE Formulation: (a) Mesh
Size= 8x Max Agg; (b) Mesh Size= lOx Max Agg; and (c) Mesh Size= 12x Max Agg
295
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1400
1200
1000
-
z
800 .::.::
-
"C
co
600
0
...I
400
200
0
0 1
=1250kN

/ .,.. ..... \
/ ;""

/ .... \
\
2 3
\
'\
- 6C_8ag (VT2)
- -6C_10ag (VT2)
- - 6C_12ag (VT2)
- 6C _ 8ag(Proposed)
- GC _ 1 Oag (Proposed)
- 6C_ 12ag (Proposed)
4 5 6
Displacement (mm)
7
Fig. 6.66 The Predicted Load Displacement of Specimen 6C using Three Different
Mesh Refinement Levels analyzed by Using VT2 and the Proposed FE Formulation
In Figs. 6.67 and 6.68, the reduction of mesh dependency in the local level is
presented. The strains experimentally measured from the two strain gages, CG7 and
SG9b, attached to the specimen 6C are selected to compare with the predicted strains
from the models analyzed by VT2 and the proposed model. Three different mesh sizes
from fine to coarse mesh are also adopted in the comparison. Based on the results
presented in Figs. 6.67 and 6.68, the use of the proposed model results in the reduction in
the mesh dependency in the predicted values of strain in concrete and reinforcement as
clearly seen in the plots of the predicted strains at the locations of strain gage CG7 and
SG9b. However, the mesh dependency in VT2 results in the significant differences
between the predicted strain values.
296
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-z
e.
"C
s
1400
1200
~ .
--'-',S ......
...... 1000
-"'="-...._
'<;--
800
.'
." 600
-CG?(Exp)
~ < 00
CG? (FEA) 8Agg
CG? (FEA) 12Agg
- CG? (FEA) 10 Agg
Proposed Formulation ~ ~
-0.001 -0.0008 -0.0006 -0.0004 -0.0002
Strain (mm/mm)
... ~ .......
,,
'
. "'I-..
~
...
...
CG? (Exp)
CG? (FEA) 8Agg
CG? (FEA) 12Agg
- CG? (FEA) 10 A
VT2
-0.001 -0.0008 -0.0006 -0.0004 -0.0002
Strain (mm/mm)
1400
1200
1000
800
0
0
Fig. 6.67 Comparison of Concrete Strain Prediction between the Proposed
Formulation and VT2 using Different Mesh Refinement in Specimen 6C
297
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1200
1000
-
800
z
.
.....
"'C
600
cu
0
_,
400
200
0
0
1400
1200
1000
z
800 .
.._.
"'C
Iii
600
0
.J
400
200
0
0
Proposed Fonnulation
-SG9b (Exp)
SG9b (FEA) 8Agg
- - SG9b (FEA) 12Agg
- SG9b FEA 10A
0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035
Strain (mm/mm)
/
/.
/ .. . ,
/
. /
/.
"" ..
-SG9b (Exp)
/ .. - SG9b (FEA) 8Agg
/, ..
- - SG9b (FEA) 12Agg
VT2
- SG9b FEA 10A
0.0005 0.001 0.0015 0.002
Strain (mm/mm)
Fig. 6.68 Comparison of Reinforcement Strain Prediction between the Proposed
Formulation and VT2 using Different Mesh Refinement in Specimen 6C
298
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.6 Influence of Orientation of Principal Stress in Concrete and Crack Angle on
Structural Strength and Responses of Complex D-Regions
In this section, the influence of the orientation of principal stress in concrete and
crack angle on structural strength and responses of complex D-Regions in both element
and structural levels is presented. As pointed out by Nagle and Kuchma (2007), the crack
direction measured from D-Regions tests can significantly deviate from the principal
stress direction and this will affect the shear demand along the crack surface. The effect
of the shear demand along crack surface will directly affect the mechanics of the slippage
of the crack plane and eventually influence to the average deformation of the small
regions (continuum element) which might contribute significantly not only to the
deformation ofD-Regions in the structural level, but also the post-cracking stiffness,
capacity, failure modes, and the ductility of D-Regions. Thus, it is worth investigating the
influence of the deviation of those components on the structural behaviors of D-Regions
at both element and structural level.
6.6.1 Element Level
The study of the influence of the deviation of the crack angle from the principal
stress direction in the concrete in the element level is performed by considering a two
dimensional continuum element that is assumed to represent a small region located in the
end region of a girder. The element is reinforced only in the vertical direction as usually
found in the web and end region of a girder and subjected to the combined shear and
compressive stress acting at the boundary of the element. Three combinations of the shear
and compressive stress acting on the element are considered in the study: 1)
-r: crx:cry=l :0:-0.39; 2) -r: crx:cry=l :-0.39:-0.39; and 3) -r: crx:cry=l :-0.69:-0.69. The first case
299
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
is to mimic the stress condition occurring in the end of a girder when the applied load
was transferred to the support in the end region by direct strut action. The second case is
to capture the effect of the prestressed force that is applied to the end girder. The third
case is the same as the second case; however, with the higher applied stress level.
Figs. 6.69 to 6. 71 illustrate the influence of the deviation of the crack angle from
the principal stress direction in the concrete on the structural responses in the element
level for the first, second, and third load cases, respectively. In overall, the structural
responses of the shear panel are significantly affected by the deviation of the crack angle
from the principal stress direction. The predicted shear strength of the panel is degraded
and the panel behaves more brittle (less ductile) when the level of deviation of those
angles increases as clearly illustrated in Figs. 6.69a, 6.70a, and 6.71a. In addition, the
post cracking stiffness of the panel is predicted to be slightly stiffer when subjected to the
higher level of deviation of the angles. The behavior can be completely explained by
considering the plots between the ratio of the shear demand to the capacity and the
transverse tensile strain as given in Figs. 6.69c, 6.70c, and 6.71c. The transverse tensile
strain in concrete (Ec1) is relatively lower than the panel that is subjected to the higher
level of deviation of the angles. This is because the higher level of deviation of the angles
leads to the higher shear stress and slippage along crack surface (Figs. 6.69b, 6.70b, and
6.71b). The slippage along the crack contributes to the total or apparent element strain (i::)
as the average strain component ( E
5
) of the rigid translation of the crack surface which is
not affected to the concrete strain component (Ee) since the concrete strain component is
affected only from the stress induced strain (Eqn 6-32a). Thus, while the apparent strain
(i::) increases due to the rigid slippage along crack, the concrete strain component (Ee) as
300
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
well as the principal tensile strain in concrete (Ec1) does not significantly increase. This
thus leads to fewer cracks being developed and less degradation of the post-cracking
stiffness.
The effect of the combined shear and biaxial compressive stress on the structural
responses can is illustrated in Figs. 6.69 to 6. 71. As expected, the biaxial compressive
stress applied on the panels increases the capacity of the panels as illustrated in Figs.
6.69a to 6.71a. The biaxial compressive stress also reduces the influence of the deviation
of those angles since the capacity of the panels that is subjected to the different levels of
deviation of those angles is slightly different, particularly, at high level of biaxial
compressive stress (Fig. 6.71a). In addition, the biaxial compressive stress affects the
failure modes of the panels. As illustrated in Figs. 6.69e, 6.70e, and 6.71e, the mode of
failure of the panel in case 1 is crushing of the concrete and then is followed by slippage
along the crack surface while the rest of the panels fail by slippage along the crack
surface. This is due to the capacity of concrete of the panel in case I was more degraded
and reduced by the higher level of the principal tensile strain in concrete (Ec1) as
illustrated in Figs. 6.69c, 6.70c, and 6.71c.
The influence of the deviation of the crack angle from the principal stress
direction in the concrete on structural responses in the element level is also investigated
by considering the interaction diagram of a selected shear panel. The selected shear panel
is reinforced only in the vertical direction with the reinforcement ratio of I. 79% and is
subjected to the combination of shear stress and vertically applied compressive stress. As
illustrated in Fig. 6. 72, the interaction diagram of the shear panel is plotted based on the
FE results. Three different lines are used to draw the interaction diagrams for a shear
301
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
panel that is subjected to three different levels of the deviation of the angles. The
deviations of angles of 0, 5, and 7.5 are plotted by using solid, dash, and dot lines,
respectively. From the figure, the interaction diagram is shown to be affected by of the
deviation of the crack angle from the principal stress direction in the concrete. An
increase of the deviation of those angles results in the decrease of the capacity of the
shear panel as illustrated by the smaller interaction diagram.
In addition, the predicted failure mode of the shear panel is significantly
influenced by the deviation of those angles. As illustrated in Fig. 6. 72, the interaction
diagram is separated into three small regions: 1) i to a; 2) a to b; and 3) b to c. The i to a
region denotes the failure mode of the shear panel that occurs by sliding shear failure
along the crack surface. The failure mode by crushing of the degraded and softened
concrete is denoted in the a to b region. It should be noted that the degradation and
softening of the concrete are due to cracks and transverse tensile strains in concrete. The
b to c region denotes the failure mode by crushing of the concrete without any crack
development. Based on the interaction diagram, region no.2 for the mode of failure of the
shear panel decreases while regions no. I and 3 increase when the deviation of the angles
increases. Thus, in the overall, the shear panel tends to have lower capacity and more
brittle failure mode when subjected to the deviation of cracking angles from the principal
tensile stress in concrete.
302
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
9
tu'
8
a.
7
!.
t/I 6
t/I
!! 5
-
(/)
4 ....
Ill
GI
3
..c
(/)
2
1
i:ox:oy=1:0:-0.39
- proposed-a
- proposed-5
- proposed-7.5
- proposed-9.0
t:GX:G)'1:Q:-0.39
I \
i ---- tautotal ;
I I
1 tautotal-5 i
1- tautotal-7
tautotal-9.9

.................................... ,
0 0.005 0.01 0.015 0.02
Shear Strain(mm/mm)
0.75 0 75
0.3
Slip( mm)
0.4 0.5 06
t:ax:ay=1:0:-0.39
><
e 0.5
>
>
0.25
0
0 2 3
i:c1/cc0
1.2
0.8
Q.
u
l!:i0.6

04
0.2
0
0
<:GX:cry=1:0:-0.39
I :!::: i 5
I proposed-9.0
4 5 6
><
e 0.5
.:::
>
0.25
--proposed
0
T
0 0.2 04 0.6 0.8
wcr(mm)
i:ox:oy1 :0:-0.39
-,

proposed-5
+ proposed-7.5
proposed-9.0
2 3 4
eC2/eO
Fig. 6.69 Influence of the Deviation of the Crack Angle from the Principal Stress
Direction in Concrete on the Structural Responses in the Element Level, Case 1
('t: O'x:cry=l:0:-0.39)
303
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
9
';"
8
a. 7
i
6
l!:!
.....
5
Ill
4 ...
..
Cll
3 .c
Ill
2
1
0
0
0.75
)(
0.5
-s
0.25
0
.:ax:ay-1 :.0.39:.0.39
. J
........ proposed-5-
"*'- proposed-7 .5-
- O
-r-- ---- 1
0.002 0.004 0.006 000
Shear Strain(mm/mm)
2
ec1/eco
1.2
0.8
a.
u


04
0.2
--proposed
+ proposed-5
_ .. __ proposed-7.5
3 4
5
4.5
0
0
0.1 0.2 0.3
Slip(mm)
t:<rx:ay=1:.0.39:.0.39
-+- tautotal
: tautotal-5 i
i i
; - tautotal- 7 5
i -- tautotal-9 q
0.4 0.5
t:ax:ay=1:.0.39:.0.39
0 1 0.2 0.3 04
crack width (mm)
t:ox:oy=1:-0.39:-0.39
_.,_ proposed
- proposed-5
- proposed-7 .5
proposed-9.0
.. proposed-0-
1
- proposed-5-
-- proposed-7 .5-
I proposed--9.0 l
05 0.6
0 - ----------.,--------- -------,---------------1
0 0.5 1.5
sc2fo0
Fig. 6. 70 Influence of the Deviation of the Crack Angle from the Principal Stress
Direction in Concrete on the Structural Responses in the Element Level, Case 2
('t: O'x:cry=l :-0.39:-0.39)
304
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
..
'l l 0..
6 1
"'
5
f!

4 ..
..
..
3 .::
II)
2
0
0
1 !
0.75
><
05
>
>
0.25
0
0

.. proposed-0
proposed--5
proposed-7 5
-proposed-9.0
0 001 0.002 0 003 0 004 0.005 0.006
Shear Strain(mm/mm)
t:ax:oy=1:..0.69:-0.69
-proposed
propo5ed-5
proposed-7.5'

0 75
0.5

0.25
0 0.05 0 1 0.15 0.2 0.25
Slip(mm)
t:ax:ay:1 :.(),69:-0.69
-proposed
2
ec1(mm/m)
3 4
__. proposed--5
--proposed-7.5!
"-.-

1.2
0.8
0.4
0.2
0.2 04
0
06
&C2Jc0
0.1 0.2
wcr(mm)
t:ax:ay:1 :-0.69:-0.69
0.8
i _,_ proposed
I proposed-5
I - proposed-7 5:
I
12
03 04
Fig. 6. 71 Influence of the Deviation of the Crack Angle from the Principal Stress
Direction in Concrete on the Structural Responses in the Element Level, Case 3
('t: O'x!O'y=l :-0.69:-0.69)
305
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
6.5
6
5.5
5
- 4.5 <U
fl.
4
:e
- 3.5

3
2.5
2
1.5
1
0.5
0
-35 -30 -25 -20 -15 -10
cry(MPa)
Acr=O

-5

I
0
Fig. 6. 72 Influence of the Deviation of the Crack Angle from the Principal Stress
Direction in Concrete on the Interaction Diagram of a Shear Panel.
6.6.2 Structural Level
The study of the influence of the deviation of the crack angle from the principal
stress direction in the concrete in the structural level is performed by using the numerical
examples of two types of complex D-Regions: 1) specimens 8A to 8C and 2) simply
supported deep beams. By extending the study of the deviation of the angles to the
structural level, the complexity of D-Regions in the structural level could be included in
the study; for example, structural geometry, boundary condition, and the complex load
transfer mechanism.
Fig. 6.73 illustrates the effects of mesh size and difference between crack and
principal stress angles on capacity of specimens 8A, 8B, and 8C. The specimen fabricated
with only distributed reinforcement is also analyzed and plotted in the figure for
306
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
comparing with the rest of the specimens. All specimens are analyzed by using two levels
of mesh refinement: 1) the coarse mesh (dashed line in Fig.6. 73) with the element size
approximate to 90 mm or 9 times the maximum aggregate size used in the concrete; 2)
the fine mesh (solid line in Fig. 6.73) with the element size of 75 mm or 7.5 times the
maximum aggregate size. The deviation of the crack angle from the direction of the
principal compressive stress in concrete is also included in the analysis of all specimens.
Five levels of the deviation of the angles are considered in the study; for example, ~ c r =
0, 0.1, 0.5, 1, and 5.
Based on the figure, the predicted capacity of specimen 8B is highest in all cases
while the specimen with only distributed reinforcement has the lowest predicted capacity.
The mesh dependency is only slightly affected the predicted capacity of all specimens.
The insignificant effect of the deviation of the crack angle from the direction of
the principal compressive stress in concrete on the predicted capacity is occurred in
specimen 8A, C, and the specimen with only distributed reinforcement; however, the
significant drop of the predicted capacity occurs in specimen 8B at the high level of the
deviation of the angles; for example, ~ c r = 5. By considering the crack pattern just before
the failure of the specimens 8A to 8C as illustrated in Fig. 6.74, the significant drop in the
capacity of specimen 8B could be clearly seen at the high level of the deviation of the
angles.
307
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
900
800
700
-a-----13-------------------------------------------0
ZBOO SC -;:-- Meshonly_75
.::.::
-.i:5_ 00 ____ 8A_75
---..-eB_75

c. -*- 8C _75
-
()300 -1'- - Meshonly_90
200
100
0
. Meshonl -+- 8A_90

--IS 88_90
--0-- 8C_90
0 1 2 3 4 5

Fig. 6. 73 Effect of Mesh Size and Difference between Crack and Principal Stress
Angles on Capacity of Specimens 8A, 8B, and 8C
In Fig. 6.74, the crack pattern in specimens 8A and the specimen with only
distributed reinforcement is shown to be locally severe at the region between the upper
and lower openings. The severe crack development and propagation in the region
eventually leads to the sliding failure of the specimens. However, the crack development
and propagation in specimens 8B and 8C is slight and this leads to the more stress
redistribution inside the specimens that caused the specimens to fail at the capacities
much higher than specimen 8A and the specimen with only mesh. In addition, the severe
crack condition in specimen 8A and the specimen with only mesh leads to the significant
reduction of the tensile stress transferred across cracks. As clearly seen in Eqn. 7-22, the
tensile stress transferability across the cracks measured by crc
1
is one of the most
significant parameters that cause the additional demand along crack surface when the
deviation of the crack angle from the direction of the principal compressive stress in
308
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
concrete occurs. The additional shear demand along crack surface will increase the level
of slippage along the crack and can significantly reduce the structural capacity of the
specimens. Thus, the more tensile stress transferability or the less severe crack condition,
the more reduction of the structural capacity due to the effect of the deviation of the crack
angle from the direction of the principal compressive stress in concrete. This is thus why
the capacity of specimen 8B is significantly lower when the high deviation of the angles
is applied.
To demonstrate this in more details, the effect of the deviation of the crack angle
from the direction of the principal stress direction for another type of D-Region is
considered as illustrated in Fig. 6.75. In Fig. 6.75, three simply support deep beams that
have different reinforcement patterns are adopted in the study. The use of these deep
beams is beneficial to the study such that the effect of the complexity of the structural
geometry could be eliminated. For example, specimens 8A to 8C have very complex
geometry due to the existence of the openings. Those openings cause the force transfer
mechanism from the loading point to the support by the flow of the compressive struts
around the opening (Fig. 6.76). The flow of the strut in this way will induce the tension
demand of the force greatly in the region between the openings which could significantly
accelerate the crack propagation and the reduction of the tensile stress transferability in
those specimens. Thus, only a slight influence of the deviation of the angles could be
observed. By eliminating the complexity of the structural geometry and using only a
simply-supported deep beam, the force transfer mechanism is simply a direct flow from
the loading point to the support (Fig. 6. 76).
309
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Specimens BA and Meshonly
r-1\- r ~
Specimens BB and BC
Fig. 6.74 The Crack Pattern just before the Failure of the Specimens 8A to 8C
310
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Deep Beam 1
Deep Beam 2
Deep Beam 3
Fig. 6. 75 FE Models of Three Simply Supported Deep Beams
311
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
---- Compression Force - Tension Force
Fig. 6. 76 Force Transfer Mechanism in Specimens SA to SC.
Based on the FE models of three simply supported deep beams in Fig. 6.75 that
includes the deviation of crack angle from the direction of the principal compressive
stress in concrete, the significant influence of the deviation of the angles could be
evaluated as illustrated in Fig. 6.77. By increasing the levels of the deviation of the
angles from ~ S e r 0 to 5, a significant drop of the structural capacity is observed in all
types of deep beams. Deep beam 1, only distributed reinforcement is provided; no
vertical tie in the web region. It has the least reduction in the structural capacity while the
deep beam 2 that is provided both distributed reinforcement and the vertical tie in the
web region has the most reduction in the structural capacity. This is reasonable since
deep beam 1 has the most severe crack condition due to the transverse tensile strain
induced from the bottle shaped strut in the web region while the rest of the deep beams is
fabricated with the vertical tension ties and thus the development of crack is more
312
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
controllable and this thus leads to the higher level of the additional demand along crack
surfaces.
Based on the observation in the influence of the deviation of the crack angle
from the principal compressive stress in concrete of the beam type structure, the
consideration of the deviation of the angles and the additional demand along crack
surface should be included in the design process and this important suggestion is also
supported by the research work by Nagle and Kuchma (2007).
500
450
...-
400 z
.::rt.
-
>- ....,
350
-
CJ

0..

