TR155 Cordova

Department of Civil and Environmental Engineering
Stanford University
VALIDATION OF THE SEISMIC PERFORMANCE OF

COMPOSITE RCS FRAMES: FULL-SCALE TESTING,
ANALYTICAL MODELING, AND SEISMIC DESIGN
By
Paul Phillip Cordova

and
Gregory G. Deierlein
Report No. 155

September 2005
The John A. Blume Earthquake Engineering Center was established to promote
research and education in earthquake engineering. Through its activities our
understanding of earthquakes and their effects on mankinds facilities and structures
is improving. The Center conducts research, provides instruction, publishes reports
and articles, conducts seminar and conferences, and provides financial support for
students. The Center is named for Dr. John A. Blume, a well-known consulting
engineer and Stanford alumnus.
Address:
The John A. Blume Earthquake Engineering Center

Department of Civil and Environmental Engineering
Stanford University
Stanford CA 94305-4020
(650) 723-4150
(650) 725-9755 (fax)
racquelh@stanford.edu
http://blume.stanford.edu
2007 The John A. Blume Earthquake Engineering Center

Copyright by Paul Phillip Cordova 2005
All Rights Reserved
ii
Abstract
Composite RCS moment frames integrate reinforced concrete columns with structural
steel beams, providing several advantages over conventional steel or concrete moment
resisting frames. Past studies have shown these systems to be efficient in both design and
construction stages while able to maintain sufficient strength and ductility necessary in
seismic applications. Despite this past research, use of this hybrid structural system in
the United States has been limited to non- or low-seismic zones, which is largely due to
the reluctance of the engineering community to accept this new system as well as a
lack of comprehensive seismic design criteria in building codes and specifications. In
addition, past studies have acknowledged that there is a fundamental need to test full
structural systems, both analytically and experimentally, in order to (1) substantiate the
knowledge that has been accumulated up to this point and (2) act as a proof of concept
for the composite RCS frames.
The primary goal of this research program is to fill this knowledge gap and facilitate the
greater acceptance and use of composite RCS systems as a viable alternative to
conventional lateral resisting systems. This research synthesizes and interprets some of
the latest provisions and past studies on RCS systems and applies the accumulated
knowledge to (1) develop and validate improved seismic design provisions for RCS
frames, (2) assess and demonstrate the seismic performance of RCS structural systems
through full-scale frame testing and analytical simulations of prototype building systems,
and (3) develop and validate modeling guidelines for nonlinear analysis and performance
simulation. The cornerstone of this study is the planning, design, and testing of a full-
scale 3-story composite RCS moment frame. Using the pseudo-dynamic loading
technique, this specimen is subjected to a series of earthquake motions ranging in hazards
from frequent to extremely rare events. In addition to providing insight into the seismic
performance and design of composite RCS frames, the frame and supporting
subassembly tests also provide a rich data set for the validation of nonlinear analysis and
damage models. This also helps address one of the broader aims of this investigation,
iv
which is to provide support in the development of performance based earthquake
engineering.
Designed to evaluate the minimum limits of current building code requirements, the full-
scale test frame exhibited excellent seismic performance up through the maximum
considered earthquake intensity level. The damage patterns after each pseudo-dynamic
earthquake event were representative of the performance expected in well-detailed
moment resisting frames designed by current building codes, with no instances of brittle
failure (e.g. fracture). Nonlinear beam-column fiber elements and two-dimensional joint
elements were independently calibrated to multiple subassembly tests and modeling
recommendations are proposed. These models are used to simulate the test frame
conditions and loading protocol. The analytical results correlate well with the
experimental response within interstory drifts up to 3%, beyond which, the physical
damage in the frame exceeds what the analytical models used in this study are expected
to accurately capture.
Using the recommendations presented herein, trial designs of three case study buildings
(3, 6, and 20-stories) are generated, analytically modeled, and subjected to a suite of
earthquake ground motions at a range of hazard levels. The response of these case study
buildings are probabilistically evaluated considering several key engineering demand
parameters. These results, coupled with the response of the test frame, validate the
seismic behavior of composite RCS frames and also help assess and improve several key
design issues that are applicable to this system, as well as others.
v
Acknowledgements
First and foremost, I would like to extend my deepest gratitude to my professor, adviser,
and friend, Dr. Greg Deierlein. You have made this experience extremely rewarding and
it would not have been the same without you.
To the rest of the professors in the Structural Engineering and Geomechanics

Department, your role in my development as a researcher and a structural engineer should
not be understated. When I first arrived at Stanford University, I thought I knew
everything; fortunately you all saved me from myself and greatly expanded my horizons
in structural and earthquake engineering.
Special thanks to Dr. Keh-Chyuan Tsai and the rest of the NCREE staff and students that
took me in for over three months during the experimental phase of this project. It was an
amazing experience to work in one of the premier laboratories in the world alongside
with some extremely talented people. In addition to Dr. Tsai, I would like to single out
and thank a few of the most visible people on this project: Chui-Hsin Chen, Wen-Chi Lai,
Kung-Juin Wang, and Wei-Chung Cheng.
I would also like to acknowledge the contribution of Professor Gustavo Parra-Montesinos

and Luis B. Fargier-Gabaldon from the University of Michigan in the development of the
updated RCS joint guidelines.
To my fellow cohorts in the GGD Quatro ( Greg Deierlein), Arash Altoontash, Amit
Kanvinde, and Rohit Kaul, you all are an unforgettable group of guys and I wouldnt
have wanted to share this experience with anyone else. The stories we have together,
both inside and outside the research world, could fill up an entire novel, but for the sake
of all of us, I will refrain.
The Blume Center is a really special place filled with outstanding researchers and
remarkable people and I have truly been blessed to be a part of it throughout the years.
vi
Unfortunately, there are way too many people to mention here, but I am confident that
those people who deserve thanks do not need to see their name in print in order to know
that I appreciate them. Beyond that, I would like to take a line to thank Racquel Hagen,
the Administrative Associate of the Blume Center, for all her time and effort in keeping
the Blume Center and its inhabitants in order.
I would also like to acknowledge and thank my family. You all have provided me with
my foundation and my wings two gifts that have helped me persevere through the good
and bad times and accomplish the things I have.
And of course, my unending thanks to my beautiful wife, Mrs. Sylvie Cordova.

Honestly, I didnt plan for the first four months of our marriage to coincide with the final
months of my Ph.D. but what a way to start out. Your support and strength throughout
my entire academic career is appreciated.
The research reported herein has been supported by the Pacific Earthquake Engineering
Research (PEER) center, Stanford University (with special thanks to Dr. No Lozano),
Earthquake Engineering Research Institute (EERI), National Science Foundation, and the
National Science Council of Taiwan.
vii
Table of Contents
Abstract .............................................................................................................................. iv
Acknowledgements ............................................................................................................ vi
List of Tables .................................................................................................................... xiv
List of Figures .................................................................................................................xvii
Chapter 1: Introduction ............................................................................................... 1
1.1 Motivation for this Study .................................................................................. 3

1.2 Objectives.......................................................................................................... 4
1.3 Scope and Organization of Report .................................................................... 5
Chapter 2: RCS System Design ................................................................................... 9
2.1 Background of RCS Composite Moment Frames............................................. 9

2.2 Previous Research ........................................................................................... 12
2.2.1 Beam-Column Composite Joints............................................................... 12
2.2.2 Small-Scale Frame Tests ........................................................................... 14
2.2.3 Trial Design Studies of System Performance............................................ 15
2.3 System and Member Design Guidelines ......................................................... 15
2.3.1 General Building Design Requirements: IBC 2003
and ASCE 7-02.......................................................................................... 16
2.3.2 Member Design Requirements: Part II of the AISC
Seismic Provisions .................................................................................... 20
2.3.2.1 Reinforced Concrete Columns ............................................................ 20
2.3.2.1.1 Strong-Column Weak-Beam.......................................................... 21
2.3.2.1.1.1 SEAOC Blue Book Provisions................................................... 23
2.3.2.1.2 Precast RC Column Splice Design................................................ 24
2.3.2.2 Composite Steel Beams....................................................................... 25
2.3.2.2.1 Plastic Strength of Composite Beams ........................................... 27
2.3.2.2.2 Design of Shear Studs ................................................................... 29
2.3.2.2.3 Steel Beam Splices ........................................................................ 31
2.3.2.2.3.1 Bearing versus Slip Design: Bolt Banging................................. 34
2.4 Composite Joint Design Guidelines ................................................................ 36
viii
2.4.1 Joint Deformations .................................................................................... 37
2.4.2 General Detailing Requirements ............................................................... 37
2.4.3 Effective Joint Width................................................................................. 38
2.4.4 Joint Strength............................................................................................. 40
2.4.5 Inner Panel Shear Strength ........................................................................ 42
2.4.6 Inner Panel Vertical Bearing Strength ....................................................... 43
2.4.7 Outer Panel Shear Strength ....................................................................... 45
2.4.8 Joint Panel Shear and Vertical Bearing Moment Capacity........................ 46
2.4.8.1 Strong-Joint Weak-Beam..................................................................... 46
2.4.9 Detailing Considerations ........................................................................... 47
2.4.9.1 Ties within Beam Depth...................................................................... 47
2.4.9.2 Longitudinal Column Bars .................................................................. 47
2.4.9.3 Face Bearing Plates and Steel Band Plates ......................................... 48
2.4.9.4 Steel Beam Flanges ............................................................................. 49
2.4.9.5 Extended Face Bearing Plates and Steel Column ............................... 49
2.4.10 Model Validation ....................................................................................... 50
2.4.10.1 RCS Joint Tests Considered ................................................................ 50
2.4.10.2 Updated Joint Guidelines Validation Results ...................................... 51
2.4.10.3 Determination of Strength Reduction (f) Factors................................ 51
2.5 Final Recommendations .................................................................................. 54
2.5.1 Reinforced Concrete Columns .................................................................. 54
2.5.2 SCWB Criterion ........................................................................................ 54
2.5.3 Composite Steel Beams............................................................................. 55
2.5.4 Precast Element Splices ............................................................................ 56
2.5.5 Composite Joint Design ............................................................................ 57
Chapter 3: Full-Scale Composite RCS Test Frame ................................................. 76
3.1 Background ..................................................................................................... 76

3.2 Rationale for Full-Scale RCS Frame Test ....................................................... 78
3.3 Design of Full-Scale Test Frame ..................................................................... 78
3.3.1 Plan and Layout of Test Frame.................................................................. 79
3.3.2 General Design Information...................................................................... 80
3.3.3 Strong-Column Weak-Beam...................................................................... 81
3.3.4 Measured Material Strengths..................................................................... 83
3.3.5 Beam and Column Splices ........................................................................ 83
ix
3.3.6 Shear Studs in Hinge Region..................................................................... 85
3.3.7 Roof Joints................................................................................................. 86
3.4 Description of Test .......................................................................................... 86
3.4.1 Test Setup .................................................................................................. 86
3.4.2 Loading Protocol ....................................................................................... 87
3.5 Construction of Test Frame ............................................................................. 89
3.6 Test Results...................................................................................................... 90
3.6.1 Global Results ........................................................................................... 91
3.6.2 Description of Damage.............................................................................. 94
3.6.2.1 EQ#1 50% in 50 year ....................................................................... 95
3.6.2.2 EQ#2 10% in 50 year ....................................................................... 96
3.6.2.3 EQ#3 2% in 50 year ......................................................................... 98
3.6.2.4 EQ#4 10% in 50 year ..................................................................... 100
3.6.2.5 Final Pushover................................................................................... 100
3.6.3 General Observations .............................................................................. 101
3.7 Conclusions ................................................................................................... 103
3.7.1 General Seismic Performance of RCS Frame ......................................... 103
3.7.2 Frame Transient Drift .............................................................................. 104
3.7.3 Frame Residual Drifts ............................................................................. 106
3.7.4 Frame Repair ........................................................................................... 107
3.7.5 Precast System Performance ................................................................... 109
3.7.5.1 Beam Splices ..................................................................................... 109
3.7.5.2 Column Splices ................................................................................. 110
3.7.6 Frame Overstrength................................................................................. 111
3.7.7 Behavior of Composite Beams................................................................ 111
3.7.7.1 Shear Stud Design ............................................................................. 113
3.7.8 Behavior of RC Columns ........................................................................ 113
3.7.9 Strong-Column Weak-Beam Criterion .................................................... 114
3.7.10 Behavior of Composite Joints ................................................................. 116
3.7.11 Differences between Subassembly and Frame Tests............................... 117
Chapter 4: Analytical Modeling and Validation .................................................... 162
4.1 Introduction ................................................................................................... 162

4.2 OpenSees Component Models ...................................................................... 162
4.2.1 Material Models ...................................................................................... 162
x
4.2.2 Flexibility-Based Fiber Beam-Column Elements ................................... 164
4.2.2.1 Reinforced Concrete Columns .......................................................... 165
4.2.2.2 Composite Steel Beams..................................................................... 172
4.2.2.3 Convergence issues ........................................................................... 176
4.2.3 Composite Joints Elements ..................................................................... 178
4.3 Test Frame Validation Study ......................................................................... 180
4.3.1 Errors within the Pseudo-Dynamic Testing Method ............................... 182
4.3.2 Comparison to Experimental Global Response ...................................... 183
4.3.2.1 Comparison of First and Second Design Level Event ...................... 186
4.3.3 Comparison to Experimental Local Response ........................................ 187
4.3.4 Residual Drifts......................................................................................... 190
4.3.5 Time Evolution versus Predetermined Analysis...................................... 191
4.3.6 Comments on Analytical Models ............................................................ 192
4.4 Damage Indices ............................................................................................. 195
4.4.1 Damage Model ........................................................................................ 195
4.4.2 Plastic Rotations...................................................................................... 198
4.4.3 Overview of Damage after Each Event................................................... 198
4.4.4 Evolution of Damage Indices and Maximum Plastic Rotations ............. 200
4.4.4.1 Columns ............................................................................................ 200
4.4.4.2 Beams ................................................................................................ 204
4.4.4.3 General Comments ............................................................................ 206
4.5 Summary of Recommendations .................................................................... 207
4.5.1 RC Column Summary ............................................................................. 207
4.5.2 Composite Beam Summary..................................................................... 208
4.5.3 Composite Joint Summary ...................................................................... 209
4.6 Conclusions ................................................................................................... 210
4.6.1 Future Work............................................................................................. 211
Chapter 5: Applications ........................................................................................... 272
5.1 Introduction ................................................................................................... 272

5.2 Case Study Buildings .................................................................................... 272
5.2.1 Beam Design ........................................................................................... 274
5.2.2 Column Design........................................................................................ 275
5.2.3 Composite Joint Design .......................................................................... 276
5.2.4 Drift Limitations...................................................................................... 277
xi
5.3 Nonlinear Dynamic Time History Analyses.................................................. 277
5.3.1 Ground Motion Scaling Techniques........................................................ 278
5.3.2 Limitations of Analytical Models............................................................ 279
5.3.3 Ground Motion Selection ........................................................................ 280
5.3.4 Weighted 20-story Scaling Technique..................................................... 282
5.4 3-Story Perimeter Frame Results .................................................................. 284
5.4.1 Interpreting the Test Frame Results......................................................... 284
5.4.2 Modeling Variations for Realistic Building Case Study.......................... 288
5.4.2.1 Static Pushover Response.................................................................. 290
5.4.2.2 Derivation of Damping and Effects on Maximum Response............ 290
5.4.2.3 Global Response................................................................................ 293
5.4.2.4 Member Plastic Rotations ................................................................. 294
5.4.2.5 Damage Indices ................................................................................. 296
5.5 6-Story Perimeter Frame Results .................................................................. 299
5.5.1 Static Pushover Response........................................................................ 300
5.5.2 Global Response...................................................................................... 300
5.5.3 Member Plastic Rotations ....................................................................... 301
5.5.4 Damage Indices ....................................................................................... 302
5.6 20-Story Perimeter Frame Results ................................................................ 303
5.6.1 Static Pushover Response........................................................................ 304
5.6.2 Global Response...................................................................................... 304
5.6.3 Member Plastic Rotations ....................................................................... 305
5.6.4 Damage Indices ....................................................................................... 307
5.7 Conclusions ................................................................................................... 308
5.7.1 Seismic Design ........................................................................................ 308
5.7.2 General Seismic Performance ................................................................. 309
5.7.3 Drift Criterion.......................................................................................... 309
5.7.4 Composite Joint Performance ................................................................. 310
5.7.6 Damage Distribution ............................................................................... 311
Chapter 6: Conclusions ............................................................................................ 382
6.1 Summary ....................................................................................................... 382

6.2 Major Findings and Conclusions................................................................... 385
6.2.1 Seismic Performance of Composite RCS Frames................................... 385
xii
6.2.2 Performance of Precast Splices ............................................................... 386
6.2.3 Structural Period Elongation ................................................................... 386
6.2.4 SCWB...................................................................................................... 386
6.2.5 Top Floor Joints....................................................................................... 387
6.2.6 IBC 2003/ASCE 7-2002 Drift Criterion ................................................. 387
6.2.7 Full-scale System versus Subassembly Test Behavior............................ 387
6.2.8 Validity of Fiber Beam-Column Models ................................................. 388
6.2.9 Mehanny Damage Index ......................................................................... 389
6.2.10 Perspective on the Performance of the Test Frame ................................. 389
6.3 Design and Analytical Modeling Recommendations .................................... 390
6.3.2 Bolted Beam Splice Design..................................................................... 391
6.3.3 Column Grouted Splice Design............................................................... 391
6.3.4 Composite beams .................................................................................... 392
6.3.5 Updated Joint Guidelines ........................................................................ 392
6.3.6 Bond-Slip in RC Columns....................................................................... 393
6.3.7 Analytical Modeling of Composite RCS Frames.................................... 393
6.4 Future Work................................................................................................... 393
6.4.1 Calibration of Deteriorating Models ....................................................... 394
6.4.2 Investigation of Subassembly Boundary Conditions .............................. 394
6.4.3 Alternative Energy-Based Damage Models ............................................ 395
Bibliography ................................................................................................................. 396
xiii
List of Tables
2.1 Joint width comparisons between ASCE and Updated Guidelines............... 58

2.2 Summary of joint details and dimensions for those in the panel shear
joint group. (Units: kN, mm)......................................................................... 59
2.3 Summary of joint details and dimensions for those in the vertical bearing
joint group. (Units: kN, mm)......................................................................... 60
2.4 Results from Update Guidelines for the joint group failing in panel shear.
(Units: kN, mm) ............................................................................................ 61
2.5 Results from Update Guidelines for the joint group failing in vertical
bearing. (Units: kN, mm) .............................................................................. 62
2.6 Summary of values required to compute phi-factor using the beta-
reliability index.............................................................................................. 62
3.1 Rationale for large-scale test ......................................................................... 120

3.2 Design loads summary .................................................................................. 121
3.3 Test frame member properties....................................................................... 121
3.4 Measured strengths of steel tension coupons ................................................ 122
3.5 Measured crushing strength of concrete cylinders ........................................ 122
3.6 Summary of maximum and minimum drifts during each pseudo-dynamic
event (both absolute drift and with the residual removed)............................ 122
4.1 Summary of RC column specimens used in calibration study...................... 214

4.2 Composite beam validation study ................................................................. 215
4.3 Material properties for composite beam test specimens ............................... 217
4.4 Plastic rotation capacity of test frame components ....................................... 218
4.5 Correlation between the Mehanny damage index and the expected
damage in the component.............................................................................. 218
4.6 Comparison of the simulated to measured maximum plastic rotation in
1st floor interior column................................................................................. 218
1st floor exterior column ................................................................................ 219
xiv
2nd floor interior column................................................................................ 219
2nd floor exterior column ............................................................................... 219
1st floor beam................................................................................................. 220
2nd floor beam................................................................................................ 220
4.12 Summary of OpenSees input parameters for definitions of Concrete02
materials (uniaxialMaterial Concrete02)....................................................... 220
4.13 Summary of OpenSees input parameters for definitions of Steel02
materials (uniaxialMaterial Steel02) ............................................................. 221
4.14 Summary of OpenSees input parameters for definitions of Hysteretic
materials (uniaxialMaterial Hysteretic)......................................................... 221
5.1 Summary of design values for each of the case study buildings................... 313
5.2 Member design schedule of 6-story case study building .............................. 313
5.3 Member design schedule of 20-story case study building ............................ 314
5.4 Summary of column and beam strengths with the corresponding SCWB
ratios for the 3-story perimeter frame. (units: kN,mm)................................. 315
ratios for the 6-tory perimeter frame (frame line 1 and 7). (units:
kN,mm).......................................................................................................... 315
ratios for the 20-story perimeter frame (frame line A and F). (units:
kN,mm).......................................................................................................... 316
5.7 Strength of composite joints and the strong-joint weak-beam ratios for
the 3-story case study frame. (units: kN,mm) ............................................... 316
the 6-story case study frame. (units: kN,mm) ............................................... 317
the 20-story case study frame. (units: kN,mm) ............................................. 317
5.10 Modal properties of 20-story frame and corresponding weights for
record scaling ................................................................................................ 318
5.11 Comparison of IDRMAX of two OpenSees models with test frame ............... 318
5.12 Measured strengths of steel tension coupons ................................................ 318
xv
5.13 Measured crushing strength of concrete cylinders ........................................ 319
5.14 Plastic rotation capacity of 3-story case study frame components................ 319
5.15 Correlation between the Mehanny damage index and the expected
damage in the component.............................................................................. 319
5.16 Probability of beam hinges being in a specific damage state given a
10/50 and 2/50 hazard level........................................................................... 320
5.17 Plastic rotation capacity of 6-story case study frame components................ 320
5.18 Plastic rotation capacity of 20-story case study frame components.............. 321
xvi
List of Figures
1.1 Photo of composite RCS frames depicting the steel beam running
continuous through the reinforced concrete column ..................................... 8
2.1 Typical construction sequence of composite RCS moment frames............... 63

2.2 RCS precast construction utilizing beam-column modules and column
and beam spliced connection ......................................................................... 63
2.3 Example of precast RCS construction in Japan (Shimizu Corporation)........ 64
2.4 Example of cast-in-place RCS system that replaces steel erection
columns with stiffened reinforcing bar cages. (Shimizu Corporation).......... 64
2.5 Connection between steel beam and reinforced concrete column................. 65
2.6 Schematic diagrams of RCS joint details tested ............................................ 65
2.7 Typical failure modes in RCS beam-column joints (Kanno et al. 2000)....... 66
2.8 Typical hysteretic response of a RCS beam-column test failing in joint
bearing failure (Kanno 1993)......................................................................... 66
2.9 Typical hysteretic response of RCS beam-column test failing in joint
shear (Kanno 1993)........................................................................................ 67
2.10 Typical hysteretic response of RCS beam-column test failing in beam
hinging (Kanno 1993).................................................................................... 67
2.11 Typical configuration of reinforcement bars in RC columns ........................ 68
2.12 (a) Illustration showing the separation of int./ext. columns for the SCWB
criterion and (b) the appropriate load factors with respect to the P-M
column curve in order to obtain the lower flexural strength ......................... 68
2.13 SEOAC Blue Book strong-column weak-beam provisions........................... 68
2.14 Details of grouted splice connections for precast RC columns ..................... 69
2.15 Differences between the definitions of effective slab width considering
lateral versus gravity loading......................................................................... 69
2.16 Schematic of a typical bolted flange plate beam splice connection .............. 70
2.17 Assumed moment diagram for composite beam ........................................... 70
2.18 Assumed hinge length and location of beam splice....................................... 71
2.19 Cross-section of the beam splice plates and the concrete slab depicting
the force couple between the lower plates and slab....................................... 71
2.20 Plot of the location of the plastic neutral axis versus the ratio of
xvii
composite to steal beam strength, 76 mm deck and slab. (1 in. =
25.4mm)......................................................................................................... 71
2.21 Normalized load versus joint distortion response (Parra-Montesinos et
al., 2001) ........................................................................................................ 72
2.22 Joint detail showing the band plate, cover plate, and the transverse beam.... 72
2.23 Definition of outer strut width ....................................................................... 73
2.24 Joint Example ................................................................................................ 74
2.25 Predicted versus measured joint shear strength for both ASCE and
Updated model............................................................................................... 74
2.26 Predicted versus measured joint bearing strength for both ASCE and
Updated model............................................................................................... 75
2.27 Ratio of the composite to the bare steel beam strength for typical W-
section beams (1in = 25.4mm, 1ksi = 6.89MPa) ........................................... 75
3.1 Plan View of Building.................................................................................... 123

3.2 Plan and elevation views of full-scale composite test frame ......................... 123
3.3 Joint detail showing the transverse beam and placement of ties ................... 124
3.4 Mc/Mg ratios at each joint assuming a) steel beams (nominal), b)
composite beams (nominal) and (c) composite beams and RC columns
with measured material properties................................................................. 124
3.5 Schematic of a typical bolted flange plate beam splice connection .............. 125
3.6 RC column cantilever tests with the grouted splice located (a) 1-meter up
the column height and (b) flush at the column-footing interface .................. 126
3.7 Response of RC column subassembly test with precast splice at (a) 1-
meter above the footing and (b) flush at the column-footing interface.
(Tsai 2002) ..................................................................................................... 127
3.8 Top joint option #1. Section AA in Fig. 3.9................................................... 128
3.9 Section A-A from Fig. 3.8.............................................................................. 128
3.10 Reinforcement cap plate ................................................................................ 128
3.11 Plate to reinforcing bar detail ........................................................................ 129
3.12 Schematic of load path between actuators, loading beams, and test frame ... 129
3.13 Final records scaled at T1 = 1sec to appropriate Taiwanese hazard levels .... 130
3.14 Construction photos of (a) a typical pre-cast beam-column module and
(b) the completion of the first floor ............................................................... 130
3.15 Roof displacement versus time for the 50/50 ................................................ 131
xviii
3.16 Roof displacement versus time for the 10/50-1 event ................................... 131
3.17 Roof displacement versus time for the 2/50 event......................................... 132
3.18 Roof displacement versus time for the final 10/50 event .............................. 132
3.19 Maximum interstory drift ratios for each floor during each pseudo-
dynamic loading event................................................................................... 133
3.20 Maximum story shear for each floor during each pseudo-dynamic
loading event.................................................................................................. 133
loading event.................................................................................................. 134
loading event.................................................................................................. 134
loading event in dotted line) .......................................................................... 135
3.24 Hysteretic response of a 1st floor interior column base for the 50/50 event .. 135
st
3.25 Hysteretic response of a 1 floor interior column base for the 10/50-1a
event............................................................................................................... 136
3.26 Hysteretic response of a 1st floor interior column base for the 10/50-1b
event............................................................................................................... 136
3.27 Hysteretic response of a 1st floor interior column base for the 2/50 event .... 137
3.28 Hysteretic response of a 1st floor exterior beam hinge for the 50/50 event ... 137
3.29 Hysteretic response of a 1st floor exterior beam hinge for the 10/50-1a
event............................................................................................................... 138
3.30 Hysteretic response of a 1st floor exterior beam hinge for the 10/50-1b
event............................................................................................................... 138
3.31 Hysteretic response of a 1st floor exterior beam hinge for the 2/50 event.
(Large rotations are due to measurement errors caused by severe local
buckling) ........................................................................................................ 139
3.32 Hysteretic response of a 1st floor beam for the 10/50-2 event ....................... 139
3.33 Hysteretic response of a 1st floor interior beam hinge for the 2/50 event...... 140
3.34 Total and outer panel hysteretic response of a 1st floor interior joint for
the 50/50 event............................................................................................... 140
the 10/50-1a event.......................................................................................... 141
the 10/50-1b ................................................................................................... 141
xix
the 10/50-1b event ......................................................................................... 142
the 10/50-2 event ........................................................................................... 142
3.39 Hysteretic response of the 2nd floor upper interior column hinge after the
10/50-1a event ............................................................................................... 143
10/50-1b event ............................................................................................... 143
2/50 event....................................................................................................... 144
3.42 Damage in 1st floor interior column base after the 50/50 event .................... 145
3.43 Damage in 1st floor interior column base after the 10/50-1b event ............... 145
3.44 Damage in 1st floor interior column base after the 2/50 event ...................... 146
st
3.45 Damage in 1 floor interior column base after the 10/50-2 event ................. 146
3.46 Yielding in 1st floor beam after the 50/50 event ............................................ 147
3.47 Yielding and local buckling in 1st floor beam (1B1S) after the 10/50-1b
event............................................................................................................... 147
3.48 Yielding and local buckling in 1st floor beam (1B1S, exterior beam
hinge) after the 2/50 event ............................................................................. 148
hinge) after the 10/50-2 event........................................................................ 148
3.50 Yielding and local buckling in 1st floor beam (1B3N) after the 10/50-1b
event............................................................................................................... 149
3.51 Yielding in 1st floor beam (1B1N, interior beam hinge) after the 2/50
event............................................................................................................... 149
3.52 Yielding in 1st floor beam (1B1N, interior beam hinge) after the 10/50-2
event............................................................................................................... 150
3.53 1st floor splice plate after the 10/50-1b event ................................................ 150
3.54 Joint 1J3 after the 50/50 Chi-Chi event ......................................................... 151
3.55 Joint 1J3 after the 10/50-1b Loma Prieta event ............................................. 151
3.56 Joint 1J3 after the 2/50 Chi-Chi event ........................................................... 152
3.57 Joint 1J3 after the10/50-2 Loma Prieta event ................................................ 152
3.58 Damage in upper hinge of 2nd floor interior column hinge after the 10/50-
1b event.......................................................................................................... 153
3.59 Damage in upper hinge of 2nd floor interior column hinge after the 2/50
event............................................................................................................... 153
3.60 Damage in upper hinge of 2nd floor interior column hinge after the 10/50-
xx
2 event............................................................................................................ 154
3.61 Frame at its maximum drift state during static push. IDR in 1st floor:10%,
2nd: 8%, and 3rd: 3%....................................................................................... 154
3.62 Damage in 1st floor exterior column base after the final pushover event ...... 155
3.63 Damage in 1st floor interior column base after the final pushover event....... 155
3.64 Damage in upper hinge of 2nd floor exterior column hinge after the final
pushover event ............................................................................................... 156
3.65 Damage in upper hinge of 2nd floor exterior column hinge after the final
pushover event ............................................................................................... 156
3.66 Slab above exterior beam hinge, 1B1S, after the final static push ................ 157
3.67 Net section rupture of lower flange plates in 1st floor beam splice ............... 157
hinge) after the final pushover event ............................................................. 158
3.69 Yielding and local buckling in 1st floor beam (1B3N, exterior beam hinge
after the final pushover event ........................................................................ 158
3.70 Spectral acceleration versus displacement for the Loma Prieta event........... 159
3.71 Spectral acceleration versus displacement for the Chi-Chi event ................. 159
3.72 Spectral acceleration graphs of final records with highlighted spectral
values at the elongated periods ...................................................................... 160
3.73 RC column base hinges after loose concrete had been chipped away........... 161
3.74 Boundary condition in subassembly tests causes slab to pull away from
beam............................................................................................................... 161
4.1 Hysteretic response for the OpenSees Steel02 material model ..................... 222
4.2 Hysteretic response for the OpenSees Concrete02 material model............... 222
4.3 Backbone and cyclic response of Hysteretic model in OpenSees ................. 223
4.4 Typical tensile softening response of reinforced concrete............................. 223
4.5 Representation of deformations from bond slip and yield penetration.......... 224
4.6 Idealized backbone response of reinforcing bar pull out. (Fillipou et al.
1983) .............................................................................................................. 224
4.7 RC Column calibration against subassembly test: Tsai 2002-FFH08, base
springs included ............................................................................................. 225
4.8 RC Column calibration against subassembly test: Tsai 2002-FFH08, no
base ................................................................................................................ 225
4.9 RC Column calibration against subassembly test: Tsai 2002-FFL08,
xxi
grouted splice within hinge............................................................................ 226
4.10 RC Column calibration against subassembly test: Tsai 2002-FRL08,
grouted splice within hinge zone (OS model with coupler influence
shown as backbone)....................................................................................... 226
4.11 RC Column calibration against subassembly test: Tsai 2002-FRL60,
grouted splice within hinge zone, high axial load ......................................... 227
4.12 RC Column calibration against subassembly test: Tanaka et al. (1990)
test #4............................................................................................................. 227
test #3............................................................................................................. 228
test #2............................................................................................................. 229
4.15 Schematics of the cantilever and double-ended test setup with respect to
bond slip and OpenSees modeling................................................................. 229
4.16 Definition of effective slab width showing both AISC-LRFD and the
column width ................................................................................................. 230
4.17 Composite beam subassembly dimensions and cross-section details............ 231
4.18 Composite beam calibration against subassembly test: Uang (1985) test ..... 231
4.19 Composite beam calibration against subassembly test: Bursi and
Ballerini (1996) test ....................................................................................... 232
4.20 Composite beam calibration against subassembly test: Tagawa (1989)
test.................................................................................................................. 232
4.21 Composite beam calibration against subassembly test: Tagawa (1989)
test.................................................................................................................. 233
4.22 Composite beam calibration against subassembly test: Lee (1987) test........ 233
4.23 Composite beam calibration against subassembly test: Cheng (2002),
specimen INUCS- East Beam........................................................................ 234
specimen INUCS- West Beam....................................................................... 234
specimen ICLCS- East Beam ........................................................................ 235
specimen ICLCS- West Beam ....................................................................... 235
specimen ICLPS- East Beam......................................................................... 236
xxii
specimen ICLPS- West Beam........................................................................ 236
4.29 Schematic of OpenSees joint element used in this study .............................. 237
4.30 Panel shear hysteretic model (backbone and cyclic model) .......................... 237
4.31 Vertical bearing hysteretic model (backbone and cyclic model) ................... 238
4.32 Calibration results for joint specimen OJB1-0, which primarily fails in
vertical bearing .............................................................................................. 238
4.36 Calibration results for joint specimen OJS1-1, which primarily fails in
panel shear ..................................................................................................... 240
panel shear ..................................................................................................... 241
panel shear ..................................................................................................... 241
panel shear ..................................................................................................... 242
panel shear ..................................................................................................... 242
panel shear ..................................................................................................... 243
panel shear ..................................................................................................... 243
4.43 Calibration results for joint specimen HJS1-0, which primarily fails in
panel shear ..................................................................................................... 244
4.44 Calibration results for joint specimen HJS2-0, which primarily fails in
panel shear ..................................................................................................... 244
4.45 Contributions from panel shear and vertical bearing spring for joint
specimen OJS3-0, OJS4-1, and OJS5-0 ........................................................ 245
4.46 Schematic of the analytical model of the test frame...................................... 246
4.47 Ramp and hold phases of pseudo-dynamic testing and the concept of
force relaxation .............................................................................................. 246
4.48 OpenSees versus test frame response for 50/50 event: (a) roof
xxiii
displacement, (b) base shear, (c) peak IDR, and (d) peak story shear........... 247
4.49 OpenSees versus test frame response for 10/50-1a event: (a) roof
4.50 OpenSees versus test frame response for 10/50-1b event: (a) roof
4.51 OpenSees versus test frame response for 2/50 event: (a) roof
4.52 OpenSees versus test frame response for 10/50-2 event: (a) roof
4.53 Contours of the power spectral density for the frequency of the analytical
response throughout the time history (10/50-1) using a 10-second sliding
window .......................................................................................................... 252
4.54 Contours of the power spectral density for the frequency of the measured
window .......................................................................................................... 252
4.55 Contours of the power spectral density for the frequency of the analytical
window .......................................................................................................... 253
4.56 Contours of the power spectral density for the frequency of the measured
window .......................................................................................................... 253
4.57 Plot of the predominate period of a sliding 10-second window over the
displacement time history of the first 10/50 event......................................... 254
displacement time history of the 2/50 event .................................................. 254
displacement time history of the second 10/50 event .................................... 255
4.60 Analytical versus experimental comparison of the roof drift versus base
shear during the final static pushover of the test frame ................................. 255
4.61 Comparison of the roof displacement response measured from the test
frame for the first (1b) and second 10/50 (2) event ....................................... 256
4.62 Comparison of the roof displacement response predicted by OpenSees
for the first (1b) and second 10/50 (2) event ................................................. 256
4.63 Analytical versus measured response of beams during 50/50 event ............. 257
4.64 Analytical versus measured response of beams during 10/50-1a event ........ 257
4.65 Analytical versus 258measured response of beams during 10/50-1b event .. 257
xxiv
4.66 Analytical versus measured response of beams during 2/50 event ............... 258
4.67 Analytical versus measured response of beams during 10/50-2 event .......... 258
4.68 Analytical versus measured response of columns during 50/50 event .......... 258
4.69 Analytical versus measured response of columns during 10/50-1b event..... 259
4.70 Analytical versus measured response of columns during 10/50-1b event..... 259
4.71 Analytical versus measured response of columns during 2/50 event ............ 259
4.72 Analytical versus measured response of columns during 10/50-2 event....... 260
4.73 Analytical versus measured response of joints during 50/50 event............... 260
4.74 Analytical versus measured response of joints during 10/50-1a event.......... 260
4.75 Analytical versus measured response of joints during 10/50-1b event ......... 261
4.76 Analytical versus measured response of joints during 2/50 event................. 261
4.77 Analytical versus measured response of joints during 50/50 event............... 261
4.78 Comparison of the residual displacements from the analytical (thin line)
and experimental (thick line) model after each pseudo-dynamic event ........ 262
4.80 Damage index values (>30%) after 50% in 50 year event ............................ 263
4.81 Damage index values (>30%) after 10% in 50 year 1a event........................ 264
4.82 Damage index values (>30%) after 10% in 50 year -1b event ...................... 264
4.83 Damage index values (>30%) after 2% in 50 year event .............................. 265
4.84 Damage index values (>30%) after final 10% in 50 year event .................... 265
4.85 (a-d) Photos of damage progression in the 1st floor int. column after each
main event. Evolution of damage indices using (e,f) OpenSees and (g)
measured data ................................................................................................ 266
4.86 (a-d) Photos of damage progression in the 1st floor exterior column after
each main event. Evolution of damage index using (e) OpenSees and (f)
measured data ................................................................................................ 267
4.87 (a-d) Photos of damage progression in the 2nd floor interior column after
each main event Evolution of damage index using (e) OpenSees and (f)
measured data ................................................................................................ 268
4.88 (a-d) Photos of damage progression in the 2nd floor exterior column after
each main event. Evolution of damage index using (e) OpenSees and
measured data ................................................................................................ 269
4.89 (a-d) Photos of the damage progression in the 1st floor beam after each
main event. Evolution of damage index using (e) OpenSees and (f)
measured data ................................................................................................ 270
xxv
4.90 (a-d) Photos of the damage progression in the 1st floor beam after each
main event. Evolution of damage index using (e) OpenSees and (f)
measured data ................................................................................................ 271
5.1 Typical floor plan of 6-story case study building .......................................... 322
5.2 Typical floor plan of 20-story case study building ........................................ 322
5.3 IBC 2003 design hazard spectra with the code-defined periods (1.2Ta) for
the 3, 6, and 20 story frames labeled on the curve ........................................ 323
5.4 Schematic of typical transverse reinforcement in RC columns for 6-story
perimeter frame.............................................................................................. 324
5.5 Schematic of typical transverse reinforcement in RC columns for 20-
story perimeter frame..................................................................................... 325
5.6 Typical plan view of the column section just below the beam and in the
beam-column joint. (section A-A and B-B in Figs. 5.7 and 5.8) ................... 326
5.7 Cross-section of typical beam-column joint depicting the location of the
joints ties........................................................................................................ 327
5.8 Cross-section of typical beam-column joint depicting the location of the
joints ties........................................................................................................ 327
5.9 (a)The traditional cloud approach of nonlinear time history analyses and
two alternative concepts for scaling ground motions using (b)
incremental scaling of single ground motions and (c) the stripe analysis
technique........................................................................................................ 328
5.10 Magnitude and distance to the rupture pairs for ground motion records
used in this study ........................................................................................... 328
5.11 Epsilon versus spectral acceleration at a period of 1 second for the 80
ground motions considered in this study ....................................................... 329
5.12 Epsilon versus spectral acceleration at a period of 1.5 seconds for the 80
5.13 Epsilon versus spectral acceleration at a period of 4 seconds for the 80
5.14 Response spectrum for selected ground motions at (a) 0.05g, (b) 0.09g, 330
(c) 0.2g, (d ) 0.3g, (e) 0.4g, (f ) 0.5g, and (g) 0.6g stripe hazard level for -
the 3-story building........................................................................................ 333
5.15 Response spectrum for selected ground motions at (a) 0.03g, (b) 0.06g, 334
(c) 0.1g, (d ) 0.2g, (e) 0.3g, and (f ) 0.4g stripe hazard level for the 6- -
xxvi
story building ................................................................................................. 336
5.16 Example of weighted average scaling technique........................................... 337
5.17 Response spectrum for selected ground motions at (a) 0.01g,(b) 0.02g, 337
(c) 0.05g, (d) 0.1g, (e) 0.18g, and (f) 0.2g stripe hazard level for the 20- -
story building ................................................................................................. 340
5.18 Plot of the measured and simulated maximum IDR from the first four
events of the pseudo-dynamic loading protocol ............................................ 340
5.19 Comparison of stripe analysis study and the measured and simulated drift
from the test frame......................................................................................... 341
5.20 Simulate IDA stripe response for selected columns (a,b), beams (c,d),
and joints (e,f) compared to the measured response from the frame test ...... 342
5.21 Static pushover curve for 3-story frame using IBC 2003 force distribution . 343
5.22 IDR profile of 3-story frame during the pushover at the design base shear
and selected roof drift ratios .......................................................................... 343
5.23 Comparison of the median response of 3-story RCS frame with zero
damping and 2% damping based on initial and last committed stiffness
matrix ............................................................................................................. 344
5.24 Relationship between damping ratio and frequency for the 3-story RCS
frame as defined by the Rayleigh equation.................................................... 344
5.25 Stripe analysis plot of maximum interstory drift versus hazard level for
the 3-story RCS frame ................................................................................... 345
5.26 Drift profile of 3-story frame at the time of maximum drift during each
event scaled to the common hazard level labeled in the x-axis.
(Sa=0.045g-0.5g)............................................................................................ 346
event scaled to the common hazard level labeled in the x-axis. (Sa=0.6g-
1.15g) ............................................................................................................. 347
5.28 Maximum drift at each floor of 3-story frame during each event in the
corresponding stripe level (bold lines: median, 16th, and 84th percentile)..... 348
5.30 Relationship between the maximum plastic rotation in the interior
columns and the scaled spectral acceleration ................................................ 350
5.31 Relationship between the maximum plastic rotation in the exterior
5.32 Relationship between the maximum plastic rotation in the beams and the
xxvii
scaled spectral acceleration............................................................................ 352
5.33 Relationship between the maximum positive (left column) and negative
(right column) plastic rotation in the beams and the scaled spectral
acceleration .................................................................................................... 353
5.34 Relationship between the maximum rotation for the interior joints and
the scaled spectral acceleration...................................................................... 354
5.35 Relationship between the maximum rotation for the exterior joints and
the scaled spectral acceleration...................................................................... 355
5.36 Summary of the median and standard deviation of plastic rotations of
3-story frame members at the 10%in50year level (Sa = 0.72g)..................... 356
3-story frame members at the 2%in50year level (Sa = 1.08g)....................... 356
5.38 Relationship between the final value of the damage index for the interior
5.39 Relationship between the final value of the damage index for the exterior
5.40 Relationship between the final value of the damage index for the beams
and the scaled spectral acceleration ............................................................... 359
5.41 Relationship between the final value of the damage index for the interior
joints and the scaled spectral acceleration ..................................................... 360
5.42 Relationship between the final value of the damage index for the exterior
joints and the scaled spectral acceleration ..................................................... 361
5.43 Summary of the median and standard deviation of damage indices of
frame members in 3-story frame at the 10%in50year level (Sa = 0.72g) ...... 362
frame members in 3-story frame at the 2%in50year level (Sa = 1.08g) ........ 362
5.46 Static pushover curve for 6-story frame using IBC 2003 force distribution . 363
5.47 IDR profile of 6-story frame during the pushover at the design base shear
and selected roof drift ratios .......................................................................... 364
the 6-story RCS frame ................................................................................... 364
0.4g) ............................................................................................................... 365
xxviii
Drift profile of 6-story frame at the time of maximum drift during each
5.50 event scaled to the common hazard level labeled in the x-axis. (Sa=0.48g-
0.8g) ............................................................................................................... 366
6-story frame members at the 2%in50year level (Sa = 0.72g)....................... 369
5.55 Summary of the median and standard deviation of damage indices of 6-
story frame members at the 10%in50year level (Sa = 0.48g) ........................ 370
5.56 Summary of the median and standard deviation of damage indices of 6-
story frame members at the 10%in50year level (Sa = 0.72g) ........................ 370
5.58 Static pushover curve for 20-story frame using IBC 2003 force
distribution..................................................................................................... 371
5.59 IDR profile of 20-story frame during the pushover at the design base
shear and selected roof drift ratios................................................................. 372
the 20-story RCS frame ................................................................................. 372
0.2g) ............................................................................................................... 373
0.3g) ............................................................................................................... 374
20-story frame members at the 10%in50year level (Sa = 0.18g)................... 377
xxix
5.69 Maximum interstory drift response for each case study building at the
10/50 hazard level.......................................................................................... 381
xxx
Chapter 1: Introduction
Composite and hybrid structures comprised of reinforced concrete and structural steel
members have gained popularity over the past thirty years, due in large part to the
efficiency that they provide in the design and construction of multi-story buildings.
These systems utilize structural steel and reinforced concrete components, wherein the
intrinsic advantages of each material are optimized in resisting the applied loads. These
composite design concepts also provide flexibility in construction, which can lead to
improved integration of construction trades and reduced construction time. Rapid
advancements in composite systems occurred in a number of cities (e.g., Dallas and
Houston, Texas; Atlanta, Georgia), during a period of rapid growth in the early 1980s.
To a large extent, the applications of these systems have been limited to regions of low-
to moderate-seismicity, in spite of a number of key advantages these systems can provide
for seismic design.
This research focuses on one type of composite moment frame system, which is
comprised of reinforced concrete (RC) columns and steel (S) beams and is referred to as
composite RCS moment frames. As shown in Fig. 1.1, this system permits the primary
steel beam to run continuous through the reinforced concrete column, thereby allowing
the beam to be spliced away from the location of maximum moments. Through past
studies on beam column subassembly experiments (Sheikh 1987, Deierlein 1988, Kanno
1993, Parra-Montesinos 2000, Liang 2004), this connection detail has been shown to
provide the necessary ductility and toughness required for use in seismic applications. In
addition, splicing the beam outside of the hinging zone avoids the fracture problems
encountered in conventional steel frames during the 1994 Northridge and 1995 Hanshin
earthquakes. Past analytical studies of system performance have demonstrated that the
inelastic dynamic response of these systems is comparable to that of an equivalent steel
moment resisting frame (Mehanny et al., 2000, Bugeja 1999, Noguchi 1998).
With the ultimate goal to facilitate greater acceptance and use of RCS systems for seismic
design, this research extends past studies on RCS systems and applies the accumulated
1
knowledge to (1) develop and validate improved seismic design provisions for RCS
frames, (2) assess and demonstrate the seismic performance of RCS frames through full-
scale frame testing and simulations of prototype building systems, and (3) develop and
validate modeling guidelines for nonlinear analysis and performance simulation. A focal
point of this research is the planning, design and testing of a full-scale composite RCS
moment frame that is pseudo-dynamically loaded to simulate earthquake motions. In
addition to the direct benefits of the test results, the frame and supporting subassembly
tests provide a rich data set for the validation of nonlinear analysis models. The
analytical portion of this study focuses on modeling implementations in the program
OpenSees (Open System for Earthquake Engineering Simulation), which has been
developed by the Pacific Earthquake Engineering Research (PEER) Center (McKenna et
al., 1999). Data from RCS subassembly tests and the full-scale frame test are used to
verify how well the nonlinear OpenSees models can simulate the earthquake-induced
response and damage to RCS building systems.
In addition to validating the seismic performance and design of RCS systems, a parallel
goal of this investigation is to provide data and interpret information that support the
development of performance based earthquake engineering (PBEE). PBEE is a recent
focus of research that has been sparked by the improved knowledge about earthquake
phenomena and the ability to realistically simulate the structural response characteristics.
This methodology provides a tool to inform owners and engineers regarding the trade-
offs of the up front cost of a structure versus the expected performance (i.e. structural and
nonstructural damage) during specific loading events and the costs associated with those
performance levels (Krawinkler and Miranda, 2004). PBEE approaches and enabling
technologies provide the means to accelerate the development and implementation of
new seismic resisting systems, such as provided by RCS moment frames. In this sense,
this investigation provides the opportunity to advance PBEE by contributing to the
development and calibration of tools for performance assessment and an application
(RCS frame design) that is ripe for greater adoption by earthquake engineers.
2
1.1 Motivation for this Study
The advantages of composite RCS systems have been well documented (Griffis 1986)
and are presented in detail in Chapter 2 of this report. Past research has shown that
composite frames can be designed with seismic deformation capacity and toughness at
least equal to traditional steel or reinforced concrete construction. Despite this research,
the use of composite RCS frames remains limited in the United States to low seismic
zones. This can be attributed to the hesitation of structural engineers, contractors, and
building code officials to accept this new system, the historical separation of
construction trades in the United States, and the lack of comprehensive earthquake
engineering design criteria in building codes and specifications.
During 1990s, considerable research on composite construction was conducted through a