300
u
250
200
0 1 2
A0cr
-G Deepbeam3_90
--,,,.,_ Deepbeam1_90
-ft- Deepbeam2_90
3 4 5
Fig. 6. 77 Effect of the Diff ere nee between Crack and Principal Stress Angles on the
Capacity of Simply Supported Deep Beams
313
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.7 Summary
In this chapter, the author initially presented a framework for automated NLFEA
and the analytical works required for development of an improved NLFEA methodology
for the automated analysis of complex D-Regions. Those analytical works are the
concepts and formulations of the embedded reinforcement element including bond stress-
slip, the improved constitutive model for reinforced concrete continuums, mesh
dependency reduction achieved by using the improved constitutive model, and the
numerical study on the influence of the deviation of the crack angle from the direction of
the principal compressive stress in concrete on structural strength and responses of
complex D-Regions.
For the embedded reinforcement element including bond stress-slip, the author
demonstrated the implementation and validation of the approach using experimental test
data including the anchorage specimens and the concrete structures. The strong
correlation between the predicted and experimental responses was presented and thus
confirmed the validity and the potential value of the proposed formulation.
The improved constitutive model specialized for complex D-Regions in reinforced
concrete structures was implemented in MATLAB and Augustus was adopted as a post-
processing tool. The model was validated by performing series of FEA at both element
levels and structural levels. All analyses were compared with the existing experimental
data from the literature and from the available experimental data of complex D-Regions
in Chapters 3 to 5. The numerical results correlated well with most of the experimental
data. In addition, the reduction of the mesh dependency in the structural level could be
obtained in most cases.
314
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Subsequently, the proposed FE formulation was adopted to perform a parametric
study on the influence of the deviation of the crack angle from the direction of the
principal compressive stress of concrete on the predicted structural capacity, ductility,
and failure modes. Both element and structural levels were included in the study. Based
on the parametric study, the important observations were: 1) the additional shear demand
along crack surface will increase the level of slippage along the crack and can
significantly reduce the structural capacity of the specimens; 2) the additional tensile
stress transferability or the less severe crack condition will lead to a greater reduction of
the structural capacity due to the effect of the deviation of the crack angle from the
direction of the principal compressive stress in concrete; 3) for the beam type structure
without any openings, the consideration of the deviation of the angles and the additional
demand along crack surface should be included in the design process to ensure accurate
predictions of response.
In the next chapter, an introduction to the proposed design and analysis
framework for effective structural performance design of complex D-Regions will be
presented. The application of the proposed framework on designing and analyzing several
types of complex D-Regions typically found in a building will be demonstrated by using
several numerical examples. The available experimental data will be used to assess the
structural performance and responses of the designed complex D-Regions.
315
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 7
Design and Analysis Framework for Effective Structural Performance
Design of Complex D-Regions
In this chapter, a design and analysis framework for effective structural
performance design of complex D-Regions is presented. The introduction to the need of
the design and analysis framework is presented in section 7.1. The details of the proposed
design and analysis framework of complex D-Regions is presented in section 7.2. In
section 7.3, the application of the proposed design and analysis framework will be
demonstrated by using the design examples based on several structural components that
are often found in a typical building. These examples include a propped cantilever deep
beam with an opening, a simply supported deep beam with an opening, a simply
supported deep beam with a <lapped end and an opening, and a shear wall with openings.
A summary will be given in section 7.4.
7.1 Introduction
As discussed in chapter 1, complex D-Regions are a type of D-Regions where
strain distribution across the section is highly nonlinear and those complex D-Regions
usually have a very complex geometry, supports, and loading conditions. In addition,
those regions could be designed by using multiple possible STM shapes. The multiple
possible STM shapes in the design of these complex D-Regions and the deficiency in the
current design practice based on STM creates difficulty and uncertainty in the design
process.
The multiple possible STM shapes could lead to the design of D-Regions with the
poor structural performance, particularly, during the service load as clearly seen in the
316
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
experimental validation of STM as presented in chapter 4. However, based on the study
of the influence of STM shapes on the structural responses and performance of complex
D-Regions in chapter 5, the significant improvement of the structural performance during
service loads could be obtained by adopting the STM shapes generated from the topology
optimization technique.
Based on experimental validation of STM in chapter 3 and 4, the deficiency of the
current STM design practice can be stated as follows: 1) different codes of practice
recommend different stress limits for STM components; 2) limited guidance is available
for proportioning the forces in indeterminate STMs; and 3) distributed reinforcement is
needed to be provided to ensure sufficient ductility as required per the plasticity theorem.
These deficiencies affect the structural performance of designed D-Regions. Thus, to
ensure that the structural performance of designed D-Regions satisfies desirable
performance criteria, reliable analytical methods are required to assess the structural
performance of D-Regions after the D-Regions have been designed by STM. The results
of these predictions can be used to revise and inform these designs.
As addressed in chapter 2, the current analytical methods usually adopted to
analyze D-Regions are nonlinear STM and nonlinear FEA. Those current methods are
still limited when used to predict the response of complex D-Regions. A nonlinear STM
for the analysis of complex D-Regions is still overly complex for use in practice. It
requires careful selection of the STM shape, dimensions, constitutive models, and stress
limits, particularly when completing a complex and indeterminate STM design. The use
of existing nonlinear FEA methods typically requires substantial effort and time to
construct the structural geometry of the FE model, and further requires that the user is
317
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
able to make appropriate selections for material models and can correctly interpret the
results of analyses. However, based on the newly proposed formulation of nonlinear FEA
for reinforced concrete structure specialized for complex D-Regions presented in chapter
6, the analysis of the complex D-Regions is more straight-forward. In addition, the
accuracy and validity of the proposed FEA in analyzing the complex D-Regions were
also confirmed by a series of the experimental programs.
By combining the shape selection technique based on topology optimization and
the proposed nonlinear FEA to the current code of practice for the STM design of D-
Regions, the proposed design and analysis framework for complex D-Regions can be
obtained. The schematic view of the framework previously illustrated in Fig. 1.5 is
repeated here in Fig. 7.1. The presentation of the framework is given in section 8.2.
Start

Simple D-Regions
+-
Complex D-Regions
a+



lllllllP"' Adjust or Refine

Yes
Fh1.ish
,,,!:
Fig.7.1 The Proposed Design and Analysis Framework
318
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7 .2 Proposed Design and Analysis Framework
As stated in chapter 1, the overall objective of this research study is to improve
the current approach for the design and analysis of simple and complex D-Regions. The
improvement of the design and analysis of complex D-Regions could be achieved by
expanding the current design and analysis framework to include three major parts (the
colored components in Fig. 7.1). These components are aimed at ensuring good structural
performance under service loads and under overloads. Three major parts are: 1) Methods
and guidelines based on topology optimization to facilitate improved STM shape
selection (chapter 5), 2) Nonlinear STM analysis for simple D-Regions (chapter 2), and
3) Nonlinear FEA for automated analysis of complex D-Regions (chapter 6).
As presented in chapter 5, the good structural performance during service loads of
complex D-Regions can be obtained by adopting the STM shape selection technique
based on Topology Optimization. In topology optimization, D-Regions with the
completed descriptions of geometry, loading, and boundary conditions are discretized
into small FE domains by using either manual or automatic mesh generation available in
most of the pre-processing tools, e.g. Patran, for high quality meshes. The discretized
structures are used as the initial domain where the optimal topology can be obtained by
gradually removing the underutilized elements from the initial domain. The optimal
topology is then adopted to characterize the appropriate STM shape used in the STM
design provisions.
Subsequently, the predicted structural responses based on either nonlinear STM or
FEA are assessed to ensure the achievement of the structural performance targets which
are generally determined by parameters relevant to the structural responses; including
319
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
crack widths, displacements, material stresses and strains, as well as overall capacity and
ductility. For complex D-Regions, which are the focus in this chapter, involving much
more complexity of geometry, loading and boundary conditions, an automated nonlinear
FEA model validation methodology as previously presented in chapter 6 is adopted.
Based on the assessment of the structural performance, if the structural
performance targets are not satisfied, the redesign process will be performed iteratively
until the performance targets are satisfied. This includes the possible tasks of using new
material properties, proportioning the structural dimensions, revising the reinforcement
details, etc. In this chapter, only the application of the design and analysis framework to
achieve the most effective structural performance design of complex D-Regions will be
addressed since the details and the approaches of the redesign process mainly depend on
the type of the structures, the structural conditions, and the experience of the designers.
To demonstrate the achievement in the design of complex D-Regions for the most
effective structural performance, the design examples based on complex D-Regions
usually found in a typical building are given in the next section.
7.3 Application of Proposed Design and Analysis Framework
In this section, the application of the proposed design and analysis framework for
achieving the effective structural performance design of complex D-Regions typically
found in a building is presented. Fig. 7.2 illustrates a typical building composed of
several structural components considered as complex D-Regions. These include the
beam-column connections, transfer girders with and without openings, shear walls with
and without openings, and a mat foundation. The complex D-Regions used in the design
examples are a propped cantilever deep beam with an opening, a simply supported deep
320
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
beam with an opening, a simply support deep beam with a <lapped end and an opening,
and a shear wall with openings.
a
1
Beam-Column Connection

: t--------' o D
DD
DD
DOD
DOD
DOD
QDD
D
b

' ' '
'----t. t--------' D D
:
0
: DOD
'----t ' '

'-------' < C? : D D D
... DOD
' D '
'----I q <I <J <I f----'
DOD
'----1<1<1 <I <I

D
. . . n

c D ----..------. f
. '\
D ' '\
, , r Jirder
D

0 <I <l<I
.__ ___ __, ,___ ___ --I t-------'
I /Mat Founr8 I
0
' 'o __ Shear wall D D G.L.
JIA AA A 4 A A 4 4 A
A A A4 4 A A A A A A A A A A A A A A A A A A A A A A A A A 4 A A ''4A A A A A .'I.
ii iii iii iii iii iii iii iii iii iii iii iii iii i
Fig.7.2 Complex D-Regions in a Typical Building
7.3.1 A Propped Cantilever Deep Beam with an Opening
The propped cantilever deep beam with an opening is adopted as the first
design example to demonstrate the application of the proposed design and analysis
framework. In the first example, the demonstration will mainly focus on presenting the
general process in design and analysis of a complex D-Region by using a selected STM
321
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
shape based on conventional approach, designing the complex D-Region by using STM
based on Appendix A of ACB 18-08 code provisions, and analyzing the designed
complex D-Regions by using both nonlinear STM and FEA.
In this example, STM is used to design for a large storey-high propped
cantilever beam that supports a single concentrated load near midspan and has a large
rectangular opening between the point of loading and the fixed-ended support. The STM
design is complicated by the use of an externally and internally statically indeterminate
truss to carry the imposed loading to the supports. This produces a situation in which the
calculated capacity is largely affected by the relative stiffness of truss members and the
degree of plasticity attributed to the truss. In addition, the shortcoming and advantages of
both nonlinear STM and FEA analysis techniques will be also presented using this
example.
As illustrated in Fig. 7.2, the propped cantilever deep beam can be considered as
a part of the continuous deep beam that spanns from point "a" to point "c". Subjected to
dead and live loads, by designing the continuous deep beam "ab c" to form the
mechanism by firstly yielding of the top reinforcement at the supporting column "b", the
continuous deep beam becomes a simply supported deep beam with an opening on the
left span and a propped cantilever deep beam with an opening on the right span.
By considering the propped cantilever deep beam with an opening on the right
span and assuming 10 times larger in the dimension of the deep beam than the dimension
given in Fig. 3.lb in chapter 3, a large storey-high propped cantilever beam with a large
rectangular opening can be shown in Fig. 7.3 and the beam is to be designed using the
STM. The beam supports a single factored concentrated load from a column of 1304 kips
322
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(5800 kN). The deep beam with a clear spanned of 36 ft (11 m) has a width of 24 in.
(609 mm) and a 158 in. ( 4 m) overall depth. The width of the column at the location of
the concentrate load and at the left support is 40 in. (1016 mm) and 28 in. (711 mm),
respectively. The self-weight of the member is included in the factored loads.
5800 kN
13'-2" (4.00 m) -i
t=24"(609 mm)
28"(711 mm)
40"(1016 mm)

;shear Wall
L 59" l 108" (2750 mm) _;
1(15oommT :
36'-0" (11.00 m)
Fig. 7.3 Member and Loads for the Propped Cantilever Deep Beam with an
Opening
The material properties uses in this design and analysis example are taken as the
material properties of the 1/10 scale specimen (Fig. 3.lb) that was used to evaluate the
design, as presented in Table 7.1 below.
323
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7.1 Summary of Material Properties.
Material
Design and Analysis
Properties concrete Meshing bar (#7) bar(#9) bar (#10)
f c (psilMPa) 5858140.4 NIA NIA NIA NIA
Agg (in/mm) 1125.4 NIA NIA NIA NIA
Fy (ksilMPa) NIA 102/703.4 67.11463 67.11463 49.41341
Fu (ksilMPa) NIA 102/703.4 107.6/742 107.6/742 71.61494
Es (ksilGPa) NIA 276951191 27695/191 276951191 30301209
Esh (ksi,GPa) NIA 0 66714.6 66714.6 39212.7
esh (me) NIA 1.8 2.9 2.9 10
The design of this propped cantilever was completed to satisfy the requirements
of Appendix A of ACI 318-08. The STM design is complicated by the use of an
externally and internally statically indeterminate truss to carry the imposed loading to the
supports. This creates a situation in which the calculated capacity is largely affected by
the relative stiffness of truss members and the degree of plasticity attributed to the truss.
The STM provisions in Appendix A are an ultimate strength design approach, and
thereby do not ensure that the structure will exhibit adequate performance under service
load levels. Both the results from non-linear analyses and a physical experiment are used
to assess the complete performance of the designed structure.
The selected truss model for use in the design of this propped-cantilever is shown
in Fig. 7.4a. The shape of this STM was selected from one of several possible shapes.
This selection was guided by the conventional approach based on the results from a
linearly elastic FEA and the desire to use an orthogonal (and presumable more practical)
layout of reinforcement. In this model, the force transfer from the column loading to the
left support is provided by one diagonal direct struts AF which is horizontally
324
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
equilibrated by strut AB at the top and tie FG at the bottom. The main force transfer in the
right side of the column force is transferred by prismatic strut AB and diagonal bottle-
shaped strut AG to flow around the opening which is equilibrated by using a smaller truss
system around the opening, the dimensions for which are fully presented in Fig. 7.4b.
The design procedure consisted of the following steps:
1) Construct STM model using software CAST (Computer-Aided Strut-and-Tie)
by Tjhin (2004)
2) Determine the effective dimension of struts and amount of reinforcement in the
ties to satisfy the requirements in Appendix A of ACI 318-08 where the STM is
intentionally designed to fail in the ductile manner by yielding of ties. To solve for the
forces in the indeterminate STM, all members were assumed have the same axial stiffness
(EA).
3) Arrange the reinforcement and check that adequate anchorage has been
provided.
4) Evaluate the structural performance and capacity based on nonlinear analysis
techniques: 1) Plastic truss analysis; 2) Nonlinear elastic truss analysis; and 3) Nonlinear
FEA.
5) Investigate the adequacy of the design and non-linear analysis by conducting a
scale-model test of the designed structure.
From Steps 1 and 2, the STM of the deep beam is constructed using the CAST
software and the design process adopts the iterative steps for selecting the effective
dimensions and quantity of reinforcement of STM members to fail by yielding of ties.
325
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
E :
; f.- 157" (4000 mm) -t>j+- 62"-t1+4CY+l+- 80" -t>j+4CY+j+- 53"
e726 kips(3229.5 kN) (1590) (1015) {2030) {1015)(1350)
c.; mm mm mm mm mm
N
(a)
B c D E
T
____ .... _ .... _ .... _ .... _ .... '"--e'f0' _' N mm)
\
\
' 60"(1525 mm)
Q \R
'' '' I
' '
G H I J
40" 80" 40"
(2030)
mm mm mm
(b)
823 kips
(3659 kN)
Fig. 7.4: Strut-and-Tie Model and Forces: (a) Overview; and (b) Details around the
Opening.
326
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
To deal with the indeterminate STM, all STM members are assumed to have
equal relative stiffness (EA), where E is modulus of elasticity and A is effective area. Fig.
7.5 shows the distribution of the forces in the strut-and-tie model. A positive value is
used to denote tension, and a negative value denotes compression. Fig. 7.6 shows the
effective widths selected for the struts and ties and the utilization rates (Fdemand/Fcapacity) in
STM component in this deep beam. From Fig. 7.6, the strength of STM is shown to be
1686 kips (7500 kN) with the ductile failure caused by yielding of tension tie ES as the
utilization rate of the tie approaches 1.0.
,
,
,
~ ?
..,,
~
o-
Fig. 7.5 Force Distribution in Struts and Ties.
Fig. 7.6 The Effective Widths selected for the Struts and Ties and the Utilization
Rates (Fdemand/Fcapacity) in STM of the Deep Beam.
327
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
After completing this design step, the next step is to validate this design and the
assumption of equal relative stiffness in all STM members. To do this, nonlinear analyses
based on the plastic truss analysis, nonlinear elastic truss analysis, and nonlinear FEA
analysis were performed.
The plastic analysis of STM is performed using CAST by assigning the linear
elastic perfectly plastic stress-strain relationship (Fig. 7.7) to all struts and ties which are
assumed to behave elastically up to the ultimate strength (0.85 ~ s c for struts and Fy for
ties). After reaching the yielding stress, the truss members are assumed to have very large
plastic deformation capacity which is represented by a substantial long plastic range. To
model plastic behavior of tension tie, no contribution of the concrete under tension is
considered in this section, e.g. no consideration of tension stiffening model. The results
from the plastic truss analysis are illustrated in Figs. 7.8 and 7.12. As illustrated in Fig.
7.12, modes of failure of STM are predicted to begin by yielding of the tie NR located at
the far right side of the opening at the load magnitude of 2082 kips (9264 kN). A
subsequent yielding of tension tie FG is predicted to occur at a load of 2087 kips (9284
kN). Eventually the deep beam is predicted to reach it ultimate strength at the load
magnitude of 2090 kips (9299 kN) by yielding of tension tie KO and this results in a very
large displacement to be predicted at failure. Based on this plastic truss analysis, the STM
is predicted to reach its capacity by yielding of ties located adjacent to the opening (Fig.
7.8) and the top tension tie ES is not approaching the yielding stress with the utilization
rate of 0.639 as designed to yield in the designed model. The model is thus able to predict
the ductile failure the same as the designed model; however, the locations of the STM
members approaching yielding stress and mechanism are not the same and this is thus
328
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
due to the influence in nonlinear characteristics of the model that adopts the plastic
behavior in STM members.
500
-400
cu
Q..
:E
iif300
en


"'200
100
0
I
I
'
1- - - - - - -
/ --Prismatic Strut
I
I
J
I
- Bottle Shape Strut
---Tie FG and LP
- -Tie KO and ES

0.002 0.004 0.006 0.008 0.01 0.012
Strain( mm/mm)
Fig. 7. 7 The Stress-Strain Relationship of "Key" Struts and Ties for Plastic Truss
Analysis.
Fig. 7.8 The Effective Widths selected for the Struts and Ties and Their Utilization
Rates (Fdemand1Fcapacity) in STM of the Deep Beam (Plastic Truss Analysis).
Next, a nonlinear elastic truss analysis was performed to more fully evaluate the
structural performance and capacity of the STM. Fig. 7.9 illustrates the stress-strain
relationship of struts and ties. For struts, the relationship is assumed to follow the average
329
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
curve obtained from concrete test cylinders with the ordinates being scaled down to the
point of the peak strength of 0. 85 ~ f c where ~ is the factor to account for cracking and
strut shape as suggested by Appendix A in ACB 18-08. For tension ties, a simple tension
stiffening curve is adopted and the reinforcement is assumed to be distributed uniformly
within the effective area of the tie.
At a load of 2092 kips (9305 kN), the fist failure mechanism of the STM is
predicted to occur by yielding of the tie NR at the far right side of the opening and this is
followed by yielding of the tie KO at the left side of the opening at a load of 2117 kips
(9418 kN). A large displacement is subsequently predicted to occur at a load of 2124 kips
(9449 kN) due to yielding of the tie FG. This sequence of the failure is illustrated in Figs.
7.10 and 7.12 in which the utilization rates of those truss members are shown to approach
1.0. Based on the nonlinear elastic truss analysis, the deep beam is also predicted to fail
in a ductile manner with the same mechanism as predicted by the plastic analysis.
Subsequently, nonlinear FEA of reinforced concrete structure is performed using
the Vector2 program to predict the structural behavior of the deep beam and to compare
this with the existing nonlinear STM analysis as previously presented. The FE mesh of
the deep beam is generated using the 1355 quadrilateral elements and 226 truss elements
as illustrated in Fig. 7.11. The constitutive models adopted in the analysis are DSFM
(included explicit cracked shear stress-slip components), Popovic (NSC) for concrete pre-
peak model, 1992 Vecchio model A for compression softening model, 2003 Bentz model
for tension stiffening, bilinear model for tension softening model, and perfect bond
condition for the truss elements in the analysis.
330
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The results from the FE analysis are illustrated in Figs. 7.12 to 7.15. The
predicted load-displacement relationships for the deep beam are presented in Fig. 7.12 in
which the predicted condition of the beam at key points is illustrated. Potential slip crack
surface is predicted to initiate at the load about 3372 kips (15000 kN) and this is shown to
results in an abrupt change in the stiffness of the response. At a load of about 3709 kips
(16500 kN), the bottom reinforcement (equivalent to the tension tie FG in Fig. 7.4a) is
predicted to yield (Fig. 7.15). The deep beam is finally predicted to fail by shear
compression failure of the concrete regions above and below the opening (Fig. 7.14) at
the capacity of 3 736 kips ( 16622 kN). The failure regions above and below the opening
correspond to the location of the compression struts CM and Pl in Fig. 7.4a. The vital
signs, the ratio of the principal compressive stress to the maximum available compressive
stress, of the deep beam at the failure load is also illustrated in Fig. 7.13 and illustrate a
shear compression failure above and below the opening.
Not surprisingly, the designed behavior and that predicted by the nonlinear FE
analysis are quite different, as the STM design requires a coarse idealization where non-
linear continuum analysis methods aim to capture more fully all aspect of behavior. This
includes the influence of cracking on stiffness, non-linear constitutive properties, and slip
along crack surfaces. Slippage is a particular concern when little distributed
reinforcement is provided as a single crack could dominate the response.
331
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
40
35
30
-
ca 25
Q.
:i:
Cir20
ti)
CV
b 15
U)
10
5
0
- Prismatic Strut
- Bottle Shape Strut
0 0.001 0.002
Strain( mm/mm)
-----Tie FG
- -Tie KO
----Tie LP
--Tie ES
0.003
Fig. 7.9 The Stress-Strain Relationship of "key" Struts and Ties for
Nonlinear Elastic Truss Analysis.
0.004
Fig. 7.10 The Effective Widths selected for the Struts and Ties and the Utilization
Rates (F demand/F capacity) in STM of the Deep Beam (Nonlinear Elastic Truss Analysis).
332
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c
Deep Beam (as-built) ACl-208SP Truss Bars Ab Displacement Factor; 2.20
- 000 - 1556100
-580500
-1064700
- 10836 00
Fig. 7.11 FE Mesh of a Reinforced Concrete Deep Beam constructing with 1355
Quadrilateral Elements and 226 Truss Elements
333
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-
18000
16000 __ __
14000 --l-----
12000
-
I
I
s 10000 iFEA-
......
0
8000
"'C -:------------ -
.3 6000 -+.-f-f---""-1'-11-------+-----+---_:__--+"'"----I
.