cooperative US-Japan Cooperative Earthquake Engineering Research Program on
Composite and Hybrid Structures (Goel 2004, Deierlein and Noguchi 2004). During the
latter stages of this initiative it was recommended to conduct tests and complementary
analyses on full structural systems to validate the knowledge that had been gained
through the program. System testing would also provide a platform to study the 3-
dimensional, indeterminate aspects of complete structure behavior and allow the
investigation of many practical aspects of composite structures, such as economy and
constructability. The scope of this research investigation, which incorporates a full-scale
composite RCS frame test conducted through another international collaboration with
researchers from Taiwan, was devised to address this need. In addition to the planning,
design, analysis and execution of the frame test, this investigation synthesizes much of
the past research on RCS components and systems. An underlining goal is to promote
greater utilization of composite RCS frames in the United States through development of
seismic design guidelines, validation of seismic response, and modeling validation for
nonlinear analyses.
An integral element in the performance assessment of structures is the ability to

accurately model the time history response of a building to an earthquake excitation. The
3
full-scale testing program provides valuable data to validate the analytical models used to
simulate the seismic response of structural systems. This is something of great
importance to the current state of structural engineering; particularly as the push for
performance based earthquake engineering requires the simulation of building response
out to large interstory drifts and damage states.
1.2 Objectives
The main objectives of this work can be summarized in the following points:
1. Synthesize and interpret current design specifications for composite RCS moment
frames and investigate alternative design approaches based on emerging
performance-based design concepts and recommendations from related research
and development studies.
2. Contribute to the development of an updated strength model for composite RCS
beam-column joints. This will be accomplished by reinterpreting the original
tests used to create an earlier set of guidelines (ASCE 1994), examining more
recent subassembly tests that have studied a wider variety of joint details and
configurations, and evaluating the results from the subassembly and full-scale
testing program described below. This portion of the research is part of a broader
collaborative effort that is aimed toward updating the 1994 ASCE Guidelines for
Design of Joints between Steel Beams and Reinforced Concrete Columns.
3. In a collaborative project with researchers from the National Center for Research
on Earthquake Engineering (NCREE), develop and implement a testing program
to investigate the design, constructability, and seismic performance of a full-scale
composite RCS moment resisting frame.
4. Investigate differences between the response of beam-column subassembly and
full-scale system testing and evaluate how this affects the interpretations from
these tests.
5. Using the results and observations from both the full-scale test as well as other
subassembly tests, calibrate and validate analytical models for composite RCS
moment frames. These simulation models consist of (a) flexibility-based beam-
4
column elements that integrate fiber sections along the length of the member in
order to capture distributed plasticity and (b) a two-dimensional joint model with
dual springs in series to capture the deformation mechanisms of composite joints.
Modeling recommendations will be proposed to facilitate the accurate simulation
of the behavior of RC columns, composite steel beams, and composite joints.
6. Validate damage indices for use in conjunction with analytical models to correlate
the physical damage of a test specimen to the simulated nonlinear response of the
components.
7. Exercise the recommended design provisions and modeling guidelines in the
design and performance assessment of three case-study buildings under various
seismic hazard levels.
8. Using the information from both the full-scale test and results from analytical
models, validate the current design methods for RCS frames and propose
improved techniques where current methods are found to be insufficient or where
information is unavailable.
1.3 Scope and Organization of Report
In Chapter 2, a brief summary of the background on the rationale and development of

composite RCS systems is provided. Previous research on these systems is discussed,
including beam-column subassembly tests, small-scale frame tests, and trial design
studies of system performance. Building system guidelines are summarized from the
International Building Code (ICC 2003) and the ASCE-7 Standard Minimum Design
Loads for Buildings and other Structures (2002). Member design requirements for the
RC columns and steel (or composite) beams are evaluated from the AISC-LRFD
Specification (1999), the AISC Seismic Provisions for Structural Steel Buildings (2002),
and the ACI-318 code (2002). In addition to these standards and provisions, a review of
the latest research on alternative design recommendations and aspects of component
performance is also provided. A detailed review of the current standard for the design of
the composite RCS joints (ASCE 1994) is provided as well as the concurrent
development of a proposed update to these guidelines. This chapter is concluded with a
5
summary of a recommended methodology for the system and component design of
composite RCS frames.
In Chapter 3, the testing program for a full-scale 3-story composite RCS frame, which
was conducted at the National Center for Earthquake Engineering in Taipei, Taiwan, is
presented. An overview of the design of the test frame along with several of the key
design issues is outlined. The test setup, the pseudo-dynamic loading protocol, and the
final quasi-static pushover are reviewed. A brief overview of the construction process,
which was handled by an outside contractor to provide as much realism to the process as
possible, is also discussed. A description of the test is provided with a discussion on
global and local results, as well as corresponding photos of the damage incurred by the
frame. Conclusions and implications regarding design, construction, and performance
are discussed, as well as some of the observed differences between full-scale frame and
subassembly testing.
Chapter 4 summarizes the modeling recommendation for composite RCS frames and the
corresponding calibration and validation studies. Nonlinear analyses are conducted using
the OpenSees platform and the analytical results are compared to data from subassembly
tests and the full scale test frame. Component models for the RC columns, composite
steel beams, and composite joints are developed and calibrated against subassembly data.
Individual modeling recommendations are proposed for each of these components. These
recommendations are used to create an analytical model of the test frame and simulate the
response of the frame to input earthquake loading. The measured and simulated
responses are compared at the global and local level, and the similarities and differences
are discussed. The analytical models are extended using damage indices to determine the
predicted damage states from the analytical models and then compared to the physical
damage observed in the test frame. Conclusions on the validity and limitations of the
simulation and damage models are presented.
Chapter 5 applies the design recommendations from Chapter 2 and generates the design
for three case study buildings with heights of 3, 6, and 20 stories. These frames are then
6
modeled in OpenSees using the recommendations presented in Chapter 4 and the
structural performance is assessed at multiple seismic hazard levels, which represent
frequent to extremely rare earthquake loading events. This probabilistic study
investigates the response of these frames in terms of three engineering demand
parameters: (1) maximum interstory drift, (2) inelastic component rotations, and (3)
cumulative damage indices. These models are used to assess the performance as well as
investigate some of the design issues raised in other chapters of the report.
An overview and summary of the major findings and conclusions of the study, and ideas
for future research are discussed in Chapter 6.
7
Composite
Joint
Gravity
Steel
Beam
Beam
Bolted Beam
RC Column Splice
Figure 1.1 Photo of composite RCS frames depicting the steel beam running continuous
through the reinforced concrete column.
8
Chapter 2: RCS System Design
This chapter begins with a brief discussion of the motivation and development of
composite structures in the United States and Japan. A summary of past research on
composite RCS frames is also presented and will serve as the foundation for the current
study. What follows is a detailed review of the current knowledge for the design of
composite RCS frames in building code provisions, as well as alternate approaches
proposed from recent research. Current joint design guidelines (ASCE 1994) are
reviewed and compared to an updated model that has been developed and validated as
part of this research. A recommended design methodology is summarized in the
conclusion of this chapter.
This research applies and evaluates the provisions of the 2003 International Building
Code (ICC 2003) that are relevant to composite RCS frames. This code, either directly or
indirectly, references the ASCE-7 Standard Minimum Design Loads for Buildings and
other Structures (2002), the AISC-LRFD 1999, AISC Seismic Provisions for Structural
Steel Buildings (2002), and the ACI-318 (2002). The 1994 ASCE Joint Design
Guidelines constitute the current standard for composite joints. While there exists more
current versions of these codes and provisions during the time of publication of this thesis
(IBC 2005, ASCE-7 2005, AISC-LRFD 2005, and AISC Seismic 2005), the earlier codes
mentioned herein were the governing standards when this research was completed.
Where applicable, the newer versions of the code are mentioned when the design method
has had an appreciable change.
2.1 Background of RCS Composite Moment Frames
The RCS composite moment frame systems began to gain popularity in both the United
States and Japan in the late 1970s and early 1980s. In the US, this system came about
as an attractive modification of traditional steel moment frames for mid- to high-rise
buildings in relatively low seismic zone (e.g. Houston, Texas). Replacing the heavy
wide-flange columns of a typical steel moment frame with the more cost effective
9
reinforced concrete columns to resist the high axial compressive loads brings about an
economical advantage to the RCS system (Griffis 1992). The economic advantage of RC
columns over steel columns increases as buildings get taller and stiffness (drift) criteria
tend to control the design (Leon and Deierlein, 1995).
In Japan, RCS composite systems were developed as an alternative to low-rise reinforced

concrete moment frames in high seismic zones. Here the goal was to take advantage of
the long-span capabilities of steel beams to provide column-free spaces for the low-rise
office buildings and retail stores. When steel beams were incorporated with a composite
slab, the advantage over typical deep reinforced concrete beams became even more
obvious since the typical depth and weight of a floor could be reduced, resulting in cost
savings in the foundation design. In Japan, the tradition of mixing reinforced concrete
and steel members was already established with the common steel reinforced concrete
(SRC) construction, which is characterized by full concrete encasement of structural steel
frames (columns and beams). In general, composite systems are much more developed in
Japan due to the large amount construction companies that have a longer tradition of
mixing the trades of steel and reinforced concrete, something that is not as common in the
US.
Another important advantage of these composite systems is the ability to accommodate

innovative construction techniques that lower the overall cost and expedite the process of
erecting buildings. In the US, a typical high-rise construction sequence utilizes small
steel erection columns to advance steel framing several floors ahead of placing reinforced
concrete columns (Fig. 2.1). This staggered staging allows for the separation of the steel
and concrete trades (as well as other construction activities), which gives each of the
groups enough space to work independently of each other. For example, a typical
composite frame construction sequence may read as follows if we start from the upper
most floor and work our way down (Griffis 1992):
1. Upper stage: placement of the bare steel frame using small steel erection columns.
2. 2nd tier: Follow with any necessary welding (steel erection columns) and
placement of metal deck for composite slab.
10
3. 3rd tier: Follow with placement of shear studs and slab reinforcement.
4. 4th tier: Pouring of the concrete slab.
5. 5th tier: Placement of the reinforcing bar cage around steel erection columns.
6. 6th tier: Setting of the column formwork and casting of the concrete columns.
This sequence can be further accelerated by the use of innovative jump-form framing
systems. Optimization of this sequence requires close coordination and control of the
overall construction process.
An alternative precast construction method can also be implemented in the erection of

composite frames. In this scheme, the steel beam is cast integral with the RC column and
field spliced a short distance away from the column face. The columns are spliced
together with grouted sleeve couplers and the beams are spliced together with bolted
flange plates and shear tabs to generate a continuous system, as shown in Fig. 2.2 and
2.3. The size of these precast modules can be varied to span several stories and/or bays to
accommodate the optimum configuration for ease of construction. Variations to these
methods, such as utilizing the column reinforcing bar cage as the erection column have
been developed by construction companies in Japan (Fig. 2.4).
Perhaps one of the biggest advantages provided by RCS systems lies within the
composite beam-column joints, an example of which is shown in Fig. 2.5. In these
composite joints the steel beam runs continuous through the concrete column, thereby
eliminating the need to interrupt the beam at the column face. This type of detail avoids
welding or bolting of the beam at the location of maximum moment, which mitigates
some of the fracture problems encountered in conventional steel moment frames during
the 1994 Northridge and 1995 Hanshin earthquakes. The longitudinal reinforcement of
the column is also continuous through the joint and can be spliced away from the joint.
This beam-column joint detail is common to both the cast-in-place and precast forms of
construction.
11
2.2 Previous Research
RCS composite moment frames have been studied for over twenty years, with a primary
focus on understanding the behavior and design of the beam-column joints. In addition
to research on connections, there have also been system design and performance
assessment studies with the intent to evaluate and benchmark the behavior of these
innovative frames against more conventional moment frames. This section will review
some of the important past research that have helped develop the understanding of
composite joints and systems.
2.2.1 Beam-Column Composite Joints
The reinforced concrete column and the steel (or composite) beams are relatively
straightforward to design following provisions for members in conventional steel or
reinforced concrete construction. A primary challenge in the design of RCS frames lies
within the unique connection between the steel beam and the RC columns, and thus, the
primary focus of past research has been on these composite connections. During the
1980s, over 400 RCS connection subassemblies were tested in Japan, and 17 were tested
in the United States. Many of the early joints tested in Japan were of proprietary details
sponsored by Japanese construction companies with the primary goal being to validate
specific joint details. While the results of these tests are interesting, they are of limited
research value to quantify the internal force transfer mechanisms of the joint. The
contributions from the U.S. consisted of two series of subassembly tests that were
performed at the University of Texas Austin by Deierlein et al. (1989) and Sheikh et al.
(1989). Based on these two investigations, Deierlein (et. al 1989) proposed design
equations to quantify the strength and stiffness of composite connections, which later
came to form the basis of the 1994 ASCE Guidelines for Design of Joints between Steel
Beams and Reinforced Concrete Columns (ASCE 1994).
During the 1990s, the United States-Japan Cooperative Earthquake Engineering

Research Program on Composite and Hybrid Structures began and provided the impetus
to test about 56 more connection subassemblies (33 in Japan and 23 in the U.S.) with the
12
specific intent of investigating and quantifying the internal joint transfer mechanisms.
Deierlein and Noguchi (2000, 2004) summarize the subassemblies tested under the US-
Japan program. Apart from the formal US-Japan program, Kanno and Deierlein (1993,
1997) tested 19 RCS beam column joint subassemblies and 50 small concrete blocks
loaded with steel bearing plates to investigate the bearing strength in RCS joints above
and below the steel beam. Both the tests covered by the US-Japan program and those by
Kanno (1993, 1997) provided a significant amount of new data to fill knowledge gaps for
various connection configurations and force transfer mechanisms.
Summarized in Fig. 2.6 are several examples of the types of RCS joint details that have
been tested. There are two distinctive types of RCS joints; one is the through-beam
detail (details 1-7 in Fig.2.6) where the steel beam runs continuous through the concrete
column, the advantages of which have been already discussed. The second is the
through-column detail (details 8-11 in Fig. 2.6) where the beam flanges are interrupted
at the column face to accommodate a variety of reinforcing bar arrangements and to
facilitate the placement of concrete in the joint. Detail 12 in Fig. 2.6 is an example of a
hybrid detail, where the SRC concept of encasing the steel beam in concrete is applied for
a short segment of the beam framing into the joint. While the through-beam type detail is
preferred in the United States, both types are used in Japan particularly among the
proprietary details developed by the construction companies. Other than the through-
beam versus through-column difference, the joint details shown in Fig. 2.7 differ by the
wide variety of stiffeners, cover plates, and bearing plates that are implemented to ensure
adequate force transfer between the steel beam and the concrete column.
Overall, the subassembly tests have shown that when detailed to ensure proper force
transfer between the beam and column, RCS composite joints are capable of providing a
reliable amount of strength and ductility necessary for seismic design. In fact, in order to
investigate joint failure in these subassembly tests, the dimensions of the steel beam had
to be significantly altered from regular practice to ensure that the beam possessed enough
moment strength to avoid beam hinging (large flange thickness), while limiting the joint
panel shear strength (small web thickness). For standard rolled W-shape beams, fairly
13
simple joint details can be provided to ensure that the joints have enough strength to force
plastic hinging to occur in the steel or composite beams.
The beam-column subassembly tests that have been specifically designed to cause failure
in the composite connections generally exhibit either one or a combination of two
primary failure modes (Fig. 2.8). Vertical bearing failure is characterized by rigid body
rotation of the beam through the joint resulting from concrete crushing both above and
below the joint. This localized crushing of the concrete causes gaps to open up between
the steel beam and the concrete column. The deformations caused by this vertical
bearing failure leads to more of a pinched hysteretic response, as shown in Fig. 2.8. The
second type of joint failure is panel shear failure, which is very similar to the
corresponding behavior in joints within conventional all steel or RC moment frames. The
difference in RCS joint shear failure is that the benefit of both steel web yielding and the
development of concrete struts exist within the joint region. Although the deformation
response for this type of failure does contain some slight pinching, it is more associated
with larger, energy dissipating hysteretic loops, as shown in Fig. 2.9. Again, keep in
mind that these specimens were intentionally designed to fail in the joints, whereas in
traditional seismic design it is recommended that the majority of inelastic action occur
within the beam. In a properly detailed RCS joint, it is fairly easy to ensure beam
hinging and the associated response shown in Fig. 2.10. Nevertheless, even if failure
were to occur in the joint, the hysteretic response has proven to be quite stable.
2.2.2 Small-Scale Frame Tests
The US-Japan program included two reduced-scale RCS moment frames one at the
Osaka Institute of Technology (Baba and Nishirmura 2000) and the second at Chiba
University (Noguchi and Uchida 2004). Both are about 1/3-scale two-bay two-story RCS
frames with through-beam type connections with differences only in the joint details (one
had cover plates and band plates while the other had face bearing plates and band plates).
The frame was designed such that the plastic strength of the beams was nearly equal to
the ultimate shear strength of the joints, so as to provide information on the interaction
14
between frame and connection response. Both test specimens were subjected to reverse
cyclic loading and withstood story drift ratios in excess of 5% without significant
strength or stiffness degradation, thus confirming the reliable seismic behavior of RCS
framing systems.
2.2.3 Trial Design Studies of System Performance
Several groups of researchers have developed trial designs of RCS frames based on a
common theme building devised for the US-Japan program (Mehanny and Deierlein
2000, Bugeja 1999, Noguchi 1998). The goal of these studies were to first apply the
proposed seismic design provisions for RCS systems and then evaluate the seismic
performance of resulting designs using nonlinear analyses and advanced performance
assessment techniques. Traditional steel frames were also investigated in these studies to
benchmark the performance of conventional frames compared to the composite RCS
frames. Using a common floor plan, the building heights varied as well as the
implementation of perimeter versus space frame systems. These design studies have
shown that the steel beam sizes tend to be similar for the RCS and steel system and that
the main differences lie in the column and connection designs. Given the additional
stiffness provided by the RC columns, the RCS frames tended to be controlled more by
the minimum strength requirements whereas the steel frames were restricted by lateral
drift limitations. In general, these investigations have shown that the inelastic dynamic
response of the RCS frames is similar to comparably designed steel moment frames.
2.3 System and Member Design Guidelines
In the United States, the seismic design criteria for composite RCS frames are distributed
over several codes and standards (Deierlein 2000). The first formal seismic design
requirements for composite steel-concrete structure came with the 1994 edition of the
NEHRP (National Earthquake Hazards Reduction Program) Recommended Provision for
the Development of Seismic Regulations for New Buildings (BSSC 1995). The 1994
edition included a new chapter that contained a comprehensive set of design criteria for
composite systems, members, and connections as well as defined the seismic design
15
coefficients (R and Cd) necessary for calculating the design earthquake loads and
deformations. In these provisions, the Building Seismic Safety Council (BSSC) defined a
list of seven composite systems that have the likeliest practical applications; included on
this list was composite special moment frames. The strength and detailing requirements
for the composite members were largely based on the already existing AISC and ACI
standards, but also included clarifications of how they should be applied and
supplementary information when certain issues were not covered. These provisions are
important not only because they were the first of their kind for composite structures, but
they also set the foundation for subsequent codified publications.
The current set of documents that control the design of composite RCS structures begins
at the International Building Code (IBC) (ICC 2003) and the ASCE 7 Standard Minimum
Design Loads for Buildings and other Structures (ASCE 2002), both of which specify the
general seismic loading and design requirements, much like that of the 1994 NEHRP
Provisions. ASCE-7 (2002) adopts Part II of the AISC Seismic Design Provisions (2002)
for specific detailing requirements for RCS frames (composite special moment frames),
which again are largely based on the original 1994 and 1997 NEHRP Provisions. Given
that these systems are made up of components of steel and reinforced concrete, the AISC
Seismic Provisions extensively reference both the AISC-LRFD Specifications
(1999/2001) and the ACI 318 Building Code (2002) for guidance in designing the
appropriate members. In the commentary of the AISC Seismic Provisions, users are
referred to the 1994 ASCE RCS Joint Design Recommendations.
2.3.1 General Building Design Requirements: IBC 2003 and ASCE 7-02
To determine the general system design information, loading, and the seismic design
criteria of a building one must first turn to the IBC 2003. Information such as the source
of dead and live loads, load combination factors, and wind loading is common to all
structural systems. It is not until the determination of the seismic design base shear that
specific seismic system types, such as composite frames, are identified. The 2000 edition
of the IBC was pretty much self-contained in that all of the information necessary to
16
determine the earthquake loading and other seismic design criteria was included in the
structural design chapter (16). In the 2003 edition, most of the detailed criteria have been
removed and are now included by reference to the 2002 edition of the ASCE 7 Standard
Minimum Design Loads for Buildings and other Structures. The IBC 2003 basically
permits the design of a building if it fully adheres to the provisions of ASCE 7-02
Sections 9.1 through 9.6, 9.13, and 9.14.
The seismic design base shear can be determined using the following general equation
defined by ASCE 7 (2002):
V S S D1
= DS (2.1)
W (R I ) T (R I )
but the result should not be less than:

V
= 0.044S DS I (2.2)
W
where:
V = design base shear
W = weight (based on seismic mass)
SDS, SD1 = 2/3 SMS, 2/3 SM1
where SDS and SD1 are the design spectral accelerations and SMS and
SM1 are the maximum considered earthquake spectral accelerations for
short (0.3 sec) and long (1.0 sec) periods, respectively.
SMS,SM1 = FaSS, FvS1
where Fa and Fv are site soil factors and SS and S1 are the mapped
spectral accelerations for short and long periods, respectively.
T = fundamental (first mode) period
I = importance factor based on building type
R = structural response modification factor
The R-value, coupled with the importance factor, I, adjusts the elastic spectral
acceleration demand (SDS or SD1) to account for the amount of nonlinearity that is
expected to occur in a design level earthquake. Both ASCE 7 and IBC 2003 define the
R-value according to a buildings seismic force resisting system. For larger values of R,
17
the corresponding system should be able to accommodate larger levels of inelastic action.
Special composite moment frames are specifically noted in this section and assigned an
R-value of 8, which is identical to convention steel or reinforced concrete special moment
frames. The fundamental first mode period, T, can be computed by structural analysis or
by the following code defined equation:
Ta = CT hnx (2.3)
where:
Ta = approximate fundamental period
CT = building period coefficient (e.g. 0.016 and 0.028 for RC- and steel-
MRFs, respectively)
hn = the height above the base to the highest level of the building (ft)
x = building period coefficient (e.g. 0.9 and 0.8 for RC- and steel-MRFs,
respectively)
For the purpose of calculating the seismic base shear, the maximum period permitted by
ASCE 7 is Cu times the period given by 2.3. The period coefficient, Cu, depends on the
design spectral response acceleration at 1-second period and varies from 1.4 at SD10.3g
to 1.7 at SD10.1g. This cap on the natural period deters users from intentionally
generating a flexible analytical model in order to reduce design level forces. This upper
limit can result in the introduction of additional overstrength in the design (Mehanny et
al. 2000 CCIV).
An additional base shear is superimposed to the lateral resisting system to account for
accidental torsion caused by an assumed 5% offset of the center of mass of each floor.
This is directly added to the design base shear (Eq. (2.1)) and can potentially be an
apparent source of overstrength when evaluating the performance of a 2-dimensional
analytical model (Mehanny et al. 2000 CCIV).
Using the earthquake load, in conjunction with the dead, live, wind, snow, etc., the
structural system must be designed to resist the most critical of the load combinations
specified in Section 1605.2 of the IBC 2003. This stage of the design process can be
18
referred to as the strength design and results in the initial sizing of the elements in the
lateral resisting system. For the composite RCS moment frame, this is the stage where the
composite beams, joints, and RC sections are sized to satisfy the strength requirements
specified by the appropriate codes, the details of which will discussed later in this
chapter.
The IBC 2003 also refers to the ASCE 7 (2002) to stipulate a minimum stiffness criterion
that limits the amount of drift that occurs during the design level event. Deflections
calculated by an elastic analysis under the seismic design base shear and are related to the
expected inelastic deflections by the following equation:
Cd design
inelastic = = ( Cd R ) elastic (2.4)
I
where:
inelastic = expected inelastic deflection under design earthquake
design = deflection calculated by elastic analysis under the design base shear, V
elastic = deflection assuming elastic response under the design earthquake
Cd = deflection amplification factor
The deflection amplification factor, Cd, is similar to the R-value in that they both are
defined according to the lateral resisting system. For special composite moment frames,
Cd is assigned a value of 5.5. For moment resisting frames, the drift criterion in IBC
2003 states that the inelastic drift computed for each floor by Equation (2.4) should be
less than h/40 if four stories or less in height and h/50 for all other buildings (where h is
the story height). For purposes of the drift analysis only, the seismic design forces (Eq.
(2.1)) used to calculate design are not subject to the minimum base shear obtained by
Equation (2.2) nor by the upper limit on the period (i.e., T CuTa ). Both of these
exceptions can greatly impact the drift criterion since it generally allows for the reduction
of forces when computing deflections.
For ductile moment frames, it is more common for the design to be controlled by seismic
drift rather than the strength criterion. When the drift unit controls, it is usually most
effective to increase the stiffness of the beams, (which also increases their strength). The
19
resulting increase in beam strength in turn impacts the design of the columns since the
strong-column weak-beam criterion must also be satisfied. This results in a final design
that has a higher capacity than that required by the strength criterion and the seismic
design base shear defined in Equation (2.1).
ASCE 7-02 requires that the design, construction, and quality of the composite steel and
concrete components that comprise the lateral resisting system adhere to the requirements
of the AISC-LRFD (1999), ACI-318 (2002), and the AISC Seismic Provisions (2002).
These are discussed next.
2.3.2 Member Design Requirements: Part II of the AISC Seismic Provisions
Part II of the AISC Seismic Provisions (2002) outlines the requirements for the design
and construction of composite structural steel, reinforced concrete members, and
composite connections. A majority of the requirements are referenced in from other
existing design provisions, but, the Seismic Provisions supplement missing information
and clarify ambiguities on how to combine other published requirements for steel and RC
systems. In this section, the design requirements for each of the components of
composite RCS frames will be discussed.
2.3.2.1 Reinforced Concrete Columns
The AISC Seismic Provisions state that the design of the reinforced concrete columns in
a composite moment frame shall be in accordance with ACI-318, including all of Chapter
21 (Special Provisions for Seismic Design) except for 21.10. For composite RCS frames,
the reinforced concrete columns are generally designed as they would be in a
conventional reinforced concrete moment frame. One of the major differences in these
composite frames is that the longitudinal steel must be arranged such that it can
accommodate the passage of the steel beams through the joint region. A typical
configuration that demonstrates this situation is shown in Fig. 2.11.
20
Another significant issue that needs to be addressed is how the calculation of internal
forces and deformations of composite RCS systems are affected by the relative stiffness
of the reinforced concrete columns and the composite steel beams. Given that the
seismic design is based on elastic analysis, the effective stiffnesses of the elements should
be representative of the conditions at the onset of significant yielding in the building.
This suggests that the stiffness of the reinforced concrete columns should be
appropriately modified to represent the conditions of an effective cracked section. An
effective stiffness for reinforced concrete columns, which accounts for the amount of
axial load present in the member, is given by Mehanny (2000) in the following equation:
EI eff P
= 0.4 + 0.6 0.9 (2.5)
EI g ,tr 2.4 Pb
where:
EIeff = effective stiffness of the cracked reinforced concrete section
EIg,tr = transformed stiffness of the gross reinforced concrete section
P = expected axial load in reinforced concrete column
Pb = balance axial load taken from the RC column P-M interaction diagram.
This stiffness modification will influence the distribution of forces throughout the system
and also change the total amount of drift in the building. This is especially important in
composite RCS frames since the stiffness of the composite steel beams are not expected
to change as much as the RC columns. The recommendations on computing the effective
stiffness of the composite steel beams are developed in the following section.
2.3.2.1.1 Strong-Column Weak-Beam
As in all special moment resisting frames, the columns in composite RCS systems must
be designed to meet the strong-column weak-beam (SCWB) criterion in order to
minimize the potential for the development of a story or multi-story mechanism. The
AISC Seismic provisions reference the ACI 318 (2002) document for the design of RC
column, which in turn specifies the following SCWB criterion:
M ( 6 5 ) M
c g (2.6)
where:
21
Mc = sum of nominal flexural strengths of the columns framing into the joint
Mg = sum of nominal flexural strengths of the girders framing into the joint
The (6/5)-factor in Equation (2.6) implies that the columns framing into the joint should
be at least 20% stronger than the beams framing into the joint. The column flexural
strength should be calculated considering the factored axial load (according to the load
combinations specified in Section 1605.2 of the IBC 2003) that would result in the lowest
flexural strength. This implies that for columns that are designed below the balance point
(Pbal as shown in Fig. 2.12b), the load combination that would likely control the SCWB
design is that of 0.9D+1.0E, since this would produce the least amount of compression
(or perhaps even tension) and reduce the flexural strength of the column. The opposite is
true for columns designed above the balance point, where the more compression in the
column the lower its flexural strength, implying that the 1.2D+0.5L+1.0E load
combination should control the SCWB design. The best way to approach the SCWB
criteria is to group the design of the interior and exterior columns, as shown in Fig. 2.12a,
since the factored axial loads in the exterior columns are likely to be much different than
those in the interior columns due the overturning moment caused by the lateral loads.
In the development of Equation (2.6), it is assumed by ACI 318 (2002) that both the
columns and beams are reinforced concrete members. While this applies to the RC
columns in composite RCS frames, it is not suitable for the composite steel beams. For
this reason it is also necessary to consider the SCWB criterion for steel structures to
evaluate how the case of a steel (or composite) beam is handled. The AISC Seismic
Provisions (2002) stipulate the following SCWB criterion for steel frames:
M *
pc
=
Z ( F P A ) > 1.0
c yc uc g
(2.7)
M *
pb (1.1R M + M )
y p v
where:
M *
pc = the sum of moments in the column above and below the joint at the
intersection of the beam and column centerlines
M *
pb = the sum of moments in the beams at the intersection of the beam and
column centerlines.
22
Zc = plastic section modulus of the column
Fyc = specified minimum yield stress of column
Puc = required compressive strength using LRFD load combinations
Ag = gross area of column
Ry = ratio of expected yield strength to minimum specified (1.1 for Grade
50 steel)
Mp = nominal plastic flexural strength of beam section
Mv = additional moment due to shear amplification from the location of the
plastic hinge to the column centerline, based on factored load
combinations.
For beams with grade 50 steel, where Ry=1.1, the resulting ratio of column to beam
strength in Equation (2.7) is approximately 1.2, which is the same as in Equation (2.6).
The AISC Seismic Provisions (2002) do not specify whether one should consider the
composite action of the beam in Equation (2.6), although the beam strength definition of
1.1Ry M p implies that only the steel section be considered. Researchers have recognized
that the current approach in design codes to ignore the contribution of the concrete slab in
the strength and stiffness of the structural system may be unconservative and could shift
hinging from the beams to the columns, causing an undesirable failure mechanism (Leon
and Hajjar 1998). For example, the composite strength of a W27x84 with a 2.5 inch slab
over a 3 inch metal deck can be up to 40% stronger than the bare steel section, which can
ultimately have a great impact on the balance of strengths between the beams and
columns in a moment resisting frame. This issue is addressed in further detail in Section
2.5.1.
2.3.2.1.1.1 SEAOC Blue Book Provisions
The SEAOC Seismology Committee has drafted a proposed Blue Book (Maffei 2004)
provision for the SCWB provision, which recommends a criterion based on the strength
of the beam hinges over an entire floor to the strength of the column hinges below that
23
floor. This proposed SCWB criteria, shown graphically in Fig. 2.13, should be applied to
all floors except at the roof level. This criterion can be summarized by the following
equation:
M M
c b (2.8)
where:
M c = sum of nominal moment strength in all columns framing into the

underside of a level, taken at the center of the joints, for all columns
acting to resist lateral forces in the direction under consideration
M b = sum of nominal moment strengths at each end of each beam, taken at