: I
2000 o mear Elas 1 -.Ahruysis---1-----------------1----
I I
0
0 20 40 60 80 100
Displacement (mm)
Fig. 7.12 The Predicted Load and Displacement of the Deep Beam from
Nonlinear Analysis Techniques: Plastic Analysis, Nonlinear Elastic Analysis,
and Nonlinear FEA.
334
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
fem
Deep Beam (as-built) ACl-208SP Vital S12ns fem Displacement Factor:= 20.80
-to 0.04 -to 0 21 -to 0.38
-to
0.54 -to 0 71
-to
0.88
-to 0 08 -to 0 25 -to 0 42 -to 0 58 -to 0 75
-to
0 92
-to 0.13 -to 0.29 -to 0.46 -to 0 63 -to 0.79
-to
0 96
-to 0.17 -to 0.33 -to 0.50 -to 0 67 -to 0.83
-to
1 00
Fig. 7.13 Concrete Compressive Vital Sign in the Deep Beam: Ratio of Compressive
Stress to Compressive Stress Capacity.
Deep Beam (as-built) ACl-208SP
Combined View
' /,.,./ \.'''
// ,,. r
\
\.
Displace/Cracks. Combined View Displacement Factor= 20 80
Displacement Magn1f1cat1on 20 00 x
Crack Widths thin< 1 OD mm. mid. thick> 2 00 mm
Fig. 7.14 Combined Deformed Shape and Crack Pattern at the Failure State of the
Deep Beam
335
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
cr-truss at crack
H
Deep Beam (as-built) ACl-208SP Stress (steel) o-truss at crack Displacement Factor::: 20 80
-to 4606 - to 11989 - lo 19373 - to 267 56 - to 34139 - lo415 23
- lo 64 52 - lo 138 35
-1021218 - to28602 - to35985 -1043368
-to 8298 -to15681
- to23064 - to 304 48 - to 378 31 - lo 452 14
- to 10143 - to 17527 - to24910 - to32293 - to39677 -to47060
Fig. 7.15 Stress (at Crack Locations) in Reinforcement of the Deep Beam at the
Failure State
To validate the designed STM and nonlinear analysis techniques presented
previously, a scale-model of the deep beam was constructed, instrumented, and tested to
failure. The geometry of the scale-model, as 1/10 of the prototype design structure, was
shown in Fig. 7.3. The thickness of the scale-model was selected to be 6 in. (152 mm).
The material properties used in the scale-model are the same as the designed model was
given in Table 7.1. The reinforcement cage for the scale-model is shown in Fig. 7.16. A
welded wire mesh with a reinforcement ratio similar to that required in ACI3 l 8-08 for
this deep beam was provided as to provide a minimum level of ductility.
336
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. 7.16 Reinforcement Details of the Scale-Model
Based on the experimental results, the first diagonal crack was observed at the
web region of the specimen between the left support and the loading point at a load of 40
kips (178 kN), which is approximated to 40 x 10 x (24/6) = 1567 kips (6972 kN) in the
actual scale model and then propagated upward to the loading point. At the load of 94
kips (418 kN), or 3680 kips (16372 kN) in actual scale, the failure mechanism of the
specimen formed by crushing and splitting at the top concrete region just below the
loading point, shear compression failure at the regions above the opening, and crack slip
failure under the opening. The observed crack pattern of the scale-model at the failure
state and the close-up look of the shear compression failure of the specimen are
illustrated in Figs. 7.17 and 7.18, respectively.
337
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. 7.17 Crack Pattern of the Scale-Model at the Failure State.
Fig. 7.18 Shear Compression Failure above and below the Opening of the Scale-
Model.
The experimental results presented previously are now adopted for a comparison
of the design, analyses, and experiment. The summary of the nominal capacity and the
failure mode of the deep beam predicted using different analysis techniques is given in
338
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7 .2 and compared with the designed nominal capacity by indeterminate STM and
the measured capacity and observed failure mode from the experiment.
Table 7.2 Capacity and Failure Mode Comparison of Designs, Analysis, and Experiment
Model Pn (kips) Failure Mode
(kN)
STM (Designed)
(Pu/0.75) 1303 Yielding of the top reinforcement
(5800)
Plastic Analysis STM 3044 Crushing of strut above and below the opening
(13542) Yielding of the bottom reinforcement
Nonlinear elastic STM 3056 Crushing of strut above and below the opening
(13596) Yielding of the bottom reinforcement
Nonlinear FEA 3736 Crushing/spliting at the top
crushing/potential slips above and under the
(16622) opemng
Experiment
(small seal e) 94 crushing/ spliting at the top
(418) shear compression above the opening
Experiment
(actual seal e) 3680 crack slip failure under the opening
(16372)
From Table 7.2, using indeterminate STM with an assumption of equal relative
stiffness in all STM members result in a conservative design, as evaluated by the nominal
capacity predicted by nonlinear analysis techniques and observed by experiment. The
nominal capacity predicted from nonlinear STM analyses (plastic analysis and nonlinear
elastic analysis) is highly underestimate when comparing to the measured capacity from
the experiment with the ratio of the predicted capacity to measured capacity of 0.567 and
0.568, respectively. The inaccuracy of the prediction from nonlinear STM analyses is due
to the fact that the nonlinear STM adopts only the nonlinear characteristics of two-force
339
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
members (trusses) to approximate the load transfer characteristics inside the continuum.
Thus, the complete responses of the structure; for examples, redistribution of stress due to
cracks and slippage along a crack surface, are unable to be captured by the nonlinear
STM. In addition, the inaccuracy is also due to the use of yielding stress (Fy) as the
maximum limit in the stress-strain relation of STM members while, in the actual
structure, the capacity of the reinforcement could increase up to the fracture point (Fu) as
shown in Table 7.1.The predicted capacity from the FE model is most accurate with the
predicted to measured capacity ratio of 1. 015 and this is not surprising since the model
can capture many of the complexities of the observed behavior of the test structure
including the full material properties of the reinforcement.
Regarding the mode of failure, the nonlinear STM analysis were unable to predict
the experimentally observed failure mode, in which the brittle failure mode of the deep
beam was caused by the shear compression failure above and below opening governing
with the bottom reinforcement of the deep beam approaches yielding stress. This is thus
due to the complexity of the deep beam in which the nonlinear material models adopted
in the nonlinear STM analysis is insufficient to capture the true behavior. The predicted
mode of failure from FE model is highly accurate and is able to capture dominant and
critical failure mechanism observed in the experiment. FE model predicts crushing and
splitting of concrete at the top of the deep beam just under the loading plate. In addition,
crushing and potential slips above and under opening are accurately captured by the
model.
It should be noted that nonlinear STM has no capability in modeling and
capturing a distinct slip along crack surface as accurately as by the non-linear FEA. This
340
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
is due to the fact that STM is formulated as two-force members where only compression
and tension characteristics could be modeled. Thus, to capture more complex mechanics
in the structures, FEA should be adopted as the main analysis tool.
7.3.2 Simply Support Deep Beams with an Opening
The simply support deep beam with an opening is adopted as the second design
example to demonstrate the application of the proposed design and analysis framework.
In this example, the demonstration will mainly focus on applying the proposed
framework for design a complex D-Region subjected to non-reverse static loads to
achieve the most effective structural performance. The example will follow the design
and analysis steps as illustrated in Fig. 7.1. The topology optimization will be adopted to
generate the optimal STM shape used in design of the deep beam by using STM based on
satisfying the provisions of Appendix A of the ACB 18-08 code provisions. The designed
deep beam will be analyzed using Vector2 to assess the structural performance. The
predicted structural performance of the deep beam will be compared with the available
experimental data to validate the accuracy of the prediction. In addition, the structural
performance comparison between the deep beam designed by the framework and other
deep beams designed by conventional approach will be given.
As was illustrated in Fig. 7.2, the simply supported deep beam with an opening
can be considered as the beam in the left span of the continuous deep beam "ab c" after
the yielding of the top reinforcement at the column "b" occurs.
By using the small scale deep beam having the same dimension of the deep beam
as the dimension given in Fig. 7.la in chapter 3 in this design example, a simply
supported deep beam with a rectangular opening is selected as shown in Fig. 7.19 and the
341
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
beam is to be designed using STM. The deep beam with a clear spanned of 1200 mm has
a width of 100 mm and a 600 mm overall depth. The beam supports a single concentrate
load (P) from a column. The self-weight of the member is included in the concentrated
load. The objective of this design example is to adopt the proposed design and analysis
framework to obtain the most effective structural performance design of this small scale
deep beam; for example, the designed deep beam with the maximum capacity, highest
stiffness, maximum cracking load, and maximum ratio of the capacity to the volume of
the reinforcement.
t-400-t
o ~
0
0
N
0
l.O
("")
- - - - - ~ ~ ~ ~ ~ ~ ~ - - ~ J _
0
l.O
100H200
1
---1 ooo---1'--111 oo
Fig. 7.19 Members and Loads of a Simply Supported Deep Beam with a
Rectangular Opening
The material properties used in this design and analysis example are taken from
the material properties of the small scale specimen 4A to 4D (Fig. 3.la) that was used to
evaluate the design, as presented in Table 3.2 and 3.3.
Based on Fig. 7.1, the design procedure consisted of the following steps:
1) Discretize the structure into small FE meshes
342
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2) Perform the topology optimization based on the scheme given in section 5.2.1.
3) Construct STM model based on the optimal topology using the CAST
(Computer-Aided Strut-and-Tie) program by Tjhin (2004)
4) Determine the effective dimension of struts and the amount of reinforcement in
the ties to satisfy the requirements in Appendix A of ACI 318-08 where the
STM is intentionally designed to fail at the maximum possible load. This
involves adjusting the amount of reinforcement and the dimensions of struts,
nodes, and ties until key struts are limited by the dimensions of the structure
and stressed to the limit permitted by the ACB 18-08 provisions. To solve for
the forces in the indeterminate STM, all members are assumed have the same
axial stiffness (EA).
5) Arrange the reinforcement and check that adequate anchorage has been
provided.
6) Evaluate the structural performance and capacity based on the nonlinear FEA;
Vector 2 is adopted in this example as a computational tool.
7) Compare the structural performance and capacity of the deep beam with other
deep beams designed by using other STM shapes and conventional approaches.
It should be noted that no comparison of the predicted structural responses of the
deep beam designed by using the topology optimization with the experimental program is
performed in this type of the structure since no experimental program is available as
343
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
stated previously in chapter 5. Thus, only the calibrated FE model is adopted to evaluate
the structural performance and capacity of the deep beam.
From steps 1 and 2, the optimal topology is illustrated in Fig. 7.20a. The
corresponding performance index (Pl) obtained for the topology is also given in Fig.
7.20b. The optimal topology is then used to guide the construction of STM model using
the CAST software. The STM model from CAST is illustrated in Fig. 7.20c. From Fig.
7.20c, the orientation of STM members is selected as closely as possible to the optimal
topology, particularly, the orientation of the tension tie because the optimal topology
theoretically forms the skeleton structure inside the design domain which follows the
principal compressive and tensile stress direction of the material. Thus, using the
orientation of tension tie based on the optimal topology will be most effective in resisting
the tension demand and controlling the crack development and propagation.
Fig. 7.21a shows the distribution of the forces in the strut-and-tie model. A
positive value is used to denote tension, and a negative value to denote compression. Fig.
7.21b shows the effective widths selected for the struts and ties and the utilization rates
(FctemanctlFcapacity) in STM of the deep beam. From Fig. 7.21b, the strength of STM is
shown to be 332 kN with the failure caused by crushing of struts spanning between nodes
N39 and N53 and between nodes N39 and N54 as the utilization rate of the struts
approaches 1.0.
344
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.8
Perform;;tn.C8 Ind.ex. l:ilstory
1.6
1.4
1.2
1
a:
0.8
0.6
0.4
0.2
o
o 20 40 60 80
Iteration No
(a) (b)
(c)
Fig. 7.20 The Optimal Topology for a Simply Supported Deep Beam with a
Rectangular Opening: (a) The Optimal Topology; (b) The Performance Index; and
(c) STM model from CAST
345
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
97.8 kN
~ < Y - r - - - - - - ~
I 111rl.r.
~ ~
______ ... ~
'
~
(a)
(b)
Fig. 7.21 STM Model from CAST for a Simply Supported Deep Beam with a
Rectangular Opening: (a) The Force Distribution in Struts and Ties; (b) The
Effective Widths selected for the Struts and Ties and the Utilization Rates
(Fdemand!Fcapacity) in STM of the deep beam
346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In the next step, the deep beam with the reinforcement details (Fig. 7.22a)
obtained from the STM is analyzed by using Vector2 to predict the structural responses
and to assess the structural performance. The FE model (Fig. 7.22b) of the deep beam
was constructed using 1281 quadrilateral 4 nodes elements. Based on the calibration
technique used in section 5.5.2, the FE mesh size is approximately 2.67 times the
maximum aggregate size (3/8" or 9.5 mm) of the mixed concrete used in the
experimental program for validating the deep beam after completing the design process
(chapter 3). The mesh generation available in the pre-processing tool "Patran" is used to
generate the mesh oriented in the orientation of the reinforcement as illustrated in the
reinforcement detail in Fig. 7.22a. The constitutive models adopted in the analysis are
DSFM (included explicit cracked shear stress-slip components), Popovic (NSC) for
concrete pre-peak model, 1992 Vecchio model A for compression softening model, 2003
Bentz model for tension stiffening, and a bilinear model for tension softening model. The
main reinforcement was modeled by using the 183 truss elements and the perfect bond
condition is adopted in the model. Material properties for concrete and reinforcement
adopted in the FE models are obtained from Table 3.2 and 3.3, respectively.
347
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
p
(a)
(b)
Fig. 7.22 Reinforcement Details and FE model of a Simply Supported Deep Beam
with a Rectangular Opening: (a) Reinforcement Details; and (b) FE Model
The results from the FE analysis are illustrated in Figs. 7.23 to 7.26. The
predicted load-displacement relationships for the deep beam (4PBO) are presented in Fig.
7.23 in which the predicted condition of the beam at key points is illustrated. The first
348
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
crack is predicted to initiate at the load about 75 kN and this is shown to result in a
slightly change in the stiffness of the response. The slightly change in the stiffness after
the first crack is because of the effectiveness of the reinforcement provided in the
orientation of the optimal topology in resisting the induced tension demand that
potentially causes more propagation of the damages and degradation of the stiffness.
From the analysis, no yielding of the reinforcement is predicted to yield (Fig. 7 .26). The
deep beam is finally predicted to fail by crushing of the concrete region just above the
opening (Fig. 7.24) at the capacity of 346.5 kN. The failure regions above the opening
correspond to the location of the compression struts N39 and N53 in Fig. 7.4a. The vital
signs, the ratio of the principal compressive stress to the maximum available compressive
capacity, of the deep beam at the failure load is also illustrated in Fig. 7.25 and illustrates
crushing of the concrete just above the opening.
By comparing the predicted capacity of the deep beam with the nominal design
load from STM model, the capacity calculated by the STM model is found to slightly
underestimated the capacity of the specimen and thus the deep beam is safely designed.
In addition, the predicted failure mode of the deep beam also agrees well with the failure
mode obtained from STM model.
For the structural performance evaluation of the deep beam, the predicted
capacity and structural responses of the deep beam are compared with the predicted
values of the other deep beams designed by using different STM shapes as illustrated in
Fig. 7.23. Overall, the deep beam (4PBO) has the highest post cracking stiffness and
capacity. The mode of failure by crushing of concrete just above the opening is predicted
to occur in the deep beam and this is similar to the failure modes predicted to occur in
349
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
other deep beams. In addition, the parameters to measure the structural performance of
the designed deep beam are plotted in Fig. 7.27; for examples, capacity, reinforcement
volume, and ratio of capacity to reinforcement volume. From the plot, the deep beam
designed using the proposed design and analysis framework has the highest capacity and
maximum ratio of capacity to reinforcement volume. Those parameters thus confirm the
achievement of the most effective structural performance design of the deep beam by
using the proposed design and analysis framework.
400
----------------------- - - - - - - ~ - - - - - - -
!
350
Crushing of concrete ab ve the opening!
l
300
- - - - - - - - - ~
--- ...._ I
250
-
2 0 0
-
/ ... -
~ .k
x- ....... crushing of concrete above t
0.. "x/.,.,.k"'
150 ....:: > Crushing of concrete above he support
100
-+-4PBO
--4A
50
------48
- -x- 4C
----40
0
0 0.5 1 1.5
Displacement(mm)
Fig. 7.23 The Predicted Load and Displacement of the Deep Beam from
Nonlinear FEA by Vector 2
350
2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
to 021
- to 0.25
- to 029
- to 033
fem
to O 54
- to 0.58
- to 063
-to067
s: fem Displacemert Feictor" 1.BO
to 088
- to 092
- to 096
-to1.00
Fig. 7.24 Concrete Compressive Vital Sign in the Deep Beam: Ratio of Compressive
Stress to Compressive Stress Capacity.
Combined View
I I I \ -- /J/ '- "
1111-rttnH+
1
t\+-J-L
1
+-1-LJ:',,.; / ,/
/
' \ \ ,_
I -
Specimen 7 A (Abltqus irp.i:) D1splaceiCracks. Combined YICM' Oispl&eemeri factor"' 1 60
Displacemerf tk9'1ficotion: 20.00 x
CrliCkVVdhs: thin< 1.00 mm, mid, thick> 2.00 mm
Fig. 7.25 Combined Deformed Shape and Crack Pattern at the Failure State of the
Deep Beam
351
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
,_,__
~ ......
c--,_
""""
....
,_
--
'-'
v
v
-
v
v
-----
/
-....
,___
r--
i--.._
,___
-
---
,__ __
,___
,___
--
to 61.71
- to104.25
- to12679
-10149.33
er-truss at crack
... !:;:;: -,__
t- t-
-- ............ i--._._
--i--
-
,__ ...
~
~
-,._,__
,___
-
--
-
~ ~ ... -c-- ,___
,___
r-
~
~
to171.66
- to19442
- to21696
- to239.50
/ / /
-v
'-'"'
/
I
//
I I I
I I
rfilfj!
v ':. 1---.
-
... _
to 262.04
- to264.56
- to30713
- to329.67
~
';X
".)..-
/!
_ ....
.---
I-
........
......
'---
V'
S'tress steel
to 352.21
- to374.75
- to39729
- to41963
~ ..
~
-
c-
~
~
';;"
"'v
>-- -
_'--1
.--'
'"'
,...__
-
_,,v
v
------
~
v"'-
,...
~ l l l
----
.....
-- -
o-truss al crack Disptacemert Factor = 1 .BO
to 442.38
-to464.92
- to48746
!lli!lill to51000
Fig. 7.26 Stress (at Crack Locations) in Reinforcement of the Deep Beam at the
Failure State
0.4
0.35
...
.2! 0.3
~
~ 0 . 2 5
cu
~ 0.2
cu
c: 0.15
cu
!
0 0.1
0.05
0
~ Strength(kN)*1000 RebarVolume(mm"3)x10"7
o Strength/Rebar Volume (N/mm"3)
4a 4b 4c
Specimen No.
4d 4pbo
Fig. 7.27 The Structural Performance Evaluation of Designed Simply Supported
Deep Beams with an Opening.
352
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7.3.3 Simply Support Deep Beams with a Dapped End and an Opening
The simply-supported deep berun with a <lapped end and an opening is adopted as
the third design example to demonstrate the application of the proposed design and
analysis frrunework for achieving the most effective structural performance design of
complex D-Regions subjected to non-reverse static loads. The example will follow the
design and analysis steps as illustrated in Fig. 7.1. The topology optimization will be
adopted to generate the optimal STM shape used in design of the deep beam by using a
STM design to satisfy Appendix A of ACB 18-08 code provisions. The designed deep
berun will be analyzed by using FEA with the proposed constitutive models and FE
formulations developed in chapter 7 to assess the structural performance. The predicted
structural performance of the deep beam will be compared with the available
experimental data to validate the accuracy of the prediction. In addition, the structural
performance comparison between the deep berun designed by the framework and other
deep beams designed by conventional approach will be given.
The simply support deep berun with a <lapped end and an opening adopted in this
design example could be illustrated in Fig. 7.2 as a part of the continuous deep beam "de
f'. Clearly, by designing the continuous deep beam to allow yielding of the top
reinforcement above the column "e", the simply supported deep berun with a <lapped end
and an opening can be considered as the beam in the right span of the continuous deep
beam "def'.
By using the small scale deep berun having the srune dimension of the deep berun
as the dimension given in Fig. 3.lc in chapter 3 in this design example, a simply
supported deep beam with a rectangular opening can be shown in Fig. 7.28. The deep
353
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
beam with a clear span of 1800 mm has a width of 150 mm and a 1000 mm overall depth.
The rectangular opening has a width and a depth of 400 mm and 200 mm, respectively.
The beam supports a single concentrated load (P) from a column. The self-weight of the
member is included in the concentrate load.
D
1
t.,()
_L
8
N
-,--
0
~
..._ __________________________ ,_.__,____!__
50
350-f-soo-----+--I ~
150 150
Fig. 7.28 Members and Loads of a Simply Supported Deep Beam with a Dapped
End and an Opening
The material properties uses in this design and analysis example are taken as the
material properties of the small scale specimen 6A to 7C (Fig. 3.lc) that was used to
evaluate the design, as presented in Tables 3.2 and 3.3.
Based on Fig. 7.1, the design procedure consisted of the following steps:
1) Discretize the structure into small FE meshes
2) Perform the topology optimization based on the scheme given in section 5.2.1.
3) Construct STM model based on the optimal topology using the CAST
(Computer-Aided Strut-and-Tie) program by Tjhin (2004)
4) Determine the effective dimensions of struts and the amount ofreinforcement in
354
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the ties to satisfy the maximum possible capacity design and the requirements
in Appendix A of ACI 318-08. In addition, the indeterminate STM is solved by
assuming that all members have the same axial stiffness (EA).
5) Arrange the reinforcement and check that adequate anchorage has been
provided.
6) Evaluate the structural performance and capacity based on the nonlinear FEA
with the proposed constitutive models and FE formulation as described in
Chapter 6.
7) Investigate the adequacy of the design and non-linear analysis by conducting a
scale-model test of the designed structure.
8) Compare the structural performance and capacity of the deep beam with other
deep beams designed using other STM shapes and conventional approaches.
From steps 1 and 2, the optimal topology is illustrated in Fig. 7.29a. The
corresponding performance index (Pl) obtained for the topology is also given in Fig.
7.29b. The optimal topology is then used to guide the construction of STM model using
the CAST software. The STM model from CAST is illustrated in Fig. 7.29c. The model
was constructed to have the orientation of STM members as closely as possible to the
optimal topology for the most effectiveness in structural performance during the service
state as pointed out in chapter 5.
However, the orientation of the reinforcement based on the optimal topology
might be impractical for fabrication and construction. Thus, the more practical
reinforcement pattern may need to be slightly adjusted from the orientation of the optimal
355
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
topology to be in the orthogonal orientation could be also possibly obtained as illustrated
in Fig. 7.29d.
1.6
Performance Index History
1.4
1.2
ii:0.8
0.6
0.4
0.2.
0
0 20 40 60 80 100
Iteration No
(a) (b)
(c) (d)
Fig. 7.29 The Optimal Topology and STM for a Simply Supported Deep Beam with
a Dapped End and a Rectangular Opening: (a) The Optimal Topology; (b) The
Performance Index; (c) STM Model (Optimal); and (d) STM Model (More
Practical)
Fig. 7.30a shows the distribution of the forces of the STM model in Fig.7.29c. A
positive value is used to denote tension, and a negative value denotes compression. Fig.
7.30b shows the effective widths selected for the struts and ties and the utilization rates
(Fctemanct!Fcapacity) in the STM. From Fig. 7.30b, the strength of STM is shown to be 1231
kN with the failure caused by crushing of struts spanning between nodes N21 and N25
based on the utilization rate of the strut equal to 1.0.
356
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
For the STM model in Fig. 7.29d, the force distribution and the utilization rates
(FctemanctlFcapacity) are illustrated in Fig. 7.31a and 7.31b, respectively. The strength of the
STM is equal to 731 kN and the failure of the model is due to crushing of the
compression strut spanning between nodes N2 and N32. It is very interesting that the
strength suggested from the STM model when the more practical reinforcement pattern is
used is significantly lower than the strength of the STM model with the optimal
reinforcement pattern is used. This is due to the difference of the orientation of the truss
members and the assumption of the relative stiffness used in solving the indeterminate
STM. The more accuracy in predicting the strength, structural responses, and structural
performance by using nonlinear FEA is to be performed in the next step to ensure the
accuracy in quantitative comparison between both designs.
In the next step, the deep beam with the optimal reinforcement details (Fig. 7.32a)
obtained from the optimal shaped STM referred as "7B" is analyzed by using the
proposed nonlinear FEA to predict the structural responses and to assess the structural
performance. The FE model (Fig. 7.32b) of the deep beam was constructed by using 198
quadrilateral 4 nodes elements. The proposed behavior model in chapter 7 is adopted in
the model. No consideration of the deviation of crack angle from the principal
compressive stress direction is used in the analysis. The main reinforcement was modeled
by using the proposed embedded reinforcement including bond stress-slip. The local
bond-slip relation employs the model from CEB-FIP model code for monotonic loading
and unconfined concrete. Good bond conditions are assumed in the model.
357
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
(b)
Fig. 7.30 STM Model (Optimal) from CAST for a Simply Supported Deep Beam
with a Dapped End and a Rectangular Opening: (a) The Force Distribution in
Struts and Ties; (b) The Effective Widths selected for the Struts and Ties and the
Utilization Rates (Fdemand1Fcapacity) in STM of the Deep beam
358
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(a)
(b)
Fig. 7.31 STM Model (More Practical) from CAST for a Simply Supported Deep
Beam with an Dapped End and a Rectangular Opening: (a) The Force Distribution
in Struts and Ties; (b) The Effective Widths selected for the Struts and Ties and the
Utilization Rates (Fdemand!Fcapacity) in STM of the Deep beam
Material properties for concrete and reinforcement adopted in the FE models are obtained
from Tables 3.2 and 3.3, respectively.
359
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Similarly, the FE model of the deep beam with the more practical
reinforcement (Fig. 7.33a) referred to as "7C" is also constructed by using the proposed
constitutive models and FE formulation in Chapter 7. With the advantage of using the
embedded reinforcement, the same structural mesh as the deep beam with the optimal
reinforcement pattern is adopted. Thus, the FE model (Fig. 7.33b) of the deep beam was
also constructed by using 198 quadrilateral 4 nodes elements. The same local bond-slip
relation and the material properties for concrete and reinforcement as the deep beam with
the optimal reinforcement are adopted in the model.
The predicted load displacement responses from the FE analysis of deep
beams 7B and 7C are illustrated in Fig. 7.40. The responses from other deep beams
designed using different STM shapes are also plotted in the figure for comparison. The
results from the FE analysis of the deep beam with the optimal reinforcement denoted as
7B are additionally illustrated in Figs. 7.34 to 7.36 while the additional analysis results of
the deep beam with the more practical reinforcement denoted as 7C are illustrated in
Figs. 7.37 to 7.39. Overall, the predicted load displacement responses from deep beams
7B and 7C are significantly stiffer than other deep beams designed using different STM
shapes, particularly, the post cracking stiffness. The predicted cracking loads of both
deep beam 7B and 7C are also higher than other deep beams. By comparing the predicted
load-displacement for the deep beam 7B with that of the deep beam 7C, as presented in
Fig. 7.40, the first crack of the deep beam 7B is predicted to initiate at the load about 470
kN while the observed cracking load of deep beam 7C was about 325 kN. After cracking,
deep beam 7B is predicted to have much higher post cracking stiffness than deep beam
7C. The significant difference in the cracking load and the post cracking stiffness of both
360
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
deep beams (7B and 7C) is considered to be due to the orientation of the reinforcement
used in fabrication of those beams. While deep beam 7C has more practicability and
constructability of the reinforcement, deep beam 7B has much more structural
performance during the service load. From the analysis, no yielding of the reinforcement
is predicted to yield (Figs. 7.36 and 7.39) in deep beams 7B and 7C. The predicted stress
in the reinforcement of deep beam 7B is more uniform and less than the predicted stress
of deep beam 7C. Deep beams 7B is finally predicted to fail by crushing of the concrete
regions just underneath the loading point and no critical cracks in the specimen at the
capacity of 1190 kN (Fig. 7.34) while deep beam 7C is predicted to fail by crushing of
concrete underneath the loading point and have some critical cracks and slips underneath
the opening at the capacity of 1140 kN (Fig. 7.37), respectively. The failure regions
underneath the opening of specimen 7B correspond to the location of the compression
struts N2 and N40 in Fig. 7.30. For the specimen 7C, the failure regions correspond to the
compression struts N2 and N28 in Fig. 7.31.
By comparing the predicted capacity of the deep beams with the nominal design
load from STM model, STM models of both deep beams 7B and 7C result in
conservative designs. In addition, the predicted failure mode of the deep beam 7B agrees
well with the failure mode obtained from STM model. However, the failure mode
obtained from STM model of deep beam 7C is inaccurate; for example, no slippage is
possible to capture in STM.
To validate the designed STM and nonlinear analysis techniques presented
previously, specimens of deep beams 7B and 7C were constructed, instrumented, and
tested to failure. The geometry and thickness of the specimens are the same as the
361
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
prototype design structures as shown in Fig. 7.28. The material properties used in the
specimens are the same as the designed model as given in Tables 3.2 and 3.3. The
reinforcement cages for the specimens 7B and 7C are shown in Figs. 7.41 and 7.42,
respectively. A welded wire mesh with a reinforcement ratio similar to that required in
ACB 18-08 for deep beam was provided as to provide a minimum level of ductility.
Based on the experimental results, in specimen 7B, the first cracks form at the
corner of the <lapped end at the load of 60 kips (267 kN). Then new cracks form at the
corners of the opening at the load of 75 kips (333 kN). At a load of 160 kips (711 kN),
the diagonal crack above the opening forms and causes the cantilever action of the part of
the deep beam above the opening. Finally, at the load of 292.5 kips (1300 kN), the
specimen fails by sudden crushing of the concrete adjacent the loading point. The
observed crack pattern of specimen 7B at the failure state is illustrated in Fig. 7.43.
In specimen 7C, the first crack was observed at the corner of the opening at a load
of 30 kips (133 kN) and then propagated upward to the loading point. At the load of 35
kips (155 kN), new crack forms at the <lapped end. The critical shear crack underneath
the opening forms at the load of 110 kips ( 488 kN). At the load of 234 kips (1040 kN),
the failure mechanism of the specimen formed by crushing at the concrete region just
above the opening and sliding shear failure at the regions underneath the opening. The
observed crack pattern of specimen 7C at the failure state is illustrated in Fig. 7.44.
The experimental results presented previously are now adopted for a comparison
of the design, analyses, and experiment. The summary of the nominal capacity and the
failure mode of the deep beam predicted using the proposed FE analysis technique is
362
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
given in Table 7 .3 and compared with the designed nominal capacity by indeterminate
STM and the measured capacity and observed failure mode from the experiment.
From Table 7.3, as expected in both specimen 7B and 7C, using indeterminate
STM with an assumption of equal relative stiffness in all STM members results in a
conservative design. The predicted capacity from FE model is highly accurate with the
predicted to measured capacity ratio of 0.92 and 1.09 for specimens 7B and 7C,
respectively. The predicted modes of failure from FE models of both specimens are
exceptionally accurate since those FE models could capture the types and locations of the
failure modes very reasonably as shown in Table 7.3. This is not surprising since the FE
model include many of the complexities and nonlinear behaviors in the formulation as
presented in chapter 6.
For the structural performance evaluation of the deep beam, based on the
experimental program, the measured capacity and structural responses of the deep beam
are compared with the measured values of the other deep beams designed by using
different STM shapes as illustrated in Fig.7.45. In overall, the deep beam (7B) has the
highest capacity. The observed crack pattern in deep beam 7B (Fig. 7.43) shows
evidently less damage than in deep beam 7C (Fig. 7.44), particularly at, service load and
up to failure of the specimen. In addition, the parameters to measure the structural
performance of the designed deep beam; for examples, capacity, reinforcement volume,
and ratio of capacity to reinforcement volume confirm the achievement of the most
effective structural performance design of the deep beam 7B by using the proposed
design and analysis framework since the deep beam has the highest capacity and
maximum ratio of capacity to reinforcement volume.
363
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
p
(a)
~ ~ -
\. ..;"
\
/
I\. ./
\ v
\ /
/
,,
/
v
'\.
.1'
"'
- ~
\
\.
--