the center of joints, for all girders acting to resist lateral forces in the
direction under consideration
This SCWB provision is more transparent than those defined in Equations (2.6) and (2.7)
in that it is trying to prevent a story mechanism by ensuring that the strength of the
columns over the entire floor can provide the strength to cause hinging in the steel beams.
The major difference between this and the previously described SCWB criteria is that
only the columns below a joint are considered in the design check. If two consecutive
floors have the same column design, this new provision will essentially increase the
previously defined column to beam strength ratios from 1.2 to 2.0. Support for this
provision can be found in research by Dooley and Bracci (2001), which recommends a
minimum SCWB ratio of 2.0 to provide a significantly high probability of preventing
story mechanisms under design basis seismic loading. 1
2.3.2.1.2 Precast RC Column Splice Design
RC column splices are required when the precast method of construction is used. These
grouted splices with mechanical reinforcement bar couplers are common in conventional
precast RC moment frames, an example of which is shown in Fig. 2.14. The high
1
The SEAOC Blue Book Provision for the SCWB criterion (Equation (2.8)) shows promise by capturing
the ultimate two-story failure mechanism in the 3-story, 3-bay test frame (Chapter 3), whereas the
traditional SCWB criteria (Equations (2.6) and (2.7)) were unsuccessful in protecting the frame from a
story mechanism.
24
strength grout is pumped under high pressure into the couplers and splice region and is
designed to develop full plastic strength of the reinforcement bars.
The location of these splices along the length of the column is also an important design
variable. In Japan, it is common to locate these splices directly above the floor level, as
is shown previously in Fig. 2.3. From a structural performance point of view, this
practice is risky considering the potential for severe hinging in this region of the column.
It seems logical to move these splices out of the possible hinging zones and into the mid
height of the column to avoid any adverse effects in the inelastic response of the
columns. 2 The author has consulted with industry engineers regarding this precast
composite RCS system, and they have agreed that mid-height splices are preferable from
a structural point of view. They have also argued that the mid-height splices make the
constructability of the system easier since construction workers will not have to crouch
down to work with the splice. The precedence for this mid-height splice location exists
since it is also common in steel columns.
2.3.2.2 Composite Steel Beams
The beams in a composite RCS moment can be either designed as bare steel or composite
beams. It is recommended that one take advantage of the many benefits incorporating the
composite slab, especially considering that even bare steel beams will require shear studs
to transfer seismic shears between the beam and slab. The focus in this section will be on
composite steel beams, since the design of bare steel beams is fairly straightforward and
well documented. The AISC Seismic Provisions require that the design of the composite
steel beams shall be in accordance with the AISC-LRFD Specification Chapter I. In
order to ensure an adequate amount of ductility in the steel beam prior to crushing of the
2
These precast grouted splices were implemented and tested in the full-scale composite RCS moment
frame and performed exceptionally well up through severe levels of inelastic action. The location of the
splices were also evaluated in the test frame and in preliminary subassembly tests, with results indicating
that both the splice within the hinge zone (Japanese practice) and sufficiently out of the hinge zone
performed quite well, although there was a bit more strength deterioration during repeated cycles with the
splice in the hinge zone. These tests have shown that these splices are very reliable for use in precast
systems in high seismic zones.
25
concrete slab, the Provisions also set the following limit on the distance from the top of
the concrete slab to the plastic neutral axis:
Ycon + db
(2.9)
1700 Fy
1+
Es
where:
Ycon = distance from the top of the steel beam to the top of the concrete slab, in.
Db = depth of the steel beam, in.
Fy = specified minimum yield strength of steel beam, ksi.
Es = elastic modulus of the steel beam, ksi.
The AISC Seismic Provisions also provide a more stringent width to thickness ratio for
the steel beam flanges ( b t 52 Fy , ksi) than those defined by AISC-LRFD. This is
meant to ensure that the steel sections are able to achieve ductilities up to 6 or 7 without
the occurrence of severe local buckling.
The composite slab provides a significant amount of restraint to the upper flange of the
steel beam and largely prevents the occurrence of large local buckles. This behavior has
been observed in several subassembly tests, including Liang et al. (2004) and Civjian et
al. (2000). When the flutes of the metal deck are oriented parallel to beam (where the
slab is in constant contact with the beam), as in the tests by Liang et al. (2004), local
buckling occurs only in the lower flange of the beam while the upper flange remains
intact as the slab provides full restraint. There is slightly less restraint provided by the
slab if the flutes of the metal deck are perpendicular to the beam, since the slab is only in
contact every other flute width. This was the case in the tests reported by Civjian et al.
(2000), where local buckles were observed in both the upper and lower flanges, but those
in the upper flange are not nearly as significant due to the presence of the slab. 3
Regardless of the orientation of the flutes, the presence of the composite slab minimizes
the amount of local buckling in the upper flange, thereby increasing the level of ductility
that can be achieved by the member.
3
In the RCS test frame, where the flutes of the metal deck were oriented parallel to the steel beam, the slab
provided full restraint to the upper flange and therefore local buckling was limited to the lower flange and,
in severe excursions, the lower portion of the web panel.
26
While the effective stiffness of the RC columns are dependent on the axial load and the
percentage of section that is considered cracked, the effective stiffness of the composite
steel beams for the elastic analyses used for design purposes should consider what
portions of the beam are in positive moment (i.e., composite section bending) and
negative moment (i.e., steel section only). Under earthquake loading, it is typically
assumed that the beams in a moment frame system will predominately remain in double
curvature. For this reason, it is reasonable to assume that the effective stiffness of
composite steel beams can be taken as the average stiffness of the composite and steel
section, which presumes that approximately one-half of the beam is in negative bending
while the other half is in positive bending. The effective width of the slab for this
stiffness calculation should be determined from Section I3 in the AISC-LRFD (1999),
which states that beff is equal to the sum of the widths for each side of the beam
centerline, each of which shall not exceed:
(1) one-eighth of the beam span, center-to-center of supports
(2) one-half the distance to the center-line of the adjacent beam (2.10)
(3) the distance to the edge of the slab
This definition of composite beam stiffness, coupled with the effective stiffness for the
RC columns presented in Section 2.3.2.1, is a reasonable assumption for modeling the
stiffness of the frame at the onset of significant yielding and to determine the distribution
of forces and deflections in the structural frame.
2.3.2.2.1 Plastic Strength of Composite Beams
The plastic strength of composite beams can be defined in different ways depending on
the loading and boundary conditions. Figure 2.15 depicts two of the most common
conditions in moment frames: (1) lateral side-sway where the plastic moment develops in
the hinge zone adjacent to the column and (2) gravity conditions where the maximum
moment is reached somewhere in the middle of the span of the beam. Under gravity
conditions, the effective width and stress in the slab are described by the AISC-LRFD
(2002) recommendations for beff (2.10) and 0.85 f c' . For the lateral sway case, the
27
boundary conditions are very different than the gravity conditions and it is unclear
whether or not the same effective width and stress should be applied to compute the
capacity of the composite section.
Several researchers have studied the effective width and stress of the slab in the lateral
sway case in order to correctly model the ultimate moment capacity of a composite beam
section. It has been reported by du Plessis (et al.1972) that the effective width of the slab
at the ultimate strength of the section can be related to the slab area that is in direct
contact with the column flange. This recommendation is logical considering the high
stiffness in the region of the slab and column interface and that localized crushing in the
slab along the width of the column is often observed in subassembly tests (Liang et al.
2004). 4
It has also been verified that the concrete compressive strength can reach an effective
stress of greater than f c' due to the high confinement of the slab near the column flange.
Researchers have shown that the concrete within this region can attain an ultimate
effective stress in the range 0.85 to 1.8 f c' (du Plessis et al. 1972, Tagawa 1989, Lee
1987, Civjan et al. 2001, and Cheng et al. 2002). The reason for the large amount of
variation can be largely attributed to whether or not the shear studs in the subassembly
tests were designed for full composite action. If the studs are designed for a partially
composite section, then the strength and ductility of the composite section is governed by
the behavior of the studs. This issue is investigated more in detail in the calibration study
of composite beams in Chapter 4. Final recommendations on computing the plastic
strength of composite beams for sway frames are withheld until the conclusion of this
chapter.
4
Local slab crushing along the width of the column was also observed after the conclusion of the final
pushover of the RCS test frame, as discussed in Chapter 3.
28
2.3.2.2.2 Design of Shear Studs
The design of the shear studs to provide composite action between the concrete slab and
the steel beam is covered by the AISC-LRFD Specifications. The nominal shear strength
of headed shear studs is defined by the following equation:
Qn , stat = 0.5 Asc f c' Ec Asc Fu (2.11)
where:
Qn,stat = nominal strength of a single shear stud embedded in concrete
Asc = cross-sectional are of shear stud connector
Ec = modulus of elasticity of concrete
Fu = specified minimum tensile strength of a shear stud connector.
This equation for shear stud capacity, which is based on work by Ollgaard et al. (1971), is
intended for studs subjected to static shear loading. There are two reduction factors, Rg
and Rp, applied to the upper limit of the shear strength (i.e., the last term in Eq. (2.11))
which account for (1) whether the steel deck ribs are oriented parallel or perpendicular to
the steel beam, (2) how many studs are placed within each perpendicular rib, and (3)
whether the studs are welded through the steel deck or directly to the steel beam. What
Eq. (2.11) does not account for is the cyclic loading of shear studs under seismic loads.
Given the poor performance of shear studs in composite beams witnessed in numerous
subassembly tests (Cheng 2002, Civjan et al. 2001, Bugeja et al. 2000, Leon and
Flemming 1997), it is evident that the current design provisions seem to overestimate the
capacity of shear studs under inelastic cyclic loading.
The AISC Seismic Provisions acknowledge that the effects of reverse cyclic loading on
the strength and stiffness of the shear studs should be considered in the design of
composite beams and recommend that the strength of headed shear studs should be
reduced by 25% unless a higher strength is substantiated by cyclic testing. 5 Based on this
recommendation, this would lead to the following modification of Equation (2.11) with
cyc assumed as 0.75:
5
This requirement has remained unchanged in the newer version of the AISC Seismic Provisions (2005).
29
(
Qn ,cyc = cyc Qn , stat = cyc 0.5 Asc )
f c' Ec Asc Fu (2.12)
where:
cyc = cyclic strength reduction factor
It is later recommended in the commentary of the Provisions that an inspection and
quality assurance plan be implemented to insure the proper placement of the shear studs
on the steel beam and that the use of additional shear studs beyond that required in AISC-
LRFD may be necessary to ensure composite action. These provisions, as well as FEMA
350 (2000), do not allow the placement of shear studs from the column face to one half
the beam depth beyond the theoretical hinge point to avoid the introduction of any
imperfections to the beam flange that may lead to early fracture. 6
Civjan (2003) performed a series of experimental and analytical tests on modified

composite push-out specimens and bare steel studs to investigate the behavior of shear
studs subjected to reverse cyclic loading. He found that a cyclic loading history resulted
in an approximate reduction in shear strength capacity of 40% as compared to the static
case proposed by Equation (2.11). This reduction is a result of low-cycle fatigue in the
shear stud and weld materials as well as local concrete degradation in the proximity of
the stud. Based on the recommendations from Civjan (2003), the cyc defined in
Equation (2.12) would be further reduced to 0.60. 7
Whereas fully composite beams cause plastification of the steel beam and/or crushing of
the concrete slab, partially composite beams rely on the strength of the studs for the
strength and ductility of the member. Based on the previous discussion, the reliability of
shear studs under seismic loading is questionable; and, if a composite beam is designed
with the studs as the weak link, one must recognize that the probability of early stud
failure and loss of force transfer between the slab and the beam is likely to occur. One
6
This recommendation was inadvertently put to test in the 3-story composite RCS moment frame tested at
the National Center for Research in Earthquake Engineering in Taipei, Taiwan (Chapter 3). No fractures
occurred despite the repeated inelastic cycles experienced in the regions where studs were placed. This
may be due to the fact that upper flange yielding was limited due to composite action of the section.
7
The shear studs in the 3-story composite RCS test frame, which will be described later in Chapter 3,
adhered to the recommendations of Civjian (et al 2003) and exhibited excellent behavior throughout large
levels of repeated inelastic excursions and experienced no instances of shear stud fracture.
30
may argue that even if composite action is lost because of failure of the studs, the steel
beam is still capable of resisting the seismic loads. While this may be true, one must also
keep in mind that the earthquake loads are transmitted through the slab and into the beam
via the shear studs and direct bearing on the RC column. Loss of the force transfer
between the slab and the beam could result in other types of failures as the slab attempts
to drag the force into the moment resisting frame. Therefore, this becomes a complicated
issue that must be addressed if the beams are designed as partially composite.
2.3.2.2.3 Steel Beam Splices
In either the precast or cast in place method of construction, there will be a need to splice
the steel beams together and ensure the continuity of force transfer. Obviously there are
several ways to splice steel beams together, but here the focus will be on a typical bolted
beam splice using flange plates and a shear tab, an example of which is shown in Fig.
2.16. This bolted beam splice is attractive in that it is easy to construct and avoids field
welding.
Just as it was an issue for the RC column splices, the location of the beam splice raises
several questions. From a structural performance standpoint, it is better to locate the
splice as far away from the expected hinging zones as possible to avoid any unwanted
effects on the plastification of the composite steel beam. For the cast-in-place method,
this is fairly easy to accomplish since the steel beams are shipped to the site as signle
(straight pieces) and their length can easily be managed to provide mid-span splices. For
precast construction, the issue is less straightforward since the precast beam-column
modules are prefabricated and shipped to the site. From a constructability point of view,
the closer the beam splice is to the column the easier it is to transport from the
prefabrication plant to the construction site. Beam-column modules with long steel
beams may be more difficult to ship and handle than those with short beam stubs
protruding from the columns. This is entirely a constructability issue and the ultimate
decision may vary from job to job, especially considering that one may find it more
economical to have precast modules that span multiple stories and/or bays. Nevertheless,
31
it is important that the splice not be located within the expected hinging zone, and
therefore it is recommended that the edge of the splice should be at least two times the
depth of the beam ( 2db ) away from the face of the column.
The beam splice should be designed to develop the expected plastic moment ( 1.1Ry M p )
of the steel beam. The moment demand on the splice under lateral loading can be
determined from Fig. 2.17a, assuming a linear moment diagram between the two
theoretical hinge zones. The moment demand on the splice due to gravity loading is
dependent on the location of the splice and the distribution of gravity load, but the
general shape of this moment diagram is shown in Fig. 2.17b. When these two loading
cases are superimposed, the resulting demand on the beam splice is likely to be close to
the expected plastic moment of the beam ( 1.1Ry M p ). The fact that the slab is neglected
in the required splice demand (i.e. ignoring composite action) is thought to be reasonable
since the additional capacity at the splice location provided by the slab is also neglected.
The location of the edge of the splice is at least two times the depth of the steel beam
away from the RC column face, assuming that the hinge zone of the steel beam is fully
contained within one beam depth from the face of the column (Fig. 2.18). This
accommodates at least one full beam depth between the end of the severe hinging zone
and the edge of the splice plate. The design of the beam splice is based on methods
described in the AISC-LRFD (2001) with the following steps:
1. Design Force: The splice should be designed to develop the expected plastic
moment of the steel beam ( 1.1Ry M p ).
2. Plate Thickness: Assuming that the upper and lower flange plates resist the full
moment demand, the thickness of these plates can be determined based on tension
element design concepts using the appropriate phi-factor, = 0.9 . It is possible
to use a single or double plate configuration, as shown in Fig. 2.16, to
accommodate the design forces. The tension force carried in the flange plates can
be determined from the following equation:
1.1Ry M p
TFlangePlates = (2.13)
db t f
32
3. Bolt Design: The size and number of bolts should be determined according to the
bearing strength, as opposed to slip-critical. Further discussion on this topic is
presented in Section 2.3.2.2.3.1.
4. Bolt Shear: The bolts should be checked against bolt shear failure, per the
Specifications for Structural Joints (2000), by assuming either single or double
shear planes (depending on the plate configuration) and the appropriate phi-factor,
= 0.75 .
5. Net Section: Both the flange plates and the beam flange should be checked
against fracture along a net section accounting for the bolt holes. The ultimate
tensile stress, Fu, and a phi-factor of = 0.75 should be used in this design check.
6. Bearing Strength at Bolt Holes: The stresses in the interface between a bolt and
the connected material should be checked against the design moment according to
the Specifications for Structural Joints (2000), which will limit excessive bolt
elongation from occurring during the life of the bolted connection
( Rn 2.4dbtFu ).
7. Block Shear: All appropriate cases of block shear must be considered in both the
flange plates and the beam flanges.
An alternative approach to design the beam splice is to consider the composite strength of
the beam and the splice. This would imply that in step 1, the required design force
should consider the additional strength provided by the composite slab in positive
bending according to the recommendations in AISC-LRFD (1999), which implies an
effective width equal to Equation (2.10) and an effective stress of 0.85 f c' . This
definition of composite strength controls given that the splice location is sufficiently
away from the column face. The design force in the plates can be determined by
computing the required force couple between the slab and the lower flange plates, as
shown in Fig. 2.19, in order to develop plastic moment of the composite beam. This
simplifies the design approach by neglecting the contribution of the upper flange plates
and the shear tab to the strength of the splice in positive bending. This is a reasonable
simplification of the design process given that when composite action is significant
33
enough to affect the design of the splice plates (large ratios of Mp,comp/Mp,steel), the
expected location of the plastic neutral axis (PNA) in the region of the beam splice is
close to the upper flange plates, which would limit the amount of strain in these elements.
Figure 2.20 shows the location of the PNA for W-flange beams within the groups of W21
through W44 assuming that the effective slab width is equal to about 1525mm, which is
calculated by the AISC-LRFD provisions for span lengths of about 6.4 meters long. This
figure shows that for beams with large ratios of Mp,comp/Mp,steel (1.3), the PNA is
relatively close to the location of the upper flange plates (80% height of steel beam).
When the PNA is closer to the mid-height of the beam, then the influence of the
composite slab is lessened and the design of the splice plates reverts back to the steel
beam design. This figure tends to support the assumption to ignore the contribution of
upper flange plates in the design. 8 Once the required force couple is determined (Fig.
2.19), then the plates and bolts can be appropriately proportioned according to the
previously defined steps 2-6. Despite the fact that the upper flange plates are ignored in
positive bending, they still contribute to the negative bending strength of the splice. One
option is to simply use the same design as the lower flange plates to avoid unnecessary
confusion during the construction process. The second option is to design the upper
flange plates to develop the expected plastic moment of the steel beam, 1.1Ry M p , at the
column face, which could achieve savings in both the number of bolts used and plate
size.
2.3.2.2.3.1 Bearing versus Slip Design: Bolt Banging
The AISC Seismic Provisions (2005) state that all bolts should be design using bearing
strength values, but that the bolts should be fully tensioned with faying surfaces prepared
as for Class A or better slip-critical conditions. This implies that even thought the bolts
are designed for bearing resistance, there should be a minimum amount of slip-critical
resistance ( = 0.33 ) to limit slip in moderate earthquakes. This concedes that above
8
In the test frame, discussed in Chapter 3, there was physical evidence that very little yielding occurred in
the upper flange plates due to the presence of the composite slab, and is the reason why it is chosen to
ignore the contribution of these plates in the design.
34
certain earthquake intensities (or other types of extreme loading) the bolts will slip and
activate bearing against the edge of the bolt hole. This slip will result in a loud and sharp
noise, often described as sounding like a high-powered rifle shot. While this bolt
banging phenomenon is generally benign to the connection, it can raise the concern of
building occupants as to the safety of the building (Schwein 1999, Tide 1999). In the
precast composite RCS system, where the splices are generally located close to the
location of high moment reversals, this bolt banging phenomenon would occur
repeatedly during an earthquake. 9
This brings about the question as to whether steel beam splices such as those in RCS
frames should be designed to avoid or at least minimize the occurrence of bolt banging.
A standard slip-critical beam splice would double the amount of bolts required for a
typical bearing connection, and therefore double the length of the splice itself. In
addition to the increase in cost and time of construction, the location of the beam splice
would have to be reconsidered to ensure that the flange plates and bolts do not interfere
with the expected hinging zone. There are alternatives that would make this connection
more attractive, such as the Hyper Splice plates produced by Nippon Steel that doubles
the coefficient of friction and therefore significantly reduces the amount of bolts required
to provide a slip-critical connection. Perhaps one of the more attractive solutions may be
to design the splice not to full slip-critical conditions, but rather to some intermediate
level that would minimize the occurrence of slip during events less than the design level
earthquake (i.e. 10% in 50 year hazard). This design methodology would ensure that
during more frequent events (i.e. 50% in 50 year hazard, immediate occupancy) bolt
banging would be much less of a problem and likely would not alarm the building
occupants.
9
As will be discussed in Chapter 3, the composite RCS test frame, which was a precast system with splices
approximately 2db away from the column face, experienced a significant amount of bolt banging even
during the immediate occupancy (50% in 50 year hazard) event. The bolt slippage did not cause any
detrimental effects on the performance of the building, but it was clear to the lab participants that during a
real time earthquake the cannon-like sound generated during each of the hundreds of slips would
undoubtedly frighten the building occupants. The moment at which the splices slipped was calculated to be
approximately 0.5Mp,steel, which implies that slip could potentially begin at very frequent events of the order
of 50% in 5 years.
35
This issue is not unique to RCS systems, but rather all steel framing systems with bolted
connections similar to what is presented here (Schwein 1999, Tide 1999, Committee on
Steel Structures 1992, Mann 1984). Ultimately, this is more of a comfort issue for the
building occupants since it does not cause any negative effects to the structural system.
Nevertheless, design engineers should be aware of this issue and may consider to
minimize the occurrence of bolt-banging during lower intensity earthquakes.
2.4 Composite Joint Design Guidelines
The 1994 ASCE Guidelines for Design of Joints between Steel Beams and Reinforced
Concrete Columns is referenced by the AISC Seismic Provisions (2002) as the
recommended design methodology for composite RCS joints. While the 1994 ASCE
Guidelines represent the culmination of research on these joints prior to 1990, there have
been several key studies since then that have further developed the state of art for the
design of these composite joints. These recent studies have lead to several important
improvements to the original 1994 ASCE guidelines, some of which include:
1. They have since been validated and extended for use in high seismic zones.
2. Reduced requirements for transverse ties within joint height.
3. Extend the original model to cover a wider variety of joint details.
4. Modifications to address the differences between interior and exterior joints.
5. Allowance for the use for high strength concrete
6. Performance-based requirements to limit the expected deformation and damage.
Since the 1994 ASCE guidelines have been published, there have been several other
proposals made which incorporate these improvements and better capture the expected
strength and stiffness of composite joints (e.g., Parra-Montesinos et al. 2001a, 2003;
Kanno et al. 1993, 1997, 2002; Kuramoto and Nishiyama 2004; Kuramoto 1996). Based
on the work that has been accomplished since the development of the ASCE Guidelines,
an updated RCS joint design guidelines is currently being developed by a group of
researchers (including the author) that is meant to replace the current ASCE Guidelines
and will likely be implemented within the year of 2005. This section will discuss both
36
the 1994 ASCE Guidelines and the Updated Guidelines as well compare and contrast
some of the major differences between the two.
2.4.1 Joint Deformations
The ASCE guidelines instruct users to account for joint deformations by one of three
typical approaches: (1) neglecting finite joint size and computing member stiffnesses
based on centerline dimensions, (2) employing a modified finite joint dimension (i.e.
50% rigid joint), or (3) adjusting the beam and column stiffnesses to account for the
effect of the joint. All of these are common techniques used for other types of moment
frames and were recommended given the lack of conclusive data on the actual
deformation response at the time of publication. The Updated Guidelines improve upon
the original recommendations by defining an idealized joint shear force versus total joint
distortion envelope response (Parra-Montesinos et al. 2001), as shown in Fig. 2.21, which
has been calibrated with test results of interior and exterior connections, failing in both
shear and bearing (Kanno 1993, Parra-Montesinos and Wight 2000, Parra-Montesinos et
al. 2003, Liang et al. 2003). This idealized force-deformation response is intended for use
in modern frame analysis programs that allow the explicit modeling of connection
behavior.
2.4.2 General Detailing Requirements
As in the 1994 ASCE Guidelines, the Updated Guidelines will focus only on the
through-beam type of connection, where the steel beam runs continuous through the
RC column. The 1994 ASCE Guidelines were also limited in scope to the following joint
details: the face bearing plate (FBP), extended face bearing plate (E-FBP), small column,
vertical joint reinforcement, and headed stud details. There have been numerous other
details proposed by researchers, industry engineers, and construction companies since the
development of the original guidelines, some of which have been developed to facilitate
construction and others that have improved the joint strength and performance. The
Updated Guidelines will be extended to include design provisions for the following new
details:
37
1. Transverse beam (Fig. 2.22): this is required detail when one utilizes RCS joints
in a space frame system.
2. Steel band plate (Fig. 2.22): this has emerged as one of the most effective details
for increasing the bearing strength and providing confinement above and below
the beam.
3. Cover plate (Fig. 2.22): this replaces the joint ties and can act as formwork for the
joint when casting the concrete during the construction phase. This joint detail
came about as a logical extension of the face bearing plates in a two-way space
frame joint, but can also be used in a one-way joint as shown in Fig. 2.22
As in the 1994 ASCE Guidelines, it is assumed that all RCS joints will, as a minimum,
have face bearing plate stiffeners within the beam depth, with a width at least equal to the
flange width. These FBPs have proven to significantly increase the joint strength and
improve the overall joint performance by increasing joint stiffness, delaying the onset of
localized cracking and crushing, and providing the necessary confinement to develop the
inner concrete strut.
2.4.3 Effective Joint Width
The effective joint width, bj, is equal to the summation of the inner and outer panel
widths (bi and bo):
b j = bi + bo (2.14)
where:
bj = effective joint width
bi = inner panel width
bo = outer panel width
While the 1994 ASCE guidelines considers the inner panel width equal to the width of
the FBP bp even if it is wider than the beam flange b f , the updated guidelines does not
acknowledge this extra width and sets the inner panel width equal to the beam flange
width, bf:
38
1994 ASCE Guidelines Updated Guidelines
bi = max(bp , b f ) (2.15) bi = b f (2.16)
where:
bp = width of FBP
bf = width of beam flange
The outer panel width bo should be calculated according to the shear keys used to
mobilize the concrete outside of the width of the steel beam flanges (i.e. steel columns,
steel band plates, extended FBPs, etc.) and the connection geometry. The updated
guidelines assume that when steel columns or extended FBPs are used, the outer panel
width can be determined by assuming a 1:3 slope projecting from the edges of the shear
key, as shown in Fig. 2.23 and defined in Equation (2.18)a, with dimensions defined by x
and y. For the band plate detail, the assumption is that x = h and y = bf, except that where
the band plate replaces the ties within the bearing region, then the maximum outer panel
width is controlled by the stiffness of the band plate and should be taken as bo = 12tbp (tbp
= thickness of band plate), regardless of the other shear keys present. The reason for this
reduction in the outer panel joint width is based on strain measurements from
subassembly tests (Parra-Montesinos 2000) that indicate the band plates transfer shear
force to the outer concrete regions primarily through direct bearing on the concrete over
the width of the beam flange, as opposed to the formation of a horizontal strut and tie
mechanism. If no shear keys are provided, a minimum outer panel width is allowed by
assuming x = 0.7h and y = 0 in Equation (2.18)a to account for the outer strut formation
between the joint ties and friction in the regions of beam flange subjected to high bearing
stresses (there is also likely a strut developed between the FBPs and the joint ties).
The 1994 ASCE Guidelines define the outer joint width as a ratio of the average of the
beam flange width and the column width according to the types of shear key used in the
joint. If no shear keys are provided, the 1994 ASCE guidelines recommend that the outer
panel width be taken as zero. Both of these definitions for outer panel width are
compared to the new guidelines equations below:
39
bo = C ( bm bi ) 0.5db (2.17)a bo = y + 2 3 x b f b b f (2.18)a
bm = ( b f + b ) 2 b f + h 1.75b f (2.17)b band plates w/no joint hoops
C = ( x h) ( y bf ) (2.17)c bo = 12tbp b b f (2.18)b

where:
C = joint mobilization coefficient
bm = maximum effective width of joint region
db = depth of beam
b = width of concrete column measured perpendicular to the beam
h = height of the concrete column measured parallel to the beam
x,y = effective dimensions of shear keys defined in Fig. 2.24
tbp = thickness of the band plate
In order to compare the differences in joint width for each method, consider a typical
joint with a steel column and FBPs, as shown in Fig. 2.24. The calculated joint widths
summarized in Table 2.1 show that the Updated Guidelines predict an outer panel width
about 2.6 times larger than the ASCE Guidelines. This difference reflects the fact that
the joint width determination from the ASCE Guidelines was largely based criteria for
reinforced concrete joints, whereas more recent research has shown that the composite
RCS joints create internal stress transfer mechanisms with larger effective joint regions.
2.4.4 Joint Strength
Joint panel shear strength is handled the same way in both models and is defined as the
summation of the contribution of the: (1) inner panel strength, Vin, which includes the
capacity of both the steel web panel and inner concrete strut, and (2) the outer concrete
panel strength, Von. Each of the individual components making up the total shear strength
have been refined in the Updated Guidelines, reflecting the most recent test data
available. These changes are discussed in detail in the following sections. The phi-factor
for the shear strength (s) has also been increased in the new model to 0.85 from the
original value of 0.7. The rationale for the updated phi-factors will be discussed later in
40
Section 2.4.10.3. A new strength reduction factor, k = 0.85, has also been introduced into
the design check to control joint distortions such that joint damage is limited to moderate
cracking with web panel yielding under ultimate loads. Shear design checks for each set
of guidelines are shown in Equations (2.19) and (2.20), respectively.
In the 1994 ASCE Guidelines, joint bearing strength is treated as a total joint failure,
separate from joint shear, and is computed and checked against the appropriate design
forces, as shown in Equation (2.21). The Updated Guidelines assumes that vertical
bearing failure occurs only over the inner panel width, bi, and that the outer panel is more
likely to fail in outer panel shear. Test data has shown that latter approach is more
representative of the true behavior in these composite RCS joints. Kanno (1993)
observed that the outer panel fails in relatively the same manner regardless of whether the
inner panel fails in joint shear or vertical bearing. Therefore, the Updated Guidelines
limits the amount of shear that can be developed in the inner panel by the vertical bearing
strength, as shown in Equation (2.20)b. In the Updated Guidelines, the factor in
Equation (2.20)a depends on the governing mode of failure, s = 0.85 for shear and b =
0.70 for bearing. These differences in phi-factors reflect the fact that it is more desirable
to design the joint to fail in shear to avoid the undesirable pinched response associated
with bearing failure.
ASCE Guidelines Updated Guidelines

V j s (Vin + Von ) (2.19)a V j k (Vin + sVon ) (2.20)a
b ( M vb Vb h )
Vin = Vspn + Vicn (2.19)b Vin = s (Vspn + Vicn ) (2.20)b
dj
Vj =
M b
Vc (2.19)c M
dj Vj = b
Vc (2.20)c
dj
M c + 0.35hVb b M vb (2.21)
where:
Vj = joint shear demand imposed by adjacent beams and columns
s = joint shear phi-factor (ASCE: 0.7, Updated: 0.85)
41
b = joint bearing phi-factor (ASCE: 0.7, Updated: 0.75)
Vin, Von = inner and outer panel shear strength
Vspn = steel web panel shear strength
Vicn, Von = inner and outer concrete panel strength
Mb, Mc = summation of beam or column moments transferred into the joint
dj = depth of joint equal to distance between beam flange centerlines
Vc = average column shear
Vb = difference in beam shears
Mvb = moment derived from vertical bearing strength of joint
k = deformation-based strength reduction factor, set to 0.85
2.4.5 Inner Panel Shear Strength
The shear strength of the inner panel, Vins, is simply summation of the capacities of the
steel web panel, Vspn, and the inner diagonal concrete strut, Vicn. This can be described as
follows:
Vins = Vspn + Vicn (2.22)
where:
Vins = nominal horizontal shear strength of inner panel
The horizontal shear strength provided by the steel web panel is defined by the ASCE
and Updated Guidelines as follows:
Vspn = 0.6 Fysp tsp jh (2.23) Vspn = 0.6 Fysptsp h (2.24)
where:
Fysp = nominal yield strength of steel web panel
tsp = thickness of steel web panel
jh = horizontal distance between bearing force resultant (found by
iteration, 0.7h)
= coefficient differentiating interior (0.9) and exterior joints (0.8)
42
Test data was used to simplify Equation (2.23) by removing the iteratively determined jh-
term and replacing with constants that account for the differences in strain distribution for
interior versus exterior configurations. As a result, even if it assumed that jh = 0.7 h in
Equation (2.23), the Updated Guidelines (Equation (2.24)) predict a stronger steel panel
zone by approximately 29% (reduced to 14% for exterior joints).
The capacity of the inner diagonal concrete strut is defined by each of the guidelines as
follows:
Vicn = 1.7 f c' bp h 0.5 f c'bp d w (2.25) Vicn = ki f c' bi h 0.5 f c'b f d j (2.26)
where:
f c' = nominal compressive strength of concrete (MPa)
bp = effective width of FBP: bp b f + 5t p 1.5b f
tp = thickness of FBP
= strength factor depending on connection type (1.0 interior, 0.6
exterior)
ki = strength factor set to 1.7
dw = depth of the beam web
For interior joints, both the 1994 ASCE and Updated guidelines yield the same strength
for the inner concrete strut. Test data has shown that this strut is less effective for
exterior joints, therefore the strength factor, , reduces the nominal strength to 60% of
interior joints. To prevent bearing failure at the ends of the strut, both models limit the
amount of shear developed in the concrete strut by a bearing stress of 2.5 f c' over an area
at the top and bottom of the FBPs equal to 0.25b f d j
2.4.6 Inner Panel Vertical Bearing Strength
The 1994 ASCE and Updated Guidelines define the vertical bearing strength of the inner
panel of the joint as follows:
43
M vb = Ccn h (1 1 2 ) + (
M vb ,i = Ccn h 1 1* 2 + )
(2.27)a (2.28)a
hvr (Tvrn + Cvrn ) hvr (Tvrn + Cvrn )
Ccn = fb b j ( 1h 2 ) (2.27)b Ccn = fb b f ( 1* h 2 ) (2.28)b
fb = 2.0 f c' (w/min. tie reinf.)

fb = 2.0 f c' (2.27)c (2.28)c
fb = 2.5 f c' (w/min. band plate reinf.)
where:
Ccn = nominal compression strength of bearing zone
1* = stress block depth coefficient taken as the 0.85 for f c' 27.6MPa, and
reduced by 0.05 for every 6.9MPa increase in strength, with a
minimum limit of 0.65.
fb = effective bearing stress block intensity
The biggest difference between the models is the effective joint width over which the
effective bearing stress acts. The ASCE Guidelines assume that the bearing stress acts
over the entire effective width of the joint, bj, where on the other hand, the Updated
Guidelines assume that the bearing stress is effective only over the inner panel width, bf,
of the joint. As previously discussed, this latter approach has been confirmed in the
subassembly tests (Kanno 1993) and is a better way to model the joint failure
mechanisms. The ratio of the Updated bearing strength to the ASCE bearing strength
yields the following result:
M vb ,i ,Updated fb ,Updated bf 1* (1 1* 2 )
= (2.29)
M vb , ASCE f b , ASCE bj 1 (1 1 2 )
If the joint contains standard joint hoops above and below the beam, then the first ratio
(fb) in Equation (2.29) will simply be unity. If steel band plates are present, then this ratio
will be 2.5. The second ratio in Equation (2.29) will typically be less than 1; with the
only exception being equal to 1 if no shear keys are provided to mobilize the outer
concrete strut. If we take the example developed in Section 2.4.3 (Fig. 2.24) and assume
that only joint hoops are provided for confinement in the bearing zone and
44
f c' = 41.3MPa (6ksi) , then the result of Equation (2.29) is 0.81, indicating that the
Updated Guidelines predicts a bearing strength that is about 20% less than the 1994
ASCE prediction. This is expected since the difference in effective joint width in bearing
is considerable. However, the net effort on the joint design is minimal since the Updated
Guidelines include the outer panel shear strength in the total joint bearing strength, as
shown in Equation (2.20).
2.4.7 Outer Panel Shear Strength
The ASCE and Updated Guidelines define the strength of the outer concrete strut as
follows:
Von = 1.7 f c' bo h (2.30) Von = ko f c' bo h (2.31)
where:
ko = confinement strength factor for outer concrete strut
= 1.5 (cover plate or joint hoops and band plates)
= 1.25 (all other cases with either joint hoops or band plates)
Unlike the 1994 ASCE Guidelines, the Updated model varies the strength factor
depending on the type of details in the joints and also accounts for the differences due to
interior versus exterior joint configuration. The ratio of the predicted strengths from the
Updated to the 1994 ASCE Guidelines yields the following results for interior joints:
Von ,Updated bo ,Updated
= (0.74 or 0.88) (2.32)
Von , ASCE bo , ASCE
Again, if we take the example joint from Section 2.4.3 (Fig. 2.24), then Equation (2.32)
would yield a ratio of 1.95 to 2.32, depending on the joint details used. For exterior
joints, these ratios are reduced by an additional 40%. Nevertheless, this shows that the
Updated Guidelines predict a much larger contribution from the outer panel concrete strut
compared to the original 1994 ASCE Guidelines.
45
2.4.8 Joint Panel Shear and Vertical Bearing Moment Capacity
Up to this point, the strength of the joint has been represented by joint shear and divided
between the inner and outer joint panels. It is often necessary to combine each of these
contributions to obtain an equivalent joint moment capacity for panel shear and vertical
bearing mechanism. This information can be used to compare with the moment
capacities of the surrounding beams and columns or in analytical models of these
composite joints, such as those described in Chapter 4. The shear strength provided by
the inner concrete strut, the steel web panel, and the outer concrete strut can be combined
in the following equation to compute the moment capacity of the joint in panel shear:
M ps = Vspn d w + Vicn d j + 1.25Von d j (2.33)
where:
Mps = equivalent joint panel shear moment capacity
The inner vertical bearing strength can be combined with the outer panel shear strength to
obtain the equivalent joint vertical bearing moment capacity:
M vb = M vb,i + 1.25Von d j (2.34)
where:
Mvb,total = equivalent joint vertical bearing moment capacity
The smaller of these two moment capacities will control the design strength of the joint.
2.4.8.1 Strong-Joint Weak-Beam
For the seismic design of composite joints, the nominal moment strength of the joint must
be at least equal to the summation of the nominal strengths of the beams framing into the
joint. The nominal strength of the beam in positive bending should consider the
composite strength of the beam, with nominal steel properties and an effective slab stress
of 1.3 f c' . This is referred to as the strong-joint weak-beam criterion and is found in the
AISC Seismic Provisions (2002).
46
2.4.9 Detailing Considerations
In this section, the design requirements of each of the joint details are proposed to
provide the capacity to develop the full strength of the composite joint.
2.4.9.1 Ties within Beam Depth
Several studies (Kanno 1993, Parra-Montesinos et al. 2000, and Liang et al. 2004) have
shown that the requirements for the horizontal reinforcing bar ties within the depth of the
steel beam can be relaxed from the requirements specified by the 1994 ASCE Guidelines
(Equation (2.35)). Based on these tests, the Updated Guidelines have decreased the tie
requirements to Equation (2.36), which is similar to the minimum requirements for
reinforced concrete joints (ACI 318).