~ . . . . .
. . . . . ~
\.
_..,,..
~
-
(b)
Fig. 7.32 Optimal Reinforcement Details and FE model of a Simply Supported Deep
Beam with a Dapped End and a Rectangular Opening: (a) Reinforcement Details;
and (b) FE Model
364
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3#4
(a)
~
\.
'
\.
\
\
'
\
'
--
\
'I\
\
\
- -
(b)
Fig. 7.33 More Practical Reinforcement Details and FE model of a Simply
Supported Deep Beam with a Dapped End and a Rectangular Opening: (a)
Reinforcement Details; and (b) FE Model
365
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
No fie Loaded
- to 0.04
- lo 0.08
-to0.12
-to0.16
to 020
-to0.24
- to 028
-to0.32
to 036
-lo0.41
-to045
- to 0.49
fem
to 0.53
-to0.57
-to0.61
- to 0.65
to 0.69
-to0.73
- to 077
-to0.81
to 0.65
lll!Bllto0.69
llll!llllm to 0.93
lllm to 0.97
Fig. 7.34 Concrete Compressive Vital Sign in the Deep Beam (Optimal
Reinforcement): Ratio of Compressive Stress to Compressive Stress Capacity.
Combined View
~ I / /
-
rTTT1 -
I
\
I I
/
I
I
\ I
/ / I
I
\
I I
/
\
\
\
/
I
//j
/
I I
I I I / / I I \
\
\
\
\
///
I I I I
I I \ \
\
"'
~
\
I I
----
\
~ / /
/ /
I I I \ \
\ ''\
,__ ,__
I \
\ \
- -
I
\ \
//
I I I I \ \ \
\
\ \
\ \
I I I I I
I I / I
""
\
\
I I I I I I I \
\
\ \
\
\ \
\ I
\
"'- \
I I
I / I I \ \
\ \
\
I \
\
c----
No file Loaded Displece/Crtds Combined View Lo&d Factor 4 20
Oi$pl!JCemeri Ma{lnificlll:ion: 20.00 x
Crack Ylo4cthS. thi1 c: 1 00 mm, mid, thick :> 2 00 !Tiil
Fig. 7.35 Combined Deformed Shape and Crack Pattern at the Failure State of the
Deep Beam
366
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I\.
\
"
\
~ ~
\
'f\.
'\
\
to-24043
-to-215.93
-lo-19144
- to-166.94
'
er -truss at crack
-
I\
\
'
-to-142.44
-to-117.94
-to-9344
- ICJ-68.94
-
/
v
_,,,
to-4444
- to-19.94
-to4.56
- to29.06
./
~
/
v
v
v
--
_,...--
....---
....--
- -
to 5355 to151.55
- to 7805 - to176.05
- to102.55 ll!lllllB to20'.J.55
- to 127 05 G!illll to 225.0S
Fig. 7.36 Stress (at Crack Locations) in Reinforcement of the Deep Beam at the
Failure State
to 038
-to 0.42
-to046
-to0.51
fem
to 0.55
- to 0.59
- to 0.63
-to0.67
to 072
- to 0.76
-tooao
-toO.a4
fem load Factor" 3 40
to 069
-to0.93
l!mmto0.97
11111111 to 1.01
Fig. 7.37 Concrete Compressive Vital Sign in the Deep Beam (More Practical
Reinforcement): Ratio of Compressive Stress to Compressive Stress Capacity.
367
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Combined View
\!__//I/ I
I I
\
ii<:_ I I I I
"'
'
I /
'
\\
I I
""
'"'
77< I/// I I I I \ \
"" ""
\

I
I I \
\

""
""

""
\
,__,___<I I
I I
I I I
I I \
\
\
\
""
\ I./ \
/ /
I I
I I
I \ \
\
\
\ \
I
I I
I I
I
\\ \ \ \
\ \
I
I I
I I
I I
I \

I I I
I I
I I
I
I
I I
\ I
I
I I
I I
I I
\
I


Displace.Qecks: Combined View load Factor .. 4 60
Displ&cemert Magnificttioo: 20.00 x
Crack \Afdlhs, thin< 1.00 mm, mid, thick" 2.00 rTJTI
Fig. 7.38 Combined Deformed Shape and Crack Pattern at the Failure State of the
Deep Beam (More Practical Reinforcement)
cr-truss at crack
'\.
\
"
\
I\
'
'
'

\
'
\.
\
\.

No file Loaded stress (steel): at crack Load Factor .. 4.60
- to-203.00 - lo-10485
-to-669
- to91.46
to189.61 - to2B7.n
- to-17846 - lo-80.31 - to1784 - to116.00 - 10214.15 - 1031230
- to-15392 - lo-55.77 - to4238 - to140.54 - to23889 - 1033684
- to-12938 - to-3123 - to66.92 - to165.07 - to263.23 - 1038138
Fig. 7.39 Stress (at Crack Locations) in Reinforcement of the Deep Beam (More
Practical Reinforcement) at the Failure State
368
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-
z
~
-
'C
cu
0
..J
1400
1200
1000
800
600
400
200
78, Exp=1300 kN ,
...................................... nUnoHUUUUHuHUHuu ............ u ....................... uunoHUHOUUUH!
6C, Exp=1292 kN
7B,ACl=1150kN
0 - e - ~ ~ ~ - - . ~ ~ ~ ~ . . . . . . - ~ ~ ~ - - . ~ ~ ~ ~ . . . . . . - ~ ~ ~ - - ;
0 1 2 3 4 5
Displacement (mm)
Fig. 7.40 The Predicted Load and Displacement of the Deep Beams by Using
the Proposed Nonlinear FEA
369
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. 7.41 Reinforcement Details of Specimen 7B
Fig. 7.42 Reinforcement Details of Specimen 7C
370
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. 7.43 Crack Pattern of the Specimen 7B at the Failure State.
Fig. 7.44 Crack Pattern of the Specimen 7C at the Failure State.
371
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7.3 Capacity and Failure Mode Comparison of Designs, Analysis, and Experiment
Model 7B Pn (kN) Failure Mode
STM (Designed) Crushing of bottle shaped strut above the left
(Pu/0.75) 1231 support
Crushing of concrete underneath the loading
Nonlinear FEA 1190 point
Experiment Crushing of concrete adjacent to the loading
(small scale) 1300 point
Model 7C Pn (kN) Failure Mode
STM (Designed) Crushing of bottle shaped strut under the
(Pu/0.75) 731 loading point
Crushing of concrete underneath the loading
Nonlinear FEA 1140 point
Critical shear crack underneath the opening
Experiment
(small scale) 1040 Shear compression above the opening
Slide shear failure underneath the opening
372
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.4
II Strength(kN)*1000
o Stren th/Rebar Volume
Rebar Volume(mm"3)x10"7
... 1.2 -+---------
.f!
1

ns
a..0.8
"C
cu
c:::0.6
cu
t.n
(30.4
0.2
0
6A 68 6C 60 78 7A
Specimen No.
7C
Fig. 7.45 The Structural Performance Evaluation of Designed Simply Supported
Deep Beams with a Dapped End and an Opening.
7.3.4 Shear Wall with Openings
The shear wall with openings is adopted as the last design example to
demonstrate the application of the proposed design and analysis framework for achieving
the most effective structural performance design of complex D-Regions subjected to
reverse static loads. This example will follow the design and analysis steps as illustrated
in Fig. 8.1. The scheme of topology optimization for multiple load cases as presented in
section 5.2.2 will be adopted to generate the optimal STM shape used in design of the
shear wall by completing an STM design to satisfy the provisions in Appendix A of the
ACB 18-08 code provisions. The designed shear wall will be analyzed by using the static
push over analysis in order to estimate the ultimate strength and ductility capacity of the
shear wall. The shear wall will be modeled by using FE with the proposed constitutive
models and FE formulations developed in chapter 7 to assess the structural performance.
373
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The predicted structural performance of the shear wall will be compared with the
available experimental data to validate the accuracy of the prediction. In addition, the
structural performance comparison between the shear wall designed by the framework
and other shear wall designed by conventional approach will be given.
The shear wall with openings adopted in this design example could be illustrated
in Fig. 7.2 as a part of the multi-storey shear wall. Only the first storey shear wall is
considered in this design example. By considering the lateral loads applied to the whole
building, it is straightforward to obtain the total lateral load that is transferred to the first
storey shear wall as illustrated in Fig. 7.46. In the design process, all load cases are
required to be considered in the shear wall; for example, the gravity load, the lateral loads
acting on the left side of the shear wall, the lateral loads acting on the right side of the
shear wall, and the combination of the gravity load (dead and live) and the lateral loads.
However, in this design example, only the lateral load transferred from the upper storey
to the right side of the shear wall will be considered for the simplicity of the comparison
with the available experimental data as previously presented in chapter 5.
By using the small scale shear wall having the same dimension as the right side
of the specimen given in Fig. 3.ld in chapter 3 in this design example, a shear wall with
rectangular openings can be shown in Fig. 7.47. The shear wall has a height of 900 mm, a
width of 1125 mm, and a thickness of 175 mm. Both rectangular openings have a width
and a height of 300 mm and 265 mm, respectively. The shear wall supports a single
lateral concentrated load (P) transferred from the upper storey to the right side of the
shear wall. The self-weight of the member, other gravity loads (dead and live load), and
374
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
transferred bending moment from the upper storey are not included in this design
example.



D

D

D
.___
D
+--

\J
Fig. 7.46 The Lateral Loads transferred from the Upper Storey to the Shear Wall
25
0
("')
265
.....
900 300
65
150
1125
Fig. 7.47 Members and Loads of a Shear Wall with Openings (No Consideration of
the Applied Gravity Load and Bending Moment)
375
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The material properties uses in this design and analysis example are taken as the
material properties of the small scale specimen 8A to 8C (Fig. 3.ld) that was used to
evaluate the design, as presented in Tables 3.2 and 3.3.
Based on Fig. 7.1, the design procedure is extended for the complex D-Regions
subjected to reversed loading as the followings:
I) Discretize the structure into small FE meshes
2) Perform the topology optimization based on the scheme given in section 5.2.2.
3) Construct STM model based on the optimal topology using the CAST
(Computer-Aided Strut-and-Tie) program by Tjhin (2004)
4) Determine the effective dimension of struts and amount ofreinforcement in the
ties that is needed to satisfy the maximum possible capacity design and the
requirements in Appendix A of ACI 318-08. In addition, the indeterminate STM
is solved by assuming that all members have the same axial stiffness (EA).
5) Arrange the reinforcement and check that adequate anchorage has been
provided.
6) Evaluate the structural performance and capacity by using the static pushover
analysis with the use of nonlinear FEA included the proposed constitutive
models and FE formulation as described in Chapter 7.
7) Investigate the adequacy of the design and non-linear analysis by conducting a
scale-model test of the designed structure.
8) Compare the structural performance and capacity of the shear wall with other
376
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
shear walls designed using other STM shapes and conventional approaches.
From steps 1 and 2, the optimal topology of the shear wall subjected to reversed
loading is illustrated in Fig. 7.48a. The corresponding performance index obtained for the
topology is also given in Fig. 7.48b. The optimal topology is then used to guide the
construction of STM model using the CAST software for loads acting on the shear wall
in either the left or right direction. The STM models from CAST for loads acting on the
left and right side of the shear wall are illustrated in Fig. 7.48c and 7.48d, respectively.
The models were constructed to have the orientation of STM members as closely as
possible to the optimal topology for the most effectiveness in structural performance
during the service state as pointed out in chapter 5.
In addition, in this design example the shear wall designed based on the
conventional approach is also presented. The conventional approach generally selects the
STM shape based on the principal stress direction from the linear elastic FEA. The STM
shape based on the conventional approach is illustrated in Fig. 7.49. The orientation of
the reinforcement based on the conventional approach is also selected to be in the
orthogonal direction for the structural performance comparison with the shear wall
designed by using the optimal topology.
377
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.