Ash Fysh
ka f c' bo (2.35)a s 0.01 (2.36)a
sh
Ash 0.004bsh (2.35)b sh = min(0.25d j , 0.25h) (2.36)b

where:
Ash = cross-sectional area of reinforcement parallel to beam with spacing sh.
Fysh = nominal yield strength of column ties
sh = center to center spacing of column ties
ka = strength factor set to 1.89 if joint is subject to tension force, 1.44 if no
tension
s = volumetric ratio of hoop volume within beam depth to joint volume
( d j hb )
2.4.9.2 Longitudinal Column Bars
The limitation on the size of the longitudinal bars passing through the joint is unchanged
from the 1994 ASCE Guidelines, and is specified in Equation (2.37). This requirement is
47
meant to limit the amount of bar slip in the joint associated with possible large changes in
bar forces due to the transfer of moments through the joint (ACI-ASCE 352).
d j ( 420 ) dj
d bar (2.37)
20 Fyr 20
where:
dbar = diameter of the longitudinal column bar
Fyr = yield strength of longitudinal column bars in MPa
2.4.9.3 Face Bearing Plates and Steel Band Plates
The requirements on the thickness of the FBPs did not change from the original ASCE
guidelines. The FBP thickness should be sized to for the capacity to resist the nominal
shear strength of the inner concrete strut, Vicn.
3
tp
b f Fup
(
Vin b f tw Fyw ) (2.38)
3Vicn
tp (2.39)
2b f Fup
Vicb f
t p 0.20 (2.40)
Fyp d w
where:
Fyp, Fup = nominal yield strength and ultimate strength of bearing plate
dw, tw = depth and thickness of beam web
Equations (2.38)-(2.40) are semi-empirical formulas developed from joint tests by Sheikh
et al. (1987). The following minimum thickness requirement should also be adhered to:
t p b f 22 (2.41)
The Updated Guidelines also specify thickness requirements for the steel band plates,
which were derived from the FBPs requirements given the similarities in behavior.
48
3 1
tp Von (2.42)
Fup 2dbp + b f

4Von dbp
tbp 0.20 (2.43)
Fybp h
where:
dbp, tbp = depth and thickness of the band plate
Fybp, Fubp = nominal yield strength and ultimate strength of band plate
2.4.9.4 Steel Beam Flanges
Requirements for the thickness of the steel beam flanges have remained the same from
the original ASCE guidelines. These requirements are in place to ensure that the flanges
are able to resist the vertical bearing force associated with the joint shear in the steel
panel. The flange thickness requirement is as follows:
b f tsp dFysp
t f 0.30 (2.44)
hFyf
where:
Fyf = nominal yield strength of the beam flanges
2.4.9.5 Extended Face Bearing Plates and Steel Column
The extended FBPs and/or the steel columns should be designed to resist a force equal to
the joint shear carried by the outer concrete compression strut, Von. The average concrete
bearing stress against these elements is assumed to be approximately 2.5 f c' , which acts
over a height of 0.2dj. These elements may be considered capable of resisting the joint
shear forces if their thicknesses adhere to the following equation:
Vonbp'
t f 0.12 (2.45)
0.25d j Fy
where:
49
b p' = the flange width of steel column or width of extended FBP
Fy = nominal yield strength of plate

In addition to this, the thickness of the extended FBPs should not be less than that of the
FBPs.
2.4.10 Model Validation
In this section, data are summarized to validate the Updated Guidelines against RCS
beam-column subassembly tests. For each of the applicable tests, both the 1994 ASCE
and Updated Guidelines will be computed and compared to the actual measured strength
of the joint specimen.
2.4.10.1 RCS Joint Tests Considered
The following joint test series were considered in this validation study: (1) Sheikh 1989,
(2) Deierlein 1989, (3) Kanno 1993, (4) Parra-Montesinos and Wight 2000, and (5)
Liang et al. 2003. This includes a total of forty-nine RCS joint tests, which of these,
thirty-eight are of interior configuration and eleven are exterior configurations. Since it is
a requirement of both guidelines, all of the joints that did not have FBPs were not
considered in this study (five in total: Sheikh 1, 3, 9 (1989) and Deierlein 12, 14 (1989)).
The joints were separated into the following failure mechanisms: (1) joint panel shear, (2)
joint vertical bearing, and (3) other (i.e. column failure, beam failure, etc.). There are a
total of twenty-one joints in the joint shear failure group and six in the joint bearing
failure group. The details for these joint tests are summarized in Tables 2.2 and 2.3. For
further details on each of these tests, please refer to the original references (Sheikh et al.
1989; Deierlein et al. 1989; Kanno 1993, 1997; Parra-Montesinos et al. 2000; and Liang
et al. 2003).
50
2.4.10.2 Updated Joint Guidelines Validation Results
Using the information presented in Tables 2.4 and 2.5, the predicted strengths of each of
the subassemblies are computed in terms of beam shear (Vbc) using the 1994 ASCE and
Updated Guidelines and are then compared to the maximum beam shear reached during
the test (Vbe). The results of this procedure for the joint panel shear and joint bearing
failure groups are shown in Figs. 2.25 and 2.26, respectively. The filled circles in these
figures represent the predicted to measure ratio using the Updated Guidelines, while the
unfilled circles are the results from the 1994 ASCE Guidelines. In the joint shear
validation (Fig. 2.25), one can see an overall increase in the ratios from the 1994 ASCE
values to the Updated values. The average values of predicted to measured strengths for
the joint shear group are 0.96 and 0.80 for the Updated and 1994 ASCE Guidelines,
respectively, indicating that the Updated model is more accurately picking up the
measured strength of these specimens. The coefficient of variation from the Updated
Guidelines (13.8%) is also slightly better than the 1994 ASCE guidelines (16.0%).
The predicted to measured ratios for those joints failing in vertical bearing are shown in
Fig. 2.25. The Updated Guidelines greatly improve the accuracy of the strength model,
increasing the mean strength ratios from 0.76 to 0.92. The consistency of the strength
prediction for the vertical bearing model is also improved, as shown by the decrease in
the coefficient of variation from 15.4% to 8.0%.
This validation study shows that for both the joint shear and joint bearing failure modes
the proposed changes incorporated into the Updated Guidelines improves both the
accuracy and the consistency of the original ASCE strength model.
2.4.10.3 Determination of Strength Reduction () Factors
The strength reduction () factors for the two strength models are intended to account for
the variability in the capacity due to (1) discrepancies between the nominal and measured
material properties, (2) differences in fabrication and erection, and (3) inconsistency of
51
the design equations. The phi-factor used for the original ASCE joint guidelines was
taken conservatively as 0.7, since at the time of the publication there was only limited
data to validate the provisions. Given that more experimental results are available, the -
factors can now be determined using the beta-reliability method, which is the technique
used to develop the -factors for the AISC-LRFD (2002). Using this method, the
expression for the -factor can be derived as follows:
Rm
= exp ( VR ) (2.46)
Rn
where:
= linearization approximation constant used to separate the resistance
and demand uncertainties, taken as 0.55.
= beta reliability or safety index defined as the number of standard
deviations from the mean
R R
ln ln m
Q m Qm
=
ln( R / Q ) VR2 + VQ2
where Rm and Qm are the mean values of the resistance and load effect
and VR and VQ are the corresponding coefficients of variations.
VR = coefficient of variation of the resistance variable
= VM2 + VF2 + VP2
VM, VF, VP = coefficient of variation representing the material strength,

uncertainties in fabrication, and uncertainties in assumptions (i.e., the
professional factor)
Rm
= ratio of the means of the measured to the nominal resistance.
Rn
Traditionally, the target beta-value is assumed to be 3.0 for members and 4.5 for
connections. A larger value is traditionally targeted for connections, given that the
ductility of members in is more reliable than that of connections. Using the same
rationale for composite joints, the joint panel shear check is assigned a beta-value of 3.0
52
while vertical bearing check is assigned a value of 4.5. The coefficient of variation
accounting for the material strengths and the uncertainties in fabrications are presented in
Table 2.6, following the assumptions used by Ravindra and Galambos (1978) to develop
-factors for the AISC-LRFD (2002). The coefficient of variation that accounts for the
uncertainties in assumptions is largely related to the confidence in the strength model that
was developed in Section 2.4.4 through 2.4.7. This value can be assumed as the
coefficient of variation found in the predicted to measured data presented in Section
2.4.10.2 and is summarized again in Table 2.6. The measured to nominal resistance term
( Rm / Rn ) in Equation (2.46) is simply the inverse of the predicted to measured ratios
presented in Section 2.4.10.2 multiplied by the ratio of the expected to nominal material
strengths. The latter ratio is necessary since the predicted strengths in Section 2.4.10.2
are based on measured material properties, whereas the Rm / Rn term requires the ratio of
measured to nominal mean resistance. The expected to nominal material strength ratio
will be assumed as 1.1, which is equivalent to the assumed value for Grade 50 steel (Ry).
Using the values summarized in Table 2.6 and Equation (2.46), the phi-factors for the
panel shear and vertical bearing strength models are computed as 0.84 and 0.85,
respectively. While this process seems reasonable, there are some reservations about
recommending such a high -factor for the vertical bearing strength model, given the
limited amount of available data on which this is based (only 6 data points for bearing
model). The limited data set for the vertical bearing strength reduces the level of
confidence in the low coefficient of variation (0.08), which has a large impact on the
calculated -factor. A reasonable approach may be to assume that the coefficient of
variation of the bearing model would be roughly equivalent to the panel shear model
(0.138) if a larger data set was available. If the vertical bearing phi-factor is recalculated
using this assumption, then the phi-factor is decreased to 0.77 (as shown in italics in
Table 2.6). Given the limited data set and the low confidence in the coefficient in
variation, this assumption is considered reasonable and the decreased phi-factor is
accepted for the vertical bearing model. For the sake of consistency with current codes,
53
the final recommended phi-factors for the panel shear and vertical bearing model are
assumed to be 0.85 and 0.75, respectively.
2.5 Final Recommendations
This chapter has reviewed the current knowledge in design standards for composite RCS
frames as well as some of the latest research that can have an impact on these systems.
This section will summarize the recommended design methods that will be used and
validated through the rest of this study. In addition to the recommendations presented in
this section, the general building design requirements presented in Section 2.3.1 from the
IBC 2003 and the ASCE 7 (2002) will be adopted in this study.
2.5.1 Reinforced Concrete Columns
The design of the RC columns follows the conventional seismic design as recommended
in Chapter 21 of ACI 318 (2002). The only special consideration is to arrange the
longitudinal reinforcement bars to accommodate the passage of steel beam (and
potentially a cross-beam for a two way joint). The recommendations of Mehanny (2000)
are adopted to model the effective stiffness of the RC columns, as presented in Equation
(2.5).
2.5.2 SCWB Criterion
The SCWB criteria from ACI 318 (2002) and the AISC Seismic Provisions (2002) were
introduced in Section 2.3.2.1.1. When applying either of these to composite RCS frames,
problems arise with these criteria since they are written for applications to all-concrete or
all-steel moment resisting frames. A natural extension would be to combine the
applicable portions of each of the SCWB criterions into one that could be applied to the
design of composite RCS frames. The updated SCWB criterion is proposed as follows:
M c
> 1.0 (2.47)
M *
g
54
where:
M c = sum of nominal flexural strengths of the RC columns framing into the

joint, as defined per Equation (2.6)
M *
g = sum of the expected beam strengths, where applicable, considering the
composite section of the beam.
Equation (2.47) uses the definition of column strengths from the ACI 318 (2002)
(Equation (2.6)) as described in Section 2.3.2.1.1. The sum of the beam strengths in
Equation (2.47) now requires that the strength of the composite beams also be included in
the SCWB check, recognizing that the beam in positive bending will attain the full
composite strength while the beam in negative bending will achieve the expected plastic
moment of the bare steel beam ( 1.1Ry M p ). 10 Calculation of the plastic strength of the
composite section should follow the recommendations presented in Section 2.3.2.2.1.

This proposed SCWB criterion is used for the design of the frames presented within this
study.
While the SCWB provision proposed by the SEOAC Blue Book (Equation (2.8)) is not
used for design in this study, this provision will be investigated with respect to the
seismic performance of the frames.
2.5.3 Composite Steel Beams
As mentioned in Section 2.5.1, it is recommended that the composite strength of the steel
beams be considered in the seismic design of composite RCS systems. The amount of
overstrength that a composite slab provides to a bare steel beam is significant enough to
disrupt the intended sway mechanism of a moment frame and therefore should not be
ignored. This is clearly shown in Fig. 2.27, which shows the increase in capacity that a
composite slab provides for typical W-section beams. In addition to the SCWB criteria,
the composite strength and stiffness should also be considered in the strength design of
10
This conclusion is supported by the results of the full-scale test frame, discussed in Chapter 3, which
shows the durability and performance of the slab significantly impacted the specimen behavior.
55
the beams and in meeting the drift limitations. Recommendations on modeling the
stiffness of composite beams have been proposed in Section 2.3.2.2.
It is recommended that the shear studs be designed to provide a fully composite section
according to the recommendations from the AISC Seismic Provisions (2002) described in
section 2.3.2.2.2. Partially composite sections are not recommended given the limited
ductility of shear studs, particularly under cyclic loading. For a fully composite section,
the maximum plastic strength of a composite section can be modeled according to the
recommendations of du Plessis (et al. 1972) with an effective concrete stress of 1.3 f c'
acting over an effective width equal to the column width. 11 To remain consistent with
the recommendations for a bare steel beam, the plastic force of steel beam should
consider the expected yield strength, including strain hardening, of the steel, which is
defined as 1.1Ry Fy .
2.5.4 Precast Element Splices
Design recommendation for the steel beam splices were presented in section 2.3.2.2.3. It
is sufficient to consider the simplified version of the splice design by ignoring the
composite slab since it factors into both the demand and capacity. It is suggested that the
location of the beam splice be at least twice the depth of the beam away from the column
face to avoid any interaction with the hinging zone of the steel beam, which is likely to be
contained within one depth of the beam away from the column face. The slip capacity of
these bolted connections and the bolt banging phenomenon will be investigated further
in Chapter 3, but for the time being, it is recommended that the bolts be designed for
bearing capacity
The RC columns are spliced using the common precast construction technique of grouted
splices with mechanical reinforcement bar couplers. These splices are typically designed
by the construction company to develop the entire plastic moment of the RC column
11
Both the effective width and effective stress of the composite slab presented within this section are
investigated and validated in the calibration study presented in Chapter 4.
56
section. It is suggested that these splices be located in the middle-third of column height
to avoid any interaction with potential hinging regions at the column ends and also to
make the workability of the splice more convenient for construction workers.
2.5.5 Composite Joint Design
The proposed updates to the 1994 ASCE Joint Design Guidelines, as presented in Section
2.4, are recommended for the design of composite RCS joints. As shown in Section
2.4.10.2, this new model has been shown to be more accurate and consistent in predicting
both the panel shear and vertical bearing strength of these joints. The updated provisions
now cover a wider variety of joint details that have proven to be effective in maintaining
the strength and stiffness of composite joints as well as improving the constructability.
New phi-factors have been proposed using the beta-reliability method, which is the same
technique used to develop the phi-factors for the AISC-LRFD (2002) design approach.
57
Table 2.1 Joint width comparisons between ASCE and Updated Guidelines.
Joint Dimension ASCE Guidelines Updated Guidelines
bi 266 mm 266 mm
bo 102 mm 269 mm
bj = bi + bo 368 mm 535 mm
58
Table 2.2 Summary of joint details and dimensions for those in the panel shear joint group. (Units: kN, mm)
Name h b bf db tf tsp f'c Fysp Tvrn Cvrn hvr Lc Lb Joint Details
KANNO OJS1-1 406 406 127 355 10.21 6.5 41.20 379.0 0 0 0 3391 3048 JH
KANNO 0JS2-0 406 406 102 229 19.55 6.66 42.60 407.0 0 0 0 3391 3048 JH,SC
KANNO OJS3-0 406 406 153 355 25.93 6.66 42.60 407.0 324.9 324.9 305 3391 3048 JH,DB,E-FBP (d=102,t=13.45,w=153)
KANNO OJS5-0 406 406 153 355 25.25 6.53 48.30 393.0 324.9 324.9 305 3391 3048 JH,DB,BP (dbp=102,tbp=13.45)
KANNO 0JS7-0 406 406 153 355 25.25 6.53 48.30 393.0 324.9 324.9 305 3391 3048 JH,TB,DB,E-FBP (d=102,t=13.45,w=153)
KANNO HJS1-0 406 406 153 355 25.93 6.66 102.00 407.0 0 0 0 3391 3048 JH,E-FBP (d=102,t=13.45,w=153)
KANNO HJS2-0 406 406 153 355 25.93 6.66 102.00 407.0 0 0 0 3391 3048 JH,ST,E-FBP (d=102,t=13.45,w=153)
DEIERLEIN 10 508 508 203.2 450.9 22.4 6.68 32.41 249.6 0 0 0 3708 4877 JH
DEIERLEIN 11 508 508 203.2 450.9 22.4 13.39 32.41 249.6 433.9 433.9 381 3708 4877 JH,DB
DEIERLEIN 13 508 508 203.2 450.9 22.4 6.68 34.47 249.6 0 0 0 3708 4877 JH,ST
DEIERLEIN 15 508 508 203.2 450.9 22.4 6.68 28.27 249.6 0 0 0 3708 4877 JH,SC
DEIERLEIN 17 508 508 203.2 450.9 22.4 6.68 26.89 249.6 433.9 433.9 381 3708 4877 JH,SC,DB
SHEIKH 4 508 508 203.2 444.5 19.05 6.35 29.65 247.5 0 0 0 3708 4877 JH
SHEIKH 5 508 508 203.2 444.5 19.05 6.35 29.65 247.5 0 0 0 3708 4877 JH
SHEIKH 7 508 508 203.2 444.5 19.05 6.35 25.6 247.5 0 0 0 3708 4877 JH
SHEIKH 8 508 508 203.2 444.5 19.05 6.35 24.8 247.5 0 0 0 3708 4877 JH,E-FBP (d=101.6,t=22.23,w=203.2)
PARRA 1 400 400 203 241 32 12.7 43.40 279.0 0 0 0 2240 2440 JH,SC
PARRA 6 400 400 172 390 29.08 7.94 29.60 381.0 0 0 0 2240 2440 JH,SC
PARRA 9 400 400 172 390 29.08 7.94 29.00 320.0 0 0 0 2240 2440 TB,SC,BP (dbp=100,tbp=13)
59
Table 2.3 Summary of joint details and dimensions for those in the vertical bearing joint group. (Units: kN, mm)
Name h b bf db tf tsp f'c Fysp Tvrn Cvrn hvr Lc Lb Joint Details
KANNO OJB1-0 406 406 153 355 16.1 25.93 48.70 345.0 0 0 0 3391 3048 JH,SC
KANNO OJB4-0 406 406 153 355 16.1 25.93 48.50 345.0 0 0 0 3391 3048 JH,SC
KANNO OJB5-0 406 406 153 355 16.1 25.93 46.00 345.0 228.8 228.8 305 3391 3048 JH,SC,DB
KANNO OJB6-1 406 406 153 355 16.1 25.93 46.00 345.0 0 0 0 3391 3048 JH,SC
SHEIKH 2 381 381 101.6 304.8 11.11 6.35 24.48 383.4 0 0 0 2743 2438 JH
SHEIKH 6 508 508 203.2 444.5 19.05 6.35 27.58 247.5 0 0 0 3708 4877 JH
60
Table 2.4 Results from Update Guidelines for the joint group failing in panel shear. (Units: kN, mm)
Updated ASCE
Name bo Vspn Vicn Vohn Mvb Ccn x Vb,PanelShear Vb,Bearing
Vmeas/Vpred Vmeas/Vpred
KANNO OJS1-1 0.69 0.60 0 540.1 562.6 0 362200 1275 284.2 136.8 132.3
KANNO 0JS2-0 0.86 0.65 177.7 594.3 455.1 588.5 300800 1058 284.2 126.7 159.1
KANNO OJS3-0 0.89 0.76 253.0 594.3 689.2 838.0 649400 1588 284.2 273.1 361.2
KANNO OJS4-1 0.83 0.70 253.0 594.3 689.2 838.0 649400 1588 284.2 273.1 361.2
KANNO OJS5-0 0.75 0.81 161.4 562.6 733.9 683.1 837700 2250 284.2 252.5 406.8
KANNO OJS6-0 1.01 0.71 253.0 562.6 733.9 892.3 709800 1800 284.2 283.9 391.6
KANNO 0JS7-0 0.80 0.67 253.0 562.6 733.9 892.3 709800 1800 284.2 283.9 391.6
KANNO HJS1-0 1.04 0.87 253.0 594.3 1067.0 1297.0 1080000 3802 284.2 386.8 586.4
KANNO HJS2-0 1.04 0.78 253.0 594.3 1067.0 1297.0 1080000 3802 284.2 386.8 586.4
DEIERLEIN 10 0.94 0.81 0 457.4 999.0 0 713800 2007 355.6 142.3 165.5
DEIERLEIN 11 1.16 0.90 194.7 916.8 999.0 704.0 1044000 2007 355.6 272.9 329.5
DEIERLEIN 13 1.01 0.81 152.4 457.4 1030.0 568.2 759200 2135 355.6 215.9 246.6
DEIERLEIN 15 1.01 0.80 137.5 457.4 933.0 464.4 622600 1751 355.6 193.4 202.0
DEIERLEIN 17 1.14 0.80 194.7 457.4 910.0 641.2 922900 1665 355.6 213.1 293.6
SHEIKH 4 0.86 0.75 0 431.2 955.5 0 653000 1836 355.6 134.8 151.3
SHEIKH 5 0.89 0.77 0 431.2 955.5 0 653000 1836 355.6 134.8 151.3
SHEIKH 7 0.66 0.78 0 431.2 887.5 0 563400 1584 355.6 128.0 130.5
SHEIKH 8 1.00 0.81 304.8 431.2 874.3 964.2 546600 1537 355.6 245.5 245.4
PARRA 1 0.86 0.94 138.8 680.3 545.6 274.3 592000 2114 280.0 138.4 300.0
PARRA 6 1.21 1.23 169.8 580.8 381.7 277.2 342000 1222 280.0 222.6 228.2
PARRA 9 1.08 0.80 228.0 487.8 377.8 368.4 418900 1496 280.0 226.9 285.9
61
Table 2.5 Results from Update Guidelines for the joint group failing in vertical bearing. (Units: kN, mm)
Updated ASCE
Name bo Vspn Vicn Vohn Mvb Ccn x Vb,PanelShear Vb,Bearing
Vmeas/Vpred Vmeas/Vpred
KANNO OJB1-0 0.89 0.87 151.7 1961 736.9 537.1 515800 1815 284.2 404.8 271.0
KANNO OJB4-0 0.78 0.76 151.7 1961 735.4 536.0 513700 1808 284.2 404.4 270.0
KANNO OJB5-0 0.90 0.87 160.5 1961 716.2 552.4 626800 1714 284.2 404.6 313.8
KANNO OJB6-1 0.82 0.80 151.7 1961 716.2 522.0 487200 1714 284.2 399.9 258.2
SHEIKH 2 0.60 0.60 0.0 500.8 325.6 0.0 151600 568.6 266.7 108.9 69.6
SHEIKH 6 0.63 NaN 0 431.2 921.6 0 607400 1708 355.6 131.4 140.7
Table 2.6 Summary of values required to compute phi-factor using the beta-reliability index.
Failure Recommended
- value Rm/Rn VM VF VP VR
Mechanism
Panel Shear 3.0 1.1*(0.96) -1 = 1.15 0.10 0.05 0.138 0.178 0.84 0.85
0.08 0.137 0.85

Vertical Bearing 4.5 1.1*(0.92)-1 = 1.20 0.10 0.05 0.75
(0.138) (0.178) (0.77)
62
Steel
Beams
Steel Erection
Column
Formwork Reinforcing
Bar Cage
RC Column
Figure 2.1 Typical construction sequence of cast-in-place composite RCS moment

frames.
Rebars Steel Beam
Beam Splice
Band Plate
RC Column
Grouted Splices Bolted Beam

Splices
Precast Beam-Column Module

Figure 2.2 RCS precast construction utilizing beam-column modules and column and
beam spliced connection.
63
Figure 2.3 Example of precast RCS construction in Japan (Shimizu Corporation).
Figure 2.4 Example of cast-in-place RCS system that replaces steel erection columns
with stiffened reinforcing bar cages. (Shimizu Corporation)
64
Longitudinal
Reinforcement
Stiffener
Steel Band
Plate
Steel
Beam
Joint Ties
Face Bearing
Plate Reinforced
Concrete Column
Figure 2.5 Connection between steel beam and reinforced concrete column
Figure 2.6 Schematic diagrams of RCS joint details tested.
65
concrete steel web
crushing yielding
gap
concrete
strut
crushing
Figure 2.7 Typical failure modes in RCS beam-column joints (Kanno et al. 2000).
500
Kanno 1993
400
OJB3-0
300
200
Beam Shear (kN)
100
-100
-200
-300
-400
-500
-0. 1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1
Drift Angle (rad)
Figure 2.8 Typical hysteretic response of a RCS beam-column test failing in joint
bearing failure (Kanno 1993).
66
150
Kanno 1993
OJS2-0
100
50
Beam Shear (kN)
-50
-100
-150
-0. 1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1
Drift Angle (rad)
Figure 2.9 Typical hysteretic response of RCS beam-column test failing in joint shear
(Kanno 1993).
250
Kanno 1993
200
OB1-1
150
100
Beam Shear (kN)
50
-50
-100
-150
-200
-250
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
Drift Angle (rad)
Figure 2.10 Typical hysteretic response of RCS beam-column test failing in beam
hinging (Kanno 1993).
67
Steel Beams
RC Section
Transverse Ties Reinforcing Bars
Figure 2.11 Typical configuration of reinforcement bars in RC columns.
EQ Loads
SCWB Interior Axial 1.2D+0.5L+1.0E
Pbal
Min. Flexural
Strength
Moment
SCWB Ext SCWB Ext 0.9D+1.0E
Min. Flex.
Strength
(a) (b)
Figure 2.12 (a) Illustration showing the separation of int./ext. columns for the SCWB
criterion and (b) the appropriate load factors with respect to the P-M column curve in
order to obtain the lowest flexural strength.
Mp,beam
Mp,col M p , col M p ,beam
Figure 2.13 SEOAC Blue Book strong-column weak-beam provisions.
68
Rein. Bar
High Strength
Mortar
Column
Top Sleeve
Flow Out Flow Out
Flow Out Flow In
Column
Figure 2.14 Details of grouted splice connections for precast RC columns.
(P,M,V)C1
Lateral loading:
sway condition
P
M+
M+
(P,M,V)B2
A
A
(P,M,V)B1
P/2 P/2
M- Simply supported beam:
gravity mid-span condition
(P,M,V)C2
Slab stress concentrated

over column width bcol
beff
Figure 2.15 Differences between the definitions of effective slab width considering
lateral versus gravity loading.
69
Flange Plates
Shear Tab
Figure 2.16 Schematic of a typical bolted flange plate beam splice connection
1.1RyMp+
Lateral load
moment diagram RC Column
xsplice
db/2 db/2
Lb, clear - db
Theoretical 1.1RyMp-
hcol center of hinge
(a)
Gravity load
moment diagram
(b)
+
1.1RyMp
Gravity + lateral
moment diagram
1.1RyMp-
(c)
Figure 2.17 Assumed beam moment diagram from the lateral and gravity loads with
respect to beam splice location.
70
db
db
xsplice-db/2
xsplice
Figure 2.18 Assumed region of severe hinging and location of beam splice.
dslab Cslab = Fplates

ddeck
neglected
Fplates = AplatesFy
Figure 2.19 Cross-section of the beam splice plates and the concrete slab depicting the
force couple between the lower plates and slab.
1.9
W21-W44 Beams
1.8 Steel Deck (3 in)
Slab Depth (3 in)
1.7 Slab Width (60 in)
1.1*R y*F y (60.5 ksi)
1.6 0.85f'c (3.4 ksi)
M p,Comp/Mp,Steel
1.5
1.4
1.3
1.2
1.1
1
0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Location of PNA from Bottom (%db)
Figure 2.20 Plot of the location of the plastic neutral axis versus the ratio of composite
to steel beam strength, 76mm deck and slab. (1in = 25.4mm)
71
1.2
1
(0.045, 1)
Normalized Joint Load (0.023, 0.95)
0.8
(0.008, 0.7)
0.6
0.4
(0.0015, 0.3)
0.2
Total Joint Distortion
Vertical Bearing Distortion
Panel Shear Distortion
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
Distortion (rad)
Figure 2.21 Normalized load versus joint distortion response
(Parra-Montesinos et al., 2001)
Steel Band
Plate
Transverse
Beam
Steel
Beam
Face Bearing
Plate Reinforced
Concrete Column Cover Plate
Figure 2.22 Joint detail showing the band plate, cover plate, and the transverse beam.
72
hc
dscol
bo/2
bc bf y = bscol
3
1
bo/2
x
(a) Steel column
x = hc
bo/2
y
bc
3
bo/2 1
(b) E-FBP
bo/2 6tbp
bc
bo/2 6tbp
(c) Steel band plate
Figure 2.23 Definition of outer strut width
73
V bc/V be
0
0.2
0.4
0.6
0.8
1
1.2
1.4
KANNO OJS1-1 750 mm
KANNO 0JS2-0
266 mm
KANNO OJS3-0
ASCE
KANNO OJS4-1
Updated
KANNO OJS5-0
KANNO OJS6-0
KANNO 0JS7-0
Joint Panel Shear Failure

KANNO HJS1-0
202 mm
W 250 X 49
KANNO HJS2-0
DEIERLEIN 10
74
DEIERLEIN 11
model
DEIERLEIN 13
750 mm
250 mm
DEIERLEIN 15
DEIERLEIN 17
SHEIKH 4
16-35 mm bars
Figure 2.24 Joint design example.
SHEIKH 5
SHEIKH 7
SHEIKH 8
PARRA 1
PARRA 6
Mean = 0.96
COV = 0.138
PARRA 9
Figure 2.25 Predicted versus measured joint shear strength for both ASCE and Updated
1.4
Joint Bearing Failure

1.2
0.8
V bc/V be
0.6
0.4
0.2 Mean = 0.92

ASCE
Updated COV = 0.08
0
KANNO OJB1-0
KANNO OJB4-0
KANNO OJB5-0
KANNO OJB6-1
SHEIKH 2
SHEIKH 6
Figure 2.26 Predicted versus measured joint bearing strength for both ASCE and
Updated model
1.8
W21-W44 Beams
Plastic Strength of Composite Beam (%M p,steel)
1.7 Steel Deck (3 in)

Slab Depth (2.5 in)
Column Width (30 in)
1.6 1.1*Ry*F y (60.5 ksi)
1.3f'c (5.2 ksi)
1.5
1.4
1.3
1.2
1.1
1
20 25 30 35 40 45
Depth of Beam (in)
Figure 2.27 Ratio of the composite to the bare steel beam strength for typical W-section
beams (1in = 25.4mm, 1ksi = 6.89MPa).
75
Chapter 3: Full-Scale Composite RCS Test Frame
This chapter reviews the background, preparation and design, and execution and results
of a full-scale pseudo-dynamic and quasi-static test of a 3-story composite RCS frame.
This testing program is a collaborative project between researchers at the National Center
for Research on Earthquake Engineering (Dr. Keh-Chyuan Tsai, Chui-Hsin Chen, and
Wen-Chi Lai) and Stanford University (the author and Dr. Greg Deierlein).
Only the relevant information from the testing program that are later used to draw
conclusions in this and in other chapters of this thesis are presented herein. This testing
program is discussed in a more detailed manner in Cordova et al (2006), which reviews
more of the intricacies of the test design, setup, instrumentation, and a more complete
discussion of the results. There are several aspects of the frame test, including the
preliminary analyses, selection of loading events, introduction of loads, etc., which are
discussed in Cordova et al. (2006) but are either condensed or removed in this chapter.
3.1 Background
Near the latter part of the US-Japan Cooperative Earthquake Engineering Research
Program on Composite and Hybrid Structures, it was widely acknowledged that testing
work and design implications studies on full structural systems needed to be undertaken
in order to validate the knowledge that had been gained through the program. This would
also provided the platform to study the 3-dimensional, indeterminate aspects of complete
structure behavior and also allow the investigation of many practical aspects of composite
structures, such as economy and constructability. In the past, there have been a number
of full or near full scale structural tests (Foutch et al 1988, Roeder et al 1988, Lee and Lu,
1990, Atkan and Bertero, 1987, Liang and Ding, 1995) and each of these have uncovered
many nuances and provided a number of insights about the structural system (i.e.
unexpected failures, additional strength capacity, etc.) that could not have been found
through simple subassembly tests.
76
In 2001, researchers from the National Center for Research in Earthquake Engineering
(NCREE) in Taipei, Taiwan and Stanford University began collaboration on developing
the plans for a full-scale composite RCS moment frame to be tested in the NCREE
laboratory. The test specimen is a three-story, three-bay composite RCS moment frame,
consisting of reinforced concrete (RC) columns and composite steel (S) beams.
Measuring 12 meters tall and 21 meters long, the frame is among the largest frame tests
of its type ever conducted.
The three-story prototype structure is designed for a highly seismic location either in
California or Taiwan, following the provisions for composite structures in the
International Building Code 2000. The frame is loaded pseudo-dynamically using input
ground motions from the 1999 Chi-Chi and 1989 Loma Prieta earthquakes, scaled to
represent 50%, 10%, and 2% in 50years seismic hazard levels. Following the pseudo-
dynamic tests, quasi-static loads are applied to push the frame to interstory story drifts up
to 10 percent, which provides valuable data to validate simulation models for large
deformation response.
The test has three primary objectives. First, it provides data to evaluate and validate
design provisions for composite moment frames. Particular topics of investigation
include strong-column weak-beam criterion, composite action of concrete slab and steel
beams, integrity of the pre-cast column and composite beam-column connections, and
overall frame response. Second, the test provides valuable information to validate models
and computer codes for nonlinear simulation and performance assessment. Finally, the
full-scale test provides validation to support the use of innovative composite moment
frames as alternatives to conventional steel and concrete systems for high-seismic
regions. Apart from these direct benefits, the frame provides the impetus to explore
international collaboration and data archiving envisioned for the NEES initiative.
77
3.2 Rationale for Full-Scale RCS Frame Test
As mentioned in Chapter 2, the concept for the RCS moment frame system has been
around in some form for nearly thirty years. A majority of its implementation in the
United States has been limited to high-rise buildings in low seismic regions. Engineers
are quick to take advantage of the many benefits of composite framing in high-rise
applications, which range from optimal material usage to expediting construction time
with innovative staging or precast techniques. Unfortunately, its use in high seismic
zones has been limited in the United States despite the fact this system is widely
implemented in Japan. There has been research (Kanno 1993, Mehanny 2000, Liang et
al. 2004) that has shown RCS systems to be equivalent, if not superior, to the seismic
behavior of all-steel moment frames. As discussed in Chapter 2, the seismic design
provisions are currently available to implement a code-based design of an RCS moment
frame. It is believed that a full-scale test has the potential to be a proof of concept for
this structural system. The design, construction, and ultimately the testing and
performance of this frame should be viewed as a complete example of this system from
start to finish. It is hoped that with this testing program, a majority of the issues
concerning this system can be resolved in some way to ensure that the engineering
community will be able to accept this innovative system as a viable alternative to the
conventional steel or concrete moment frames.
Table 3.1 lists some of the issues that that are investigated within this testing program, as
well as those that are neglected. These issues will be further developed and discussed
throughout this remainder of this Chapter (and also in Chapter 4). Obviously this is not
an exhaustive list, and hopefully this will continue to grow as researchers re-examine this
data in the future.
3.3 Design of Full-Scale Test Frame
The loading and design of this test frame is meant to be as realistic as possible to ensure
that the experiment produces meaningful data that can be applied to real world structures.
This achievement of realism must also be balanced with laboratory limitations and
78
constraints, some of which include the available lab space, actuator capacities, and
funding. With this in mind, this section presents the general overview of the design of
the test structure. In addition to this, some of the important design parameters and their
impact of the behavior of the frame are discussed in further detail. Refer to Cordova et
al. (2006) for further information on the design of this test frame.
3.3.1 Plan and Layout of Test Frame
The size of the test frame is restricted by the 15-meter tall strong wall, the 1-meter
spacing of the tie-down holes on the strong floor, and general space limitations due to
other experiments in the NCREE lab. Based on these considerations, the overall plan and
profile is based on a typical rectangular 3-story office building with a 6 by 4 grid of 7-
meter bays and 4-meter story heights. Two of the seven short direction frames are
considered in the lateral resisting system, with the rest acting only as gravity resisting
frames. The plan and location of the RCS moment frames are shown in Fig. 3.1. The
lateral load resisting frames were configured as interior moment frames in order to
provide a symmetric frame (i.e. when the slab is equal on each side of the frame).
The test specimen, shown in Fig. 3.2, is a three-story, three-bay frame meant to represent
one of the two RCS frames within this theoretical plan. In order to capture the composite
slab effect in the behavior, a 2150mm wide slab is integrated with the floor beams on
each level of the frame. To help support this slab as well as provide restraint for the main
beam against lateral torsional buckling, transverse beams are provided at the third points
along the span of the beams. Loading beams are provided at the north end of the frame
and along each edge (east and west) of the slab. The force transfer mechanism between
the actuators and the frame is described in detail in Section 3.4.1.
79
3.3.2 General Design Information
The test frame is designed for a highly seismic region in either California or Taiwan,
following the provisions for composite structures previously outlined in Chapter 2 1 .
Seismic design forces are based on mapped spectral accelerations of Ss = 1.5g and S1 =
0.72g. The soil condition at the building site is assumed to be that of site class D (Fa =
1.0), while the building itself is assigned a Seismic Use Group I and a Seismic Design
Category D according to IBC/ASCE 7-02. The SAP2000 model of the frame calculated
the fundamental period to be 1.0 seconds, while the code-defined period is only 0.47
seconds, with an upper period cap of 0.56 seconds 2 . Given this information, the design
base shear coefficient defined in equation (2.1), including the additional shear to account
for accidental torsion, is computed to be V W = 0.13 , which equates to a required design
base shear of 1160kN. A summary of the design loading, including the dead and live
loads for a typical office building, is presented in Table 3.2.
The frame was intentionally designed to the minimum limits of the building code so as to
represent the minimum expected performance and to interrogate system design
parameters. This approach was also necessary to ensure that the strength of the frame did
not exceed the capabilities of the laboratory. As Taiwans seismic design codes adopt
similar requirements to those in the United States, the frame is equally representative of
design standards in both countries. Composite beams were sized based on strength
requirements in negative bending (1.2DL+0.5LL+1.0E), while the reinforced concrete
columns were controlled by the strong-column weak-beam (SCWB) criterion specified in
Equation (2.47). This SCWB criterion was one of the key design parameters in the frame
1
Given that the test specimen was designed during the years of 2001 and 2002, the prevailing building
code was the IBC 2000, which referenced the 1997 AISC Seismic Provisions, ACI 318-99, and the AISC-
LRFD (1997). It can be verified that the more current design provisions (i.e. IBC 2003, etc.), as discussed
in Chapter 2, would result in a nearly identical frame considering that the changes to the codes that apply to
this design have been relatively minor. Therefore, unless otherwise noted, the reader is to assume that the
applicable design provisions are those specified in Chapter 2.
2
The approximate fundamental period equation changed slightly from IBC 2000 to 2003, but this did not
factor into the final design forces since both periods were in the constant acceleration segment of the design
spectra.
80
and will be discussed further in Section 3.3.3. The drift criterion from the ASCE 7-02,
specified in equation (2.4), is automatically satisfied and did not require any resizing of
the beams or columns. The final member sizes are presented in Table 3.3. It should be
mentioned here that the 1st floor beam was originally sized to be 596x199x10x15mm, but
given its lack of availability during construction, a larger beam was substituted
(600x200x11x17mm), which increased the moment capacity of the beam by
approximately 15.5%. This change will have implications in both the SCWB ratios and
the design of the beam splices.
The 1st and 2nd-floor composite joints were designed based on the 1994 ASCE Joint
Guidelines, with some modifications from the work of Kanno (2000) to account for some
of the details not considered in the older guidelines. The joints were designed to develop
the nominal moment in the adjacent composite beams following the strong-joint weak-
beam criterion of the AISC Seismic Provisions (2005). Joint details include face bearing
plates, a steel band plate above and below the beam, and transverse beams. Since space
frames are more common in Japan and Taiwan, it was decided to incorporate a transverse
beam into the joints to simulate space frame conditions. This detail provides additional
concrete confinement within the joint region and increases the strength and stiffness of
this connection (Kanno 1994). Based on subassembly tests prior to the testing of the full-
scale frame (Cheng 2002, Cordova et al. 2006), it was decided to construct the joint
without complete hoops and simply provide small ties for longitudinal bar support
through the joint, as shown in Fig. 3.3. These ties were considered to be sufficiently
anchored into the concrete that is confined by the beams and face bearing plates. Note
that these ties are not considered effective in resisting joint shear. The design of the 3rd-
floor joints required special details to anchor the longitudinal column bars, as will be
discussed in Section 3.3.7.
3.3.3 Strong-Column Weak-Beam
An important design provision, which often controls the column sizes in special moment
frames, is the strong-column weak-beam (SCWB) criterion. As discussed in Chapter 2,
81
the nominal column to beam strength ratio is generally assumed to be approximately 1.2
(ACI 318-02, ASCI Seismic Provisions 2005). Since the intent of the frame test was to
push the limits of current code provisions, we took a liberal interpretation of the SCWB
criterion with respect to whether the calculation is made using properties of the steel or
composite beam for the SCWB check. Differences can be significant, since the
difference in nominal strengths between the steel and composite beam assumption is
about 30%. The calculated strength ratios using the expected steel beam strengths
(1.1RyMp= 1.21Mp) to the nominal strengths of the columns all equal or exceed the
specified ratio of 1.0 (1.0 passes SCWB criterion), as shown in Fig. 3.4a. However,
when calculated based on composite beam strengths according to equation (2.47), several
of the joints (shown in dashed boxes) violate the 1.0 limit (Fig. 3.4b). The values shown
in Fig. 3.4b are based on the assumption that beams flexed in positive bending will act
compositely and those in negative bending will act as bare steel beams. Note that the 1st
floor interior joint ratios are approximately 15% below the required SCWB ratio, which
stems from the substitution of the original 596x199x10x15mm to the 600x200x11x17mm
beam during construction, previously discussed in Section 3.3.2. This last minute change
in beam size has the potential to greatly impact the behavior of the frame, but fortunately,
the measured properties of the beam and column helped mitigate this problem. Figure
3.4c summarizes the SCWB checks based on composite beam behavior using measured,
as opposed to nominal, material strengths of the beams (RyFy,nom = Fy,meas) and columns
( f c', Meas and Fy,rebar,meas). Being as the main intent of the SCWB criterion is to avoid story
mechanisms, the fact that one joint in a story violates the criterion is not necessarily
detrimental to the frame behavior. For example, one could interpret the values for the
first and second floors in Fig. 1b as meeting the intent of the SCWB criterion, since the
average ratio for each of these floors exceeds 1.0. However, as will be seen later, the
SCWB ratios alone do not provide an accurate gage of where inelastic deformations will
occur under earthquakes.
82
3.3.4 Measured Material Strengths
The material properties of the beam and reinforcement steel were measured prior to the
frame test. The results of these tensile tests are shown in Table 3.4. It was surprising to
learn of the high amount of overstrength in the steel beams, especially in the 2nd floor,
which was found to be 40-50% higher than nominal properties. The 1997 AISC Seismic
Provisions specifies that the expected yield strength of grade A50 steel should be
assumed as RyFy, or 1.1Fy. This expected yield strength was obviously exceeded in all of
the steel beams, which is undesirable considering that we are attempting to control the
failure mechanism of the frame through the SCWB criterion.
The measured strengths of the concrete cylinders were obtained on the first day of testing,
the results of which are shown in Table 3.5. Again, the measured strength properties are
much greater than the nominal concrete properties, with an average increase in strength
of 67% for the columns above the splice at the 1st floor. The 1-meter column stubs below
this splice turned out to have a measured strength approximately 115% higher than the
nominal. Given the low axial load in these RC columns, the strength of these members is
largely controlled by the reinforcement bars. Therefore, these large differences in
nominal to measured concrete strength only resulted in approximately a 10% increase in
the moment capacity.
While some material overstrengths were anticipated in the design (e.g. 10-25% in steel),
the large differences discussed here were not expected. Fortunately, the balance between
the column and beam strengths remained roughly intact and the SCWB ratios were not
severely affected, as reviewed in section 3.3.3.
3.3.5 Beam and Column Splices
Given the popularity of the precast RCS construction in Taiwan and Japan, this method
was chosen for the erection of the test specimen. In discussion with industry engineers
and contractors in the United States, the precast method garnered the most interest based
on the ability to prefabricate these elements and expedite the construction process. This
83
method of construction requires splices in both the steel beams and the RC columns. In
the test specimen, the steel beams are spliced 1500 mm away from the face of the
column. As discussed in Chapter 2, this location is a compromise between competing
desires to minimize the offset for ease of construction (fabrication and shipping of the
precast column-beam assemblies) and maximizing the offset so as to reduce the design
moment for the splice. In Chapter 2, it is recommended that the beam splice be designed
to develop the expected plastic moment of the steel section ( 1.1Ry M p ); however, the
method implemented in the test frame was slightly different. The beam splice was
originally designed based on requirements interpreted from the FEMA 350 design of
bolted flange plate moment connections. FEMA 350 requires that the flange plate is able
to develop the expected yield moment of the beam at the column face
( M yf = R y Fy S = (1.1) Fy S ), which means that yielding is expected in the flange plates
and is part of the preferred mechanism. The reason for the difference in the design
philosophy from Chapter 2 is really due to the fact that the appropriate design
methodology of the beam splices was still being considered at the time and the frame test
provided a means to test out an alternative design option.
In accordance with U.S. construction practice and the AISC Seismic Provisions, the
connection is proportioned based on bearing strength of the bolts, as opposed to slip
critical values. As will be discussed later in this Chapter, this design practice resulted in
slipping of the bolts during all of the pseudo-dynamic tests. The main consequence of the
slipping was loud bolt banging during the test; otherwise, the bolt slippage did not have
any appreciable effect on the overall frame behavior. Figure 3.5 shows a typical detail of
one of these bolted beam splices. Unfortunately, the last minute resizing of the 1st floor
beams was not accounted for in the design of the beam splice, and as a result this
imposed particularly high demands on the splice plates. As discussed later, this led to
rupture of the 1st floor connection splice plates during the final pushover test of the frame.
Precast column splices are located 1-meter above the foundation footing and directly
above the connection band plates (at the top of slab) at floors one and two. The influence
of the splice location on performance at the column base was examined through
84
subassembly tests, conducted prior to the frame study (Cheng 2002, Chen 2002, Cordova
et al. 2006). Two splice locations were investigated in these tests the first was flush
against the column-footing interface (Fig. 3.6a) and the second was 1-meter up the height
of the column (Fig. 3.6b). Both tests showed that the precast connection could develop
and maintain the full column strength through large inelastic deformations (Fig. 3.7a),
although the specimen with the splice adjacent to the footing experienced more strength
degradation at high drifts (Fig. 3.7b). In the end, due to the critical nature of the 1st floor
column hinges, it was decided to place the splice at the 1-meter location. The splice
location for 2nd and 3rd floor columns is less critical because analysis studies have shown
that severe column hinging is not likely to occur above the floor beam except in
instances where the SCWB criterion is not met and a severe story mechanism could
occur. Since it is common practice in Japan to splice the column right above the floor,
and since the intent of the frame test was to investigate the limits of performance relative
to standard practice, it was decided to locate the grouted splices directly above the beam-
column joints.
3.3.6 Shear Studs in Hinge Region
The AISC Seismic provisions and FEMA 350 does not permit the placement of shear
studs within the expected plastic beam hinge region, which is assumed to extend from the
column face to one-half beam depth away from the column. The concern is that the shear
studs will lead to strain localization and premature beam flange fracture during plastic
hinging. During construction of the test frame, the contractor inadvertently placed shear
studs within the hinge region of the 2nd and 3rd-floor beams. This construction mistake
was noted at the time, but it was decided against trying to remove the studs due to
concerns about further damaging the beam flange. As it turned out, the studs in these
regions did not lead to any detrimental behavior of the hinge, which is largely due to the
fact that the upper flange was protected against large inelastic action due to the presence
of the composite slab (i.e. shift of the neutral axis resulted in less strains in the upper
flange).
85
3.3.7 Roof Joints
The 1994 ASCE Joint Recommendations do not provide explicit guidance for designing
the roof joint connection (i.e. where the column does not extend above the joint).
Therefore, it was necessary to develop and evaluate a unique detail for this condition to
ensure adequate force transfer from the beam to the column. Three alternative joint
details were proposed and investigated through subassembly tests conducted prior to the
frame test (Tsai et al., 2002). The design that was selected, shown in Fig. 3.8, employs a
steel cap plate on top of the column to provide anchorage for the longitudinal reinforcing
bars and bearing force for the steel beam. To make the assembly of this joint easier, the
contractor preferred to extend the cap plate to cover the entire top face of the column.
The longitudinal reinforcement bars are plug welded to this bearing plate, as shown in
Figs. 3.9 through 3.11. This joint is assumed to be able to provide the full amount of
bearing resistance similar to that of a typical interior joint. As described later, like all the
joints in the frame, these top details worked very well and were able to develop beam
hinging adjacent to the column with only minimal damage in the joint region.
3.4 Description of Test
3.4.1 Test Setup
Three 1000kN-capacity actuators with a stroke limit of 500mm are provided at each
floor and are connected to the transverse loading beam as shown in Fig. 3.2. The purpose
of the transverse loading beam is to distribute the load from the actuators equally to both
the longitudinal loading beams. The longitudinal loading beams are cast integrally with
the slab and connected via shear studs interwoven with spiral hoop reinforcement. These
studs and reinforcement help transfer the load from the loading beam into the slab. The
assumption here is that the load transfer from the loading beams to the slab occurs at a
relatively uniform distribution along the entire length of the frame. This was ultimately
proved a valid assumption based on strain measurements in the loading beams during the
test (Cordova et al. 2006). Figure 3.12 depicts the probable load path between the
86
loading beams and the moment frame. When a load is applied to the loading beams the
force enters the slab and is distributed to the frame through two main load paths. The
first path requires development of a compression strut between the columns and the
loading beams. This path is mobilized through direct bearing of the column face,
compression of the slab and reinforcement, and the shear force transferred by the loading
beam studs. The second path develops a compression field through the slab between the
main beam shear studs and the loading beam shear studs. Both of these loading paths are
shown in Fig. 3.12.
An interesting result of the load transfer from the slab into the frame is the fact that it
alleviates the axial stresses induced in the slab due to positive and negative composite
bending moments. When the beams are in positive bending (i.e. in composite action) the
load transfer stresses tend to pull the slab away from the column while the opposite is
true when the beam is in negative bending. This mechanism is thought to represent the
conditions present in real world structures given that we are simulating a realistic load
path by imposing loads directly into the slab. The implications of this behavior will be
discussed later with the results of the frame test.
3.4.2 Loading Protocol
Extensive nonlinear static and dynamic time-history analyses were performed using the
analysis program OpenSees (McKenna et al. 1999) prior to testing. These analyses were
used to select suitable records from a suite of ground motions and predict the ultimate
lateral strength and the inelastic demands imposed on the frame specimen under the
simulated earthquake effects. The main focus of these studies was to predict the possible
peak responses of the frame during the test while verifying that both the force and stroke
limitations of the actuators were not exceeded. These analyses are similar to the post-
testing analyses presented in Chapter 4 and will not be presented here. For further
information on these analyses and the procedure to select the records, please refer to
Cordova (2006).
87
Ultimately, two earthquake records, TCU082-EW and LP89G04-NS, were chosen from
bins of the 1999 Chi-Chi and 1989 Loma Prieta earthquake records, respectively. The
analytical model predicted that the peak responses from these events would be within the
limitation of the actuators and suggested that the hinging was fairly well distributed in the
frame.
Using the pseudo-dynamic loading methodology described in Cordova (2006), these

records are inputted as a series of four earthquake loading events, which are defined as
follows:
1. 50% chance of exceeding in 50 years (50/50) using the 1999 Chi-Chi (TCU082)
record ( Sa (T1 ) =0.408g): Immediate occupancy or service event
2. 10/50 1989 Loma Prieta (LP89G04) event ( Sa (T1 ) =0.68g): Design level event
3. 2/50 TCU082 event ( Sa (T1 ) =0.92g): Maximum considered event
4. Repeat of the 10/50 LP89G04. ( Sa (T1 ) =0.68g)
The scaled response spectrums for these events are shown in Fig. 3.13. Following the
four earthquakes, a final static pushover using a triangular loading pattern is applied to a
maximum roof drift ratio of 8%.
For the pseudo-dynamic loadings, the records are scaled based on the spectral
acceleration at the first mode period of the building to represent the range of different
hazard levels. The natural period of the frame is determined by assuming cracked
stiffnesses of RC columns and composite beams and realistic joint size and stiffness.
This period was later validated in the test frame by using the actuators to determine the
stiffness matrix via the flexibility method. One important distinction to make here is that
these hazard levels reflect a highly seismic Taiwanese site. The 10/50 event
(Sa(T1)=0.68g) matches that of the US site, but the 2/50 Taiwan event (Sa(T1)=0.92g) is
slightly less intense and corresponds to approximately a 4/50 hazard level for the US site.
This is apparent from the differences between the 2/50 IBC2000 hazard curve and the
2/50 TCU082 response spectrum curve in Fig. 3.13 at a period of 1 second. The seismic
mass used in the pseudo-dynamic algorithm is consistent with that used in the design, as
88
are the gravity loads (1.0DL+0.25LL, approximately 3500kN at each floor) included into
the geometric stiffness (P-) effects.
3.5 Construction of Test Frame
In addition to gaining insight into the performance of these composite RCS frames, the
frame test provided the opportunity to investigate the constructability issues regarding
these innovative systems. As mentioned before, the test frame was detailed and built in
the precast method of construction in conformance with standard industry practice. The
strength and durability of these precast spliced connections are issues that have not been
widely investigated for RCS systems prior to this testing program.
At the precast fabrication plant, single story precast column assemblies were integrally
cast with steel beam stubs to create a beam-column module that can be seen in Fig. 3.14a.
This process was optimized and repeated to generate the beam-column modules for the
entire frame. The beam stubs are pre-drilled and fabricated to accommodate the bolted
splice connections to connect to the central beam spans. The lower region of the precast
column is outfitted with standard mechanical couplers for the grouted splice connections.
The upper region contains the embedded steel beam as well as the face bearing plates,
band plates, and gravity beam as shown in Fig. 2.22. Longitudinal reinforcing bars
protrude beyond the top of the column to allow for the grouted connection of the upper
column. The foundation-column stub assemblies were prefabricated in the same manner.
To begin the construction process, the foundation-column assemblies were first placed
and tied down to the strong floor. This then allowed the workers to begin to build up the
test specimen floor by floor. The beam-column modules were shipped to NCREE and
assembled in the following process:
1. Beam-column modules are fitted into the lower columns (or footings) and
propped up with lateral bracing.
2. The center spans of the steel beams are placed between the beam-column trees,
and the beam splices are bolted into place.
89
3. The longitudinal loading beams are erected along the length of the frame and
attached to the embedded transverse beams protruding from the concrete columns.
The transverse loading beam is also attached to the northern end of frame between
the actuators and the east and west loading beams.
4. Transverse steel beams are then placed at the third points between the main span
beam and the bottom of the loading beam.
5. The story is plumbed to construction tolerances and steel bolts are tightened and
RC columns splices are grouted into place.
6. The steel deck is laid out and tack welded to the beams.
This process was repeated until all three floors were completed, with each floor requiring
about 2 days worth of work. Figure 3.14b is a photo of the finished first floor. Once the
3-story, 3-bay bare frame was constructed, workers attached the shear studs to the floor
beams, laid out the wire mesh reinforcement, and then poured the slab. The entire
specimen was constructed in the NCREE lab within two weeks from the positioning of
the precast footings to pouring of concrete slabs. A more complete photo documentation
of the construction process can be found in Cordova et al. (2006).
3.6 Test Results
The NCREE RCS frame test was conducted in October of 2002 and was witnessed by
about thirty researchers from the US, Taiwan, and Japan. Testing was broadcasted live
via the Internet (http://rcs.ncree.gov.tw), including real-time data plots and live video.
The response of the frame was monitored and documented with over 300 data channels,
visual inspections, and photographic images. Instrumentation was set up to measure the
hinge rotations in the RC columns and composite beams, composite joint rotations, slab
rebar strain, slip and lift off between slab and steel beam, elastic strains in beams, footing
slip, interstory drifts, and axial strain in the loading beams. A detailed description of the
instrumentation plan is withheld here and the reader is referred to Cordova et al. (2006)
for information regarding this matter of the test frame. The following subsections
summarize some of the key observations and data that have emerged from this testing
program that have direct applications to the performance assessment and seismic design
90
of composite RCS frames.
3.6.1 Global Results
The roof displacement histories for each of the pseudo-dynamic events are shown in Figs.
3.15 through 3.18. Maximum and minimum interstory drift ratios and story shears are
plotted for each earthquake in Figs. 3.19 and 3.20, respectively. During the 50/50
TCU082 event the roof experienced a maximum displacement of about 200mm, with
fairly uniform interstory drift values that ranged from 1.5 to 2.0%. A maximum base
shear of about 3000kN occurred during this event, which is about 2.6 times that of the
design value (1160kN). Residual drifts after this event were negligible. System
identification of the recorded time history show that the fundamental vibration period
elongated from its initial value of 1 second to 1.3 seconds after the 50/50 event,
indicating that the effective stiffness reduced to 60% of its initial value. This period shift
can be easily seen by comparing the elapsed time between displacement peaks in the
beginning and the end of the record in Fig. 3.15.
As can be seen in Fig. 3.16, the first design level event (10/50 LP89G04-1) had to be
stopped at about 7 seconds into the record and then restarted from the beginning. The
first segment of this event produced much larger excursions than the 50/50 event, and
began to cause hinging at the 1st-floor column bases. The frame initially has very little
stiffness in the transverse direction and the development of these hinges resulted in
further reduction of this stiffness. The external steel braced frame is designed to support
the RCS frame in the transverse direction, but a problem occurred during the setup of this
frame that did not permit a fully fixed connection to the strong floor. Therefore, when the
RCS frame required the support of the external frame during this event, the external
frame rocked sideways (in the transverse direction), causing out-of-plane moments at the
base of the RCS columns. This led to out-of-plane hinging and damage of these columns.
Ultimately, the frame rocked out to 1.5% roof drift in the transverse direction before the
external frame provided adequate restraint. Given that the frame had already experienced
the major excursions of the 10/50 event, it was decided to rerun the same event scaled to
91
80% of the original intensity. We felt that together, these two events would adequately
represent the intensity and damage of the original 10/50 (design level) event, had it been
run through without incident. Given this assumption, the maximum roof displacement
that occurred during this design level event was about 300mm, with maximum interstory
drifts ranging from 1.5% to 3.0%. The base shear was approximately 3800kN, which is
about 3.3 times larger than the design level event. The residual roof drift after this event
was approximately 0.3%. System identification on the response shows that the period
further elongated to approximately 1.5 seconds during the 10/50 event, indicating that the
frame was now at approximately 45% of its initial stiffness.
Under the maximum considered earthquake (2/50) the frame experienced a maximum
roof displacement of 500mm at about 28 seconds into the record. This displacement
exhausted the maximum stroke of the actuators at the roof, so that loading could not be
continued beyond this point. Examination of the pseudo-velocities and accelerations of
the frame at this time step suggested that the frame had reached its maximum drift and
was beginning to reverse direction. Analytical simulations showed that this was the
maximum excursion of the event and that subsequent cycles were smaller. Given these
observations, it was decided that this event did indeed represent a maximum considered
event (2/50) even though it was subjected to only 28 of the full 45 seconds of the event.
Deformations began to concentrate in the first two stories in this event, with a maximum
IDR of 5.5% occurring in the first floor. The maximum base shear remained about
3800kN, equivalent to the peak value attained in the design level (10/50) event. The
residual roof drift after this event was about 2.7%, with a largest contribution from the
first floor (3.4%). System identification on the response shows that the period elongated
to approximately 1.7 seconds after the 2/50 event, indicating that the frame was now at an
effective stiffness of 35% the initial value.
Given the large amount of residual drift after the 2/50 event, there was concern that the
frame would again hit the maximum actuator stroke during the final 10/50 event.
Therefore the decision was made to straighten the building as much as possible before
continuing with the next loading event. This realignment was accomplished using the
92
actuators and reduced the residual roof drift to approximately 0.3%. The final pseudo-
dynamic event was then executed, which involved re-running the 10/50 design level
event. Here the decision was made to repeat the scaled (80%) version of the 10/50 event,
so as to repeat the record used for loading the second segment of Fig. 3.16. Thus, this
event had two purposes. One was to help gage the amount of damage caused by the 2/50
(maximum considered) earthquake. The second was to simulate the possible effects of an
aftershock to the 2/50 event. The maximum roof drift during this record was about
300mm, with interstory drifts remaining within 3%. If the residual deformations are
removed from both the first and second scaled version of the 10/50 event, the transient
response can be directly compared, as is shown for the roof displacement in Fig. 3.21 and
the maximum interstory drifts in Fig. 3.22. Comparison of the roof displacement of the
two events show that the frame responds much differently in the second event given that
the stiffness of the frame has drastically changed from the damage incurred during the
previous events. Despite these differences throughout the time history, the maximum
interstory drift during each of the events was measured as 2.7%, although this occurs in
the 2nd floor in the first event and the 3rd floor in the second. The measured base shear
was considerably lower in this final event than in the previous three, with values peaking
at approximately 2200kN (58% of maximum seen in 2/50 event), presumably due to
softening under the multiple earthquakes. The residual roof drift was about 1.1%.
The final pushover of the frame resulted in a maximum base shear of 3200kN (Fig. 3.23),
which is about 2.8 times that of the design base shear. This reveals that even after
sustaining significant damage through four major earthquakes, the frame still maintained
a very large overstrength value. It was necessary to complete the pushover in two stages
so that the actuators could be readjusted to accommodate the large drifts. This is
apparent in the unloading-reloading sequence that occurs at about 4% drift. During this
test, deformations continued to concentrate in the first and second floors, creating a fairly
pronounced two-story mechanism, characterized by column hinging at the base and
beneath the second floor beams and flexural yielding of the first floor beams. Recall that
the second floor beams were found to have measured yield strengths 40% larger than the
specified minimum strength (484 MPa versus the minimum specified yield strength of
93
345 MPa), which led to limited yielding in the second floor beams and hinging in the
columns.
Figure 3.23 also shows the final pushover curve if the P- effects from the gravity loads
(1.0DL+0.25LL) from the tributary area of the moment frame and the interior gravity
resisting columns had been considered. This reduces the maximum base shear by
approximately 30% and also introduces a negative slope to the response curve at about
3.5% drift. Note that this response does not consider the resistance of the interior gravity
resisting frames and other sources of resistance, which would tend to strengthen and
stiffen the realistic response of the building.
3.6.2 Description of Damage
This section will outline some of the typical damage that was observed in the test frame
during each loading event. For the sake of brevity, photos and hysteretic response plots
will only be shown for selective components of the frame. The moment versus rotation
response plots shown in this section are interpreted from tiltmeters located within the
hinging zones of the beam and column components. The rotation in the 1st floor base
column hinges also includes the rotation due to bond slip between the column and the
footing. Joint outer rotation is obtained from pi-gauges on the outer panel, while total
joint rotation is inferred from adjacent tiltmeters in the columns and beams. Given the
indeterminacy of the frame, component forces were inferred from strain gauges in elastic
regions of the beams coupled with assumptions of the inflection point location from the
analytical models. For more complete information on details of the frame response,
derivation of component forces, and instrumentation results please refer to Cordova et al.
(2006).
The hysteretic response plots for selected columns, beams, and joints are shown for the
first four pseudo-dynamic events (50/50, 10/50-1a, 10/50-1b, 2/50) in Figs. 3.24 through
3.41 and will be referred to throughout this section. Corresponding photos of damage
incurred by these components are presented in Figs. 3.43 through 3.60. Photos depicting
94
the damage of components after the final quasi-static pushover are shown in Figs. 3.61-
3.69.
3.6.2.1 EQ#1 50% in 50 year
The 50/50 event, which represents the frequent or immediate occupancy event, caused
relatively little damage to the components of the frame. There were very minor hairline
flexural cracks observed in all of the reinforced concrete columns. These cracks were
most noticeable at the 1st floor where they were spaced at 150-200mm over the bottom
1m of the column. Figures 3.24 and 3.42 show a photo and the hysteretic response,
respectively, of an interior 1st floor column hinge (1C3) after this 50/50 event. The +
mark on Fig. 3.24 represents the estimated plastic moment of the RC section, indicating
the moment at which the base hinges are expected to have reached the yield moment of
the section. These base hinges reach rotations of up to 0.014 radians and begin to show
some small amounts of energy dissipation through hysteretic loops. The rest of the
columns in the frame remain relatively elastic during this event, which is to be expected.
The steel beams in all floors showed very minor yielding in the flanges, with the beams
framing into the exterior joints showing slightly more developed yielding than those into
the interior joints. Figures 3.28 and 3.46 show the photo and hysteretic response,
st
respectively, of a 1 floor beam framing into an exterior joint (1B1S). All of the
instrumented composite beams responding in a relatively elastic manner to this event,
with hinge rotations reaching a maximum of 0.007 radians in the 1st floor. The composite
slab showed relatively no observable damage after this event.
The interior joints of the first two floors experience the most deformation, with outer
joint distortions reaching approximately 0.4% and total distortion reaching a peak just
shy of 1%. While the estimated shears for these joints are just below their predicted
maximum strengths, there was very little observable damage to the composite joints after
this event, which suggests that these shear values may be overestimated. The hysteretic
response of a 1st floor interior joint (1J3) is shown in Fig. 3.34 with a corresponding
95
photo in Fig. 3.54. The remaining joints in the frame experience very little distortion and
remain relatively elastic during this event. Note that the total joint rotation may be
slightly overestimated in all the loading events given that the tiltmeter measurements are
taken from the surrounding beams (on FBP adjacent to joint) and columns (100mm above
and below the joints) and are likely to include some additional rotation from these
elements.
During this loading event, and throughout all subsequent loading events, there was loud
bolt banging associated with slippage of bolts in the beam splices. Aside from the loud
sound, the bolt slippage and banging did not detrimentally affect the frame.
Upon thorough inspection, researchers and engineers witnessing the test agreed that the
frame met the performance target for immediate occupancy, where the structural
stability was not compromised and the frame required little, if any, repair.
3.6.2.2 EQ#2 10% in 50 year
The 10/50 event, which is considered the design level event, caused moderate damage
throughout the frame. Crack widths near the base of the 1st floor columns opened to
about 2mm and were accompanied by some minor spalling of the cover concrete. Some
of the damage within these columns was caused by the 1.5% out of plane rotation due to
problems with the lateral restraint system. Figure 3.43 shows a photo of a damaged
interior column base (1C3) after segment 1b of the design level event. The hysteretic
response plots for this hinge during events 1a and 1b are shown in Figs. 3.25 and 3.26,
respectively, with hinge rotations reaching as high as 0.037 radians.
Minor cracks appeared in the upper portion of the 2nd floor columns, with crack spacing
between 100-300mm (see Fig. 3.58). In addition to these cracks, there was some minor
spalling below the steel band plates of the 2nd floor joints. Figures 3.39 and 3.40 show
the hysteretic response of a typical 2nd floor interior upper column hinge for both of the
segments of the 10/50 event. This shows that while the deformation demands are not
96
nearly as large as the 1st floor column base hinges, there are definite signs of yielding and
dissipation of hysteretic energy.
The steel beams in all floors yielded, although the second floor beams experienced much
less yielding than the other stories. Local buckles appeared in the first and third story
beams in the lower flanges and slightly into the web. The largest buckling distortions at
the flange tips measured up to 15 to 25 mm, which occurred at the first and third floor
beams framing into the exterior joints (i.e. exterior beam hinges). These hinges also
developed much more yielding than the interior framing hinges. The damage in both the
exterior beam hinges (1B1S and 1B3N) in the 1st floor can be seen in Figs. 3.47 and 3.50.
Beam hinge 1B1S reached rotation values of approximately 0.01 radians, with definite
signs of yielding and energy dissipation, as seen in Figs. 3.29 and 3.30. There was also
some slight yielding in the lower flange beam splice plates of the first floor beams (Fig.
3.53).
The design level event continues to push the interior joints in the first two floors to higher
distortions levels, reaching maximum distortions in joint 1J3 (Figs. 3.35 and 3.36) of
0.75% and 1.4% for the outer panel and total joint, respectively. The interior joints
experienced very minor cracking as shown in Figs. 3.57. The third floor joints, as well as
all of the exterior joints, are not pushed quite as far as these two interior joints of the first
two floors. This makes sense in the third floor considering that nearly all of the
deformation was concentrated in the plastification and local buckling of the steel beam
hinges. Similarly, the beams framing into the exterior joints were more susceptible to
local buckles and therefore the deformation was concentrated within this region.
The first and second floor slabs experienced some minor cracking along the length of the
frame, but in general the slab was still very much intact. At this stage, the building was
considered to meet the life safety performance target, where the level of damage
required repairs but had not significantly affected the safety of the structure. Repairs to
the structure would likely involve epoxy injection of cracks, patching of spalled concrete,
heat straightening of local flange buckles, and re-plumbing to reduce the residual
97
interstory drift.
3.6.2.3 EQ#3 2% in 50 year
The 2/50 event, considered the maximum credible event, produced the most significant
damage of all the pseudo-dynamic events. The base of the 1st floor columns experienced
the most severe hinging, with distributed cracks up to 4mm in width and significant
spalling of the cover concrete. In addition, large cracks measuring about 10mm opened
up between the bottom of the column and the footing - presumably due to yield
penetration of the longitudinal bars in the column footing. Figure 3.44 depicts the
damage on the interior 1st floor base column (1C3). The hysteretic response of this hinge
is shown in Fig. 3.27, with maximum rotations reaching 0.06 radians. Measurements
show that the majority of plastic deformation occurs within the first 650mm of the
column height, which is equal to the column depth. The assumption that the plastic hinge
length is equal to the column depth is a common one, and is validated through this data.
During this event, the second floor columns visibly began to experience more
deformation than in the previous events, particularly in the upper hinge zone. The
flexural cracks within these hinges grew to widths of 4mm and were accompanied by
minor spalling just below the beam-column joint, as shown in Fig. 3.59. Plots of the
interior 2nd floor upper hinge (2C3) in Fig. 3.41 are able to pick up this nonlinearity and
show that maximum rotations reached up to 0.02 radians.
First floor beams (1B1S) reached rotations of slightly past 0.06 radians in negative
bending, as shown in Fig. 3.31. This measurement is not completely accurate
considering that there were extreme local buckles occurring at this location (Fig. 3.48),
which may have influenced the tiltmeter readings. These local buckles were more severe
within the exterior beam hinges, concentrating in the bottom flanges and the lower half of
the web, with flange tip distortions up to 70 mm. The first and third floor beams
experienced extensive yielding, with flange yield penetrations reaching as far as 1.25
times the depth of the beam. Again, the second floor beams remained relatively elastic,
98
which can most probably be attributed to the large material overstrength in the steel of
this beam.
Measurements show that there was much more nonlinearity occurring in the beams
framing into exterior joints than those into interior joints (a portion of which is likely due
to measurement error given the more severe local buckling in the exterior beam hinge).
For example, compare the response of the exterior hinge in Fig. 3.31 to that of the interior
hinge in Fig. 3.35. This phenomenon was also apparent to the researchers present and is
shown Figs. 3.48 and 3.51 by the large difference in visible damage. This behavior can
be attributed to the difference between the continuity provided at the interior joints (i.e.
continuous slab and beam) versus the truncation of the beam and slab at the end of the
exterior joints. In the interior joint, the beam hinges on each face of the joint work
together to help restrain the rotation of the joint, which reduces the amount of rotation
demand in the beam and mitigates the amount of local buckling that occurs in the hinge
in negative bending. This behavior does not exist in the exterior joints, which allows for
larger local buckling in these beam hinges.
As shown in Fig. 3.37, the interior joints reach outer panel distortions reaching 1% and
total joint distortions measured up to 2.7%. This level of total distortion is too high given
the limited amount of damaged observed in this joint (Fig. 3.56) and may be a result of
problems within the tiltmeters (total joint rotation was inferred from the surrounding
beam and column tiltmeters and may include some rotation from these members as well).
There were some visible cracks within these joints, but nothing to suggest that the
structural integrity these connections had been compromised in any way. While the large
difference between the inner and outer joint panel may be partially due to measurement
errors, this also suggests that the outer panel zone may not be completely mobilized by
the joints details. This is likely a result of the presence of the transverse beams in the
joint that tends to divide the outer joint into two sections thereby reducing the
effectiveness of the outer panel strut. There was very little observable damage in the
remaining joints with measured joint distortions remaining well within 1%.
99
The first and second floor slabs experienced some local crushing on the interface of the
slab and the column, as well as cracking, which occurred on all three floors. With
significant local damage and a residual interstory drift of 3.4% (140 mm) in the first story,
the consensus of participants who witnessed the test was that the frame had reached its
collapse prevention performance level, implying that the structure would need
significant repairs to restore its strength and stiffness.
3.6.2.4 EQ#4 10% in 50 year
The post event inspection for the final design level event revealed little additional
damage beyond that observed after the maximum considered (2/50) event. The column
base hinges at the 1st floor were subjected to several large cycles that resulted in further
deterioration of the concrete, as is apparent in Fig. 3.45. There was a modest level of
shear and flexural cracks up the height of these 1st floor columns with some local spalling
beneath the joints steel band plates. The damage patterns within the upper hinges of the
2nd floor columns remained relatively the same as that after the 2/50 event, as depicted in
Fig. 3.60. The spread of plasticity in the steel beams remained within the limits of the
2/50 event, while local buckling continued in the 1st and 3rd floor beams (Fig. 3.49).
These buckles were concentrated in the lower flange and web, and tended to straighten
out once beam was pushed into a positive bending cycle, which can be easily seen by
comparing photos of the beam hinge 1B1S after the 10/50-2 event (Fig. 3.49) to the final
pushover (Fig. 3.68). Figure 3.52 shows the damage in an interior beam hinge (1B1N),
which consists of distributed yielding in the lower flange and up two-thirds the height of
the web. This particular hinge did experience some local buckling, but this had been
straightened out by the end of the event. The composite joints exhibited some cracking,
although this was more severe in the interior connections (Fig. 3.57).
3.6.2.5 Final Pushover
During the final static push the deformations concentrated in the lower two stories of the
frame, with a peak interstory drift ratio of 10% reached in the first story. By the
100
conclusion of the test, there was severe hinging at the column bases, with extensive
cracking, spalling, and crushing of the concrete. The concrete was fully deteriorated in
this region and the rebars were primarily acting in dowel action to carry the large amount
of shear transmitted through these columns. Figs. 3.62 and 3.63 show an exterior and
interior base column, respectively.
There was significant yielding and local buckling in the 1st -floor beams (Figs. 3.68 and
3.69) with beam lower flange distortions in the exterior joint reaching approximately 90
mm. The 1st floor slab experienced localized crushing and slight bulging around the
columns (Fig. 3.66). The upper hinges at the second floor columns experienced flexural
cracks up to 8 mm wide and developed a 10mm gap below the joints steel band plate.
Figures 3.64 and 3.65 show the damage in an exterior and interior 2nd floor column,
respectively. Yielding and local buckling of the steel beams dominated the behavior in
the 3rd floor, with very little damage occurring in the columns or the composite joints.
The final two-story mechanism is apparent in Fig. 3.61, which shows the frame near its
maximum drift state. The sudden strength drop apparent in Fig. 3.23 at a roof drift ratio
of 6% (corresponding to about 9% interstory drift in the first story) was caused by a net
section rupture in one of the lower beam flange splice plates for one of the 1st floor beam
splices (Fig. 3.67). This first rupture precipitated subsequent ruptures in neighboring
splices, which are evident in the strength drops under continued loading (see Fig. 3.23).
This failure can be traced back to the last minute change in the beam size that was not
accounted for in the beam splice design, which increased the moment capacity of the
beam by 15.5%. Nevertheless, the splice fracture is not considered a serious concern,
given that it only occurred after 10% story drift with significant distortions in the first-
floor framing.
3.6.3 General Observations
The concrete slab performed surprisingly well throughout the entire loading protocol. It
was expected that there would be much more concrete crushing and more severe cracking
101
that would compromise the integrity of the slab. Instrumentation confirmed that the slab
did not slip or pull off of the beam, and subsequent dismantling of the slab showed no
evidence of shear stud fracture. This is in stark contrast to many subassembly tests that
have described premature fracturing and shearing of the studs which ultimately leads to
rapid deterioration of composite beam action (Cheng 2002, Civjan 2001, Leon 1998,
Bugeja 2000).
Throughout the test, local buckling of the steel beam was concentrated in the lower beam
flanges and the lower portion of the beam webs. It was evident that the slab provided
considerable restraint to prevent distortions from developing in the upper flanges. The
yielding in the steel beams was most significant in the lower flange where yield
penetration extended out as far as 1.1 and 1.25 times the depth of the beam in the 1st and
3rd floor beams, respectively. Web yielding extended as far as 0.7 to 0.75 times the depth
of the beam near the bottom of the beam and then linearly tapered off up to the upper
flange. The damage in the upper flange was much less than the rest of the section and
showed signs of yielding out to 0.4 times the depth of the beam. The limited damage (i.e.
yielding and local buckling) in the upper flange is largely due to the shift in the neutral
axis due to the contribution of the concrete slab.
It was apparent that the most severe local buckling took place at the beam hinges framing
into the exterior joints. It seems that the continuity of the beam and the slab in the
interior joints provided additional restraint which did not allow the amount of buckling
that was evident at the exterior joints. In addition to this, it was common to see large
buckles form in the beam hinging zones during a negative (steel only) bending excursion,
only to be straightened out later when the earthquake produced an equal positive
(composite) bending excursion. These are particularly interesting issues because they do
not arise within most subassembly tests with steel beams, which points to the
shortcoming associated with proper modeling of boundary conditions in typical isolated
beam-column tests.
102
The composite joints behaved very well throughout all phases of the loading protocol.
Even through the final pushover, the composite joints, with standard details, experienced
only minor to moderate cracking and forced hinging to occur in the steel beams.
The RC column base hinges were subjected to significant plastic rotations throughout the
loading history of the test frame. As a result, the concrete within these hinges was
significantly deteriorated to the point that during the final pushover test the rebars were
carrying nearly all the moment and shear. This type of behavior seemed to be a result of a
combination of extensive plastic hinging and progressive shear failure. If the beam
splices had not failed, it is reasonable to state that the test would have been stopped
shortly thereafter due to the severity of damage within the column base hinges.
3.7 Conclusions
3.7.1 General Seismic Performance of RCS Frame
Even when designed to the minimum requirements of current building codes, the RCS
test frame showed excellent seismic behavior under various seismic hazards. Until the
final stages of the static pushover, the behavior and the distribution of damage in the RCS
frame was rather predictable with ductile hinging concentrating in the composite beams
(yielding, local buckling, and slab crushing) and the 1st-floor RC column bases
(distributed cracking, crushing, and spalling). The connections in the test frame has
shown superior ductility through drifts levels that far exceed the requirements of
conforming steel connections as specified by SAC Building Committee (2000).
Although there is not a general consensus on the link between the seismic hazard and
structural performance, there are quite a few documents (IBC 2003, FEMA 273/356) that
specify some broad requirements suggesting that the structure should exceed near-
collapse performance for 2% in 50 year hazard and protect life safety for 10% in 50 year
hazard. More specifically, documents such as FEMA 356 generally describe the
expected damage within particular elements (RC columns, steel beams) after the range of
103
seismic hazards investigated in this test program. The overall damage incurred by the
frame after each event, as described in Section 3.6.2, correlates well with FEMA 356
definitions of immediate occupancy (50/50), life safety (10/50), and collapse prevention
(2/50).
3.7.2 Frame Transient Drift
Current building codes, such as IBC 2003 and ASCE 7-02, require that the design story
drift shall not exceed 2.0% of the story height for typical moment frames with standard
interior walls and partitions. As stated in Section 3.3.2, the design of this frame adhered
to this drift criterion, but nevertheless, the drifts attained by the test specimen exceeded
those predicted by the code design procedure. As reported in Table 3.6, the test frame
reached a peak drift of 2% in the 2nd floor during the immediate occupancy event (50/50)
and 3% in the 1st floor during the design level event (10/50).
An argument can be made that the frame became more flexible after enduring the cycles
of the 50/50 event, which then had an impact on the drifts in the subsequent design level
event. As was reported in Section 3.6.1, it was determined that the fundamental vibration
period elongated from its initial value of 1 second to 1.3 seconds after the 50/50 event.
The Loma Prieta event was scaled to the 10/50 hazard level (Sa = 0.68g) based on the
initial period of 1 second. The spectral shape of this event is such that the spectral value
at 1 second is at a local minimum of the spectrum, beyond which at higher periods the
values increase. This is clearly evident in Fig. 3.70, which shows spectral acceleration
versus spectral displacement for the Loma Prieta event scaled to the 10/50 hazard level.
When the period lengthened to 1.3 seconds the spectral acceleration and the spectral
displacement increased by 26% and 114%, respectively, thus making the event much
more intense than a design level earthquake. With this increase in intensity and the
additional flexibility of the frame, it seems that it should have been expected that the
frame would exceed the drift limitation of 2% during this design event.
104
When the record was scaled down by 80% to represent the second segment of the design
event, the spectral values were also scaled by the same factor, and are represented on Fig.
3.70 by the white circles along the natural period lines. This scaled design level event
seems to represent the same spectral acceleration level (Sa = 0.68g) as the design level
hazard, but the spectral displacement is still approximately 70% larger. This provides
some insight to explain the high drift levels during this design level event, but given that
the frame had already reached 2% drift during the immediate occupancy event, it seems
highly probable that it would still exceed the design level drift (2%) during the 10/50
event even if it had not gone through the first event.
Therefore the question remains as to why the design methodology with an imposed drift
limit failed to minimize the amount of drift to 2% during the design level event? This
brings up an important issue of how accurate the current design methods are in ultimately
predicting the behavior of the system. These codes are empirical in nature and therefore
it is difficult to rely on their estimates of performance, especially since we are using static
elastic analyses to estimate the dynamic nonlinear behavior of systems. This is the
reason that current research is pushing for a performance-based engineering approach to
design and assessment rather than the current status quo. Therefore, part of the answer to
the initial question may be that the current codes are simply not robust enough to
accurately predict the behavior of systems designed given the current procedure.
After the first design level event, the period had softened to approximately 1.5 seconds.
Figure 3.71 shows the relationship between spectral acceleration versus spectral
displacement for the Chi-Chi record scaled to the 2/50 hazard level at the initial period of
1 second. The shift to 1.5 seconds, as shown in Fig. 3.71, decreases the original spectral
acceleration by 30% and increases the spectral displacement by 54%. The shape of the
response spectrum of the Chi-Chi event is relatively close to that of the IBC 2000 curve
between 1 and 2.5 seconds, which is evident in Fig. 3.72. So it can be argued that even
though the frame had significantly softened from its original MCE point, the final result
was still close to representing a true 2% in 50 year event and the peak drift of 5.4% is
representative of the expected drift in such an event.
105
This trend occurs again in the final Loma Prieta event, where the period of the frame at
the beginning of the record is around 1.7 seconds (shown in Fig. 3.72). This raises the
issue as to whether the pseudo-dynamic events should have been scaled differently by
perhaps accounting for this expected softening of the natural period. Figure 3.72 shows
the spectral acceleration graphs of the scaled earthquake records with the IBC 2000 10/50
and 2/50 design spectra. This figure reveals that the first design level event (10/50) is
actually slightly more intense that the IBC 2000 2/50 hazard when considered at 1.3
seconds (light square on Fig. 3.72). This also shows what was discussed earlier that the
imposed maximum credible event is approximately equal to the 2/50 hazard at 1.5
seconds (light diamond on Fig. 3.72).
3.7.3 Frame Residual Drifts
FEMA 351 (2000) and FEMA 356 (2000) both make some references to the amount of
residual drift expected for each structural performance levels (i.e. immediate occupancy,
life safety and collapse prevention). For the 50/50 event (immediate occupancy), the
expected value of residual drift is defined as negligible (FEMA 356) to less than 1%
(FEMA 351). This performance is clearly met by the performance of the test frame. The
10/50 (life safety) event yielded a residual drift in the test frame of approximately 0.3%,
which easily satisfies the 1% limit required by the FEMA 356 document. For the 2/50
event, a residual drift of 3.4% in first floor and 2.7% in total roof drift was recorded in
the test frame. This falls within the requirements of both documents, where FEMA 356
expects between 4-5% residual drift while FEMA 351 simply states that one should
expect a large permanent offset in the building.
One important issue to note here is the interior gravity resisting frames were not
accounted for in the pseudo-dynamic algorithm. Recent studies have shown that the
contribution of interior gravity frames in the seismic response of steel moment resisting
frames can lead to smaller deformation demands, even if the columns and beams in these
frames are considered as pinned connections (Lee and Foutch, 2002). Additional strength
106
and stiffness is provided by the interior columns and the simple shear tab connections
acting with a concrete slab. This additional capacity would definitely apply to composite
RCS moment frames and could help reduce both the residual drift and the maximum
transient drift.
Despite being permitted by current standards (e.g. FEMA 356, etc.), it should be
recognized that large residual deformations are extremely difficult to repair and may even
result in the demolition of the building after a seismic event. As performance based
earthquake engineering is becoming more of a reality, ductile systems such as moment
resisting frames and buckling-restrained braces need to be able to control the amount of
residual drifts in the building in order to mitigate the cost of repair. This is an issue that
has not been fully resolved and is an area that requires some further study. One such
approach to reduce the amount of residual drift may be to incorporate these composite
frames into a dual system with either a braced or reinforced concrete wall core, as is
shown by Uang et al (2003).
3.7.4 Frame Repair
Damage patterns after each loading event were discussed in detail in section 3.6.2.
Numerous researchers were on hand to witness this test and were able to make comments
on the damage and the likely repairs. It was largely concluded that the damage incurred
by the frame during the 50% in 50 year event, was minimal and required little, if any,
repair. This hazard level is often correlated with an immediate occupancy damage state,
which implies minimal damage and no interruption to the function of the building. Given
the limited amount of damage to the frame following this event, it is reasonable to state
that the frame met the general conditions for immediate occupancy.
After the first design level event, the frame had some noticeable damage, none of which
affected the immediate safety of the structure but, nevertheless, still required some form
of repair. The 1st-floor RC columns would likely require epoxy injection of some of the
cracks and some minor patching of the spalled cover concrete. Some local buckles that
107
occurred in the 1st and 3rd-floor may require heat straightening to recover some of the
initial stiffness of the element. Perhaps the most difficult repair would be straightening
or plumbing of the frame to reduce the residual interstory drift, which amounted to
approximately 0.3% roof drift.
The maximum credible earthquake resulted in severe cracking and spalling of the RC
columns and a large amount of yielding and local buckling in the steel beams. In
addition to the local member damage, there was a large amount of permanent drift in the
frame that would undoubtedly need to be straightened to resume operation of the
building. Researchers present agreed that although the frame could be repaired using
similar techniques described for the design level event, it would most likely be
economically unfeasible to fund the level of repairs required.
For each of the hazard levels examined in this testing program, the resulting damage
within the frame seems to correlate well with the structural performance expected by
current building codes, such as FEMA 356. A majority of the damage seen in this
composite frame would also occur in the more traditional all concrete or all steel moment
frames. The advantage within this system is that some of the more serious problems of
these traditional systems, including early steel beam fracture, have been avoided.
This test is considering only the performance of the bare lateral resisting frame, but in the
context of performance based engineering, one should also account for the level of
damage expected in nonstructural elements (i.e. partitions, cladding, building contents,
etc.) and the gravity resisting elements. In fact, these components will ultimately
comprise a majority of the cost of the building damage (Aslani 2005). While it is
acknowledged that this is an important aspect of the performance assessment of
buildings, it is not in the scope of this study, where the focus is primarily on the structural
performance of these composite RCS frames.
108
3.7.5 Precast System Performance
As discussed in Section 3.3.5, the prefabrication of the column-footing and beam-column