(c)
D
'
'
'
'
1.6
1.4
1.2
0::
0.8
0.6
0.4
0.2
0
0
10 20 30lteratio'WNo. 50 60 70
(b)
D
(d)
Fig.7.48 - Optimal Topology, STM, and Performance Index History of the Shear
Wall Side of D-Regions in Fig. 3.ld: (a) Optimal Topology; (b) Performance Index
History; (c) STM generated from CAST for Load acting on the Left Side; and (d)
STM generated from CAST for Load acting on the Right Side.
378
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
'
D
'
'
'
Conventional approach
'
'
'
'
'
'
'
'
Load Case1 Load Case2
(a) (b)
Fig.7.49 The Conventional Approach for Design of Complex Regions subjected to
Multiple Load Cases (Tjhin and Kuchma, 2002): (a) STM generated from CAST for
Load acting on the Left Side; and (b) STM generated from CAST for Load acting
on the Right Side.
Fig. 7.50a shows the distribution of the forces of the STM model in Fig.7.48d. A
positive value is used to denote tension, and a negative value denotes compression. Fig.
7.50b shows the effective widths selected for the struts and ties and the utilization rates
(FctemanctlFcapacity) in the STM. From Fig. 7.50b, the strength of STM is shown to be 334.7
kN with the failure caused by crushing of the prismatic struts spanning between nodes
N35 and N36 based on the utilization rate of the strut is equal to 1.0. This prismatic strut
transferred the load from the loading point into the shear wall.
For the STM model in Fig. 7.49b, the force distribution and the utilization rates
(FctemanctlFcapacity) are illustrated in Fig. 7.Sla and 7.51b, respectively. The strength of the
STM is also equal to 3 34. 7 kN. The capacity of the shear wall is also limited by the
prismatic strut spanning between nodes N35 and N36 that transfers the lateral load into
the shear wall. This is not surprising since both shear walls are designed for the
379
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
maximum possible load resistance where the design approach is intentionally limiting the
capacity of the D-Regions by the stress limits and dimensions of the struts. At this load
resistance, an increase of the reinforcement quantity will not increase the capacity of D-
Regions.
In the next step, the shear wall with the optimal reinforcement details (Fig. 7.52a)
obtained from the optimal shaped STM is analyzed by using the static push over analysis
to estimate the structural responses, capacity, ductility, and structural performance. The
shear wall was modeled by using the proposed nonlinear FEA. The FE model of a half of
specimen 8B was adopted for modeling the shear wall as illustrated in Fig. 7.52b. The
whole specimen 8B was constructed by using 524 quadrilateral 4 node elements. The
proposed behavior model of concrete element in chapter 7 is adopted in the model. No
consideration of the deviation of crack angle from the principal compressive stress
direction is considered in the analysis. The main reinforcement was modeled using the
proposed embedded reinforcement including bond stress-slip. The local bond-slip relation
employs the model from CEB-FIP model code for monotonic loading and unconfined
concrete type. Good bond conditions are assumed in the model.
380
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
, ~ ,
/
(a)
(b)
Fig. 7.50 STM Model (Optimal) from CAST for a Shear Wall with Rectangular
Openings: (a) The Force Distribution in Struts and Ties; (b) The Effective Widths
selected for the Struts and Ties and the Utilization Rates (Fdemand!Fcapacity) in STM of
the Shear Wall
381
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[
=o
1
(a)
(b)
Fig. 7.51 STM Model (Conventional Approach) from CAST for a Shear Wall with
Rectangular Openings: (a) The Force Distribution in Struts and Ties; (b) The
Effective Widths selected for the Struts and Ties and the Utilization Rates
(Fdemand!Fcapacity) in STM of the Shear Wall.
382
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Material properties for concrete and reinforcement adopted in the FE models are obtained
from Tables 3.2 and 3.3, respectively.
For the FE model of shear wall with openings designed by using the
conventional approach, one half of the FE model of specimen 8C is adopted as illustrated
in Fig. 7.53b. The reinforcement details of the shear wall are also illustrated in Fig.7.53a.
Similarly, the FE model of the shear wall is also constructed by using the proposed
constitutive models, FE formulation in Chapter 7, and embedded reinforcement. The FE
model of the whole specimen 8C was constructed by using 524 quadrilateral 4 node
elements. The same local bond-slip relation and the material properties for concrete and
reinforcement as the FE model of specimen 8B are adopted in this model.
The predicted load displacement responses from the static push over analysis
of both shear walls are illustrated in Fig. 7.60. The results from the FE analysis of the
shear wall with the optimal reinforcement denoted as "Optimal" are additionally
illustrated in Figs. 7.54 to 7.56 while the additional analysis results of the shear wall
designed by using the conventional approach and constructed with the orthogonal
reinforcement denoted as "Conv" are illustrated in Figs. 7.57 to 7.59. It should be noted
that since these FE Models are constructed based on the whole specimens 8B and 8C,
only the half right side of the models will be used for the shear walls. Overall, the
predicted load displacement responses from shear wall designed with the optimal
reinforcement is significantly stiffer than the shear wall designed by using the
conventional approach, particularly, the post cracking stiffness although the predicted
cracking load (approximately 100 kN) of the shear wall with optimal reinforcement is
slightly higher than the cracking load (approximately 85 kN) of the shear wall designed
383
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
by the conventional approach. The significant difference in the post cracking stiffness of
both shear walls is apparently due to the orientation of the reinforcement used in
fabrication of these beams. From the analysis, no yielding of the reinforcement is
predicted to yield (Figs. 7.56 and 7.59) in both shear walls. Shear wall with the optimal
reinforcement is finally predicted to fail by crushing of the concrete regions just adjacent
to the loading point (the concrete region above the right support of specimen 8B as
illustrated in Fig. 7.54) and no critical cracks in the specimen at the capacity of 462 kN
(Fig.7.55). Similarly, the shear wall with the orthogonal reinforcement is predicted to fail
by crushing of concrete adjacent to the loading point (the concrete region above the right
support of specimen 8C as illustrated in Fig. 7.57) and having more damage in the
regions between the openings due to cracks at the capacity of 405 kN (Fig. 7.58).
By comparing the predicted capacity of both shear walls with the nominal
design load from STM model, STM models of both shear walls are found to result in the
conservative designs. In addition, the failure regions of both shear walls (Figs. 7.54 and
7.57) agree well with locations of the prismatic struts that were suggested to fail by STM
models (Figs. 7.50b and 7.51b).
Next, the experimental programs, for example, the simply supported deep beam
with openings (specimens 8B and 8C) as illustrated in Fig. 3.8 are adopted for validating
the design and nonlinear analysis of shear walls. As explained in details in chapter 5,
since the specimens 8B and 8C were set up to test in the simply supported beam
configuration, the right side of those specimens behaves similarly to the shear wall with
the openings where the right support of those specimens is in the location of the loading
point acting on the right side of the shear walls. Thus, the experimental data collected
384
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
from the right sides of specimens 8B and 8C as illustrated in Figs. 3.8 are used to validate
the design and nonlinear analysis of shear walls designed with the optimal and orthogonal
reinforcement, respectively. The specimens were constructed, instrumented, and tested to
failure. The geometry and thickness of the specimens are shown in Fig. 3.8. The material
properties used in the specimens are the same as the designed model as given in Tables
3.2 and 3.3. The reinforcement cages for the specimens SB and SC are shown in Fig. 7.61
and 7.62, respectively. A welded wire mesh with a reinforcement ratio similar to that
required in ACl3 l S-OS for deep beam was provided as to provide a minimum level of
ductility.
Based on the experimental results, in specimen SB, the first cracks form at the
corner of the lower opening at the total applied load of 60 kips (267 kN). This total
applied load is the load that was applied to specimen SB. The portion of the load
transferred to the shear wall side of the specimen can be estimated from the equilibrium
as illustrated in Fig. 4.8. Then new cracks form at the corner of the upper opening at the
total applied load of 90 kips ( 400 kN). The first diagonal crack forms at the region
adjacent to the upper opening at the total applied load of 152 kips (675 kN). At the total
applied load of 166 kips (737 kN), the concrete above the support in the shear wall side
was observed to distress. Finally, at the total applied load of 1 SO kips (SOO kN), the
specimen fails by sudden crushing of the concrete adjacent the loading point. The
observed crack pattern of the right side of specimen 7B at the failure state as illustrated in
Fig. 7 .63a also correlates well with the predicted crack pattern from the FE model as
illustrated in this figure.
3S5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In specimen 8C, the first crack was observed at the comer of the lower opening at
a total applied load of 50 kips (222 kN). At the total applied load of 80 kips (355 kN),
new crack formed at the upper opening. The severe diagonal cracks adjacent to the
openings occur at the total applied load of 120 kips (533 kN). These diagonal cracks
degrade the post cracking stiffness of the specimen so significantly as apparently
illustrated in Fig. 7.60. At the total applied load of 160 kips (711 kN), the concrete above
the support in the shear wall side was observed to distress. At the total applied load of
170 kips (756 kN), the failure mechanism of the specimen formed by crushing at the
concrete region just above the opening and sliding shear failure at the regions underneath
the opening. The observed crack pattern of the right side of the specimen 8C at the failure
state is illustrated in Fig. 7.63b and compared well with the predicted crack pattern from
FEA of the specimen.
The experimental results presented previously are now adopted for a comparison
of the design, analyses, and experiment. The summary of the nominal capacity and the
failure mode of the shear walls predicted using the proposed FE analysis technique is
given in Table 7.4 and compared with the designed nominal capacity by indeterminate
STM and the measured capacity and observed failure mode from the experiment.
From Table 7.4, as expected in both specimen 8B and 8C, using indeterminate
STM with an assumption of equal relative stiffness in all STM members results in a
conservative design. The predicted capacity from FE model is fairly accurate with the
predicted to measured capacity ratio of 0.95 and 0.90 for specimens 8B and 8C,
respectively. The predicted modes of failure from FE models of both specimens are
386
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
accurate since those FE models could capture the types and locations of the failure modes
very reasonably as shown in Table 7.4.
For the structural performance evaluation of the shear walls based on the
experimental program, a comparison of the measured capacity and structural responses
between shear wall designed by using the proposed design and analysis framework (the
right side of specimen 8B) and shear wall designed by using the conventional approach
(the right side of specimen 8C) is performed. Overall, the shear wall designed by the
proposed design and analysis framework (8B) has the highest post cracking stiffness and
capacity (Fig. 7.60). The observed crack pattern in this shear wall (Fig. 7.63a) shows
apparently less damages than the shear wall designed by using the conventional approach
and the orthogonal reinforcement (Fig. 7.63b ), particularly, during the service load and
up to failure of the specimen. In addition, the parameters to measure the structural
performance of the designed shear wall as illustrated in Fig. 7.64; for example, capacity,
reinforcement volume, and ratio of capacity to reinforcement volume confirm the
achievement of the most effective structural performance design of the shear wall by
using the proposed design and analysis framework since the shear wall has the highest
capacity and maximum ratio of capacity to reinforcement volume.
387
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Half of S ecimen 88
(a)
(b)
Fig. 7.52 Optimal Reinforcement Details and FE Model of a Shear Wall with
Rectangular Openings (7B): (a) Reinforcement Details; and (b) FE Model
388
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Half of Specimen 8C
-
~
....
~
I
(a)
(b)
Fig. 7.53 Orthogonal Reinforcement Details and FE Model of a Shear Wall with
Rectangular Openings (7C): (a) Reinforcement Details; and (b) FE Model
389
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
No file Loaded
- to 0.04
-to 0.21
- to 037
- to 0.08 - to 025
-to0'1
-to0.12
- to 0.29 - to 0.45
-to0.16
-to 0.33 - to 0.49
fem
- to 0.54
- to 058
- to 0.62
- to 0.66
to 070
-to 074
-to 078
- to 0.82
Vial Signs. fem Lood Factor 2.10
- to 0.86
-to091
- to 095
- to 0.99
Fig. 7.54 Concrete Compressive Vital Sign in the Shear Wall (Optimal
Reinforcement)-The Right Side of the Specimen 8B: Ratio of Compressive Stress to
Compressive Stress Capacity.
Combined View
l I
/ \
I \
\ \ I
I \ \ \
\ \
I
\ \ \ \ \
\ \ \ \ \\\
// \ \ \
' \ \
// // / \ \ \\ \ \
// // // / / //
" \
\ \
"
\ \ \
// // / / / // /I / \ \
\ \
'\_ '\_ \ \ \ \ I
/ / / / // / /I I I I I / \ \ \ \
'\_ \
\\\
I / / / / / / / / I I I/ I I I I \ \ \ \ \ \ \ ~ ~
/ / / / / / / / / I / / I I / I I I I \
\ '\ \ \ \ ~ ~
I / / I / / / / I / / / I I I I I I I \ \ \\ \ \
\
\
\ T / I I / / / / I I I I / I I \ \ \ \ \ \ \ \ \\ \ \
/ / / / / / / / / / I I I
\ \ \ \ \ \ \ \ \ \ I \ I
\ \, ''.,, '\ \ I
I I I I I I / / I / I I I I I I I I I I I \ J I I
I
\\,\. I l
-
-
1-1-
--
~ I
No hleLOGded DlsptaceJCreicks: Combned View Lot.ad factor .. 2.40
Crack\'\4dfhs thn1.00mm,mld,thlc:k200mm
Fig. 7.55 Combined Deformed Shape and Crack Pattern at the Failure State of the
Shear Wall (Optimal Reinforcement)-The Right Side of the Specimen 8B
390
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a-truss at crack
,,,,
A ~
<?.
"/
l;! ~
~
~
,, e!F
-
-
./.
-
-
~
~
-
~
f-- - >-- -
f--
-
-
~
No file Loaded stress steel): o-truss at crack Load Fllcior = 2.90
to -50.05
-to-39.91
- to-2977
- to-19.63
to -9.49
-to 0.65
- to 10.79
- to20.94
to 31.06
- to41.22
-to51.36
-to61.50
to 71.64
- to 61.76
- to 91.92
- to102.06
to112.20 to152.76
- tot22.34 - to16290
- 1013246 - to173CM
- to14262 111111111 to163.16
Fig. 7.56 Stress (at Crack Locations) in Reinforcement at the Failure State of the
Shear Wall (Optimal Reinforcement)-The Right Side of the Specimen SB
No fie Loaded
to 0.20
- to 0.24
- to 026
- to 0.32
to 0.36
-toD.41
- to 0.45
- to 0.49
fem
to 053
-to057
-to061
-to065
Vial Signs fem Load Factor 2 50
to 08.5
- to 0.89
- to 093
-to097
Fig. 7.57 Concrete Compressive Vital Sign in the Shear Wall (Orthogonal
Reinforcement)-the Right Side of Specimen SC: Ratio of Compressive Stress to
Compressive Stress Capacity.
391
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Combined View
I
/ I \ \\\\\
/ /
\ \ \
\ \ \ \
\ '\. "\ \
\ \
\ \
'
I
\ \
\ '
\ \ \ \
// / / I \ \ \ \.
\ \
\\\
\
// / // // \\\'-.\ \ \
\\\\ \ I
// / / / / / / /
\ \ \ \ \
\ '\
\\\\ I I
///////
/ / / I I \\\.I\
\ \
\
/////////
/ / I I I I I I \\\\'\
\. \ I
}
//////////
/ / I I I I I I
\\\\' \
\
T I ///////////
/ I I I I I I I I \ \ ', \ \ \
///////////
I I I I I I I I I I I I \ \ I