modules and the ultimate construction of the test frame were handled by the Ruentex
Construction and Development Company of Taiwan. Ultimately, this was a very
successful exercise in both the precast fabrication and construction of composite RCS
frames. Discussions on the constructability of these frames with Pankow Builders
(www.pankow.com) have shown that this precast option is likely the most attractive
alternative to make these systems competitive in the United States market. It is
recognized that the economy of a system plays perhaps even a more critical role than
performance in the acceptance and application of new building systems in todays
marketplace.
3.7.5.1 Beam Splices
As described in section 3.3.5, the design philosophy for the beam splices was taken from
FEMA 350 for prequalified bolted fully restrained moment connections. This
methodology allows some yielding in the flange plates by setting the design force equal
to the expected yield moment of the beam at the column face ( M ye = R y M y ). As was
indicated in section 3.6.2, the lower flange splice plates did experience some minor
yielding during the earthquake events (10/50 and 2/50). The yielding became much more
extensive in the 1st-floor as the frame was pushed out to extremely large drifts during the
latter part of the final pushover, and as described earlier, ultimately resulted in net section
rupture near 10% story drift.
Although they performed adequately during the test, in retrospect, it seems that these
beam splices should be designed to experience only minimal yielding during the hinging
of the beam at the column face. Rather than adopting the FEMA 350 philosophy and
reducing the design moment to the expected yield moment at the column face, perhaps
one should design these splices for the expected plastic moment ( M pr = 1.1Ry Z e Fy =
109
(1.1)(1.1) M p ). This design approach would ensure the relatively elastic behavior of these
splices and minimize the potential for unexpected failure within these components. It is
for this reason that design for 1.1Ry M p is recommended in Chapter 2.
Another design consideration that has arisen from the test behavior is whether or not the
bolts in these connections should be designed as bearing or slip critical. As discussed in
Section 3.3.5, these bolts were designed as a typical bearing strength connection that
would allow slip to occur. This decision led to a significant amount of bolt banging
during each of the simulated earthquake events, including the operational or immediate
occupancy 50/50 event. While the bolt slippage did not cause any detrimental effects on
the performance of the building, it was clear to those witnessing the test that these loud,
sharp noises generated during each of the hundreds of slips could terrify the building
occupants. This brings about the question as to whether these connections should be
designed as slip-critical or at least not to slip during the more frequent events. The
occurrence of bolt banging throughout the test ultimately led to the design debate
presented in Section 2.3.2.2.3.1. This is not a new issue and has been observed in both
field and laboratory studies of steel framing systems with bolted connections (Schwein
1999, Tide 1999, Mann et al. 1984, Committee on Design of Steel Building Structures of
the Committee on Metals 1992).
3.7.5.2 Column Splices
The precast column splices were standard grouted splices designed by the Ruentex
Construction and Development Company of Taiwan. These splices, which are sized to
fully develop the expected plastic strength of the reinforcement bars, performed
exceptionally well throughout all phases of the test program. The behavior of these
splices during this frame test and the RC column-footing subassembly tests (Cordova
2006) have proven their strength and durability and should be considered as an extremely
robust connection between precast RC columns.
110
3.7.6 Frame Overstrength
The overstrength of the frame is defined as the ratio of the maximum base shear and the
design base shear. The AISC Seismic provisions (1997) specify the upper bound of the
overstrength factor for moment-frame systems to be 3. During the final pushover, the
RCS test frame reached a maximum base shear of approximately 2.8 times the original
design value (2.0 including P loads), indicating that the frame possessed a large amount
of overstrength even after enduring significant damage throughout the four pseudo-
dynamic events. The peak base shear attained during the pseudo-dynamic events
corresponds to an overstrength value of approximately 3.2. Although this peak value
does not necessarily equate to the maximum strength of the frame, it does show that the
damage in the frame caused the overstrength to reduce by at about 13% (from 3.2 to 2.8).
The final pushover also revealed that the stiffness of the test frame had been reduced to
approximately 50% of the original undamaged stiffness. Although the pseudo-dynamic
events had caused a significant amount of damage and drastically reduced the stiffness of
the frame, there was still a large amount of overstrength present in the test specimen,
which emphasizes the robust nature of the RCS moment frame.
3.7.7 Behavior of Composite Beams
As described in section 3.6.3, the composite beams performed exceptionally well

throughout all phases of the loading protocol, with the performance and durability of both
the slab and the shear studs far exceeding the initial expectations. The 1st and 3rd-floor
beams successfully endured repeated cycles of large plastic rotations without
experiencing any kind of premature fracturing, even when subjected to the large drifts
imposed by the final pushover. There were no instances of shear stud fracture and
instrumentation verified that no differential slip or lift off occurred between the steel
beam and the slab.
The concrete slab greatly enhanced the ductility of the beam section by providing
adequate restraint to the upper flange of the steel beam, thereby preventing the
111
occurrence of local buckling within the upper quarter of the beam depth. In addition to
this, the presence of the slab shifted the neutral axis into the upper half of the beam depth,
which significantly minimized the amount of yielding that occurred in the upper flange.
This is likely the reason that the shear studs that were accidentally placed within the
hinge zones of the 2nd and 3rd floor beams did not lead to any fracture-related incidents in
the upper beam flange.
As discussed in section 3.6.2, local buckling of the web and lower beam flanges began to
occur during the design level (10/50) event and became even more severe during the
maximum credible (2/50) event. In the FEMA 350 (2000) document, the structural
performance levels for steel girders are defined for the both the immediate occupancy
(50/50) and the collapse prevention event (2/50). For the 50/50 event, some of the steel
beams are expected to undergo minor local yielding and buckling, which is consistent
with what occurred in the test frame. In the 2/50 event, the extensive local yielding and
buckling present in the test frame is a damage state that is anticipated by FEMA 350 for
steel beams given this hazard level. It is also anticipated that some girders may
experience partial fractures for this event, which did not happen in any of the beams in
the test frame.
As mentioned in section 3.6.3, the most severe buckling took place at the beam hinges
framing into the exterior joints. The continuity of the beam and the slab in the interior
joints seems to provide additional restraint which (1) limits the amount of buckling that
occurs in these hinges and (2) helps straighten out buckles when the hinge is in positive
moment (i.e. tension in the bottom flange and compression in the slab). In subassembly
tests, this trend does not happen but rather the local buckles build up throughout each
cycle while the length of the beam physically shortens (Civjan 2001). This is drastically
different than what happened in the frame test and points to the difficulty and unresolved
error in modeling the boundary conditions in typical beam-column subassembly tests.
The behavior of the slab and shear studs indicate that composite action was maintained
throughout the loading protocol. This is much different than the behavior that is typically
112
found in subassembly tests (Cheng 2002, Civjan 2001, Leon 1998, Bugeja 2000), where
fractures of shear studs and the loss of composite action occurs relatively early and
abruptly. This difference in behavior points to the disparity between subassembly tests
and real moment frames and will be further discussed in section 3.7.11.
3.7.7.1 Shear Stud Design
The shear studs were originally designed based on the AISC-LRFD recommendations
(defined in Section 2.3.2.2.2) to develop a fully composite section, which required
approximately 42 studs over the entire beam span. The AISC Seismic Provisions reduces
the allowable strength of shear studs to 75% of that defined by AISC-LRFD to account
for the cyclic behavior of shear studs, which would therefore require 56 studs. As
discussed in Chapter 2, more recent work by Civjan (2003) has proposed that the AISC-
LRFD predicted strength of shear studs should be further reduced to 60%, indicating that
the amount of studs required should be 64. While the number of studs was originally
selected based on the requirements of the AISC Seismic Provisions, the final number of
studs called out in the construction drawings (64 studs per beam) coincidentally met the
requirements specified by Civjan (2003). As previously discussed, behavior of the studs
surpassed all expectations from the subassembly tests, but it is difficult to attribute this to
placing 8 extra studs to meet the requirements of Civjan (2003) or rather the overall
boundary conditions and continuity of the full scale test. Considering these points, it is
recommended to follow the AISC Seismic Provisions and provide studs to develop full
composite action, based on stud strengths equal to 75% of the (monotonic) values
specified in AISC-LRFD.
3.7.8 Behavior of RC Columns
The hinges in the RC columns performed as expected throughout all of the pseudo-
dynamic loading events, with the most severe damage occurring in the base hinges where
estimated plastic rotations reached as high as 0.055 radians during the 2/50 event. As
described in Cordova et al. (2006), the tiltmeter data revealed that the hinge length was
113
approximately equal to the depth of the column, which supports the common assumption
for design and assessment purposes.
Bond slip was found to be a significant contributor to the total rotation of the RC column
base hinges. In retrospect, the instrumentation of these columns should have been
planned such that this portion of the column rotation could be monitored and separated
from the total hinge rotation.
The final static pushover provided some insight into what may have been the governing
failure mechanism for these columns. After being subjected to numerous large plastic
excursions and experiencing a large amount of cracking and spalling during the
earthquake events, the column base hinges were considered to be well within the near
collapse damage state. The pushover imposed a maximum drift of 10% in the first floor,
which resulted in further deterioration of the column base hinges. The final failure
mechanism seemed to be a combination of extensive plastic hinging followed by
localized shear failure in the hinge region. Since the concrete was severely damaged in
this region, the high shear seemed to be carried only through dowel action of the
longitudinal bars. Figure 3.73 depicts an interior column after the loose concrete has been
chipped away. It is evident in the photos that there is significant kinking of the main
longitudinal bars, which raises the question as to whether or not this sort of behavior is
acceptable. It is possible to provide additional support for these columns, perhaps in the
form of a steel jacket or fiber reinforced polymer composite wrap, but ultimately, it
seems that the demands imposed on these hinges may be too extreme to make any
conclusions on their behavior during the final pushover.
3.7.9 Strong-Column Weak-Beam Criterion
During the design stages of the testing program, one of the most important questions had
to do with whether one should consider the strength of the composite action of the
concrete slab with the steel beams in the strong column-weak beam (SCWB) checks.
The AISC Seismic Provisions (2002 and 2005) do not explicitly address this issue and it
114
is seemingly left up to the engineers interpretation. As described in Section 3.3.3, the
strength differences between a steel or a composite section has a significant impact on the
final SCWB ratios and can ultimately effect the final design of the RC columns and if not
considered can shift the failure mechanism from beam to column hinging. Based on the
overall performance of the slab and shear studs during the test (Section 3.7.7) and
reinforced by recommendations of other researchers (Leon and Hajjar 1998), it is critical
to consider the composite strength of the beam during the SCWB design check.
The intention of the SCWB criterion is to ensure that the columns in any floor are strong
enough to force hinging to occur within the beams, thus reducing the probability of the
formation of a story mechanism. It is clear that the SCWB criterion that is implemented
in current building codes (AISC 2002, ACI-318 2002) was not able to prevent the two-
story mechanism from occurring during the final pushover event (Section 3.6.2.5),
particularly if one were to consider the strength of the steel beams only (Fig. 3.4a). This
mechanism is particular interesting because the SCWB ratios at the 2nd floor are amongst
the highest in the entire frame (interior joints: M M
col beam = 1.18 ) and satisfies the
spirit of the SCWB criterion with an average column-to-beam strength ratio of 1.04 over
the entire floor considering the composite beam strengths (refer to Section 3.3.3 and Fig,
3.4c). Another issue that should also be discussed is why a first floor story mechanism
did not develop given that the average SCWB ratio over this floor is also approximately
1.08. This lack of correlation between the SCWB design criteria and the performance of
the test frame brings about some concern with these current provisions.
As discussed in Chapter 2 (Section 2.3.2.1.1.1), the SEAOC has proposed an alternate

provision for the SCWB criterion, which recommends that the sum of the strengths of the
columns below a floor should be greater than the sum of the strengths of the beams at the
floor. Considering a building where the beams on two consecutive floors are identical,
this SCWB provision has the effect of increasing the conventional column-to-beam
strength ratio to 2.0 at each joint. Using this updated design criterion, this illustrates that
both the first and second floors fall below the permissible design value of unity with 0.67
and 0.63, respectively. The test data tend to support this new provision, since the second
115
floor columns (where hinging occurred) do not meet the proposed requirement, but again
it does not explain why the first floor columns did not form a story mechanism.
Perhaps the answer to why a first floor story mechanism did not occur has to do with the
location of the inflection point in these columns. A typical assumption that is used to
estimate the force distribution in moment frames is that the inflection points (i.e. points of
zero moment) are located at mid-height for the columns and mid-length for the beams.
Computer analyses often show that while this assumption is true for the floors above the
first floor, it often does not hold for the first floor itself, where the inflection point tends
to be above mid-height and closer to 2/3 to 3/4 the height of the column. This would
suggest that it is more difficult to develop the full moment capacity of the upper hinge in
the first floor columns than it would be in any higher story. This implies that the SCWB
criterion may need to be adjusted to account for this difference. If we assume that the
location of the inflection point is closer to two-thirds up the height of the column, then
the permissible design value for the SEAOC SCWB check should be 0.67 rather than 1.0
for the first floor columns only, which would suggest that the first level in the test frame
passes the SEAOC SCWB check. Given this adjustment, it seems that the test data tends
to support this new provision, since the second floor columns (where hinging occurred)
does not meet the proposed requirement, whereas all the other floors do. Note that this
modification for the first floor columns may not apply when the column base is not fully
fixed, and some flexibility in the footing exists.
3.7.10 Behavior of Composite Joints
With standard details, the composite joints retained their strength and stiffness throughout
all phases of the loading protocol, including the final pushover. The most significant
damage consisted of relatively minor shear cracking and some spalling below the band
plates in the 1st-floor joints, which indicates that the joints had not yet reached their
maximum strength. Following the behavior of past subassembly tests (Liang 2003,
Kanno 1994, etc.), these composite joints continued to prove their inherent strength and
116
durability whether subjected to either monotonic or cyclic loading, making them
extremely attractive for use in high seismic zones.
The roof joints proved to be a unique challenge in the design of the test frame, since
details for top floor connections have not been studied extensively. The solution that we
developed employed a reinforcing bar plate detail, described in section 3.3.7, which
mobilized both the bearing resistance for the beam and developed the longitudinal
reinforcement bars. The proposed reinforcing bar plate detail for the top beam-column
joint of a composite RCS frame has been shown in both subassembly tests and the full-
scale testing program to possess adequate strength and stiffness to force hinging to occur
in the surrounding beams. This joint performed well in both tests and is recommended as
a practical detail for use in top (roof) joints.
3.7.11 Differences between Subassembly and Frame Tests
Beyond the performance and design issues brought up by this frame test, there are several
issues regarding the differences in boundary conditions between an isolated beam-column
subassembly and a complete moment resisting frame. The continuity of the slab and
beam in the test frame provides a significant amount of restraint which prevents beam
shortening, a common phenomenon in steel subassembly tests that allows physical
shortening of the beam as local buckling builds up in the hinge zone. This continuity in
the frame restricts local buckling in steel beam hinges, particularly in those hinges
adjacent to interior joints where the continuity effect has a greater impact. Conversely,
the beams framing into the exterior joints did not have the complete benefit of the
continuity of the slab and beam, and as a result experienced noticeably more local
buckling than those framing into the interior joints. This issue was discussed in detail in
section 3.7.7 and demonstrated that the boundary conditions in typical subassembly tests
tend to overestimate the amount of damage in the beam hinge due to the problems
associated with the lack of continuity.
117
Another important difference between this test frame and subassembly tests was the
introduction of the earthquake loads into the system. In the test frame, story shears were
applied through the east and west loading beams, which in turn transferred the loads
through the slab and into the frame through bearing on the columns and shear studs on
the main girder. This introduction of loads attempts to follow the load path in a real
structure under earthquake excitation. The strain gauges on the loading beam proved that
this system performed as expected and the earthquake loading was distributed relatively
uniformly along the frame length. As described in Section 3.4.1, this loading mechanism
seemed to improve the durability of the slab by counteracting the natural tendencies of
the composite beams in positive and negative bending. As illustrated in Fig. 3.12, when
the beam is in negative bending, the slab is put into tension by the natural curvature of the
beam, but the loading mechanism is offsetting this tension by pushing the slab into the
column imposing compressive bearing stresses (location A in Fig. 3.12). The opposite is
true on the other side of the beam, where positive bending is causing compression in the
slab but the loading beam is pulling the slab away from the column (location B in Fig.
3.12).
The traditional method used to impose loads into a beam-column subassembly is by

directly applying shear to the beams, as shown in Fig. 3.74. This method has the
tendency to try to pry the slab away from the beam and does not capture the beneficial
effects that occur when the loads are imposed through the slab. This seems to provide
one explanation as to why the slab and shear studs in subassembly tests seem to degrade
at much quicker rate than what was seen in the full-scale test frame. Note that this effect
has less of an impact in the lower stories of taller buildings where the shear imposed in
the slab is generally considerably smaller than the shear accumulated in the frame from
the upper floors.
This testing program has underlined some of the problems that beam-column
subassembly tests face with their inability to enforce certain boundary conditions that
seemingly have a positive effect on the behavior of the components. These deficiencies
in the boundary conditions are very difficult to implement without greatly complicating
118
the traditional test setup. Nevertheless, it needs to be recognized that these subassembly
tests are likely overestimating the severity of the damage that one could expect to see in a
complete moment frame system.
119
Table 3.1 Rationale for large-scale test
Issues Explored Issues Neglected
Investigate the often-overlooked connection details: Evaluate the response of a damaged frame. 3-dimensional framing effects
location of the beam and column splices for precast
system, top floor beam/column joint detail, and Track the change in structural period during each Effect of high axial loads on RC columns
foundation/column splice location. earthquake and its effect on the response.
The effect of non-structural elements on the
Given the state of current analyses and drift Investigate the performance-based design issues of stiffness and damping on the frame.
requirements of performance based codes, an the test.
attempt will be put forth to push the test frame to Additional damping techniques such as visco-
large drifts. This may prove to be a benchmark testing-program, elastic dampers.
with hopes to encourage widespread collaboration
Avoid the problems of b/c cruciform tests, such as through open discussions and an accessible Shaking table versus psuedo-dynamic test
unrealistic restraints that may lead to erroneous database record of the results. procedure
results (i.e. beam shortening). Comparison of
subassembly tests and actual frame. Once damaged, determine how difficult it is to High P-delta forces
repair the structure. Extrapolate the cost to a
Test and validate the importance of current design realistic structure and compare to all-steel or all- Structural Irregularities bay width, story height,
criteria (e.g. SCWB and drift requirements). concrete frames. stiffness and mass irregularities
Will test our assumptions about composite beam Validation problem for current analytical tools and Utilize advance measurement techniques beyond
behavior; beff for strength/stiffness, effects of damage indices. the standard tools.
continuity of slab, the extent of damage that we can
we count on the composite action, and with the Investigate the response of a repaired structure.
restrictions on shear studs within the plastic hinge
zone (FEMA 351) and the length of the beam
splice, how much composite action is really
occurring?
Verify our assumptions on the effective EI of the

RC columns.
Determine and work through all construction issues

for this system.
120
Table 3.2 Design loads summary
Floor: 4.4 kPa (92 psf)
Dead Load
Roof: 4.3 kPa (89 psf)
Live Load 2.4 kPa (50 psf)
Effective Seismic Weight, W 17250 kN (3878 kips)

Base Shear, V 1078 (+ 81) kN
(+ Acc. Torsion) (242.4 (+ 18.2) kips)
F1: 193 kN (43.44 kips)
Shear Distribution, Fx F2: 395 kN (88.88 kips)
F3: 570 kN (128.23 kips)
Table 3.3 Test frame member properties.

RC Columns
Steel Beams
Section
(d, bf, tw, tf) Rein. Bars
Floor fc,meas=40
Fy,nom = 345 MPa Fy,nom = 415 MPa
MPa
(Fy,meas) Fy,meas = 527 MPa
( fc,meas)
H600x200x11x17mm 650x650mm Exterior 8-#11bars
1st
(426 MPa) (89 MPa) Interior 12-#11bars
H500x200x10x16mm 650x650mm Exterior 4-#11bars
2nd
(501 MPa) (68 MPa) Interior 12-#11bars
Exterior 4-#11bars
H396x199x7x11mm 650x650mm
3rd Int. Lower 12-#11bars
(419 MPa) (68 MPa)
Int. Upper 8#11bars
121
Table 3.4 Measured strengths of steel tension coupons.
Percent
Fy
Steel Difference
(MPa)
(meas. and min. spec.)
Flange 409 19%
1st Floor
Web 442 28%
Flange 484 40%
2nd Floor
Web 517 50%
Flange 407 18%
3rd Floor
Web 431 25%
#3 bars 390 -6%
#4 bars 411 -1%
#10 bars 586 42%
#11 bars 527 27%
Table 3.5 Measured crushing strength of concrete cylinders.

Percent
Concrete f c' (MPa) Difference
(meas. and min. spec.)
89.0 (lower) 115%
1st Floor Columns
70.8 (upper) 71%
nd
2 Floor Columns 68.2 65%
3rd Floor Columns 68.4 65%
Slab 31.0 13%
Table 3.6 Summary of maximum and minimum drifts during each pseudo-dynamic
event (both absolute drift and with the residual removed).
50/50 10/50-1a 10/50-1b 2/50 10/50-2
Max Drift 1.76 2.99 2.65 3.40 1.36
(residual removed) (1.76) (3.10) (2.26) (3.07) (1.62)
Min Drift -1.91 -2.20 -2.28 -5.32 -2.85
(residual removed) (-1.91) (-2.15) (-2.67) (-5.78) (-2.75)
122
6 @ 7m = 42m
Moment Frames
4 @ 7m
= 28m
Figure 3.1 Plan View of Building
Figure 3.2 Plan and elevation views of full-scale composite test frame
123
Face Bearing Ties
Plate
Figure 3.3 Joint detail showing the transverse beam and placement of ties.
0.99 0.97 0.97 0.99
1.03 1.23 1.23 1.03
1.11 1.06 1.06 1.11
(a)
0.58 0.72 0.72 0.58
0.68 0.97 0.97 0.68
0.76 0.86 0.86 0.76
(b)
Figure 3.4 Mc/Mg ratios at each joint assuming a) steel beams (nominal) and b)
composite beams (nominal).
124
0.70 0.86 0.86 1.22
0.72 1.18 1.18 1.07
0.91 1.03 1.03 1.36
(c)
Figure 3.4 cont. Mc/Mg ratios at each joint assuming c) composite beams and RC
columns with measured material properties.
Flange Plates
Shear Tab
Figure 3.5 Schematic of a typical bolted flange plate beam splice connection
125
Splice
Splice
(a) (b)
Figure 3.6 RC column cantilever tests with the grouted splice located (a) 1-meter up the
column height and (b) flush at the column-footing interface.
126
800
FFH08
600
400
200
Force (kN)
-200
-400
-600
-800
-8 -6 -4 -2 0 2 4 6 8
Drift (%)
(a)
800
FFL08
600
400
200
Force (kN)
-200
-400
-600
-800
-8 -6 -4 -2 0 2 4 6 8
Drift (%)
(b)
Figure 3.7 Response of RC column subassembly test with precast splice at (a) 1-meter
above the footing and (b) flush at the column-footing interface. (Tsai 2002)
127
A A
see Fig. 3.10
396x199x7x11
Band Plate &

12#32 bars
FBP (20mm)
650x650mm
Figure 3.8 Top joint option #1. Section AA in Fig. 3.9.
Rein. Cap Plates Band Plate
#10 bars
Figure 3.9 Section A-A from Fig. 3.8.
PP Rein. Bar
Plate top of
concrete
Band Plate
130x20mm see
fig. 3.11 stiffener
beam flange
(11mm)
FBP beam
374x20mm web
#32 bars
Figure 3.10 Reinforcement cap plate
128
#10 bar
These reinforcing bars must
be weldable, ACI 318
defines these as type A706
Figure 3.11 Plate to reinforcing bar detail.
compressive strut compression field

against column against shear studs
Loc. A Loc. B
CL
Beam Moment
Diagram Mpos
slab in compression
Mneg Mneg
slab in tension slab in tension
Deflected
Shape
Figure 3.12 Schematic of load path between actuators, loading beams, and test frame.
129
2.5
50/50 TCU082
2/50 TCU082
10/50 LP89G04
2 10/50 IBC2000
2/50 IBC2000
50/50 Taiwan Hazard
10/50 Taiwan Hazard
2/50 Taiwan Hazard
1.5
S (g)
a
0.5
0
0 0.5 1 1.5 2 2.5
Period (sec)
Figure 3.13 Final records scaled at T1 = 1sec to appropriate Taiwanese hazard levels.
(a) (b)
Figure 3.14 Construction photos of (a) a typical pre-cast beam-column module and (b)
the completion of the first floor.
130
200
EQ1: 50/50 TCU082

150
100
Roof Displacement (mm)
50
-50
-100
-150
-200
0 5 10 15 20 25 30 35 40 45
Time (sec)
Figure 3.15 Roof displacement versus time for the 50/50 event.
400
EQ2: 10/50 LP89G04 - 1

300
200
100
-100
-200
-300
0 5 10 15 20 25 30 35 40 45 50
Time (sec)
Figure 3.16 Roof displacement versus time for the 10/50-1 event.
131
400
300 EQ3: 2/50 TCU082
Roof Displacement (mm) 200
100
-100
-200
-300
-400
-500
0 5 10 15 20 25 30
Time (sec)
Figure 3.17 Roof displacement versus time for the 2/50 event.
150
100 EQ4: 10/50 LP89G04 - 2
50
-50
-100
-150
-200
-250
-300
-350
0 5 10 15 20 25 30 35 40
Time (sec)
Figure 3.18 Roof displacement versus time for the final 10/50 event.
132
3
E Q 1: 50/50
E Q 2: 10/50-1a
E Q 2: 10/50-1b
2.5
E Q 3: 2/50
E Q 4: 10/50-2
2
Floor
1.5
0.5
0
-0.06 -0.04 -0. 02 0 0. 02 0.04 0.06
IDR
Figure 3.19 Maximum interstory drift ratios for each floor during each pseudo-dynamic
loading event.
3
E Q 1: 50/50
E Q 2: 10/50-1a
E Q 2: 10/50-1b
2.5
E Q 3: 2/50
E Q 4: 10/50-2
2
Floor
1.5
0.5
0
-4000 -3000 -2000 -1000 0 1000 2000 3000 4000
Story Force (kN)
Figure 3.20 Maximum story shear for each floor during each pseudo-dynamic loading
event.
133
300
1 st Event
2 nd Event
200
100
-100
-200
Test Frame Results
80%-LP89G04 10/50
-300
0 5 10 15 20 25 30 35 40
Time (sec)
event.
3
1 st Event
2 nd Event
2.5
2
Floor
1.5
0.5
Test Frame Results
80%-LP89G04 10/50
0
-0.06 -0.04 -0.02 0 0.02 0.04 0.06
Interstory Drift Ratio (%)
event.
134
4000
Maximum Drifts
1 st Floor: 10.1
3000 2 nd Floor: 7.9
3 rd Floor: 3.1
2000
Base Shear (kN)
1000
-1000 Measured Response

P
Measured+P
-2000
-0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Roof Drift Ratio
Figure 3.23 Base shear versus roof drift ratio for the final static pushover (p-delta loads
are superimposed in dotted line).
1C3 Base Hinge
1
Moment (1000kN-m)
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
Figure 3.24 Hysteretic response of a 1st floor interior column base for the 50/50 event.
135
3
1C3 Base Hinge
1
Moment (1000kN-m)
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
Figure 3.25 Hysteretic response of a 1st floor interior column base for the 10/50-1a
event.
1C3 Base Hinge
1
Moment (1000kN-m)
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
Figure 3.26 Hysteretic response of a 1st floor interior column base for the 10/50-1b
event.
136
3
1C3 Base Hinge
1
Moment (1000kN-m)
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
Figure 3.27 Hysteretic response of a 1st floor interior column base for the 2/50 event.
1B1S
1
Moment (10 kN-m)
3
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
Figure 3.28 Hysteretic response of a 1st floor exterior beam hinge for the 50/50 event.
137
3
1B1S
1
Moment (10 kN-m)
3
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
Figure 3.29 Hysteretic response of a 1st floor exterior beam hinge for the 10/50-1a
event.
1B1S
1
Moment (10 kN-m)
3
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
Figure 3.30 Hysteretic response of a 1st floor exterior beam hinge for the 10/50-1b
event.
138
3
1B1S
1
Moment (10 kN-m)
3
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
Figure 3.31 Hysteretic response of a 1st floor exterior beam hinge for the 2/50 event.
(Large rotations are due to measurement errors caused by severe local buckling).
1B1S
1
Moment (10 kN-m)
3
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
Figure 3.32 Hysteretic response of a 1st floor beam for the 10/50-2 event.
139
3
1B1N
1
Moment (10 kN-m)
3
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
Figure 3.33 Hysteretic response of a 1st floor interior beam hinge for the 2/50 event.
6 6
4 4
1J3 1J3
Joint Shear (10 3kN)
Joint Shear (10 kN)
2 2
3
0 0
-2 -2
-4 -4
-6 -6
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
Total Rot. (rad/1000) Outer Panel Rot. (rad/1000)
Figure 3.34 Total and outer panel hysteretic response of a 1st floor interior joint for the
50/50 event.
140
6 6
4 4
Joint Shear (10 kN) 1J3 1J3
Joint Shear (10 kN)

2 2
3
3
0 0
-2 -2
-4 -4
-6 -6
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
10/50-1a event.
6 6
4 4
1J3 1J3
Joint Shear (10 kN)
Joint Shear (10 kN)
2 2
3
0 0
-2 -2
-4 -4
-6 -6
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
10/50-1b event.
141
6 6
4 4
Joint Shear (10 kN) 1J3 1J3
Joint Shear (10 kN)

2 2
3
3
0 0
-2 -2
-4 -4
-6 -6
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
2/50 event.
6 6
4 4
1J3 1J3
Joint Shear (10 kN)
Joint Shear (10 kN)
2 2
3
0 0
-2 -2
-4 -4
-6 -6
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
10/50-2 event.
142
3
2C3 Upper Hinge
1
Moment (10 3kN-m)
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
Figure 3.39 Hysteretic response of the 2nd floor upper interior column hinge after the
10/50-1a event.
2C3 Upper Hinge
1
Moment (10 3kN-m)
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
10/50-1b event.
143
3
2C3 Upper Hinge
1
Moment (10 3kN-m)
-1
-2
-3
-80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000)
2/50 event.
144
Figure 3.42 Damage in 1st floor interior column base after the Figure 3.43 Damage in 1st floor interior column base after the
50/50 event. 10/50-1b event.
145
Figure 3.44 Damage in 1st floor interior column base after the Figure 3.45 Damage in 1st floor interior column base after the
2/50 event. 10/50-2 event.
146
Figure 3.46 - Yielding in 1st floor beam after the 50/50 event
Figure 3.47 - Yielding and local buckling in 1st floor beam (1B1S) after the 10/50-1b
event
147
Figure 3.48 - Yielding and local buckling in 1st floor beam (1B1S, exterior beam hinge)
after the 2/50 event
Figure 3.49 - Yielding and local buckling in 1st floor beam (1B1S, exterior beam hinge)
after the 10/50-2 event
148
Figure 3.50 - Yielding and local buckling in 1st floor beam (1B3N) after the 10/50-1b
event
Figure 3.51 - Yielding in 1st floor beam (1B1N, interior beam hinge) after the 2/50 event
149
Figure 3.52 - Yielding in 1st floor beam (1B1N, interior beam hinge) after the 10/50-2
event
Figure 3.53 1st floor splice plate after the 10/50-1b event.
150
Figure 3.54 Joint 1J3 after the 50/50 Chi-Chi event. Figure 3.55 Joint 1J3 after the 10/50-1b Loma Prieta event.
151
Figure 3.56 Joint 1J3 after the 2/50 Chi-Chi event. Figure 3.57 Joint 1J3 after the10/50-2 Loma Prieta event.
152
Figure 3.58 Damage in upper hinge of 2nd floor interior Figure 3.59 Damage in upper hinge of 2nd floor interior column
column hinge after the 10/50-1b event. hinge after the 2/50 event.
153
Figure 3.60 Damage in upper hinge of 2nd floor interior Figure 3.61 Frame at its maximum drift state during static push.
column hinge after the 10/50-2 event. IDR in 1st floor:10%, 2nd: 8%, and 3rd: 3%.
154
Figure 3.62 Damage in 1st floor exterior column base after the Figure 3.63 Damage in 1st floor interior column base after the
final pushover event. final pushover event.
155
Figure 3.64 Damage in upper hinge of 2nd floor exterior column Figure 3.65 Damage in upper hinge of 2nd floor exterior
hinge after the final pushover event. column hinge after the final pushover event.
156
Figure 3.66 Slab above exterior beam hinge, 1B1S, after the final static push.
Figure 3.67 Net section rupture of lower flange plates in 1st floor beam splice.
157
Figure 3.68 - Yielding and local buckling in 1st floor beam Figure 3.69 - Yielding and local buckling in 1st floor beam (1B3N,
(1B1S, exterior beam hinge) after the final pushover event. exterior beam hinge) after the final pushover event
158
2.5
IBC 10/50
Scaled Loma Prieta Record (LP89G04)
1.5
S a (g)
1 S a=0.68g, S d=16.8cm
S a=0.86g, S d=36.0cm
0.5 S a=0.42g, S d=30.0cm
0
0 5 10 15 20 25 30 35 40 45 50
S d (cm)
Figure 3.70 Spectral acceleration versus displacement for the Loma Prieta event.
2.5
IBC 2/50
Scaled Chi-Chi Record (TCU082)
1.5
S a (g)
S a=0.92g, S d=22.8cm
1
S a=0.63g, S d=35.0cm
0.5
0
0 5 10 15 20 25 30 35 40 45 50
S d (cm)
Figure 3.71 Spectral acceleration versus displacement for the Chi-Chi event.
159
2.5
50/50 TCU082
2/50 TCU082
10/50 LP89G04
2 10/50 IBC2000
2/50 IBC2000
50/50 Taiwan Hazard
10/50 Taiwan Hazard
2/50 Taiwan Hazard
1.5
LP89G04 at 1.3s
80% LP89G04 at 1.3s
S a (g)
TCU082 at 1.5s
0.5
1.0sec 1.3sec 1.5sec

0
0 0.5 1 1.5 2 2.5
Period (sec)
Figure 3.72 Spectral acceleration graphs of final records with highlighted spectral
values at the elongated periods.
Figure 3.73 RC column base hinges after loose concrete had been chipped away.
160
Pulling beam
away from slab
Actuator
Forc
Figure 3.74 Boundary condition in subassembly tests causes slab to pull away from
beam.
161
Chapter 4: Analytical Modeling and Validation
4.1 Introduction
This chapter investigates techniques to analytically model composite RCS frames. First,
the modeling issues at the material and element level are addressed, using calibration
studies on subassembly component tests, including reinforced concrete columns,
composite steel beams, and composite beam-column connections. While the focus of this
chapter is on composite RCS frame, the results and guidelines can be applied to any
systems that utilize these structural members (e.g. conventional all-steel or concrete
moment frames). Emphasis is placed on two specific types of element models: (1) a
flexibility derived beam-column element that is able to capture distributed plasticity by
employing a fiber cross-section at integration points along the member and (2) a beam-
column joint model that captures finite size kinematics and inelastic behavior of the joint.
These models are then used to analyze the 3 story test frame described in Chapter 3, and
validated against the measured pseudo-dynamic test data. Attempts are also made to
relate the analytical response to the physical damage observed in the test specimen.
4.2 OpenSees Component Models
Open System for Earthquake Engineering Simulation (OpenSees) is a software

framework for simulating the seismic response of structural and geotechnical systems
(McKenna et al., 1999). OpenSees is an open source program that is continually
evolving as researchers improve existing models and add new models and features. This
section will specifically discuss the uniaxial materials and elements in OpenSees that can
be used to represent the behavior of composite RCS frames.
4.2.1 Material Models
The uniaxial material models are the most basic components in OpenSees to model a
variety of force versus displacement (or stress versus strain) hysteretic responses.
162
Material models can be used to make up the fiber cross-sections within beam-column
elements (as described in Section 4.2.2) or can be used in a non-linear spring to represent
the response of a designated degree of freedom (i.e. rotational, axial, or shear spring).
There are a number of material models provided in OpenSees. The three used in this
study are the steel, concrete, and tri-linear hysteretic model to represent the composite
steel beams, RC columns, and composite joints in RCS frames.
The steel uniaxial material model (Filippou et al., 1983) incorporates isotropic strain
hardening and allows a smooth transition between the elastic and hardening portions of
the response, which can simulate the Bauschinger effect (Fig. 4.1). The backbone of the
steel material model is defined by the elastic modulus ( Es ), the yield strength ( Fy ), and
the hardening ratio (b). The remaining input parameters to define the transition curve and
isotropic hardening are all typically set to the default values, as defined by Filippou
(1983). This material is referred to as Steel02 in OpenSees.
The concrete material model is defined by the modified Kent and Park model (Scott et. al.
1982) and represents typical concrete crushing and residual strength behavior. It also
allows for tensile strength with linear softening that helps to represent the interaction of
the concrete and the reinforcement bars in tension (Fig. 4.2). The compression backbone
of this model is defined by the points at which the material reaches the maximum
crushing strength ( f cc' , cc ) and the point when the residual strength attained ( fc'2 , c 2 ).
The tensile segment of the backbone is defined by the ultimate tensile strength ( ft ) and
the tensile softening slope ( Ets ). This material model is referred to as Concrete02 in
OpenSees.
The hysteretic material model, shown in Fig. 4.3, has a tri-linear backbone and the
capability to introduce both pinching behavior and stiffness degradation. The tri-linear
backbone is defined by sets of three positive and negative force-deformation points,
which provides the capability to define different positive and negative backbone curves.
Pinching is defined by the x and y factors, which define the intermediate force-
163
deformation point that the model will shoot for before returning to the previous
maximum point. Damage can be introduced to the model either through a deformation
( d ) or an energy-based factor ( d E ), which causes the current excursion to shoot for a
force-deformation point that is further out than the last excursion (thus resulting in the
accelerated loss of stiffness). This material model is referred to as Hysteretic in
OpenSees.
For a RCS composite frame, there are a minimum of five distinct material models that are
required to represent the concrete and steel fibers in the RC columns and composite beam
sections. Specific modeling recommendations for each of these materials are discussed
in section 4.2.2. For the composite RCS joints, two distinct material models are
employed via rotational springs to represent the two different deformation mechanisms;
these models are discussed in section 4.2.3.
4.2.2 Flexibility-Based Fiber Beam-Column Elements
There are several options in OpenSees to represent the nonlinear behavior of beam-
column elements, one of which is the flexibility or force based element that can capture
distributed plasticity with fiber cross-sections at a number of integration points along the
element. The flexibility-based formulation is based on linear force interpolation
functions that accurately estimate the internal force distribution of members where forces
are applied at each end node (although this does not account for the P-delta moments
along the member length). This derivation does not encounter large discretization errors,
as is generally found in the stiffness-based elements that use approximate displacement
interpolation functions (Neuenhofer and Filippou, 1997). When compared on a one to
one basis, each flexibility-based element is more computationally intensive than a
displacement-based element. However, the benefit of the flexibility derivation lies in the
fact that an accurate answer can be achieved with fewer elements (often a single element
per member for beams and columns in moment frames), whereas displacement-based
elements require more discretization to reach the same answer. This flexibility-based
164
element is referred to as the nonlinearBeamColumn or the forceBeamColumn in
OpenSees.
The fiber sections at each integration point represent the cross-section of the component
being modeled (i.e. reinforced concrete column or composite steel beam) and are
composed of a mesh of fibers, each of which is assigned a uniaxial material hysteretic
property (i.e. steel, concrete, etc.). In RCS moment frames, these fiber sections are
composed of steel and concrete models to represent materials in the composite steel beam
and the reinforced concrete column. The details of each of these cross-sections and the
material models that compose the fibers are discussed in sections 4.2.2.1 and 4.2.2.2.
4.2.2.1 Reinforced Concrete Columns
The reinforced concrete column fiber section is comprised of confined and unconfined
concrete and steel reinforcement bars. The cover concrete, located outside of the
transverse reinforcement, is considered as unconfined and will quickly begin to spall after
reaching its crushing strength. The core concrete is confined on all sides by the
longitudinal and transverse reinforcement and will behave in a more ductile manner.
Confined and unconfined regions are both modeled with the Concrete02 material in
OpenSees (see Section 4.2.1).
The compression backbone of the Concrete02 model can be defined according to the
modified Kent and Park model (Scott et al. 1982), which adjusts the backbone parameters
( f cc' , cc , fc'2 , c 2 ) based on the amount of volumetric confinement provided by the
transverse reinforcement ( s ). The maximum stress attained is defined as follows:
f cc' = Kf c' (4.1)

which is assumed to be reached at a strain of:
cc = 0.002 K (4.2)
The residual strength is defined as:
165
f c'2 = 0.2 Kf c' (4.3)
which begins at a strain defined according to the following equations:

0.5
Zm = 1/ 2
(4.4)
3 + 0.29 f c' h"
+ 0.75 s 0.002 K
145 f c' 1000 sh
0.8
c2 = + 0.002 K (4.5)
Zm
where:
K = factor defined by K = 1 + s f yh f c'
s = volumetric ratio of transverse steel to core concrete

f yh = yield stress of the transverse reinforcement (MPa)
f c' = concrete compressive strength (MPa)
h " = width of confined concrete measured to the outside of the perimeter

hoop
s h = center to center spacing between hoop reinforcement sets
When calibrated against physical tests, the analytical models consistently overpredicted
the strength of the RC column subassembly tests, and it was reasoned that this was due to
the concrete strength used in the modified Kent and Park model. For this reason, rather
than entering the modified Kent and Park model with the full f c' , a more accurate
solution is obtained when the input strength is set to 0.85 f c' , recognizing that the in-situ
strength of concrete may be less than the measured f c' from concrete cylinder tests. This
difference follows the design provisions for columns and beams in ACI-318 (2002). This
adjustment improves the correlation between the predicted and measured strengths of RC
columns, as will be described later in this section.
The cover concrete is assumed to be unconfined ( s = 0 ) and as a result will provide

very little ductility. However, it is not recommended to use the modified Kent and Park
166
model to define the backbone for the cover concrete. Using s = 0 , Kent and Park
model predicts very little ductility in the concrete and does not seem to accurately model
the behavior of cover concrete. Moreover, this definition tends to cause convergence
problems in OpenSees. Therefore, it is recommended that the backbone of the cover
concrete be defined as follows: f cc' = 0.85 f c' , cc = -0.002, fc'2 = 0, and c 2 = -0.010.
This definition fits the calibrated subassembly tests and also improves the numerical
convergence. The tension strength of the cover concrete is assumed to be negligible.
The tensile cracking strength, ft, of the core concrete is defined according to the following
equation defined by Carrasquillo et al. (1982):
f t = 0.94 f c' ( MPa ) = 11.3 f c' ( psi ) (4.6)
where ft and f c' carry the unites shown. The tension stiffening effect is an important
phenomenon that accounts for the interaction between the reinforcement bar and the
surrounding concrete when subjected to tension. This effect provides a smooth transition
in the moment curvature response by allowing the concrete to reach its maximum tensile
stress and then slowly shedding the load until it reaches zero tensile strength. This effect
has been modeled by Stevens et al. (1991) and can be represented linearly between zero
and maximum tensile stress, beyond this point the stress decays as a function of the bar
diameter and the ratio of total steel in the cross-section. Figure 4.4 depicts the typical
tensile response of concrete accounting for this tensile stiffening effect. The concrete
model in OpenSees allows one to model the tension strength decay using either an
exponential decay, as suggested by Stevens (et al., 1991), or a linear decay. The linear
model performed quite well in the calibration studies in Section 4.2.2.1 and was selected
for its simplicity. The slope of the line is defined as a best fit to the curve computed by
the Stevens et al., (1991) method.
Based on calibration studies presented in Section 4.2.2.1, the unloading stiffness ratio
= Eu Ec = 0.1 is suggested for RC columns.
167
The column reinforcement steel can be modeled in OpenSees using the Steel02 material
model, as described in section 4.2.1. The yield strength of the steel should either be taken
from the measured value or as the expected yield strength, assumed equal to1.25Fy . The
1.25-factor for the expected yield strength is based on the seismic design criteria in ACI-
318-02 (2002). The strain hardening stiffness, Eh , can be taken as the slope of the line
between the measured yield stress and strain to the measured ultimate stress and strain. If
this information is not available, then a hardening slope of 2% of the initial stiffness of
the steel, E s = 200,000MPa , is assumed. The parameters that define the transition
between the elastic and hardening branch are set to the default values of R0 = 18.5 ,
c1 = 0.925 , and c 2 = 0.15 .
The fiber beam column element alone cannot capture the additional rotation in RC
columns as a result of yield penetration and bond slip near the hinge zone. This
deformation occurs due to the relative movement of the longitudinal reinforcing bar with
respect to the concrete at the face of the joint (or footing). This deformation results in the
opening of a gap between the two concrete interfaces e.g. an RC column and footing
interface, as shown in Fig. 4.5. As suggested by Fillipou et al. (1983), the reinforcing bar
bond stress versus slip behavior can be idealized as shown in Fig. 4.6. Bar pullout will
only occur when the anchorage is insufficient to develop the bar strength, which is
unlikely in the seismically detailed members of the RCS frame. However, yielding of the
reinforcement, which is not considered in Fig. 4.6, can lead to significant slip of the bar,
even when pullout does not occur. The rebar slip is calculated as a function of the strain
distribution along the embedded bar and assumed to be zero at the point of zero-bar stress
(i.e. once full bar development is attained, as shown in Fig. 4.5). Models to describe this
type of mechanism are reported by Filippou and Popov (1984) and Mazzoni (1997).
This inelastic bond-slip phenomenon will lead to a reduction in the overall frame stiffness
and less concentrated damage in the column base hinges as compared to a fixed base
model. A simplified approach is taken to capture this effect. Concentrated rotational
elastic springs are inserted between the RC columns and the fixed bases and are assigned
168
an elastic stiffness equal to the rotational stiffness of the column, which essentially
doubles the flexibility of the columns. The spring stiffness is calculated as follows:
3EI eff
K rot = (4.7)
Li
where:
EI eff = Effective stiffness of RC section, defined in Chapter 2, equation 2.5
Li = length to columns inflection point
This model is based on findings within the literature, which report that bond slip accounts
for up to 50% of flexural deformations in experiments on fixed-fixed reinforced concrete
beam columns (Saatcioglu et al, 1989, 1999). It should be recognized that this approach
is only capturing the additional flexibility of the quasi-elastic loading and unloading of an
RC column, and it does not explicitly account for inelastic effects after column hinging.
This approach is necessary given that base rotation elements are in series with the
nonlinear beam-column, and once either of them begins to plastically deform the inelastic
deformations will concentrate (localize) in one of the two components. For this reason, it
is not advisable to model the inelastic behavior of the bond-slip phenomenon using an
inelastic spring, since there is a potential to localize all the plastic deformations in either
the nonlinear beam-column or the base rotation spring. There are other more
sophisticated methods to handling this bond slip issue, some of which are discussed in
Altoontash (2004), but this simplified technique is adopted as a relatively accurate and
practical way to capture the additional flexibility introduced by bond slip into RC
columns.
Given these material definitions and the bond-slip spring, several code-conforming RC
column subassembly tests were modeled in OpenSees using the flexibility-based fiber
section beam-column element and compared against the experimental results. The test
data includes some extracted from the PEER Structural Performance Database
(http://nisee.berkeley.edu/spd/, Tanaka and Park, 1990) as well as four columns tested in
the NCREE lab (Tsai 2002), which include grouted precast column splices. Selected
details of these column tests are presented in Table 4.1, with further information in the
169
supporting references. Shown in Figs. 4.7 through 4.14 are results of this calibration
study to evaluate how well the analytical models simulate the experimental response.
The tests completed by Tsai (2002) were part of the testing program described in Chapter
3. The specimen names reflect the following attributes: the second letter indicates
whether the column is full (F) or reduced (R) scale, the third letter describes the splice
location either at the column base (L) or 1-meter above the base (H), and the number at
the end indicates the percentage of axial load on the specimen. As shown in Fig. 4.7,
OpenSees adequately captures the response of specimen FFH08. To illustrate the
influence of these elastic bond-slip springs, Fig. 4.8 shows this same FFH08 test
compared to an OpenSees fiber element model with a fixed base connection (i.e. no
springs). While the overall response is similar in Figs. 4.7 and 4.8, it is evident that the
initial and unloading stiffness of the analytical model with the fixed base connection (Fig.
4.8) overestimates that observed in the test. The secant stiffness up to the yield point in
Fig. 4.8 is about twice as large as that in Fig. 4.7. When the elastic base spring is
introduced (Fig. 4.7), the stiffness of the analytical and test model matches up quite well
compared to the element without the spring.
Specimen FFL08 experiences more strength deterioration and pinching than what is
predicted by OpenSees (Fig. 4.9). The deterioration is due to the influence of the grouted
precast spice connectors within the hinge zone and the lack of transverse reinforcement in
this region. These effects are not represented in the OpenSees model. The strength of
specimen FRL08 (Fig. 4.10) is about 20% greater than that predicted by the OpenSees
cyclic model. This strength increase is attributed to an exaggerated influence of the
grouted couplers in this reduced scale column. It is believed that the couplers provided
additional capacity to the hinge region thereby pushing the hinging zone up above the
base of the column. This additional strength provided by the couplers was considered a
in a second OpenSees model, and the monotonic backbone response is also plotted on
Fig. 4.10. This figure reveals that the strength of the specimen is enveloped by these two
different models, but the degradation of the strength is not captured.
170
Specimen FRL60 (Fig. 4.11) contains a very high amount of axial load yet was designed
to fail at roughly the same moment capacity as FRL08. Despite this approach, the
physical specimen reached strengths nearly 60% larger than the predicted value (Fig.
4.11). This overstrength may be attributed to the differences between expected versus
measured material properties, but unfortunately, these details are not available and
therefore the reason for this large difference from the predicted strength is not easily
apparent. Using the expected material properties, the OpenSees model predicts
approximately the same strength as the FRL08 specimen, which was the intention of the
original design. As the column is pushed just beyond its design strength, the concrete
loses its strength very rapidly and the section quickly degrades to residual strength of the
core concrete and the reinforcement bars. This behavior is much different than the
physical test, but given the lack of information regarding material properties, they cannot
be directly compared. This case is specific to columns above the balance point (i.e.
concrete fails prior to yielding of reinforcement bars) and will not be considered in this
calibration study. Further validation of the fiber models to represent RC columns with
high axial loads above the balance point of the section is reserved for future study. Note
that the RC columns in the 20-story building modeled in Chapter 5 does not encounter
the problem discussed here, as all the columns remain below the balance point and
yielding of the rebars precipitates concrete crushing providing a much more ductile
section.
The Tanaka and Park (1990) tests examined the effect of confinement configuration on
the ductility of RC columns, so other than differences in the transverse reinforcement,
each of the columns in tests #2-4 are the same. These tests are configured as a double-
ended setup (as shown in Fig. 4.12) whereas the previously described tests from Tsai
(2002) are the standard cantilever setup. The setups introduce a subtle difference in the
bond-slip characteristics since the rebars in the two opposing columns of the double
ended tests will tend to interact with one another, as is depicted in Fig. 4.12. The
previously described calibration is for cases where the rebars are fully developed (i.e.
rebar stress eventually goes to zero), since this is the condition at the RC column base
hinges where large inelastic rotations are expected. In the double-ended setup, it is likely
171
that the bond-slip spring would be stiffer than the cantilever case given that the rebar are
anchored and stiffened by the opposing forces in the adjacent columns. This effect is
evident in Fig. 4.13, which shows the calibration results of specimen #4 for with (Fig.
4.13a) and without (Fig. 4.13b) the bond-slip spring. These results show that the bond
slip deformations are not as significant for the double ended setup as they are in the
previous cantilever cases. The difference in response is highly dependent upon the test
setup and how large the loading block dimension (d1 in Fig. 4.12) is in the test. Given
this difference, the remaining results from the Tanaka and Park (1990) columns are
shown without the bond-slip spring in Figs. 4.14-4.15.
Overall, the analytical material, bond-slip spring, and fiber beam-column models of the
RC column experiments match the response of the experiments rather well up through
large inelastic rotations. There are some discrepancies (i.e. double-ended bond-slip issue,
high axial loads), but those are limited to cases that are not relevant for the frame
analyses in the present study.
4.2.2.2 Composite Steel Beams
The fiber section for the composite steel beams consists of steel for the beam and slab
reinforcement (when present) and concrete for the composite slab. The beam steel is
represented in OpenSees as the Steel02 material model. The yield strength of the steel
should either be taken from the measured value from a tension coupon or as the expected
yield strength, R y Fy . The R y parameter is defined by the AISC Seismic Provisions
(2002) as the ratio of the expected yield strength to the nominal yield strength and is
taken as 1.1 for typical Grade 50 steels. Additional strain hardening factors should not be
used for Fy given that the Steel02 material already has strain hardening built into the
model ( b = E h / E s ). The strain hardening factor can be set to b = 0.02 or determined

from measured properties. The isotropic hardening and curve transition parameters are
identical to the definitions for the reinforcing steel ( R0 = 18.5 , c1 = 0.925 , and
c 2 = 0.15 ).
172
Several researchers have studied how to correctly model the slab to achieve a realistic
representation of the ultimate moment capacity of composite beam sections in positive
bending. The primary modeling concern is defining the effective width and the effective
compressive stress of the slab. This information is important because it is used in the
seismic design criteria (i.e. strength check and strong-column weak beam criterion) as
well as in the analytical model. The effective width of the slab that is cited in design
standards is typically intended for the evaluation of the elastic stiffness and strength near
the mid-span of a composite beam. For example, the AISC definition of the effective
slab width ( b AISC ) is defined in equation 2.10 and shown graphically in Fig. 4.16. While
the AISC criteria is valid for composite beams under gravity loading, it is unclear
whether or not this effective width can also be applied in cases when lateral loads causes
hinging at the beam ends in a moment frame. For beam hinging in the region adjacent to
the column, the effective width of the slab at the ultimate strength of the section is more
likely related to the slab area which is in direct contact with the column flange (du Plessis
et al 1972). The validity of the effective slab width assumptions described here (and
shown in Fig. 4.16) will be evaluated later in this section.
Past research has shown that the concrete compressive strength can reach an effective
stress greater than f c' due to the high confinement of the localized bearing region of the
slab near the column flange. For example, du Plessis et al. (1972), Tagawa (1989), and
Lee (1987) report ultimate effective stresses in the range 1.3 to 1.8 f c' (du Plessis et al.
1972, Tagawa 1989, and Lee 1987). On the other hand, other researchers (Civjan et al.,
2001 and Cheng et al., 2002) who have performed beam-column subassembly tests report
that the ultimate effective stress of the concrete slab was approximately 0.85 f c' over the
column width. However, in these cases the shear studs were not designed to develop
fully composite action (25-35% for Civjan et al. [2001] and 75% for Cheng et al. [2002],
with percentages based on AISC-LRFD definition of composite action) and the beams
often failed by shear stud rupture before the concrete was fully mobilized. Based on
Bugeja et al. (2000) composite beam-column subassembly tests suggest that the effective
173
width is equal to the AISC design specification (Manual 2001) with an effective concrete
stress of 0.85 f c' . However, it can be shown that the strength calculated using the AISC
width and bearing stress of 0.85 f c' is roughly equivalent (within 10%) of that calculated
assuming 1.3 f c' over bslab = bcol .
Figure 4.17 shows the test setup and dimensions for the composite beam subassembly
experiments that were considered in this study. Table 4.2 briefly summarizes the key
failure mechanism and general observations for each of these tests. A few of these tests
were simulated in OpenSees using fiber beam-column elements to validate the effective
slab width and stress that is appropriate to use for both design and analysis purposes. In
addition to these two parameters, the strain at which the slab reached its residual strength,
C 2 , (described in section 4.2.1) is another important parameter that affects the ductility
of the slab and influences the element response. The strain corresponding to the
maximum concrete strength ( cc ) is assumed to be -0.002 and the residual stress of the
concrete after reaching C 2 is assumed to be 20% of f cc' . The tensile strength and the
tensile softening properties of the slab are defined in a similar manner to that proposed
for the core concrete (Section 4.2.2.1). The individually calibrated material properties for
each of the investigated subassembly tests are summarized in Table 4.3. Those that were
not simulated in OpenSees are reported as not analyzed, although the recommendations
of the researcher are presented in the description column of this table. Note that in the
test by Lee (1987), hinging occurred in the steel columns rather than the composite
beams and therefore the ultimate strength and effective width of the slab cannot be
deduced from this test.
The effective stress, slab width, and the slab ductility parameter ( C 2 ) were calibrated to
each test and the best-fit parameters are listed in Table 4.2 for each of the simulated
tests. For all but the Uang test, the calibration study shows that the effective width of the
slab is best represented by the width of the column ( bslab = bcol ) where the concrete
compressive strength is 1.3 f c' . The Uang test is unique given that the test is setup as a
174
cantilever where the slab not only bears against the column but also against a strong wall
just behind the column. This setup mobilizes a larger effective slab width that is better
represented by approximately 2.5 bcol when the effective stress is assumed to be 1.3 f c' .
The results from the calibration study showed that, as expected, the effective stress of the
concrete is highly correlated with the design and performance of the shear studs. Those
tests where the shear studs were designed to provide a fully composite section (Tagawa
1989, Bursi and Ballerini 1996, Bugeja et al. 2000) performed very well (i.e. higher
moment strength and larger ductility) and attained effective slab stresses of
approximately 1.3 f c' . These well-behaved tests also show that a value of C 2 = 0.05 is
appropriate to reflect the ability of the slab to achieve high ductilities during larger
excursions.
On the other hand, the tests where the shear studs were designed for a partially composite
section did not perform nearly as well (i.e. less than design strength and low ductility),
with the effective slab stresses reaching only 0.85 f c' for the tests reviewed in this study
(Cheng ICLCS, INUCS and ICLPS 2002, Civjan et al. 2001). Given that the shear studs
are the weak link in these sections, the studs limit the amount of stress that can be
developed in the slab and also control the level of ductility that the section can attain.
Therefore, the stress in the slab is really a function of the shear capacity of the studs, and
the effective stress of 0.85 f c' may deviate if the studs are designed for different levels of
composite action. The ductility of a partially composite section is much less than if it
were designed as fully composite. Accordingly, for these cases, a lower calibrated value
of C 2 as 0.012 is used. This large difference in behavior in both strength and ductility
demonstrates why it is more attractive to ensure a fully composite section rather than
permitting the shear studs to be the critical element in seismically designed beams.
Given the results of this composite beam calibration study, a fully composite beam can be
modeled with an effective slab width equal to the column width ( bcol in Fig. 4.16) and a
175
maximum effective concrete stress of 1.3 f c' . These assumptions should be applied when
computing the plastic moment strength of the composite beam section for the strong-
column weak-beam design check as well as for analytical modeling purposes. The results
from the analytical models using these assumptions are compared against the
experimental results and shown in Figs. 4.18 through 4.28. For those tests results in Fig.
4.23-4.28 that were designed as partially composite, the analytical models were treated as
a fully composite section to emphasize the adverse effects of a partially composite
member. Although partially composite beams were briefly investigated in this calibration
study, the focus in this work will remain on fully composite beams. The results from the
simulations demonstrate that OpenSees is able to capture the overall behavior of these
elements.
The behavior of a composite beam will differ based on whether or not the metal deck is
running parallel or perpendicular to the main girder. Apart from differences in the
effective slab area, when the deck is parallel to the beam, the upper flange of the beam is
continuously braced by the slab, thereby significantly reducing the probability of
buckling in this region during composite bending. When the flutes are running
perpendicular to the beam, the upper flange is only braced when the ribs of the deck are
in direct contact with the flange, which leaves the upper flange more susceptible to
buckling in the unbraced zones. Another factor to consider is the relative effectiveness of
shear studs, depending on the deck orientation. While these aspects of behavior are not
specifically investigated in this study, they are issues that should be considered when
modeling the inelastic behavior of composite beams in moment frames.
4.2.2.3 Convergence issues
A couple of important comments should be mentioned here to improve numerical

convergence of local member and global solution operations in OpenSees. If the fiber
cross-section contains a non-ductile or softening material (i.e. concrete), then it is useful
to finely discretize this region of the section (i.e. increase the number of fibers) so that
this softening effect is smoothly captured and sudden large changes in stress in the
176
section are minimized. However, this should not be overdone, as it increases the
computational time and could potentially lead to numerical instabilities. Using the
analytical model described in section 4.3 as an example, the core of the 650-mm square
RC column section is meshed with approximately twenty 30-mm thick fibers.
The second issue that affects convergence is the number of integration points specified
along the length of the fiber beam-column element. Using the Gauss-Labotto integration
scheme, the information at each of the fiber sections are used to compute the response
parameters at the two end nodes. By increasing the number of integration points, the
nonlinear curvature can be better represented along the length of the element. Increasing
the number of sections to a certain point has a positive effect on element convergence,
although if the number is increased such that multiple consecutive sections are within the
nonlinear hinging region, then convergence problems can arise (particularly when the
section exhibits softening behavior). It should also be recognized that increasing the
number of integration points is roughly proportional to the amount of operations
performed on each element. Based on the experience gained from this study, it is
recommended that the number of integration points be set to 7.
There is also an option in the OpenSees nonlinear beam column element to control the
residual error at the element level by applying iterations within the element flexibility
solution. While this option increases the local element computation, it is highly
recommended since it improves the global convergence of the OpenSees model. If this
iterative approach is not applied at the local level, then the error is filtered into the global
system of equations and dealt with at a global level. Experience in this study is that
reliance on the global system to control the error leads to convergence problems.
Therefore, it is recommended to use the optional error control routine in the flexibility
element and to specify a maximum number of iterations of 10 with a minimum tolerance
of 1x10-18.
177
4.2.3 Composite Joints Elements
The composite joint panels are represented in OpenSees using a 2-dimensional joint
element (Fig. 4.29) that can accurately model the finite joint size effect and the
kinematics of joint deformation. This element has the capability to model large
deformation response of the joint panel, however, this option was found to be
unnecessary in this study given the drift levels examined. The joint hysteretic properties
are defined by an internal rotational spring which can accommodate any number of
moment versus rotation springs.
To accurately model the deformation behavior of composite RCS joints, the internal joint
spring must represent both the panel shear and vertical bearing deformation. Panel shear
deformation possesses characteristics similar to that of the steel model presented in
section 4.2.1 (e.g. Steel02). This mechanism relies on yielding of the steel web panel and
development of diagonal concrete struts, which exhibit fat and stable hysteretic loops, as
shown in Fig. 4.30. Vertical bearing deformations exhibit a pinched hysteretic response
associated with the local crushing of concrete and the opening of gaps above and below
the steel flanges. The hysteretic material model presented in section 4.2.1 can be utilized
to fit the response of vertical bearing deformations, as shown in Fig. 4.31. These two
springs are implemented in series to model the total deformation response of composite
RCS joints.
The two spring models (joint shear and joint bearing) are calibrated against beam-column
cruciform tests from Kanno (1994). These series of tests are also used in the RCS joint
strength model validation studies presented in Chapter 2; and details of these tests are
provided in Section 2.4.9.1. The panel shear (Mps) and vertical bearing strength (Mvb,total)
of each of these tests are determined using the proposed design equations 2.33 and 2.34,
respectively. The test setup and the displacement time history are simulated in OpenSees
and the beam shear versus total joint rotation is compared to the measured test data. Four
of these joints failed in predominately a vertical bearing mode (OJB1, OJB4, OJB5, and
178
OJB6-1), with the measured panel shear rotations remaining within the elastic limits.
These tests provide the best opportunity to calibrate the vertical bearing hysteretic spring
of these composite joints. The tri-linear backbone of the hysteretic model, as well as the
cyclic pinching and damage factors described in section 4.2.1, were adjusted in a manner
to best fit all four of the tests investigated. The initial stiffness of the bearing spring
follows the recommendations from Kanno (1993):
2M vb
K eb = , bo = 0.01 0.008 p (4.8)
bo
where:
Keb = initial stiffness of the vertical bearing spring
bo = rotation at which the applied moment is equal to Mvb
p = ratio of the applied axial force to the squash load of the column.
The calibrated backbone of this model is presented in Fig. 4.31, with M vb being the
strength of the joint in bearing. The pinching factors, x and y , were determined to be
0.2 and 0.5, respectively; and the damage factors, d and d E , were chosen to be 0.015
and 0.0, respectively. The calibrated results from these joints are shown in Figs. 4.32
through 4.35.
Nine of Kannos tests can be classified as failing primarily in a panel shear mechanism,
but unlike the four previously described joints, these joints also have considerable
deformations in the vertical bearing mode. This interaction of the two failure modes
makes the calibration of the panel shear spring slightly more challenging since the panel
shear mode is not completely isolated from the bearing mode. Given that the parameters
of the bearing spring have already been set, these nine shear failure tests will be used to
calibrate the panel shear spring. The specified yield strength of the panel shear (Steel02)
model should be set to the computed joint shear strength ( M ps ), as determined from
equation 2.33 in Section 2.4.8. The stiffness of this spring follows the recommendations
from Kanno (1993) per the following equation:
2M ps
K es = , so = 0.01 0.0067 p (4.9)
so
179
where:
Kes = initial stiffness of the panel shear spring
bo = rotation at which the applied moment is equal to Mps
The kinematic hardening slope is taken as 2% of the initial stiffness, Kes. The remaining
parameters in the Steel02 model that define the transition of the curve between the elastic
and the hardening branch are set to the following values: R0 = 18.5 , c1 = 0.96 , and
c 2 = 0.15 . The calibrated results for these nine joints are shown in Figs. 4.36 through
4.44. Note that the pinching behavior in these hysteretic plots is primarily coming from
the contribution of the joint bearing spring. In several of the tests (such as OJS7-0, HJS1-
0, and HJS2-0), the pinching is not adequately captured. This phenomenon has less to do
with the calibrated hysteretic models of the joints springs and more to do with the actual
prediction of the relative joints strengths. If the panel shear strength is slightly
underestimated then it is possible that the joint deformations localize in this spring. On
the other hand, if the prediction of the shear strength had been slightly increased, the
relative strengths of the two springs may be such that the deformations become more
distributed between the two springs. This provides some insight into the variability that
is inherent in these nonlinear models. Figure 4.45 shows the contribution of each spring
to the total rotation for the OJS3-0, OJS4-1, and OJS5-0 tests. Here, it is evident that the
two springs in series predicts larger panel shear rotations and underestimates the bearing
deformations in this test. Once the panel shear spring begins to yield it limits the amount
of force transmitted through the bearing spring and therefore dominates the inelastic
deformation response. This type of behavior is a result of the fact that the actual panel
shear strength is slightly underestimated in these cases.
4.3 Test Frame Validation Study
An analytical model of the 3-story test frame discussed in Chapter 3 is created using the
elements and techniques described in Section 4.2. Every effort is taken to reasonably
account for laboratory conditions such as member dimensions, measured material
strengths, and test loading protocol. The ultimate goal is to compare the analytical to the
experimental response in an attempt to validate the nonlinear frame elements described in
180
this chapter. It should be acknowledged that differences will undoubtedly occur between
the analytical and experimental response, but ultimately the point is to verify where and
why these discrepancies occur, how the models could be improved to better capture the
behavior, and what applications are appropriate for the current models given the
limitations present.
A general schematic of the OpenSees model is presented in Fig. 4.46. The reinforced
concrete columns and composite steel beams are represented by the flexibility-based fiber
beam-columns described in section 4.2.2.1 and 4.2.2.2, respectively. Each of the
composite joint panels is represented by the 2-dimensional joint element described in
section 4.2.3. Bond-slip springs, discussed in Section 4.2.2.1, are placed between the 1st
floor RC columns and the fixed base since large inelastic rotations are expected in this
region and bond-slip will tend to play an important role in global deformations. The 3rd
floor columns are discretized into two elements to model the change in longitudinal
reinforcement at approximately one-third the height of the floor. Fictitious leaning
columns are also modeled to simulate the P-delta effect caused by the gravity loads from
one-half of the total building area (which equals approximately 3500kN at each floor).
The leaning columns are not meant to provide any flexural resistance, so each joint must
be free of rotational restraint. This is accomplished by using co-rotational truss members
(2nd floor column and 1st-3rd floor rigid links) mixed with elastic beam-column elements
(1st and 3rd floor columns). All of these elements have a very large axial stiffness to
ensure that there are no unrealistic vertical deformations in the leaning column when the
gravity loads (equal to one half the building weight) are applied.
The dead and live loading and the seismic mass in the analytical model follow the same
assumptions as presented for the test frame in Section 3.3.2. The pseudo-dynamic
loading protocol is simulated in OpenSees using the same input ground motions applied
at the base nodes of the structure (e.g. the actuator forces are not simulated, but rather the
base nodes are excited using the same earthquake acceleration time histories). These
events are applied consecutively to ensure that the cumulative damage effect of the test
frame is represented in the analytical model. Twenty seconds of zero ground acceleration
181
is provided between each ground motion excitation in order to damp out all of the
transient vibrations between each loading event.
4.3.1 Errors within the Pseudo-Dynamic Testing Method
Before proceeding with the comparison of the measured and simulated responses, it is
important to recognize that there are a couple sources of error inherent in the pseudo-
dynamic testing algorithm that could potentially lead to variations from the actual (real
time) dynamic frame response. Two of the most important errors are described below:
1. Since pseudo-dynamic testing is performed quasi-statically, strain rate effects can
have an impact on the inelastic behavior of the test frame. During the hold
phase of the pseudo-dynamic algorithm (Fig. 4.47), the data acquisition system
collects the data and the system of equations is solved to obtain the next
displacement increment. During this stage, the actuators are maintaining a
constant displacement (within some tolerance) over a period of time, which can
lead to some relaxation of inelastic hinges, particularly in RC columns where
cracking and other types of softening occur in real time. As a result, the inelastic
strength of a system is underestimated when tested quasi-statically as compared to
a system that is tested dynamically (Mahin et. al., 1985).
2. It is impossible to impose the exact computed displacements to the structure with
actuators, so a tolerance is set at each time integration step. Even with a very
small error (0.1mm), these errors can build up over the duration of the
earthquake and can cause errors in the algorithm. In multi-degree of freedom
systems, higher modes contributions are highly vulnerable to these errors and can
even make the test system unstable. The pseudo-dynamic algorithm implemented
in the test of the composite RCS frame used the explicit Newmark method as the
numerical integration scheme. This particular method has been found to
minimize the propagation of error that is a result of experimental inaccuracies
(Mahin et. al., 1985). Regardless, this is less of a problem in the 3-story test
frame given that the effect of higher modes is relatively small.
182
Therefore, it is important to recognize when evaluating the analytical models in this
section that the differences between the measured and simulated response may not be
completely due to problems within the modeling assumptions. Potential manifestations
of these errors will be discussed in the following sections.
4.3.2 Comparison to Experimental Global Response
The first-mode period of the analytical model ( T1 = 1.0sec ) matches the measured period
of the test specimen exactly, indicating that the assumptions influencing the initial elastic
properties of the frame have been adequately captured. The test frame was subjected to a
couple preliminary pseudo-dynamic events that were well within the elastic region of the
response which were also captured quite well by the analytical model. These results were
not reported in Chapter 3, but they revealed excellent agreement for imposed drift ratios
up to 0.1%. Both of these checks were used to initially validate the analytical model as
well as the test setup and the pseudo-dynamic algorithm.
The analytical model is subjected to the same loading protocol as the test specimen,
including the truncated records and the realignment (re-straightening loading) imposed
after the 2/50 event (see Section 3.4.2 for further description of the loading protocol).
The results from each of these tests are presented in Figs. 4.48 through 4.52, each of
which contains the time histories for the roof displacement and base shear as well as the
maximum and minimum of story shear and interstory drift ratios. The first event, the
50/50 Chi-Chi ground motion, is the least damaging of all events and the measured and
calculated response plots of the frame (roof displacement and base shear time history)
agree well up until approximately 25 seconds, beyond which there seems to be a phase
shift developing between the experimental and analytical histories. This shift is
attributed the change in natural period due to inelastic softening that is occurring in the
test frame but is not accurately captured in the analytical model. This softening behavior
is likely a result of cracking in the slab and columns and perhaps from slippage in the
bolted steel beam splices. In addition to this, inherent features of the pseudo-dynamic
loading algorithm, described in section 4.3.1, may be causing some of the softening
183
response due to relaxation of force in the RC column hinges. This period shift can be
tracked by performing a Fourier analysis on a sliding window of the time history
response. The results of this analysis are shown in Fig. 4.53 and 4.54 for the analytical
and experimental response, respectively. The contours in these figures represent the
peaks of the power spectral density function, which correspond to the principal frequency
(or the inverse of the period) during that time region of the event. These figures show
that while the experimental response clearly softens to 1.3 seconds (60% of original
stiffness), the analytical response hovers just above 1.0 seconds. In spite of this lack of
agreement in the change of stiffness and period shift, the maximum and minimum
interstory displacements and story shears (Figs. 4.48c and 4.48d) are captured fairly well,
with differences in displacements ranging from 2% to 30% and story shears from 3% to
10%.
Analytical versus experimental comparisons of the first design level event (10/50 Loma
Prieta 1a and 1b) are shown in Figs. 4.49 and 4.50. The residual drifts in both the
analytical and experimental models have been removed in these and all subsequent plots
so that only the transient response is compared. During the first truncated event (Fig.
4.49), the analytical model is within 35% of the peak displacements and 21% of the
maximum base shear of the test frame. The roof displacement of the second complete
event (Fig. 4.50) shows that the analytical model, other than the first large excursion,
consistently underestimates the deformations of the frame. Calculated peak
displacements are within 33% of the measured values. The base shear (Fig. 4.50b, 4.50b)
is captured rather well by the analytical model during the larger excursions of the record.
After the first couple of large excursions, the phase shift again becomes apparent in both
the roof displacement and base shear time histories. Predominate frequency versus time
graphs are shown in Figs. 4.55 and 4.56 for the analytical and experimental results,
respectively. These graphs can been further interpreted by extracting the periods at
which the peaks occur at each time step and plotted versus time as shown in Fig. 4.57.
This figure shows that while there are some times where the predominate period of the
analytical response is large; most of the time it remains around 1.0 second. This is in
contrast to the period of the experimental response which lengthens to approximately 1.7
184
seconds. Again, the cause of this shift is perhaps a combination of the stiffness
degradation in the analytical models and the force relaxation effect caused by the pseudo-
dynamic algorithm.
Figure 4.51 summarizes the analytical and experimental comparisons for the maximum
considered earthquake event (2/50 Chi-Chi). It is apparent from the roof displacement
time history (Fig. 4.51a) that the analytical model is responding with a higher frequency
(shorter period) than that of the test frame. This is evident in the predominate period
versus time plot shown in Fig. 4.58, which shows that the test frame is responding at a
period of approximately 1.7 seconds while the analytical response is predicting about 1.3
seconds. It should be noted, that severe local beam flange buckles occur during about 18
seconds into the record, which cannot be modeled by the fiber section element. This
additional flexibility in the test frame is evident in the longer period and much larger
interstory drifts observed during this event, particularly after about 20 seconds into the
record (Fig. 4.51c). The excursion at approximately 24 seconds is not picked up by the
analytical model and results permanently offset between the two responses. Differences
between the model and the test frames maximum displacements range from 2% to 73%,
while the agreement in story shears is much better, with a range from 4% to 16% between
the measured and calculated values.
The final pseudo-dynamic loading event repeated the design level event (10/50 Loma
Prieta). The results of the analytical model are compared with that of the test frame in
Fig. 4.52. As shown in Fig. 4.52a, the model drastically underestimates the first large
excursion of the test frame that occurs at about 5 seconds into the event. This is
important because beyond this point the test frame never seems to recover from this
inelastic push and simply oscillates about the -100mm position. This is particularly
evident at the end of the record (40 seconds) where the analytical model is oscillating at a
drift of less than half of that of the test frame. Just as in the 2/50 event, it is evident that
the fundamental period of the test frame has lengthened beyond that of the analytical
model (1.7 to 2.0 seconds versus 1.3 seconds, respectively), as is shown in Fig. 4.59. The
predicted base shear history (Fig. 4.52b) consistently overestimates the true value which
185
is probably a result of two things; (1) the limitation of the fiber elements to model local
buckling and strength reduction of in steel beams and (2) the idealized behavior of the
RC column base hinges and their strength deterioration. As described in Section 3.6.2,
prior to this event, most of the severely hinged elements were thought to be at a near
collapse damage state, or past the level at which repair was still an option. The
associated behavior modes, particularly the extensive beam local buckling, are not
captured by the analysis, which explains the inaccuracies in modeling the overall frame
response. The errors in predicting the maximum displacements for this event were in the
range of 2% to 30%. The story shears were predicted with an error that ranges between
3% and 50%.
The final pushover is also simulated in the analytical model and the comparison to the
experimental results is shown in Fig. 4.60. It is quite apparent that in the initial stages the
analytical model is significantly stiffer than the actual test frame (approximately 1.9
times as stiff). This again confirms that the simulated model underestimates the amount
of stiffness degradation that takes place in the test frame. Despite the large difference in
stiffness, the ultimate strength of the analytical model is within 7% of the measured
strength of the test frame. This indicates that the plastic strengths of the composite
beams, RC columns, and composite joints are captured rather well by the analytical
model.
4.3.2.1 Comparison of First and Second Design Level Event
As discussed in Section 3.4.2, the 10/50 1989 Loma Prieta event was applied before and
after the maximum considered event (1999 Chi-Chi record) to examine the response of a
heavily damaged frame. Analytically, this allows the opportunity to validate how well
the models capture stiffness and strength degradation that occurs through these events.
The actual measured roof displacements for both the first and second 10/50 Loma Prieta
events are shown in Fig. 4.61. As shown, there is a quite noticeable difference between
the measured response of the first and second event, with a significant phase shift
occurring in the latter event. The elongation of the period between these events has been
186
discussed in Section 4.3.2. This same plot is shown for the analytical results in Fig. 4.62.
One point to notice between the two simulated events is that they seem to be oscillating
about different permanent displacements at the end of the record, where differences are
first seen at the excursions between 5-6 seconds. Despite these differences, which seem
to stem from only a couple of the major excursions, the two responses look strikingly
similar and show only a very small phase shift (not nearly as large as in the measured
results). This again points to the lack of stiffness degradation in the analytical models,
where even after being subjected to the 2/50 event, the stiffness of the frame did not
experience an appreciable degradation.
4.3.3 Comparison to Experimental Local Response
Over three hundred data channels were recorded during the test, with a large majority of
these dedicated to tracking the local response of the beams, slabs, joints, and columns.
While a select number of the plots were shown in Chapter 3, a more complete set is
presented in Cordova et al. (2005). This section will extract a couple of the measured
beam, column, and joint hysteretic plots for each pseudo-dynamic event and compare
them to the analytical results predicted by OpenSees. These plots are shown in Figs. 4.63
through 4.77. The OpenSees response is represented as a dashed line while the measured
test response is shown as a solid line. The moments for the instrumented response are
estimated from strain gauges in the elastic portions of the beams and behavioral
assumptions informed by the OpenSees results (i.e. column inflection points and shear
distribution). It should be recognized that the hysteretic response plotted from the
measured values are subject to a number of assumptions in the force distributions and
when pushed to larger excursions there are some errors due to disruption of the tiltmeters
(e.g. large local buckling caused out of plane distortions in the tiltmeters).
A couple of the composite beam hysteretic plots are shown for each event in Figs. 4.63
through 4.67. In Fig. 4.63, it is evident that the analytical models are picking up the
initial stiffness of the composite beams and correctly predicts that they remain relatively
elastic during the 50/50 event. The first design level event pushes the beams further into
187
the inelastic range (Figs. 4.64 and 4.65), which seems to be accurately captured by the
fiber beam elements. The 2/50 event produce the largest inelastic excursions of all
pseudo-dynamic events, which resulted in the formation of large local buckles in the
lower flanges of the steel beams, a phenomenon that cannot be modeled with a simple
fiber section element. One of the primary assumptions in the fiber models is that plane
sections remain plane, which allows the modeling of steel yielding and concrete crushing
and cracking. This fundamental difference in behavior is apparent in Fig. 4.66, with the
1st-floor beams (1B1S050) reaching measured rotations of approximately -0.07 radians
during the largest excursion, which is underestimated by the analytical model by a factor
of 3.5. This shortcoming is something that must be acknowledged when using fiber
beam-column sections and can only be corrected by either (a) supplementing this model
with semi-empirical degrading material fibers, or (b) by using a different type of element
altogether (e.g. nonlinear hinge). The composite beam behavior in the final 10/50 event
is generally modeled well, as shown in Fig. 4.67.
Selected hysteretic response plots for the RC columns are shown in Figs. 4.68 through
4.72. As shown in Fig. 4.68, the analytical model correctly predicts the stiffness and a
moderate amount of yielding in the 1st floor column base hinges. For the first design
level event (10/50-1a and 1b in Figs. 4.69 and 4.70), the fiber models seem to capture the
strength of the columns rather well, yet fail to pick up some of the pinching that is taking
place in these hinges. This is similar to the 2/50 event (Fig. 4.71), where the strength is
captured fairly well, but the amount of pinching that occurs in the test is not modeled as
well as one would like. This difference in behavior again has to do with the fundamental
assumption in fiber models that plane sections remain plane. In the case of the RC
column hinges, there is likely slip occurring between the reinforcing bars and the
surrounding concrete within the hinge zone, which would tend to soften the response
when unloading and reloading and therefore lead to more of a pinched response before
attaining its full plastic moment. The final 10/50 event also exhibits similar trends, where
the analytical model represents the strength well but underestimates the hysteretic
pinching response. It is also apparent that while the stiffness of the RC columns has
degraded throughout each of the loading events, the analytical models roughly maintain
188
their original stiffness. This is most apparent in the comparison of response in the 2/50
event (Fig. 4.71) and the final 10/50 event (Fig. 4.72).
Samples of the joint shear versus joint rotation plots for all events are shown in Figs. 4.73
through 4.77. Results from the 50/50 event (Fig. 4.73) show that the initial stiffness of
the analytical model matches that of the measured response rather well, yet there is some
very minor inelastic action that is not being picked up by the OpenSees model. The
moments of the 1st floor joint (1J3 in Fig. 4.73) for the 10/50-1a event seems incorrectly
estimated as the stiffness of the joint is much lower than anticipated. This is thought to
be the result of errors in predicting the internal forces in the test frame based on the
estimation procedure described in Cordova et al. (2005). Unfortunately, these errors
cannot be resolved since there were no direct measurements of the forces in the
indeterminate test frame. Joint 2J3 in the first design event (10/50-1a and 1b in Figs.
4.74 and 4.75) attain roughly the same amount of joint rotation as the 50/50 event, which
seems to be sufficiently replicated in the analytical model. On the other hand, joint 1J3
seems to show slightly more joint rotation than the analytical model, which can perhaps
be attributed to the fact that these measurements were inferred from surrounding beam
and column tiltmeters and may include additional rotation from these members as well.
During the 2/50 event, joint 1J3 (Fig. 4.76) shows a considerable amount of inelastic
cycles, which is considerably underestimated by the analytical joint model. Again, given
the level of damage observed in these joints, these large measured joint rotations seem
unrealistic and are likely a result of problems with the tiltmeters. Fig. 4.77 shows that the
joint 2J3 is remaining relatively elastic during the final 10/50 event, which is adequately
simulated by the analytical joint model. The analytical model slightly underestimates the
amount of inelasticity and energy dissipation that occurs in joint 1J3. The fact that the
level of inelasticity is not well represented by the analytical model for the last two events
may have less to do with the validity of the joint model and more to do with the
interaction of these joints with the surrounding elements and the distribution and
redistribution of forces during the event. In other words, some small errors in modeling
the surrounding elements may lead to an incorrect force distribution that does not push
the joint models to large moments which can produce inelastic joint response.
189
4.3.4 Residual Drifts
The importance of residual displacements in a structure after a major earthquake and their
role in performance based earthquake engineering has recently become a topic of much
interest (Ruiz-Garcia 2005). The evaluation of these permanent displacements is
important for both the repairability of the building and its capacity to withstand
aftershock excitations. The results from this test frame provides an opportunity to verify
how well the residual deformations in a moment resisting frame can be captured by an
structural analysis program such as OpenSees. Each of the pseudo-dynamic events
produced a finite amount of residual drift in the test specimen. For all of the events
through the 2/50 loading event, the residual drifts were allowed to accumulate in the test
frame, where each subsequent event was imposed to the frame relative to its permanently
deformed state. After the 2/50 event, the decision was made to straighten the frame as
much as possible to avoid exceeding the stroke limit of the actuators during the final
10/50 loading event. This straightening procedure is simulated in the analytical model so
as to provide a realistic simulation of the full loading history.
The measured and analytical residual drifts after each pseudo-dynamic loading event are
compared in Fig. 4.78, where the thicker of the two lines is the measured value and the
thinner line is from the OpenSees simulation. The prediction error is also reported on the
figure. For the first 50/50 event, the measured residual drifts are relatively small (0.1%
residual interstory drift), which makes the comparison to the analytical predictions not
very interesting, particularly in the 1st floor where the analytical prediction is off by less
than 3 millimeters yet the error percentage is close to 200%. Measured residual
deformations increase to 0.5% in the 1st floor during the second event (10/50-1a), with
the predictions by the analytical model fairing rather decently with a general
representation of the deformed shape and maximum error of 33%. For the last three
events (10/50-1b, 2/50, and 10/50-2), the analytical prediction of the residual
displacements are not very accurate, with prediction errors ranging from 45% up to
134%. These large errors in the prediction of residual displacements should be expected
190
given the great number of factors during an inelastic time history analysis that can
influence the final deformed state of the building. At the element level, some of these
include the variability of the strength and stiffness of the models, the degradation of
stiffness, and the unloading and reloading stiffness. These in turn influence the softening
of the structural model which therefore affects how the model responds to the excitation.
It is easy to see that if any one of these parameters is altered either in the input or as a
result of the excitation, the remainder of the event will be affected and the permanent
deformations will be changed as a result. It is difficult to foresee a model that would be
able to consistently and accurately capture the residual deformations on such a large scale
given all of the variables that influence the results. This suggests that a probabilistic
approach is required to evaluate the residual drift demand on a structure where the
variability of several of the key modeling parameters is considered over a suite of ground
motion records. This sort of approach has been recently investigated by researchers such
as Ruiz-Garcia (2005).
4.3.5 Time Evolution versus Predetermined Analysis
There is an important distinction to make on how analytical models are typically