+
I I I I I I I / I / /
I I I I I I I I I I I I / / / ",,'
-'--
r--
-
>----
-
--
,____.___
No tileLCHKted Dlsplace/Cracks: Combined View Load F8ctor .. 3.10
Magnification: 20 00 x
Crack V\4dlhs: thf"I.: 1.00 mm, mid, thick 2.00 mm
Fig. 7.5S Combined Deformed Shape and Crack Pattern at the Failure State of the
Shear Wall (Orthogonal Reinforcement)-The Right Side of the Specimen SC the
Deep Beam (More Practical Reinforcement)
to 39.35
-to47.30
- to5525
- to 63.20
er-truss at crack
to 7115
- to7910
- to87.0S
- to95.00
to 102.95
- to110.90
- to118.85
- to126.80
--
I
to134.75 to 16655
- to142.70 - to174.50
- to150.65 ma 10182.45
- to 158.60 l1iii!ll to 190.40
Fig. 7.59 Stress (at Crack Locations) in Reinforcement at the Failure State of the
Shear Wall (Orthogonal Reinforcement)-The Right Side of the Specimen SC
392
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
500
450
400
..................................................................................................................................................... ;
i
-
Pexp (Conv)=453kN
.
350 I :
z-
300
__ __,.__,.""\t\-"'----P_n....:...,A_C_1_=3_3_4_.7_k--:Ni
.::.::
:c;"250
ca
_g 200
150
100
50
I
/::,.
.-----------,
u -.-Optimal-meshsize75
----- Optimal-meshsize90
.. -- Optimal-meshsize115
--.ti:-Conv-meshsize75
-- " C onv-mesh size90
_.......Conv-meshsize115
--i!!-Optimal-Exp
- !!! Conv-Exp
0
0 1 2 3 4 5 6
Displacement (mm)
Fig. 7.60 The Predicted Load and Displacement of the Shear Walls by Using the
Proposed Nonlinear FEA
Fig. 7.61 Reinforcement Details of Specimen 8B
393
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. 7.62 Reinforcement Details of Specimen SC
(a)
(b)
Fig. 7.63 Crack Pattern at the Failure State of Shear Wall with Openings: (a) Right
Side of Specimen SB; and (b) Right Side of Specimen SC
394
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7.4 Capacity and Failure Mode Comparison of Designs, Analysis, and Experiment
Shear Wall Pn
(Optimal) (kN) Failure Mode of Shear Wall
(Model 8B) (Failure Mode of Model 8B)
STM (Designed) Crushing of prismatic strut adjacent to loading
(Pu/0.75) 334.7 point
(553) (Crushing of concrete above the right support)
Nonlinear FEA 462 Crushing of concrete adjacent to loading point
(764) (Crushing of concrete above the right support)
Experiment
(small seal e) 484 Crushing of concrete adjacent to loading point
(800) (Crushing of concrete above the right support)
Shear Wall Pn
(Orthogonal) (kN) Failure Mode of Shear Wall
(Model 8C) (Failure Mode of Model 8C)
STM (Designed) Crushing of prismatic strut adjacent to loading
(Pu/0.75) 334.7 point
(553) (Crushing of concrete above the right support)
Nonlinear FEA 405 Crushing of concrete adjacent to loading point
(669) (Crushing of concrete above the right support)
Experiment
(small scale) 453 Crushing of concrete adjacent to loading point
(756) (Crushing of concrete above the right support)
395
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0.6
li!I Strength(kN)*1000 Rebar Volume(mm"3)x10"7
o Stren th/Rebar Volume N/mm"3
0
88 SC
Specimen No.
Fig. 7.64 The Structural Performance Evaluation of Designed Shear Walls with
Openings.
7.4 Summary
In this chapter, both concept and application of the proposed design and analysis
framework for effective structural performance design of complex D-Regions were
presented. The introduction to the need of the design and analysis framework was
initially presented and then the details of the proposed design and analysis framework of
complex D-Regions were given. By using the examples of D-Regions typically found in a
building, the application of the proposed design and analysis framework was completely
demonstrated by using the design examples of several components ofD-Regions. This
included a propped cantilever deep beam with an opening, a simply supported deep
beams with an opening, a simply supported deep beam with a dapped end and an
opening, and a shear wall with openings.
The concept of the proposed design and analysis framework for effective structural
performance design of complex D-Regions is straightforward and practical by combining
three significant components that help designers to handle the difficulty and uncertainty
396
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
occurring during the design process of very complex D-Regions. Those components are:
1) the shape selection technique based on topology optimization; 2) the current code
provisions in design D-Regions by using STM; for example, ACl3 l 8-08 code provision;
and 3) the proposed nonlinear FEA of reinforced concrete structures specialized for
modeling and analyzing complex D-Regions.
The shape selection technique based on topology optimization is included in the
framework to facilitate and handle the difficulty in selecting the shape of STM for very
complex D-Regions subjected to complex geometry, loading, and/or boundary
conditions. The technique was experimentally validated to ensure that the STM shape
used in design D-Regions will result in the orientation of the reinforcement details that
effectively control crack development and propagation as well as transfer and redistribute
the stress inside the D-Regions. The technique could be effectively extended to generate
the STM shape for D-Regions subjected to reversed loading or multiple loads which are
apparently beneficial to the designers since one optimal STM shape can be adopted to
design for all load cases.
The proposed FEA is included in the framework for facilitating the automated FE
analysis of complex D-Regions where the difficulty in modeling of the complex
reinforcement pattern typically found in the optimal STM can be handled easily by the
use of the embedded reinforcement element including bond stress-slip. The concrete
element in the formulation was also specialized for D-Regions by including several
important behavior models and formulation that were validated experimentally for the
improvement and more reliability in analyzing the complex D-Regions in reinforced
concrete structures (chapter 6).
397
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As presented in the design examples in the chapter, most complex D-Regions
designed using the proposed design and analysis framework have a very effective
structural performance, particularly, during the service loads. For example, exceptionally
high cracking load, less crack development and propagation, high post-cracking stiffness,
more uniform stress distribution in the reinforcement, and high capacity can be achieved.
In addition, in most cases, those complex D-Regions can be designed to achieve the
highest ratio of the capacity to the volume of the reinforcement. Thus, not only effective
structure performance, but also economical design of D-Regions is also obtained by using
the proposed design and analysis framework.
398
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 8
Conclusions and Recommendations for Future Work
As stated in Chapter 1, the overall objective of this research is to improve the
current STM design practice for the design and analysis of D-Regions by proposing a
more complete design and analysis framework that is obtained by integrating the current
design of D-Regions by using STM in the code provisions with the additional two
components: 1) Methods and guidelines to facilitate improved STM shape selection and
2) Nonlinear FEA for automated analysis of complex D-Regions. Series of both
experimental and computational programs are also adopted in this research to validate the
proposed design and analysis framework as well as its components. This chapter presents
a summary of the key components and the findings of this research that support the
research objective. Recommendations for future work are included at the end of the
chapter.
8.1 Summary and Conclusions
8.1.1 Experimental Validation of Complex D-Regions Designed by STM
Code provisions for design D-Regions by the STM approach have been
incorporated into several codes of practice and guideline documents. This includes the
Canadian Standards Association, the AASHTO LRFD Bridge Design Specifications, and
the Building Code provisions of the American Concrete Institute. The safety of these
STM code provision is dependent on whether or not the stress limits in codes of practice
and the plasticity assumption are applicable to the D-Regions being designed for the
selected shape of the STM used in the designs. The STM approach and associated code
provisions have been most thoroughly validated for simple types of structures, such as
399
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
deep beams and corbels, but there has been only limited validation for more complex D-
Regions. This experimental component of this research program is thus to provide the
more comprehensive and extensive experimental data for the more thorough evaluating
of the response, design, and analysis of D-Regions designed using a STM approach.
Based on this experimental investigation, the measurements indicate that: 1) the
strengths calculated by the ACI STM Code provisions typically yield conservative
estimates of capacity for D-Regions even when intentionally questionable truss shapes
were selected in the design; 2) distributed reinforcement must be provided to ensure the
sufficient local ductility of D-Regions that prevents the premature and undesirable failure
modes as required by the plasticity theorem; 3) the structural responses of D-Regions
under service load levels can be significantly influenced by the selected shape of the
load-resisting truss; 4) A poor selection for the shape of a STM can lead to unacceptable
levels of cracking and damage under service loads; and 5) the load transfer mechanism of
D-Regions can be significantly influenced by the selected shape of the load-resisting
truss.
8.1.2 Method and Guidelines to Facilitate Improved STM Shape Selection
The flexibility in freely selecting the STM shape in the design process
sometimes causes difficulties to designers, particularly when designers encounter D-
Regions with very complex geometry, loading, and/or boundary conditions. Based on the
experimental investigation, except for poor detailing, the shape or topology of STM is
key for ensuring a good design. A poor selection could result in the ineffective structural
performance during service states, such as uncontrolled crack widths and propagation,
low cracking loads, and low capacity at the ultimate states. Thus, the validation of the
400
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
shape selection techniques of STM and the assessment of the structural performance of
complex D-Regions after being designed by those shape selection techniques are
performed in this research thrust.
In this research thrust, both experimental and computational programs were
adopted to assess the structural performance of complex D-Regions designed by various
shape selection techniques including the topology optimization technique entitled
"Performance Based Topology Optimization (PBO)". The experimental and
computational validation confirms that using topology optimization in selecting the STM
shape in the design process can result in the complex regions with an effective structural
performance at both service and ultimate states because the orientation of the
reinforcement in the optimal STM shapes can effectively control the development and
propagation of cracks. As an alternative to placing the reinforcement in the optimal STM
shape, an orthogonal orientation can be more practical. The structural behavior and
performance of the regions should be assessed using a nonlinear FEA tool to ensure an
effective and safe design when the sufficient distributed mesh is provided. The
distributed mesh must be provided in the D-Regions designed by STM to provide the
sufficient local ductility that prevents the premature failure due to the unstable crack
propagation as required in some code of practices and plasticity theorem assumption.
8.1.3 Nonlinear FEA for Automated Analysis of Complex D-Regions
To assess the structural performance and safety of the complex D-Regions
designed using the STM approach, a nonlinear FEA is needed. As addressed in chapter 1,
the current nonlinear FEA is still limited and not practical, particularly, when adopted to
analyze the complex D-Regions where their geometry, loading and boundary conditions,
401
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and orientation of reinforcement are complex. Thus, this research thrust was to develop a
more efficient and reliable computational tool for finite-element based design validation.
The research also included a new FE formulation and behavior models that was intent to
overcome shortcomings of the most heavily implemented smeared cracking model that
was based on the Modified Compression Field Theory (MCFT) and the Disturbed Stress
Field Model (DSFM). These shortcomings are: 1) A significant dependency on mesh
size and 2) An ability of the advanced analysis tool "Vector2" implemented MCFT and
DSFM models that can only account the effects of reinforcement by using distributed
reinforcement in each element or truss bars between elements; this both increased the
modeling effort for the use and the dependency of the prediction on mesh size selection.
In brief, the important features of the new model are: 1) the embedded
reinforcement element including bond stress-slip and 2) the improved constitutive model
for reinforced concrete continuums. In addition, mesh dependency reduction achieved by
using the improved constitutive model and the numerical study on the influence of the
deviation of the crack angle from the direction of the principal compressive stress in
concrete on structural strength and responses of complex D-Regions are also conducted
in this research thrust.
For the embedded reinforcement element including bond stress-slip, the author
demonstrated the implementation and validation of the approach using experimental test
data from the anchorage specimens to the concrete structures. The strong correlation
between the predicted and experimental responses was presented and thus confirmed the
validity and the potential value of the proposed formulation.
402
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The improved constitutive model specialized for complex D-Regions in
reinforced concrete structures was implemented in MATLAB and Augustus was adopted
as a post-processing tool. The model was validated by performing series of FEA at both
the element and structural levels. All analyses were compared with the existing
experimental data from the literature and from the available experimental data of
complex D-Regions in Chapters 3 to 5. The numerical results correlated well with most
of the experimental data. In addition, the reduction of the mesh dependency in the
smeared crack model in the structural level could be obtained in most cases with the use
of this new formulation.
Subsequently, the proposed FE formulation was adopted to conduct a
parametric study on the influence of the deviation of the crack angle from the direction of
the principal compressive stress of concrete on the predicted structural response of D-
Regions. Both element and structural level investigations were included in this study.
Based on the parametric study, the following conclusions were made: 1) the additional
shear demand along crack surface will increase the level of slippage along the crack and
can significantly reduce the structural capacity of the specimens; 2) the more tensile
stress transferability or the less severe crack condition will lead to a greater reduction in
the structural capacity due to the effect of the deviation of the crack angle from the
direction of the principal compressive stress in the concrete; and 3) for the beam type
structure, the consideration of the deviation of the angles and the additional demand
along crack surface should be included in the analysis process to ensure reliable
predictions of structural performance.
403
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8.1.4 Design and Analysis Framework for Effective Structural Performance of
Complex D-Regions
By integrating the STM shape selection technique based on topology
optimization and the proposed nonlinear FEA with the current code provisions of STM,
the complete design and analysis framework for effective structural performance design
of complex D-Regions was obtained. The framework is straightforward and practical.
The shape selection technique based on topology optimization is included in the
framework to facilitate and handle the difficulty in selecting the shape of STM for very
complex D-Regions subjected to complex geometry, loading, and/or boundary
conditions. The technique was experimentally validated to ensure that the STM shape
used in design D-Regions will result in the orientation of the reinforcement details that
effectively control crack development and propagation as well as sufficiently allow the
stress transfer and redistribution inside the D-Regions. The technique could be effectively
extended to generate the STM shape for D-Regions subjected to reversed loading or
multiple loads which are apparently beneficial to the designers so that one optimal STM
shape can be adopted to design for all load cases.
The proposed FEA is included in the framework for facilitating the automated
FE analysis of complex D-Regions where the difficulty in modeling of the complex
reinforcement pattern typically found in the optimal STM can be handled easily by the
use of the embedded reinforcement element including bond stress-slip effects. The
concrete element in the formulation was also specialized for D-Regions by including
several important behavior models and the formulations were validated experimentally
404
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
for the improvement and increased reliability in predicting the response of complex D-
Regions in reinforced concrete structures (chapter 6).
As presented in the design examples in chapter 7, most complex D-Regions
designed by using the proposed design and analysis framework exhibited good structural
performance at all limit state. For example, they illustrated higher cracking loads than
specimens with non-optimal reinforcement layouts, less crack development and
propagation, high post-cracking stiffness, and more uniform stress distribution in the
reinforcement and higher capacity at the ultimate state. In addition, designs completed
with this methodology generally required less reinforcement than for designs that did not
use this framework. Thus, not only does the methodology improve service level
performance and the safety of D-Regions, they can also be more economical.
8.2 Recommendations for Future Work
Based on the work completed in this thesis, the following suggestions are
made for future work:
The available experimental data for the validation of the STM approach
for the design of complex D-Regions is quite limited. Due to the broad
applicability of the STM, a broad array of additional evidence is great
needed. In order to both investigate STM and FEA methods for D-
Regions, dense test data should be collected, that being fiend strain and
displacements.
The experimental validation of STM adopted in the research was focused
only on the aspects of the structural responses and performance from the
service to the ultimate states. The existing experimental data, in particular
405
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the data collected by the Krypton coordinate measurement machine,
should be more fully explored to more completely understand the behavior
of the tested structures.
The validation of STM for the design of complex D-Regions subjected to
reverse loading or seismic action was performed by using only the
monotonic static load test. Thus, a more extensive experimental program
would be useful in which full load reversals were applied.
In the completed experimental research program, only relatively modest
sized specimens were tested. Thus, the future experimental research
program should include series of the larger scale specimens where the size
effect could be included and considered, particularly, in the complex
Regions designed to fail by crushing of struts or experience failure due to
propagation of individual cracks.
In this research, only three STM shape selection techniques were adopted:
1) experience of designers; 2) empirical formulation; and 3) performance
based topology optimization. The future research might include other
optimization techniques; for examples, the use of genetic algorithms, the
homogenization method, grid methods, and others. This would allow the
comparative study of the structural performance of complex D-Regions
designed by different optimization schemes.
The formulation of the embedded reinforcement element including bond
stress-slip in this research was developed in terms of only the average
stress and strain of the embedded bars. Thus, the more refine formulation
406
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
should be developed by considering the stress and strain at crack locations
which might be possible by employing an internal algorithm to determine
what zone of surrounding concrete constitutes that effective embedment
zone for the embedded bar. The area of reinforcement is then smeared
over that area and the local stress at crack calculations are performed in
the same manner as is done for smeared reinforcement.
In the embedded reinforcement including bond stress-slip, no effect of the
anchorage details; for examples 90 degree hook or steel end plate, was
taken account into the model. In addition, no interaction between the
crossing reinforcement was considered. Thus, the future research should
pursue on refining the formulations to include those effects for the
improvement of the accuracy of the strain prediction in the embedded
reinforcement.
The proposed formulation of the behavior models of the automated FEA
was assumed that the difference between the crack orientation and the
direction of the principal compressive stress in concrete is known. Thus,
the future research should pursue the development of reliable techniques
or formulations for determining the difference between those angles which
could be very useful for the design and analysis of new complex D-
Regions.
The uncertainty of the material models of the automated FEA should be
considered in the formulation. This could be done after appropriate
experimental validation of the models. This is to explore the impact of
407
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
modeling assumptions with sensitivity analyses. The results of the
validation can be used to determine and report the bounded confidence in
the predicted behavior to the designer. For example, this could provide
two additional load-deformation responses corresponding to the 5% and
95% percent confidence intervals of predicted capacity.
The design and analysis framework was only introduced and presented in
this research without the complete implementation in an integrated
computer software framework. Thus, the future research should include
the complete implementation of the framework by using a reliable
programming language to combine the three proposed components into
one software package. This implemented software packages should be
distributed to the engineering community.
408
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
References
AASHTO (1994), "AASHTO LRFD Bridge Design Specifications,"
American Association of State Highway and Transportation Officials, first
edition, Washington, DC, 1091 pp., including interim revisions for
1996 and 1997.
ACI Committee 318(2002).Building Code Requirements for Structural Concrete
andCommentary (ACI318R-02), American Concrete Institute, Farmington
Hills, Michigan, 443 pp.
ACI Committee 3 I 8(2005).Building Code Requirements for Structural Concrete
andCommentary (ACI318R-05), American Concrete Institute, Farmington
Hills, Michigan, 443 pp.
ASCE. (1982). State-of-the-Art Report on finite element analysis of reinforced
concrete, New York.
Ali, M.A., and White, R.N. (2001), "Consideration of Compression Stress
Bulging and Strut Degradation in Truss Modeling of Ductile and Brittle
Corbels," Engineering Structures,Mar, Vol. 23, No. 3, pp. 240-249.
Alshegeir, A. (1992), "Analysis and Design of Disturbed Regions with Strut-
and-Tie Models," PhD Thesis, School of Civil Engineering, Purdue
University, West Lafayette, Indiana, 274 pp.
Bazant, Z.P., and Cedolin, L. (1983), "Finite Element Modeling of Crack Band
Propagation," Journal of Structural Engineering, ASCE, January, Vol.
109, No.I, pp. 69-91.
Bazant, Z.P., and Oh, B. H. (1983), "Deformation of Cracked Net-Reinforced
Concrete Walls," Journal of Structural Engineering, ASCE, January, Vol.
109, No.I, pp. 93-108.
Belarbi, A. and Hsu, T.T.C. (1991). "Constitutive Laws of Reinforced Concrete
in Biaxial Tension-Compression,"Research Report UHCEE 91-2,
University of Houston, Houston, Texas, 155 pp.
Bentz, E.C. (1999), "Sectional Analysis of Reinforced Concrete Structures,"
PhD Thesis, Department of Civil Engineering, University of Toronto,
Toronto.
Bergmeister, K., Breen, J.E., Jirsa, J. 0. (1991). "Dimensioning ofNodal Zones
and Anchorage of Reinforcement, "IABSE Colloquium Stuttgart 1991:
Structural Concrete, International Association for Bridge and Structural
Engineering, Zurich, pp. 551-564.
Bhide, S.B., and Collins, M.P. (1989), "Influence of Axial Tension on the Shear
Capacity of Reinforced Concrete Members," ACI Structural Journal, Vol.
86, No. 5, pp. 570-581.
Carlos G. Quintero-Febres, Gustavo Parra-Montesinos, and James K. Wight
(2006), "Strength of Struts in Deep Concrete Members Designed Using
Strut-and-Tie Method," ACI Structural Journal,July.-August., Vol. 103,
No. 4, pp. 577-585.
Carter, B.J., Wawrzynek, P.A., and Ingraffea, A.R. (2000), "Automated 3-D
crack growth simulation," Int. J Numer. Methods in Engrg., Vol. 47,
No. 1-3, pp. 229-253.
409
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cervenka, V. (1970), "Inelastic Finite Element Analysis of Reinforced Concrete
Panels," PhD Dissertation, Department of Civil Engineering, University
of Colorado, Boulder, USA.
Chen, B.S., Hagenberger, M.J., and Breen, J.E., "Evaluation of Strut-and-Tie Modeling
Applied to Dapped Beam with Opening," AC! Structural Journal, V.99, No. 4,
July-Aug, 2002, pp. 445-450.
Choi, C.K. and Kwak, H.G. (1990), "The Effect of Finite Element Mesh Size in
Nonlinear Analysis of Reinforced Concrete Structures," Computers and
Structures, Vol. 36, No.5, pp. 807-815.
Comite Euro-International du Beton (1993), "CEB-FIP Model Code 1990," Thomas
Telford Services, Ltd., London, UK, 437 pp.
CSA Committee A23.3 (1984), "Design of Concrete Structures (CAN3-A23.3-M84),"
Canadian Standards Association, Rexdale, ON, Canada, 281 pp.
Darwin, D., and Pecknold, D.A. (1976), "Analysis of RC Shear Panels under
Cyclic Loading," Journal a/Structural Engineering, ASCE, January, Vol.
102, No.2, pp. 355-369.
Dodds, R.H., Darwin, A.M., and Leibengood, L.D., (1984), "Stress Controlled
Smeared Cracking in RIC Beams," Journal a/Structural Engineering,
ASCE, Vol. 110, No.9, pp. 1959-1975.
Elwi, A.E., and Hrudey, T. M. (1989). "Finite Element Model for Curved
Embedded Reinforcement,"Journal of Engineering Mechanics, Vol. 115,
No. 4, April, pp. 740-753.
Exner, H. (1979). "On the Effectiveness Factor in Plastic Analysis of
Concrete,"IABSE Colloquium Kopenhagen 1979:Plasticity in Reinforced
Concrete- Final Report, International Association for Bridge and
Structural Engineering, October, Copenhagen, Denmark, pp. 35-42.
FIP Commission 3 (1999), "Practical Design of Structural Concrete,"
federation internationale de la Precontrainte, Lausanne, Switzerland, Sept.,
114 pp.
Foster, S.J. (1992), "The Structural Behavior of Reinforced Concrete Deep
Beams," PhD Thesis, School of Civil Engineering, University of New
South Wales, Kensington, Sydney, Australia.
Foster, S.J., Budiono, B., and Gilbert R.I. (1996). "Rotating Crack Finite
Element Model of Reinforced Concrete Structures,"
Computer&Structures, Vol.58, No.l,pp. 43-50.
Foster, S.J., and Marti, P. (2003). "Cracked Membrane Model: Finite Element
Implementation," Journal a/Structural Engineering, Vol.129, No.9,
September, pp. 1155-1163.
Grob, J., and Thurlimann, B. (1976). "Ultimate Strength and Design of
Reinforced Concrete Beams under Bending and Shear,"Memoires
Abhandlungen 36-11, International Association for Bridge and Structural
Engineering, pp. 105-120.
Hwang, S.J., and Lee, H.J. (1999). "Analytical Model for Predicting Shear
Strengths of Exterior Reinforced Concrete Beam-Column Joints for
Seismic Resistance," AC! Structural Journal, Vol. 96, No.5, Sept.-Oct.,
pp.846-857.
410
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hwang, S.J., Lu, W.Y. and Lee, H.J. (2000). "Shear Strength Prediction for
Deep Beams," AC! Structural Journal, Vol. 97, No.3, May-June, pp.367-
376.
Hwang, S.J., and Lee, H.J. (2002). "Strength Prediction for Discontinuity
Regions by Softened Strut-and-Tie Model," Journal of Structural
Engineering, Vol.128, No.12, Dec.
Jirsa, J. 0., Breen, J.E., Bergmeister, K., Barton, D., Anderson, R., and Bouadi,
H. (1991). "Experimental Studies ofNodes in Strut-and-Tie Models,"
IABSE Colloquium Stuttgart 1991: Structural Concrete, International
Association for Bridge and Structural Engineering, Zlirich, pp. 525-532.
Kaufmann, W., and Marti, P. (1998). "Structural Concrete:Cracked Membrane
Model," Journal of Structural Engineering, Vol.124, No.12,pp. 1467-
1475.
Kuchma, D.A., Yindeesuk, S., Nagle, T., Hart, J., and Lee, H.H. (2008).
"Experimental Validation of the Strut-and-Tie Method for Complex
Regions, AC! Structural Journal, Vol. 105, No.5, Sept.-Oct.,
pp.578-589.
Kwak, H.G., and Filippou, F. C. (1995). "A New Reinforcing Steel Model with
Bond-Slip,"Structural Engineering and Mechanics, Vol. 3, No. 4, pp.
299-312.
Lampert, P. and Thurlmann, B.(1968). "Torsionsversuche an Stahlbetonbalken,"
Bericht 6506-2, Institut fur Banstatik ETH , Zlirich, June 1968.
Lampert, P. and Thurlmann, B.(1969). "Torsions-Biege-Versuche an
Stahlbetonbalken,"Bericht 6506-3,Institut fur Banstatik ETH, Ztirich,
June 1969.
Liang, Q.Q., Uy, B., Steven, G.P. (2002). "Performance-Based Optimization for
Strut-Tie modeling of Structural Concrete," Journal of Structural
Engineering, American Society of Civil Engineers, Vol. 28, No. 6, pp.
815-823.
Liang, Q.Q., Xie, Y.M., Steven, G.P. (2000). "Topology optimization of strut-
and-tie models in reinforced concrete structures using an evolutionary
procedure," AC! Structural Journal, American Concrete Institute, Vol.
97, No. 2, pp. 322-330.
Lee, H.J. and Kuchma, D.A. (2007), "Seismic Overstrength of Shear Wall in
Parking Structures with Flexible Diaphragms", Journal of
Earthquake Engineering, Vol. 1, Issue 11, pp. 86-109.
Marti, P. (1985). "Basic Tools of Reinforced Concrete Beam Design," ACI
Journal, Proceedings, Vol. 82, No.I, January-February, pp. 45-56.
MacGregor, J. G. (1988). Reinforced Concrete: Mechanics and Design,
Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 2nd ed., 848 pp.
MacGregor, J. G. (1997). Reinforced Concrete: Mechanics and Design,
Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 3rd ed., 939 pp.
Maxwell, B.S., and Breen, J.E., "Experimental Evaluation of Strut-and-Tie Model
Applied to Deep Beam with Opening," AC! Structural Journal, V.97, No. 1, Jan.-
Feb. 2000, pp. 142-148.
Mitchell, D. and Collins, M.P. (1976). "Detailing for Torsion," ACI
411
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Journal, Vol. 73, No.9, September, pp. 506-511.
Mitchell, D., and Cook, W. D. (1991)."Design of Disturbed Regions," IABSE
Colloquium Stuttgart 1991 :Structural Concrete, International Association
for Bridge and Structural Engineering, March, pp.533-538.
Morsch, E. (1909). Concrete-SteelConstruction, E.P. Goodrich, trans., McGraw-
Hill, New York, 368 pp.
Nagle, T.J., Kuchma, D.A. (2007), "Shear Transfer Resistance in High-Strength
Concrete Girders," Magazine of Concrete Research, Accepted for
publication to the Magazine of Concrete Research.
Ngo, D., Scordelis, A.C. (1967), "Finite Element Analysis of Reinforced
Concrete Beams," AC! Structural Journal, Sept.-Oct., Vol. 64, No. 3, pp.
152-163.
Nielsen, M.P. (1971). "On the Strength of Reinforced Concrete Discs, Acta
Polutechnica Scandinavica, Civil Engineering and Building Construction
Series No.70, Copenhagen, 261 pp.
Nielsen, M.P.(1999). Limit Analysis and Concrete Plasticity, CRC Press LLC,
2nd ed., 908 pp.
Nilson, A.H. ( 1968), "Nonlinear Analysis of Reinforced Concrete by Finite
Element Method," AC! Structural Journal, September, Vol. 65, No. 9, pp.
757-766.
Okamura, H., and Maekawa, K. (1991). Nonlinear Analysis and Constitutive
Models of Reinforced Concrete, Tokyo Gihodo, Japan.
Pang, X.B., and Hsu, T.T.C. (1992). "Constitutive Laws of Reinforced Concrete
in Shear,"Research Report No. UHCEE92-1, Department of Civil
Engineering, University of Houston, Houston.
Park, J.W., Kuchma, D.A. (2007), "Strut-and-Tie Model Analysis for Strength
Prediction of Deep Beams," AC! Structural Journal, Accepted for
publication to the ACI Structural Journal.
Petersson, P.E. (1981). "Crack Growth and Development of Fracture Xone in
Plain Concrete and Similar Materials," Report No. TVBM-1006, Lund
Institute of Technology, Lund, Sweden.
Phillipps, D.V. and Zienkiewicz, O.C. (1976), "Finite Element Nonlinear Analysis
of Concrete Structures. In: (3rd Edn. ed.)," Proc. Inst. Civ. Engrs, Vol. 61,
pp. 59-88.
Polla,M. (1992), "A Study of Nodal Regions in Strut-and-Tie Models,"
MASc Thesis, Department of Civil Engineering, University of Toronto,
Toronto.
Ramirez, J.A., and Breen, J. (1983). "Proposed Design Procedure for Shear and
Torsion in Reinforced and Prestressed Concrete,"Research Report 248-4F,
Center for Transportation Research, University of Texas, Austin, Texas.
Ranjbaran, A. (1991), "Embedding of Reinforcements in Reinforced Concrete
Elements Implemented in DENA," Computers and Structures, Vol. 40,
No.4, pp. 925-930.
Reineck, K-H., "Modeling Structural Concrete with Strut-and-Tie Models-Summarizing
Discussion of the Examples as per Appendix A of ACl318-2002," Examples for
the Design of Structural Concrete with Strut-and-Tie Models, SP-208, K.-H.
412
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reineck, ed., American Concrete Institute, Farmington Hills, MI, 2002, pp. 225-
242.
Ritter, W. (1899). "Die Bauweise Hennebique," Schweizerische Bauzeitung, Vol.
33, No. 7, February, pp. 59-61.
Rogowsky, D.M., and MacGregor, J.G. (1983). "Shear Strength of Deep
Reinforced Concrete Continuous Beams,"Structural Engineering Report
No. I I 0, Department of Civil Engineering, University of Alberta,
Edmonton, November, 178 pp.
Rogowsky, D.M., and MacGregor, J.G. (1986). "Design of Deep Reinforced
Concrete Continuous Beams,"Concrete International, Vol. 8, No.8,
August, pp. 49-58.
Rots, J.G., and Blaauwendraad, J. (1989). "Crack models for concrete: Discrete
or Smeared? Fixed, multi-directional or rotating?," HERON", Vol. 34,
No.I, pp. 1-59.
Schafer, H. (1975), "A Contribution to the Solution of Contact Problems with
the Aid of Bond Elements," Computer Methods in Applied Mechanics
and Engineering,Vol. 6, pp. 335-354.
Schlaich, J., and Schafer, K. (1991). "Design and Detailing of Structural
Concrete Using Strut-and-Tie Models,"The Structural Engineer, Vol. 69,
No. 6, March, pp. 113-125.
Schlaich, J., Schafer, K., and Jennewein, M. (1987). "Toward a Consistent
Design of Structural Concrete,"Journal of the Prestressed Concrete
Institute, Vol. 32, No. 3, May-June, pp. 74-150.
Sritharan, S, and Ingham JM. (2003). "Application of Strut-and-Tie Concepts to
Concrete Bridge Joints in Seismic Regions," PC! Journal, Vol.48,
No.4,pp. 66-99.
Sundermann, W., and Mutscher, P. (1991). "Nonlinear Behaviour of Deep
Beams," IABSE Colloquium Stuttgart 1991: Structural Concrete,
International Association for Bridge and Structural Engineering, Zurich,
pp. 385-390.
Thi.irlimann, B. ( 1978). "Plastic Analysis of Reinforced Concrete Beams," IABSE
Colloquium on Plasticity in Reinforced Concrete, Copenhagen 1979,
Introductory Report, International Association for Bridge and Structural
Engineering, Vol. 28, Zurich, October, pp. 71.
Tjhin (2004), "Analysis And Design Tools for Structural Concrete using Strut-
And-Tie Models," PhD Thesis, University of Illinois at Urbana-
Champaign, USA.
Tjhin, T.N., Kuchma, D.A. (2002), "Computer-Based Tools for Design by Strut-
and-Tie Method: Advances and Challenges," ACI Structural Journal,
Sept-Oct., Vol. 99, No. 5, pp. 586-594.
To., N.H.T., Ingham, J.M., and Sritharan, S. (2000), "Cyclic Strut & Tie
Modelling of Simple Reinforced Concrete Structures," Paper No. 1249,
Proceedings of the I 2th World Conference on Earthquake Engineering,
Auckland, New Zealand, Jan-Feb.
To., N.H.T., Ingham, J.M., and Sritharan, S. (2001), "Monotonic Non-Linear
Analysis of Reinforced Concrete Knee Joints Using Strut-and-Tie
413
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Computer Models," Bulletin of the New Zealand Society for Earthquake
Engineering, September, Vol. 34, No. 3, pp. 169-190.
Vecchio, F .J. (2000), "Disturbed Stress Field Model for Reinforced Concrete:
Formulation," Journal of Structural Engineering, ASCE, Vol. 126, No.9,
pp. 1070-1077.
Vecchio, F.J., and Collins, M.P. (1982), "The Response of Reinforced Concrete
to In-Plane Shear and Normal Stresses," Publication No. 82-03,
University of Toronto, Toronto, Canada, March, 332 pp.
Vecchio, F.J., and Collins, M.P. (1986), "The Modified Compression Field
Theory for Reinforced Concrete Elements Subjected to Shear," AC!
Journal, Proceedings, Vol.83, No.2, March-April, pp. 219-231.
Vecchio, F.J., and Collins, M.P. (1993), "Compression Response of Cracked
Reinforced Concrete," Journal of Structural Engineering, ASCE,
December, Vol. 119, No.12, pp. 3590-3610.
Vecchio, F.J., and Lai, D. (2004), "Crack Shear-Slip in Reinforced Concrete Elements,"
Journal of Advanced Concrete Technology, Vol. 2, No.3, pp. 289-300.
Walraven, J.C., (1980), "Aggregate Interlock: A Theoretical and Experimental
Analysis," Report, Delft University Press, Delft, Netherlands.
Walraven, J.C., (1981), "Fundamental Analysis of Aggregate Interlock,"
Journal of Structural Engineering, ASCE, Vol. 107, No.11, pp. 2245-
2270.
Warwick, W.B., and Foster, S.J. (1993), "Investigation into the Efficiency
Factor Used in Non-Flexural Reinforced Concrete Member Design,"
UNICIV Report No. R-320, School of Civil Engineering, University of
New South Wales, Kensington, Sydney, Australia, July, 81 pp.
Xie, Y.M., and Steven, G.P. (1993), "A Simple Evolutionary Procedure for
Structural Optimization," Computers and Structures, Vol. 49, No. 5, pp.
885-896.
Yun, Y.M. (2000), "Nonlinear Strut-Tie Model Approach for Structural
Concrete," AC! Structural Journal, July-August, Vol. 97, No. 4, pp. 581-
590.
Yun, Y.M., and Ramirez, J .A. (1996), "Strength of Struts and Nodes in Strut-Tie
Model," Journal of Structural Engineering, ASCE, January, Vol. 122, No.
1, pp. 20-29.
Zhang, L. X. B., and Hsu, T.T.C., (1998), "Behavior and Analysis of 100 MPa
Concrete Membrane Elements," Journal of Structural Engineering, ASCE,
January, Vol. 124, No. 1, pp. 24-34.
414
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix A
Post Processing Tools Based on Augustus for Embedded Reinforcement Element
Including Bond Stress-Slip
In this appendix, a summary of the post-processing tool based on Augustus for
embedded reinforcement element including bond-slip is comprehensively presented. A
brief introduction to Augustus and the MATLAB computer programs specially written to
export and convert the output data from the proposed nonlinear FEA into the Augustus
input file format are also given in details.
Augustus, an advanced graphic and visualizing post processing tool for Vector 2,
was developed by Vector Analysis Group. The tool was designed and implemented to be
fully compatible with Vector 2 in comprehensively presenting the nonlinear FEA results;
for examples, global and local load-deformation responses, element level stress and strain
between and at crack locations, deflection, crack pattern, and the parameters associated
with the damage propagation. The examples of the screens and features of Augustus are
illustrated in Fig. Al to AS. Fig. Al illustrates the overall screen of Augustus where the
analysis details and parameters are given. In the analysis details, types of FE and the
numbers of the elements adopted in constructing the FE model are provided. The selected
behavior and constitutive models adopted in the analysis are also given in the analysis
parameter box. Fig. A2 illustrates the screen displaying the mesh topological data,
boundary conditions and load cases of the model. Fig. A3 illustrates one of the most
significant features of the tool in displaying the parameters associated to the damage
conditions and failure state of the structures. Those parameters are element cracking state,
crack width, and the ratio of the principal compressive stress to the available compressive
strength of the concrete (fem).
Analysis Details
Total Number of Nodes:
Total Number of Elements:
Number of Triangular elements:
Number of Rectangular Elements
Number of Truss Elements:
Number of Linkage Elements:
Number of Contact Elements:
Number of Concreta Types:
Number of Load Stages:
Analysis Type:
1212
1444
0
1102
342
0
0
3
18
VecTor2V.4
Analysis Parameters
ConvergenceCrttEn'la Olsplauments-Welghied
Compress1on9anCu!""e Hognesiad(Par<ibola)
Compress1onPos\.Peak Mot11nedParlf.1<ent
Compress1onSoltenlng Vecclllo1992A
Tens1on8t!lfen1n11 Modified Bentz
Tens1onsone11111g lJ11e<1r
Tension611hllmg Nol Considered
Conlinemen161rtin\11h NolConsldeied
ConereteD1litl;it1on Vanable-Kupfe1
C1ackmgCrttenon Mohr-Coulomb{Stress)
CrackShearChetk No!Conslde1ed
C1ackWldlhCheck Crack Umll (Aggl5)
Concrele8ond PerfetlBond
ConcreteCreepJR11laJ1 No1Consid11red
ContreteHJste1es1s Nonl1m1arwl011'$111'0
Steel Hysteresis Setkm(wlLotalY1eld)
Material Types
w w
..... ow
f'\11 U l l l l t l l ~ I U l l ~ Ill flllllllllt::LI t ~
Specimen 7A (Abaqus input)
Struct.S2E 11/01107
Fig. Al The Screen of Augustus Displaying the Analysis Details and Parameters
415
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. A2 The Screen of Augustus Displaying the Mesh Topological Data, Boundary
Conditions and Load Cases of the Model
Fig. A3 The Screen of Augustus Displaying the Parameters Associated to the
Damage Conditions and Failure State of the Structures
416
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I I / /
, I
/ ,' .. / ,'
,, / /
'I I\\',' ',,,,'
', I I
'i ,
I I ' I
1/,/-' ',//////l,.;/// \\\\\\\\ \\
i//\\\\\\\\ \'
/,!I, I!/// I',;///// /I/' If I
1
1 \ \ \ \ \ \ \
\ \
''I
', ,\
'\\
; !
'' r \, \
I I I/,' \ I : : '; i 1/ i I I I
2000x
Cr111Cklo'.tth$Pln<1 oomm,lllld,1t111;k .. 2oomm
Fig. A4 The Screen of Augustus Displaying the Global Responses of the Structures
Fig. A4 illustrates the global responses; for examples, the load displacement and crack
pattern, in the structure at all loading steps. The local responses in a specific location of
the structure can be displayed by using the tool provided in Augustus as illustrated in Fig.
AS. The examples of the local responses are strain and stress components at and between
cracks in concrete and smeared reinforcement, cracking information, and vital sign.
-4.20 -350 -280 -2.10 -1.40
mmlm
E,: 0.22 mmhn
Eyi 0.09 mmhn
y,,: -1.64 mmhn
E1: 0,97 mmhn
E2' -0,67 mmhn
a: 132,7 deg,
f.,: 51,3 MPa
r,,: 29,8 MPa
snr8: 51 mm
Vci: 1.13 MPa
-0.70
-6 0
-9 0
-12 0
b:-5,50 MPa
f,y: -6.75 MPa
v.,:-7.70 MPa
fi: 1,60 MPa
fi:-13,8 MPa
f2max: -59 .4 MP a
fucr: 98.3 MPa
fsycr: 85.0 MPa
w: 0,05 mm
0.00 MPa
0.97
v
Fig. AS The Screen of Augustus Displaying the Local Responses of the Structures
417
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Typically, Augustus need four text files to fully display the analysis results. Those
four text files are: 1) *.JOB, 2) * .S2E, 3) * .L2E, and 4) * .A2E. The JOB file contains the
general information of the FEA analysis; for examples, the job title, file name, structural
data, loading data, analysis parameters, and material behavior models. The S2E file
contains all data of the FE model of the structure; for examples, the numbers and types of
the elements used in constructing the model, the material properties, reinforcement ratio
and orientation, nodal and element topological data, support restraints, and crack
parameter. The L2E file contains the load case data adopted in analyzing the structure.
Those load cases can be; for examples, joint loads, support displacement, element gravity
load, temperature load, pre-strain induced load, etc. The analysis results corresponding to
each loading step are fully recorded in A2E files which are automatically generated by
the solver (Vector 2) at the end of each analysis step. In each analysis step, the file
records the analysis parameter, loading condition, and structural parameters associated
with the global and local responses of the structure. Since those text files are human
readable text files, a simple text file generation by using any programming languages can
be possibly used to generate the files containing the analysis parameters from a specific
solver. Thus, in this research, the author adopts MATLAB programming language to
retrieve all necessary analysis data and results from the solver (Appendix D) and
subsequently generate those four text files for displaying the analysis results in Augustus.
By generating those four text files, all analysis results can be comprehensively presented
in Augustus. In addition, since Augustus presents the truss elements by drawing the
straight line connecting two specific points in the working space, the presentation of the
embedded reinforcement can be thus directly obtained as illustrated in Fig. A6.
2
Fig. A6 The Screen of Augustus Displaying the Embedded Reinforcement Element
418
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix B
Automatic Mesh Generation by Patran
In this appendix, a brief introduction to Patran, mesh generation technique, and
procedure to generate the mesh for Vector2 and the proposed nonlinear FEA is given.
The MATLAB computer programs specially written to export and convert the output data
from Patran format into Vector2 and the proposed nonlinear FEA input file format are
also given in details.
Patran, a general purpose, 3-D mechanical computer aided-engineering (MCAE)
and visualizing post processing package that provides a complete CAE environment for
integrating engineering design, analysis, and results, was developed by MSC Software
company. The package was designed and implemented to be able to generate various
input formats for several nonlinear analysis softwares; for example, ABAQUS, ANSYS,
MSC.Nastran, LS-DYNA 3D, etc. One of the powerful features of the package is
automatic mesh generation. Both surface and 3-D solid meshers are available. In this
appendix, only the automatic surface mesher will be presented.
The automatic surface mesher in Patran provides several capabilities in the FE
mesh generation; for examples, automatic 2-D surface meshing, user specification of
element size globally and/or in individual regions, automatic mesh smoothing, mesh
density controls including curvature checks, and geometry source independent surface
meshing. In addition, the efficiency of the bandwidth or skyline of the stiffness matrix
can be simply enchanced by the available node and element re-numbering function.
In generating the input file for Vector 2, the automatic 2-D surface meshing can
be used to generate the mesh that is fully compatible to the alignment of the
reinforcement details as required when the truss elements are adopted to model the
reinforcement bars as illustrated in Fig. 7. 22b. The user specification of the element size
globally is also benefitial to the FE model in Vector2 when the effect of the mesh
dependency is crucial; for example, the nonlinear FE model of complex Regions as
illustrated in Fig. 6.30.
In generating the mesh for the proposed nonlinear FEA, only the automatic 2-D
surface meshing is necessary since the proposed nonlinear FEA includes the embedded
reinforcement in the formulation. Thus, the mesh generation is adopted to generate the
mesh associated with only geometry, boundary, and loading conditions.
The example of the overall screen and feature of Patran is illustrated in Fig. Bl.
Fig. Bl illustrates the overall screen of Patran displaying the features of Patran in
generating the geometry, meshing and elements, loads and boundary condition, material
properties, analysis, and visualization of the results.
To generate the geometry of the structure used in this research, the standard
procedure is adopted as follows: 1) constructing the node associated with the structural
419
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
geometry; 2) connecting the nodes to construct the boundary of the structural geometry;
and 3) collecting the edges of the structural geometry to construct the surfaces in the
structural domain. Fig. B2 illustrates the example of the screen displaying the features for
generating the geometry of the structure.
To generate the structural mesh, the surface of the structure obtained form the
construction of the structural geometry is adopted. The first step of the mesh generation
involves the assignment of the mesh size which can be assigned based on either the
number of elements or the element length. The type of the mesh distribution throughout
the domain can be also selected by several approaches; for examples, uniform, one way
bias, two way bias, curve based, tabular, or PCL function. In this research, only the
uniform (non-bias) type of the mesh generation is adopted. Fig. B3 illustrates the screen
displaying the features for performing the automatic mesh generation.
M$C.P9trtin 20ot
""c - "
,.,.a....11 .,.,.......,. yPnpertiA vu.tC- vfilllll ,.,. ...... , vXV,...
!Il lJllrJ 0
Fig. Bl The Screen of Patran Displaying the Overall Features for Generating the
Geometry, Mesh, Loading and Boundary Condition of FE Models.
After the mesh is fully generated throughout the structural domain, the element
types can be assigned to the overall or specific regions of the model. Patran provides
several available types of element and formulation that can be used in constructing the FE
model; for examples, 2D solid, Membrane, Shell, 2D interface, Rigid Surface, and 2D
Rebar or Gasket. In this research, 2D plane stress solid and Rebar element types are
adopted to model the concrete regions and reinforcement details, respectively. Fig. B4
illustrates the screen displaying the features for assigning the element types to the
structural mesh.
420
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. B2 The Screen of Patran Displaying the Features for Generating the Geometry
of the Structure.
Fig. B3 The Screen of Patran Displaying the Features for Generating the Mesh of
the Structure.
421
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. B4 The Screen of Patran Displaying the Features for Assigning the Element
Types to the Existing Mesh of the Structure.
Subsequently, the boundary condition and the load cases are directly assigned to the
model for completing the modeling process. In the next step, the input text file associated
with the selected FE solver can be generated. In this research, the input text file in the
ABAQUS format is used together with a small MATLAB program specially written by
the author to retrieve and to convert the data of the model from the ABAQUS input text
file into either Vector2 or the proposed nonlinear FEA format.
422
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix C
Analysis Results of Reinforced Concrete Shear Panels
In this appendix, a summary of the numerical results in the element level of the
proposed nonlinear FEA (Chapter 7) in predicting the structural responses of the selected
shear panels in PV shear panel series that were tested and reported the experimental data
by Vecchio and Collins (1982) is provided in details.
The loading and material properties of all shear panels used in the analysis are
based on Vecchio and Collins (1982). In addition, the plots for comparing the numerical
prediction of the structural responses of those shear panels with the observed data are also
provided in the appendix. The plots consist of the following items:
1) Comparison between the predicted and observed shear stress (vxy) and shear
strain (Yxy).
2) Comparison between the predicted and observed principal compressive strain
ratio (Ec2IEco) and principal compressive stress ratio (fc2lfc2max).
3) Comparison between the predicted and observed shear stress (vxy) and
principal stress direction in concrete (8c).
4) Comparison between the predicted and observed shear stress (vxy) and
principal apparent (measured) strain direction (8
1
;).
5) Comparison between the predicted and observed shear stress (vxy) and
longitudinal reinforcement strain (Ex).
6) Comparison between the predicted and observed shear stress (vxy) and
transverse reinforcement strain (Ey).
423
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
9
_8
: 7
::il
VI 6
..
g 5
4
"'
QI
s::.
"'
10
9
3
2
1
0
0
_8
:.
7
::il
6

5
4 ...
"'
QI
3 s::.
"'
2
1 .
0
0
0 0 002
2

0.0005
. ..
0 004 0 006
Shear Strain(mm/mm)
PV1
-proposed
Exp
0008 0 01
.......... '\ <:,'-. :0
4 6 8 10
Shear Stress (MPa)
"
.-t!
0.001 0.0015 0.002 0.0025
ex(mm/mm)
1 .,
08 j
I
0 0.2 04 06 08
sc2/cc0(mm/mm)
65 1
i
45 !--'-'-.......,..,, ,,.-.- .. ,....._, . .:_ .. _'' '-"-;..;. .. "" --
40
10
9
_a
"'
II.
7
::il
"ai 6
..
!!! 5
;;
.. 4
"'
QI
3 .r::
"'
2
0
0 2
"''
0 0.0005
4 6 8
Shear Stress (MPa)
.h
,.
_,,.
. .
0.001 0.0015 0002 00025
ey(mm/mm)
10
0.003
Fig.Cl Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PVl).
424
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
60

..
Ci
ci50
45
40
4
3.5

3
i25
VI
g 2
fl)
Ii 1.5
QI
.c
1 fl)
0.5
0
0
0
. - .. -:). ...... ;; -- .. ... "" .. - ... '
PV4
[ - proposed!
!
0.005 0.01 0.015 0.02
Shear Strain(mm/mm)
"f!'--1>
2 3 4
Shear Stress (MPa)
"'-"---
0002 0004 0.006 0008 0.01
ex( mm/mm)
1 .
0.8
60
50


c
C[
20
10
0 0.5 1.5 2
cc2/ccO(mm/mm)

0 2 3 4
Shear Stress (MPa)
4
3.5
-;-
3
II..
.!' ..... .
::
17125
VI

2
:U15
QI
.c
1 fl)
0.5
0
0 0002 0.004 0.006 0.008 0.01
ey(mm/mm)
Fig.C2 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV 4).
425
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 1
..
08 l\l
Q.
........ ">
..
!,4 t--'-
Ill
fr06 Ill
f3
...

;;;
j-,.,
...
"
&o.4
\P
PV6
.c
11)1 !-proposed 0.2
!->.Exp
0
-- .... r- ----- ..- - .. ------1
0
0 0.005 0.01 0.015
0 0.05 0.1 0.15 02 0.25 0.3
Shear Strain(mm/mm) ec2focO(mm/mm)
65
1
65
I
I
60
60 I
I

j

" s
..
I .s!50
8'50
QI
c
< I
. v .. . !' '" ""
.
<45
45
, ... -
\. ... .. -, . -
!
40
.....
40
35
0 2 3 4 5 6
0 2 3 4 5 6
Shear Stress (MPa)
Shear Stress (MPa)
6
6
5
5
'i'
.....
.'"--

" Ill
Vi"
Ill ,, Ill

'
i
VI ,.

....
::2
.c .c
VI VI
1 1
0
0
0 0 002 0.004 0 006 0.008 0 01
0 0.002 0004 0.006 0 008 001
ex( mm/mm)

Fig.C3 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV6).
426
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
9
_e
:. 7
::!!
'iii' 6
Ill
.,
5 ....
u;
4 ....
..
.,
3 ..::
"'
2
1
0
65 -
60-;
-
.,
a,
--
.7-
.>
"
PV?
0 0.002 0 004 0.006 0 008
Shear Strain(mm/mm)
45 -.. -.- .. -.,- .. -.-.,.----
40 ............... - --.--
0 2 4 6 8
Shear Stress (MPa)
10
9
_a
..
ll.
7
::e

, .. ,.
II! 5
0
4 ... _<
..
.,
3 ..::
"'
2 -
10
65
60
55
-;:;
$50
.,
g>45
<
40
35
30
10
9
_e
..
7
'iii' 6
5
... 4
..
.! 3
"' 2
0
1 / 1 "
0 - ---,.. ............- ....-.,. ....._ ......._ .... .,.. ............. -.-............r--------------, 0
0 0.0005 0.001 0.0015 0.002 0.0025
0
i:x(mm/mm)
04 0.6 0.8
cc2/ecO(mm/mm)
:'). .. ./ ..... - . ...
2
Stress
8 10
0.001 0.002 0.003 0.004 0.005
ey(mmlmm)
Fig.C4 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV7).
427
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
11
10
'(;' 9
8

! 6
g;
5
..
..
4 ..
.c
3 I/)
2
1
0
0
65
60

..
' ,i50
45 i
40
0
10
9
_a
..
Q.
7
::E
i6
!!! 5
;;;
4 ..
..
..
3 .c
I/)
2
1
0
0
PV8
1

... Exp
0.003 0.005 0.008 0.01 0.013 0.015 0.018 0.02
Shear Straln(mm/mm)
-'"'-,
"
2 4 6 8 10
Shear Stress (MPa)

"
, .
''
0.0005 0001 0.0015 0.002
.:x(mm/mm)
0.8
frO 6
u.
C'l
il:!o4
..
,
0.2
0
0 02 04 06 08
ec2/f:cO(mm/mm)
65
60


..
'El

45
_---',J ... ,.. .... (,..>',,,
.,f.
40
0 2 4 6 8 10
Shear Stress (MPa)
10
9
_a
..
Q.
7
::E
Iii 6
Ill

5
y
I/)
.. 4
m
3 .c
..
I/)
2
".
0
0 00005 0001 0.0015 0.002
ey(mm/mm)
Fig.CS Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PVS).
428
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
11
10
';' 9
!E 8
i 7
e
ti)
...
..
QI
&::
fl)
6
5
4
3
2
1
0
~ 5 5
QI
c,
,i50
45
5
~ 4
:.
:IE
'[3
e
....
~ 2
~
&::
fl) 1
0
PV9
I
- proposed I
.. ..r,- Exp .
0 0.003 0.005 0.008 0.01 0.013 0.015 0.018 0.02
Shear strain(mm/mm)
>- >-.
~
2 3 4 5
Shear Stress(MPa)
,,
0.8
frO 6
IL
;,.
~ 0 4
0.2
0
0
65
60
j55
QI
c,
,i50
45
40
0
5
0
0 0.0005 0.001 0.0015 0
ex( mm/mm)
'
0.5
.

1.5
ec2/ecO(mm/mm)
t,"
2 3
Shear Stress (MPa)
2 2.5
,0
4 5
j - proposed I
:.:': 'xP... J
0.0002 0.0004 0.0006 0.0008 0.001 0.0012
ey(mm/mm)
Fig.C6 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV9).
429
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
11
10 ..

8
VI 7
6
PV10
5 j
.----
j -proposed
I .. ,. Exp
1/l 3 I
2 1
1 .
0 r
t'-.
0 0.003 0.005 0.008 0.01 0.013 0.015
Shear Straln(mm/mm)
65
60
55
50
.!!
g'45
<
40
35
30
0
5 -,
!
_4
:.
:!:
:
.. i
ts I
... 2 -,
m .
.c
l/l 1
t
. '.:!.
't , ..........
:). .. . .........
2 3 4
Shear Stress (MPa)
T ..."
0.018 0.02
5
0
-
0.8
e-06


0.2
0.25 0.5 0.75 1.25 1.5 1.75 2
sc2/&cO(mm/mm)
65
60
55
-;:;

.950
..
g.45
<
40
"-.
"\
35

30
0 2 3 4 5
Shear Stress (MPa)
5
_4
Cll
IL
:!:
"iil3
en
!!!
....

Cll
..
.c
l/l1
0
..
0 0 0002 0 0004 0.0006 0.0008 0.001 0.0012 0 0.002 0.004 0.006 0.008
ex( mm/mm) cy(mm/mm)
Fig.C7 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PVlO).
430
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
11
10
ii' 9
8

6
5
4
!V:u".
PV11
-proposed
.... Exp
1.
0.8
frO 6
u.
?>I
.i!o.4
-.. )
0.2
0 0.003 0 005 0 008 0 01 0.013 0.015 0 018 0 c 0 0.25 0.5 0 75 1 25 1 5 1 75 2
65
60

.!:!
en
,i50
45
40
4
0
0
0
Shear Strain(mm/mm)
... ., . .... -... , ....... - .. " )
,.
00005
2 3
Shear Stress (MPa)
."':>- ---
0.001
ex(mm/mm)
0.0015
4
0
ec2JecO(mm/mm)
65
60
55
"
$50
.!!
g>45
c:c
40
35
30
0 2 3 4
Shear Stress (MPa)
4
3.5 -- ...)
'i
3
:IE
vr2 5
e 2
ti)
l; 15
GI
.t::.
1. I/)
0.5
0
. ----,
0 0.002 0.004 0.006 0.008
ey(mm/mm)
Fig.CS Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PVll).
431
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12 '
11
10
0.8
ti 9
8

PV12 e-o.6
5
tu 4 ;
1!
u.
,,-
i-proposed I
04
_
!-<>Exp I
I/I 3 ............ -,, . -...... .

0 0.003 0 005 0.008 0 01 0.013 0.015 0.018 O.C
Shear Strain(mm/mm)
..... ,:.
0.2
0
70 I
651
;:;60 '
;55
en :
c i
<50 I
45 i
/t ...... <>
0.25 0.5
-
0.75 1-25
ec2/&c0(mm/mm)
. ....
1-5 1.75 2

4
0
0 2 3
Shear Stress (MPa)
.............. ,

0 0002 0.0004 0.0006 0.0008 0.001
cx(mm/mm)
4 0
3.5
3
l25
i 2
GI
...
tii1.5
...
"'
GI
1 .c
Ill
0,5
0
ooo 0
.,
{;,"
,.
2 3
Shear Stress (MPa)
,:)
0.002
.........
0.004
&y(mm/mm)
---
0.006
4
0008
Fig.C9 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV12).
432
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
11
10
ii' 9
8

! 6
tiS
5
...
Cll
4 Cll
.::.
3 fl)
2
1
0
0
65
60

.!!
l:JI

45
40
0
7
6
;f 5
::e
!4

..
.! 2
fl)
0
0
0.8
frO 6
u..
N 9
PV14
&04
{
()
- proposed 0.2
. . E:l<P
0.003 0.005 0.008 0.01 0.013 0.015 0018 0.02
Shear Strain(mm/mm)
2 4 6 8
Shear Stress (MPa)
0.0005 0.001 0.0015 0.002
ex{ mm/mm)
65
60

Cll
Cl

45
0
0 0.25 0 5 0.75 1.25 1 5 1.75 2
"c2focO{mm/mm)
,J-$"'
40 ------ .,.- - - -- ------ ; .
7
6

::e

f
t;3
...
Cll
.!2
fl)
0
0 2 4 6
0 0.002
Shear Stress {MPa)
0 004
.:y{mm/mm)
0006
8
0.008
Fig.ClO Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV14).
433
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
60
~ 5 5
GI
DI
ci50
45
40
0
0
PV16
-proposed
Exp
0 003 0.005 0.008 0.01 0.013 0 015 0 018 0 o:
Shear Strain{mm/mm)
0.5 1 1.5 2 2.5 3
Shear Stress(MPa)
0.8
0 0.25 0.5
65
60
45
......... v
0.75 1.25
GC2/ccO(mm/mm)
1.5 1.75 2
40 - - - ~ - - r - - - - . . - - -----,.. . ... ---,- ..... , .... -
0 Q5 1 1.5 2 2.5 3
Shear Stress (MPa)
0 0.005 0.01 0.015 0.1 0 0.002 0.004 0.006 0.008
ex( mm/mm) ey(mm/mm)
Fig.CH Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV16).
434
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12 i
1 ~
11.
10.
0.8
ii 9
~ 8
PV18
frO 6 1 iii
7
Ill
u.
e 6
-proposed
i'l
u;
5 ~ 0 4 ~
..
..
4
..,.Exp
GI
.c
.. ,
I/) 3 <1
2
<r 0.2
>-"'"
1
0 0
0 0.003 0 005 0.008 0.01 0.013 0.015 0 018 0.02 0 0.25 0.5 0.75 1.25 1.5 1 75 2
Shear Strain(mm/mm) ec2/ecO(mm/mm)
65
65
60
60
.. /
55
f /
5 5 ;!.--
.!50
GI g)
g,
'
c
~ 5 0
<45
... >
0 " .
40 . . ...
45
_,.
35
40 30 -------.
0 2 3 4 0 2 3 4
Shear Stress (MPa)
Shear Stress (MPa)
12 12 -,
11
11 1
10 10
ii 9 ii 9
~ 8 ~ 8
J7
I: 1
e 6
iii 5
~ 5 ..
..
4 m 4., GI
.c
3 ~ 3
I/)
................ _,
2 2
,, ........ y .
1 1
0 0
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0 0.002 0.004 0 006 0008
ex( mm/mm) ey(mm/mm)
Fig.C12 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV18).
435
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0
65
60
Ql55
c;,
c
<50
45
40
0
12
11 .
10
Ci! 9
II.
::E 8
1ii'
7
111
e! 6
~
5
....
.,
4 QI
.c
3 - ~ II)
2 "
1
0
PV20
[
...
-. - .... P. r.o. p. o. -s.e .. d. i,_
---Exp
0.003 0.005 0.008 0.01 0.013 0.015 0.018 0.02
Shear Strain(mm/mm)
I
I
'>i'
-<:--
't'' -f' ~
2 3 4 5
Shear Stress (MPa)
,,, ~ .
1 l
08 J
!
frO 6 -i
~ I
~ : : v :
0
65
60
Ql55
c;,
c
<50
45
40
0
12
11
10
Ci! 9
!i 8
1ii'
7
Ill
~
6
....
5
.,
4 QI
.c
3 II)
2
y
1
0
0
0.25 0.5
..
0.002
0.75 1.25
ec2/ec0(mm/mm)
. ..
2 3
Shear Stress (MPa)
1.5 1.75 2
4 5
0.004 0.006 0.008
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 O.OOE ey(mm/mm)
ex( mm/mm)
Fig.C13 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV20).
436
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
, ....... , ....
0 - r
PV21
I-proposed

0 0 003 0.005 0.008 0.01 0.013 0 015 0.018 0.02
Shear Strain(mm/mm)
65 -
60

GI
Ci
,,iSO
45 " .......... ... .
0.8
fr0.6

&04
0.2
0.25 0.5 0 75 1.25 1.5 1 75 2
cc2/ecO(mm/mm)
65
60

40

12
10
0
0 2 3 4
Shear Stress (MPa)
5 6
0 0.001 0.002 0.003 0.004 0.005 0 006 0.007 0.008
ex( mm/mm)
0
12 -
10 ;
2 ,:
0 .
0
2 3 4 5 6
Shear Stress (MPa)
........ ...... ' ... .
0 002 0.004 0 006 0008
ey(mm/mm)
Fig.C14 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV21).
437
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
11
10
iii 9
8

.... 5
SI 4
3
2
#>jo.
"
"'
"
<'

"
PV22
-<).
1

0
65
60

GI
'iii

45 "
40
12
11
10
iii' 9
8

e 6
w 5
....
..
GI
.r.
"'
4
3
2
1
0
0
/
0.003 0.005 0.008 0.01 0.013 0.015 0.018 0 (
Shear Strain(mm/mm)
.
2 4 6 8
Shear Stress (MPa)

0.8
0.2
65 .
45
40
0
" 1 10
iii'
IL
:5 8
Iii
Ill
e 6
w
....
..
4 GI
.r.
"'
2
0.25 0.5
2
_,,
i'

"''
(
0.75 1.25
ec21EcO(mm/mm)
- .
4
Shear Stress (MPa)
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0 0.002 0.004
ex( mm/mm) ey(mm/mm)
1.5 1.75 2
6 6
0.006 0.008
Fig.C15 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV22).
438
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
11
10
:-9
::IS 8
l7
6-
5
ill 4
3
2 -
1
PV24
-proposed
_, .. Exp

0.8
frO 6
u.
,l
r;i

0.2
0
0 0 003 0 005 0.008 0.01 0 013 0 015 0.018 o.o: 0 0.25 0.5 0.75 1.25 1.5 1.75 2
65 .
60
55 -i
!50'
Shear Strain(mm/mm)
G> i <),
gi45 --1----------------
<( .
40
35 i
30
0
IV
a.
:re
Iii'
Ill
di

...
..
di
.c
r/)
12
11
I
10
j
9
I
8
j

5>' I
5
J
I
4
3
2
1
r-0
2
. . . -( . .0
4 6 8
Shear Stress (MPal
10 12
65 1
i
60 I
I
f" l
c'i50
45 -1
40L
0
12
10
;f
:re 8
: ... tl
- 6 :i
us
...
m 4
.c
r/)
2
-0 0 0.001 0 002 0 003 0.004 0.005 0.006 0.007 0.008 -0.001
ex( mm/mm)
2
cc2/cc0(mm/mm)
1.,,.............. . ":l't
0.001
4 6 8
Shear Stress (MPa)
0.003 0.005
cy(mm/mm)
10 12
0.007
Fig.C16 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV24).
439
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
~ ~ l
10
.. 9
~ 8
1 ~
.. 5
SI 4
~ 3
2
1.
0
PV25
-proposed
-<>Exp
1 .
08-
a.o 6
~
U.0.4
0.2
0 0.003 0.005 0.008 0.01 0.013 0.015 0.018 oc
65
60
45
12
11
10
0
.. 9.
~ 8
i 7
g 6
~ 5
SI 4
~ 3.
2
1
0
Shear Strain(mm/mm)
. ~ .. > ,-. . ,,, ......... , ' ...... ' . ci
2 4 6 8 1(
Shear Stress (MPa)
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.00i
ex( mm/mm)
65
60
55 ...
.....
.950
Ill
gi45 ~
<
40 ~
35
30
12
11
10
0
'\? 9
~ 8
i?
~ 6
~ 5:
m 4.
~ 3.
2 '
0
. -0
0.25 0.5 0.75 1 1.25 15 1.75 2
cc2/cc0(mm/mm)
/t ............. ' .. ' ...... ,,,, .. ,
. I!.
..
2 4 6 8
Shear Stress (MPa)
10
0.002 0.004
<:y(mm/mm)
0 006 0 008
Fig.Cl 7 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV25).
440
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
0.8

t,'
PV26
..
"
-
0 0 003 0 005 0.008 0.01 0.013 0.015 0.018 0 o:
0 0.25 0.5 0.75 1 25 1.5 1.75 2
60
50 ;


c
<
20
10
0
12
11
10
IU 9
8

6
5
m 4
i7l 3
2
1
0
I
0
Shear Strain(mm/mm)
t-
2 3 4
Shear Stress (MPa)
5 6
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008
r.x(mm/mm)
60
50
"40
..s
QI
"5130
c
<
20
10
12
11
10
m9

0
6 -
5
m 4 I>
ti 3 ,.,
2
1
0 ,.
0
EC2/ec0(mm/mm)
......... .....
2 3 4
Shear Stress (MPa)
0 002 0.004 0.006
cy(mm/mm)
-) <of
5 6
0.008
Fig.C18 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV26).
441
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
11
10
-;- 9
~ 8
0.8
l7
frO 6
i
6
.0
~
...
5
J
..
'
PV27
~ 0 4
OI
4
~
QI
.r:.
3
,.,
"'
.
2
0.2
1
0 0
0 0.003 0.005 0.008 0.01 0.013 0.015 0 018 oc 0 0.25 0.5 0.75 1 25 1 5 1.75 2
Shear Strain(mm/mm) ec2/&cO(mm/mm)
65
65
60
60
55
~ 5 5
~
$50
GI QI
g,
g45
.:i50
<(
'I' "'1\
. ., ~ ...-.T
40
45
.v ......
"
35
40 30
0 2 4 6 8 0 2 4 6 8
Shear Stress (MPa) Shear Stress (MPa)
12
12
11
11
10
10
-;;- 9
-;;- 9
~ 8
~ 8
l7
l7
.s
6
~
~ 6 ,/
II)
5 <
~ 5 ..
Ill
4 .
"
m 4
QI
.r:. .
~ 3
II) 3
2
2
1
1
0
........ T. ....... T- -T--1
0
0 0.001 0.002 0 003 0.004 0.005 0.006 0.007 O.OOE 0 0.002 0.004 0.006 0.008
ex(mm/mm) ey(mm/mm)
Fig.C19 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV27).
442
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
11
10
ti 9
a.
::E 8
iii 7
6
in 5
...
lll 4
i'-i 3
2
1
PV28
::
0
0.8
frO 6


02
0 0003 0.005 0.008 0.01 0.013 0.015 0.018 o.o:
0 0.25 0.5 0.75 1.25 1.5 1 75 2
65 1
!
60

GI
Ci
,i50
Shear Straintmm/mml
"c21f:c01mm/mml
65 -
60
7
55
.$
GI
Ci
,i50
.
45 45
... t'. - & - "\}. ('f" '/", '
40
0 2 4 6 8 0 2 4 6
12
11
10
ti 9
8
J7
e! 6
ii) 5.;
...
m 4
3
2
1 ..
Shear Stress (MPa) Shear Stress (MPa)
12
11
10
ti 9
8
i7
_g: 6
II)
...
..
GI
.c
3 II)
2
1
0 -----r-----r--T------r----r----r--r---- 0 l
0 0001 0.002 0003 0004 0.005 0.006 0.007 0.008 0 0.002 0004
ex( mm/mm) i:y(mm/mm)
0 006
8
0.008
Fig.C20 Comparison between the Numerical Prediction and Observed Experimental
Data for the Structural Responses of Shear Panel (PV28).
443
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Author's Biography
Sukit Yindeesuk was born in Bangkok, Thailand on September 14, 1979. He
graduated from Kasetsart University, Bangkok, Thailand, in 2001 with a Bachelor's
Degree (First Class Honor) in Civil Engineering. He received his M.S. in Structural
Engineering from University of New South Wales, Sydney, Australia, in 2004. Prior to
his graduate studies, he was a structural engineer with Meinhardt (Thailand) Ltd. for two
years. During his graduate studies, he served as a graduate research assistant and a
teaching assistant with the Department of Civil and Environmental Engineering at the
University of Illinois at Urbana-Champaign.
444

Você também pode gostar