calibrated versus how they ultimately perform in a nonlinear time history analysis. The
typical method to validate models is to compare their response to a series of subassembly
tests, as was done in Sections 4.2.2 and 4.2.3. The way this is typically accomplished is
by subjecting the analytical model to the displacement history that was imposed on the
subassembly test and then compare the force versus displacement hysteretic response.
Given that the displacements were predetermined in this process and the analytical model
is guaranteed to reach the same drift levels, the comparison between the model and the
test comes in the form of the maximum strengths, the initial, unloading, and reloading
stiffness, and the strength and stiffness deterioration. This is not an exhaustive list, but it
does cover some of the key parameters typically scrutinized in calibration studies. In the
end, the goodness of fit of the model is based on how well the simulated response
represents the measured response. This process is then repeated for multiple
subassembly tests and compromises are made in each of the tests in order to generate
191
final recommendations that can be generally applied to a range of component behaviors.
A typical result of this process is shown in Fig. 4.79, where the hysteretic model has been
fit to a number of composite joints tests in order to produce the results seen in the figure.
While the calibration shown in Fig. 4.79 is considered to be a good fit for the composite
joint model, there are several subtle differences that are pointed out in the figure that can
potentially lead to large differences in response during a time evolution analysis. These
subtle differences are typically always present in these types of calibration studies. In a
time evolution analysis, both the force and displacement are unknown prior to the test,
and all of these subtle differences play a more critical role in the evolution of the
response of the structure than compared to the displacement controlled simulations of
subassembly tests.
4.3.6 Comments on Analytical Models
It is evident that there are some limitations to the use of the fiber beam-column elements,
one of which stems from their fundamental assumption that plane sections remain plane
(PSRP) and normal to the longitudinal axis. This assumption ensures that all nonlinear
behavior in the section is a result of material nonlinearities (i.e. steel yielding, concrete
crushing, etc.) and all strains and stresses act parallel to the longitudinal axis of the
member. In the behavior witnessed in the test frame, this assumption is violated in three
important ways:
1. During large inelastic excursions, the steel beams experienced local buckling in
lower flange and lower half of the web. This behavior resulted in strength and
stiffness degradation in the hinge region, contributing to an overall increase in
frame flexibility that ultimately led to much larger floor displacements than what
was predicted in the analytical model. Such behavior is not represented by the
fiber models.
2. It is evident that there are cases where the pinched response of the RC column
hinges are not adequately captured by the fiber section model and it is likely that
bond deterioration and slip within the hinging zone is a large contributor to this
behavior. The rebars and the surrounding concrete do not have perfect bond and
192
slip occurs along the interface of the two materials delaying the mobilization of
the force in the rebar. This phenomenon softens the response of the hinge which
leads to the pinched behavior in the measured response.
3. The RC column base hinges were subjected to significant plastic rotations
throughout the loading history of the test frame. As a result, the concrete within
these hinges was significantly deteriorated to the point that during the final
pushover test the rebars were carrying nearly all the moment and shear. This type
of failure seemed to be a result of a combination of extensive plastic hinging
followed by progressive shear failure. The fiber element model does not capture
this sort of failure mechanism.
The impact of the first two behavioral effects can be significant, and tends to increase the
flexibility of the hinges and the entire system. Both of these effects violate the basic
PSRP assumption of the standard fiber section model and, therefore, are difficult to
incorporate into these models without introducing adhoc techniques such as manipulating
material properties. A logical alternative would be to provide an additional spring in
series with the fiber elements to account for either the bond slip or local buckling
behavior. While this conceptually makes sense, these springs would end up dominating
the response of the hinge once they began to soften therefore limiting the amount of
plastic rotations that would occur in the fiber section. This sort of approach would be
better suited to simply model all nonlinearities in a rotational spring using more of a
lumped plasticity approach. The third point is important as well, although it became
important only after significant deterioration of the column hinge during the final
pushover of the frame. It seems unreasonable to expect the fiber beam-column element
to represent the behavior of this column after such a significant amount of damage. This
type of behavior may be better suited to be represented by a P-M yield surface model
(Kaul 2004) where the hysteretic properties can be controlled up through significant
strength and stiffness loss.
In addition to the bond slip effect on the column stiffness, there was also an overall
degradation of stiffness over the series of loading events that was not adequately captured
by the analytical model. This resulted in a phase shift between the predicted and
193
measured response, which indicated that the period of the test frame was elongating more
than that of the analytical response. In general, once the first-mode period of two models
begins to deviate, this leads to a divergence in the response and it becomes quite difficult
recover and converge to a similar answer. In addition to bond slip and local buckling,
there are several likely sources that contribute to this degradation of stiffness:
1. The measured response of the RC columns shows a general degradation in
stiffness that may be a result of the deterioration of the concrete in the hinge zone.
Another contributing factor is the degradation of the unloading and loading
stiffness which seems to stem from the previously described bond slip issue.
2. Inelastic shear deformations in the RC columns may also contribute to the
flexibility of the test frame and the elongation of the natural period.
These issues can be significant in terms of the final behavior of the moment frames. The
issue with modeling bond slip has already been discussed, but the new issue of
deterioration of concrete can be handled within the material model level by incorporating
a stiffness degradation parameter such as one that might be a function of cumulative
plastic strains or other damage indices. Shear deformations can be incorporated into the
analytical model using inelastic shear springs in series with the fiber element models.
While first thought to be relatively insignificant, under the large interstory drifts some
shear cracking was observed in the 1st floor columns.
An important engineering demand parameter that directly translates to the cost of repair
is the amount of residual drift in a building following a large earthquake. In this study it
was found that the permanent deformations are perhaps the most difficult index to
capture in a frame subjected to dynamic loading. The residual displacements are
dependent on a number of different factors, all of which can significantly affect the final
results. This potential for variability suggests the need for a probabilistic evaluation of
the residual drift demands on a structure.
It is evident from the response of the test frame and the comparative studies of the
analytical models that the composite action between the steel beams and concrete slab
was maintained throughout the entire loading protocol. Recall that this was also
194
validated in the test specimen by examining the condition of the slab and the shear studs
after the completion of the test (see Section 3.6.3). The behavior of the composite beams
reinforces the recommended calibration parameters presented in section 4.2.2.2.
4.4 Damage Indices
In this section, the OpenSees results are interpreted using local damage indices and
related to the physical damage and performance limit states in the frame. These limit
states are referred to as specific damage designations used in FEMA 356 (2000) and
similar standards, such as immediate occupancy, life safety, and collapse prevention.
Immediate occupancy corresponds to the onset of structural damage while collapse
prevention is the point in which the structural system is becoming unstable. Life safety is
an intermediate level that accepts a certain level of damage to structural and non-
structural elements, but ensures the general safety of the occupants. While system
damage descriptions of this sort are helpful to describe acceptance criteria and
performance targets, it is recognized that more explicit and detailed models are currently
under development by PEER and other organizations.
4.4.1 Damage Model
The local damage index presented herein is based on the work of Mehanny and Deierlein
(2001). Chief characteristics of this index are to (1) account for cumulative damage, (2)
reflect the temporal effects of loading sequence, and (3) readily accommodate the
response of structural components with unsymmetrical behavior, e.g., composite beams.
The index is described by the following expression,

n
+
( )

+
|
p current PHC + p+ |FHC,i
+
D = i =1 (4.10)

(( ) )
+
+ n
pu + p |FHC,i
+
i =1
where inelastic component deformations (expressed symbolically as ) are distinguished
between primary and follower cycles and accumulated over the loading history. A
195
Primary Half Cycle (PHC) is a half cycle with an amplitude that exceeds that in all
previous cycles, and the Follower Half Cycles (FHC) are all the preceding and
subsequent cycles of smaller amplitude, including the previously eclipsed PHCs. The
denominator term ( pu ) is the plastic rotation capacity of the element under monotonic
+
loading in the positive deformation direction, and , and are calibration parameters. A

similar damage component D is defined for negative deformations. The positive and
negative indices are then combined into a single damage index as follows,
D =
(D ) + (D )
+

, where is a third calibration parameter. Component failure is
defined when D 1.0. The following default parameters are recommended for RC
columns and composite beams: = 1.0 = 1.5 and = 6.0 parameters. For composite
joints, the recommended values are = 0.75, = 3.0, and = 5.0.
The maximum rotation capacity, ( pu ) , of seismically detailed RC columns is

+
assumed to be a function of the ultimate compressive strain in the confined core concrete.
The ultimate compressive strain, calculated following Paulay and Preistley (1992), is
limited by the tensile rupture of the confining transverse reinforcement. The plastic
rotation capacity for steel beams is a function of the relative influence of lateral torsional
buckling versus local flange and web buckling based on an effective lateral slenderness
ratio (Kemp and Dekker, 1991). The ultimate plastic rotation for composite joints is
determined from an empirical relationship based on tests performed by Kanno (1993),
who defined this capacity as the point where the connection resistance drops to below 0.8
times its maximum strength. Further details on how to compute the plastic rotation
capacities of these components can be found in Mehanny and Deierlein (2001). Table 4.4
summarizes the plastic rotation capacities found using these methods for each of the
components of the test frame.
Note that the predicted plastic rotation capacities for the RC columns are rather large,
particularly in the exterior columns. Given the low axial load and the fact that these
columns are under-reinforced, the RC column section must experience a large amount of
196
hardening prior to the concrete reaching compressive strains that would fracture the
hoops. Therefore, while the calculated rotation capacities in Table 4.4 are consistent with
the model recommended by Paulay and Preistley (1992), this model does not consider all
possible failure modes in the column (e.g., reinforcing bar buckling or fracture).
Fardis et al. (Fardis et al. 2003; Panagiotakos et al. 2001) assembled a comprehensive
database of experimental results of RC element tests. The segment of this database
includes a total of 230 monotonic tests of rectangular columns having code-conforming
detailing and failing in a flexural mode. From this data set, they define the "ultimate"
chord rotation as a reduction in load resistance by 20% or more under either monotonic
or cyclic loading. Using this definition, approximately 30% of the columns in their
database have a rotation capacity greater than 0.1 radians (10% drift ratio) with two-
thirds of these falling within 0.1-0.2 radians (10% to 20% drift ratio). This implies that
the computed plastic rotations for the interior columns (Table 4.4) are within the range
expected in monotonic tests. However, the predicted rotation capacities of the exterior
columns (0.3-0.4 radians) are too large, and in retrospect, should be reevaluated
considering other sources of failures (i.e. bar buckling, strength deterioration, etc.). One
alternative approach may be to utilize the empirical equations to predict chord rotations
developed by Fardis and Panagiotakos, which are based on the comprehensive data set of
RC element tests previously discussed.
This damage index can be roughly distinguished into four divisions that correlate to the
damage state of the element. These divisions are typically broken up into the following
damage levels: insignificant, moderate, significant, and complete loss of resistance.
Table 4.5 shows the original ranges of damage index value and the corresponding
damage level state description, as suggested by Mehanny and Deierlein (2001). The
additional column in this table shows proposed adjustments to these damage index ranges
as validated through the observed damage in the test frame. These index values and the
corresponding damage levels are investigated in the following section.
197
4.4.2 Plastic Rotations
The simulated plastic rotations are directly obtained from an element recorder in
OpenSees. This is an internal command that computes the elastic stiffness of the element
to find the total elastic rotation at each end given the forces on each node. These are then
subtracted from the total rotations to obtain the plastic rotations in the member.
The measured plastic rotations are obtained by making an assumption on the elastic
stiffness of the components hinge, which coupled with the estimated force demand
(derived in Chapter 3) could then estimate the elastic rotation of the hinge. This is then
subtracted from the total rotation measured by the tiltmeters to obtain the plastic hinge
rotation.
4.4.3 Overview of Damage after Each Event
Using the plastic rotation output from the OpenSees model, component damage indices
for all members and joints in the frame are calculated for all of the pseudo-dynamic
loading events. The results of this process for all of the frame elements (columns, beams,
and joints) are summarized in Fig. 4.80 through 4.84 for each of these events. The
damage indices are reported as percentages (100% = 1.0); and for clarity, values below
D = 0.3 (30%) are not shown in these figures. Recall that the observed damage state of
the frame after the 50/50 earthquake is limited to minor concrete cracking and steel beam
yielding, requiring little if any repair (see Section 3.6.2 for further details). The D-
values for a majority of the elements after the 50/50 event are less than 30 40% (Fig.
4.80), which is consistent with the observed damage. However, there are several
members that reach higher D-values, such as the 3rd floor beams, which indicate that the
analytical model and the damage index are overestimating the amount of damage in these
elements. In particular, several of these 3rd floor beam hinges have D-values close to
70%, suggesting that there is some significant hinging and local buckling taking place,
neither of which physically occurred in the members. In the test frame there was no local
198
buckling of any of the steel beams during the 50/50 event and only very slight yielding,
which was only noticed due to minor flaking of the paint in the hinge zones.
Figure 4.81 and 4.82 depicts the values of the damage indices after the first 10/50 event
(segment 1a and 1b, respectively). The RC column base hinges are now reaching values
up to 78% and 60% for the interior and exterior columns, respectively, predicting that
there is some moderate to significant hinging occurring in these regions, which is a
reasonable prediction, although perhaps slightly conservative considering the level of
damage witnessed in the test (Section 3.6.2). The 1st floor beams have D-values
between 73 to 84%, implying that there is a significant amount of hinging in these beams,
while the observed damage would imply more moderate level of damage. The analytical
model predicts a significant amount of damage (67-82%) in the 2nd floor beams, whereas
the test specimen had only limited hinging occurring in these hinges. The 3rd floor beams
now reach predicted damage levels on the brink of loss of capacity (79-94%) which again
overestimates the actual damage observed in these beams.
After the 2/50 event, researchers present at the frame test concluded that that the frame
had reached the collapse prevention limit state, characterized by significant local damage
(concrete crushing, large crack openings, local beam flange and web buckling) and
residual drift that were on the verge of being technically infeasible to repair. The
analytical results in Fig. 4.83 reflect that there is severe damage to the frame after this
event, with all of the beams reaching D-values between 81% and 101% and the base
columns within the 77-92% range. This seems to be an accurate representation, because
at this point, the base columns experienced extensive hinging, with distributed cracks and
spalling of the cover concrete and the beams in the first and third floors had been
subjected to extensive yielding and significant local buckles of the bottom flange.
However, the OpenSees model overestimates the amount of inelasticity occurring in the
second floor beam. Whereas the calculated damage indices are between 82% and 93%,
this beam only experienced slight yielding with no local buckles. The OpenSees model
correctly picks up damage (concrete cracking and spalling) that concentrates in the upper
region of the second floor columns, with damage indices reaching up to 76%. Several of
199
the interior joints in the model are showing a D-value of up to 73%, which overestimates
the observed damage of limited to moderate cracking.
The final design level event continues the same trend as the previous events (Fig. 4.84),
since by this point all of the largest excursions have already occurred in the previous 2/50
event. This event pushes all of the RC column base hinges and the beam hinges at each
floor to the near collapse or the loss of capacity level (85-104%). Recall that the test
frame was subsequently pushed out to extremely high drift demands and showed quite a
bit of residual strength as a system after this final design level event, so it is apparent that
there was a significant amount of strength left within the system. Except for the 2nd floor
beams, it can be argued that the observed damage of these hinges reflect the conditions
expected for near collapse or loss of capacity damage levels. Furthermore, it may be
unconservative to rely on the strength of these elements after attaining the type of damage
that was observed at this point in the test. Therefore, it is reasonable to assume that the
analytical model is roughly capturing the point at which the frame is reaching its near
collapse limit state.
4.4.4 Evolution of Damage Indices and Maximum Plastic Rotations
In this section, the focus will shift towards tracking the evolution of calculated indices
and physical damage of a few important column and beam hinges over the duration of the
pseudo-dynamic loading events. For comparisons sake, the maximum plastic rotations
are also reported.
4.4.4.1 Columns
Figure 4.85a-d presents photos tracking the damage of a 1st-floor interior column base
hinge after each of the four main pseudo-dynamic loading events. Figure 4.85e depicts
the computed damage evolution (D) from OpenSees of this same column (1C3). In this
figure, the evolutions of the D-index for the lower (solid line) and upper hinge (dashed
line) are plotted over the four earthquake time histories. Also included is the damage
200
evolution during two time intervals, marked I and II, which stand for (I) the first 7
seconds of the LP89G04 event (1a) that had to be stopped and (II) the frame straightening
imposed after the 2/50 event. In addition to the Mehanny et al. (2001) damage index, the
ratios of the maximum plastic excursion (DImaxExc) to the ultimate capacity for the lower
(solid) and upper (dashed) hinge are in Fig. 4.85f.
Figure 4.85g depicts the damage evolution of column 1C3 using the measured tiltmeter
response from the hinges in the test frame. The tiltmeter data is processed to obtain
effective plastic hinge rotations using assumptions of elastic stiffness and the force in the
hinge. These plastic rotations are then processed through the damage index equation
presented in equation (4.10). Comparison of the analytical and measured response data is
provided to help determine whether discrepancies between the damage index, D, and the
observed physical damage are inherent to the damage index or are a by-product of
differences in the input response data (i.e., calculated versus measured plastic rotations)
used to calculate the index. Note that the tiltmeter data was not recorded during the
realignment push (event II), which explains the missing data in Fig. 4.85f during this
event. Table 4.6 summarizes the simulated and measured maximum plastic rotations for
this column during each of the pseudo-dynamic loading events as well as the percent
difference between the two ( 100 [ Measured Calculated ] Measured ).
Now that all of the information in Fig. 4.85 has been described, this information can be
used to evaluate how well OpenSees can track progression of the physical damage in the
test frame. During the 50/50 event, Figs. 4.85e and 4.85g indicate that the progression of
damage in the lower hinge as measured by the D-index is relatively the same for the
simulated and measured response and the final value of 0.44 matches quite well. This
D-value corresponds to a negligible amount of damage (Table 4.6) and seems to match
the physical behavior of this hinge quite well, as seen in Fig. 4.85a. During segment I,
both the simulated and measured D-values both experience a sudden increase near 50-
second mark, where the frame was subjected to its first major inelastic excursion, as is
evident in Fig. 4.85a. During the subsequent 10/50-1b event, both the predicted and
measured D-values increase at relatively the same rate and end at final predicted values
201
of 0.78 and 0.88, respectively. While these values suggest that the hinge is at a near
collapse damage state, Fig. 4.85b shows that this is not the case as observed damage is
limited to minor cracking and spalling. During the 2/50 event, the predicted D-index
gradually increases to 0.92, which indicates that significant amount of damage is
predicted in this hinge. This is consistent with the observed damage state shown in Fig.
4.85c. After the final 10/50 event, loss of capacity is predicted in this hinge for both the
analytical and measured D-value, which for all relevant purposes can be argued as the
approximate observed damage state (Fig. 4.85d). The more basic damage index
(DImaxExc) in Fig. 4.85f only reaches a peak value of just over 0.2, which shows how
much of an impact the cyclic damage term in equation (4.10) can have in earthquake
excitations. The maximum plastic rotations in the lower hinge are captured rather well
throughout all events except for the 2/50 event, where the error is reported as 56% in
Table 4.6.
Figures 4.85e, f, and g also show the evolution of damage for the upper hinge with the
dashed line. The damage in this hinge is largely underestimated by the analytical model,
which predicts an ultimate D-value of 37% while the test data suggests almost 90%. This
is surprising considering that the physical damage in this hinge was relatively small
(photos are not shown here) compared to the lower hinge, suggesting that the damage
index is perhaps overcompensating for the cyclic damage or rather the assumptions made
in obtaining plastic rotations lead to an overestimation of nonlinearities. Maximum
measured plastic rotations for this hinge are reported as high as 1.47% for the 2/50 event,
while the predicted values are only 0.16% (Table 4.6).
Figures 4.86 through 4.90 are organized in the same way as Fig. 4.85, except that the D
and DImaxExc-index prediction from OpenSees is now combined into one plot. Figure 4.86
reveals that the damage progression in the lower hinge of an exterior column in the 1st
floor (1C4) is roughly the same for both the predicted and measured values, despite
having some differences in the rate of damage. At the conclusion of each of the pseudo-
dynamic events, the analytical model correctly predicts the amount of damage witnessed
in the test: negligible damage for the 50/50 event (Fig. 4.86a), moderate damage for the
202
10/50 event (Fig. 4.86b), and significant damage for both the 2/50 event (Fig. 4.86c) and
the final 10/50 event (Fig. 4.86d). The simulated maximum plastic rotations compare
well to the measured values (Table 4.7), with the highest error coming in the 2/50 event
at approximately 50%.
The overall damage in the upper hinge is much less severe than the lower hinge, with the
analytical model predicting negligible damage (~0.1) while the measured values predicts
the hinge to be at the onset of noticeable damage (0.45). The latter prediction is a bit
more representative of the observed physical damage in this element. While the errors
between the simulated and measured maximum plastic rotations are rather high in this
hinge (Table 4.7), the absolute magnitudes of these rotations are fairly low.
The evolution of damage for the interior and exterior 2nd floor columns are shown in Figs.
4.87 and 4.88, respectively. Both of these figures show that the amount of damage
predicted by the analytical model for the lower hinges greatly exceed the predictions
from the measured response. This is also reflected in the comparison of the maximum
plastic rotations in Table 4.8. The damage in this lower hinge was observed as relatively
minor supporting the measured results in Fig. 4.87f. On the other hand, the damage in
the upper hinge was much more significant, and seems to be captured generally well by
the analytical model and the damage index. After the 50/50 event, the analytical model
correctly predicts an insignificant amount of damage in both the interior and exterior
upper hinge (D20%), which is validated in the photos in Figs. 4.87a and 4.88a. The
10/50 event produces a moderate level of damage, as seen in Figs. 4.87b and 4.88b, and
is captured relatively well by the simulated D-value, but are overestimated by the
simulated maximum plastic rotations. With a D-value of 76% for the interior upper
hinge at the end of the 2/50 event, the predicted results indicate a significant amount of
damage in this hinge, while the measured D-value indicates that the damage remains
relatively moderate (Fig. 4.87c). Meanwhile, the exterior hinge correctly captures the
moderate level of damage (Fig. 4.88c) with a D-value of 54% at the end of this event.
After the final 10/50 event, the damage states of the interior (Fig. 4.87d) and exterior
203
hinges (Fig. 4.87d) correlates well with the predicted D-values of 84% (significant
damage) and 64% (moderate damage), respectively.
4.4.4.2 Beams
The evolution of damage for 1st floor beam hinges are shown in Fig. 4.89, with the
exterior hinge shown with a solid line and the interior hinge shown with a dashed line.
Photos of the exterior (south) hinge are shown in Figs. 4.89a-d since this is the location of
most severe beam damage. During the first 50/50 event the beam hinge experienced
relatively minor yielding (Fig. 4.89a), which is reflected by the measured D-value of just
below 20%. The predicted D-value increases at a much quicker rate and estimates the
hinge to be at brink of moderate damage (44%). This difference between the measured
and simulated D-value is surprising considering that the total hinge rotation versus
moment response plot in Fig 4.63 (1B1S050) was captured so well. What this reveals is
that the way OpenSees is extracting the plastic rotations from the fiber beam-column
element seems to be overestimating the actual values. OpenSees estimates a peak plastic
rotation of 0.76 radians while the measured counterpart is only 0.25 radians. The reason
there is an overestimation of plastic rotations is likely due the fact that the initial stiffness
of the beam used to derive the plastic rotations includes the effect of the uncracked slab
for both the positive and negative stiffness. This would overestimate the elastic stiffness
of the beam in negative bending and lead to higher plastic rotations. This effect will be
considered again after examining the rest of the results.
After the 10/50 event, the analytical model predicts significant amount of damage (76%)
while the beam hinge experienced only moderate yielding and very minor flange
buckling. The D-value of 45% and 60% from the measured results seems to be a better
indicator of the true damage in these hinges, as seen in Fig. 4.89b. The simulated and
measured peak plastic rotations are better correlated during this event with values of
0.014 and 0.012 radians, respectively. During the final excursion of the 2/50 record, the
D-value from the measured results of the left hinge (solid line) encounters a large spike
in the response which is a result of the large local buckling that occurred during this push.
204
The tiltmeter reading at this point in the record may not be completely accurate given that
the web began to buckle and caused out of plane distortions in the tiltmeter gauge. As
discussed before, the local buckling behavior is not captured by the fiber models and at
this point the measured D-value surpasses the analytical D-value. This large increase in
the D-value is not present in the right beam hinge (dashed line) given that this section
was experience positive (composite) bending and therefore no occurrence of local flange
or web buckling.
At the end of the final event, the left hinge is correctly predicted by OpenSees to be
within the significant/loss of capacity damage region, but in general the progression of
damage was not captured as effectively as it should have been. The damage in the
northern hinge (dashed line) is overestimated by the analytical model (Fig. 4.89e) and is
better captured by the measured results (Fig. 4.89f).
Figure 4.90 shows the results for a 2nd-floor beam, which experienced only moderate
yielding throughout all of the pseudo-dynamic events as a result of the large material
overstrength in the steel (discussed in Chapter 3). This progression of damage is picked
up rather well by the measured D-value in Fig. 4.90f, and the final value of 0.48 (south
hinge) and 0.58 (north hinge) reflects the moderate level of damage seen in Figs. 4.90a-d.
Conversely, both the progression of damage and the final D-values from OpenSees
predictions overestimate that of the true damage in these beam hinges.
It seems that predicted plastic rotations for the composite beams consistently
overestimate that of the measured results, as is reflected in Tables 4.10 and 4.11 as well
as the evolution of damage index plots presented in this section. The only exception to
this is the point at which large local buckling occurs in the test frame which is not
captured by the fiber model. This suggests that the proposed theory that the negative
stiffness of the composite beam is overestimated by including the uncracked stiffness of
the slab is likely correct. The composite stiffness can be up to two to three times as stiff
as a bare steel beam, which, as shown in this section, could prove to have a large impact
in the calculation of plastic rotations.
205
4.4.4.3 General Comments
The purpose of these damage indices is to provide a link between the analytical models in
OpenSees and the physical damage observed in the test frame. This test program
provides a unique opportunity to both validate these damage indices using measured
output from the test and predicted results from the analytical models. In general,
comparisons between the processed D-values from the measured data show good
correlation to the observed damage levels in the components of the test frame when based
on the proposed levels in Table 4.5. There are cases where the measured results tend to
overestimate the level of damage (upper hinge of 1C3, Fig. 4.85), but perhaps this could
be attributed to errors in determining the plastic rotations from the tiltmeters. These
results reinforce the validity of the Mehanny (2001) damage index and shows that it is
able to accurately correlate the plastic rotations with the actual damage in an element,
which has further implications associated with repair types and cost.
The plastic rotations from OpenSees coupled with the damage indices have shown to
accurately capture the damage in those hinges that experienced heavy damage during the
frame test. There are cases where the simulated D-values tend to overestimate the
damage condition in hinges that were less damaged in the in the frame test, but these are
largely a result of over predicting the levels of plastic rotation in these hinges rather than
problems with the index itself. There is a legitimate issue regarding the estimation of
plastic rotations in composite beams given that OpenSees is accounting for the uncracked
slab in the elastic stiffness. This assumption tends to overestimate the plastic rotations in
the beams when subjected to negative bending. While this can be corrected in the post-
processing of the plastic rotations, this should really be corrected within OpenSees.
Despite this issue, these damage indices show promise in their ability to connect the
analytical response from OpenSees to a range of physical damage states seen in
experiments.
The problems with the damage indices stemming from the determination of plastic
rotations can be largely avoided with the use energy-based damage indices. Energy-
206
based indices (i.e., Kratzig et al. 1989, Mehanny 1999) are less sensitive to such subtle
issues that have been discussed within this study. These energy-based indices also avoid
the problems associated with interpreting the plastic rotations of a highly pinched
response. In this case, the deformation-based damage index would tend to overestimate
the amount of damage given the difficulty of determining the plastic deformations of a
pinched response, whereas the energy-based index would be able to more easily interpret
the energy of each cycle.
4.5 Summary of Recommendations
A brief recap of the modeling recommendations specified in this chapter for composite
RCS moment frames are presented within this section.
4.5.1 RC Column Summary
It is recommended that reinforced concrete columns are modeled using the flexibility-
based fiber elements described in Section 4.2.2. The RC fiber section can be defined
using three different materials to represent the core and cover concrete and the
reinforcement steel. The Concrete02 material model is used to represent both the core
and cover concrete. The backbone of the core concrete can be defined using the modified
Kent and Park model (Scott et al. 1982), which adjusts both the strength and ductility of
the concrete according to the amount of confinement provided. Based on the calibration
study presented in Section 4.2.2.1, it is recommended to specify the nominal compressive
strength in the modified Kent and Park model as 0.85 f c' rather than the full f c' , since
this was found to provide a better estimate of the strength of the subassembly tests
considered in the calibration study. The tension strength of the concrete can be defined
by equation (4.6) according to the research by Carrasquillo (et al., 1982). Concrete
tension stiffening effect can also be accounted for using the work from Stevens (et al.,
1991). In this research, the linear decay model in Concrete02 was used for simplicity and
fit to the curve derived by Stevens (et al., 1991). Modeling recommendations for the core
concrete are summarized in Table 4.12.
207
The backbone of the cover concrete is defined according to the following backbone
parameters: f cc' = 0.85 f c' , cc = -0.002, fc'2 = 0, and c 2 = -0.010. This definition stems
from both the calibration studies and the practical aspects of numerical convergence. The
complete definition of the cover concrete material is summarized in Table 4.12.
It is recommended that the longitudinal reinforcement steel be modeled using the Steel02
material model in OpenSees. The yield strength should be assigned either as the
measured strength or the expected yield strength of 1.25Fy . Strain hardening can be
determined experimentally or assumed as 2% of the elastic modulus of steel. The

hysteretic parameters that define the cyclic properties of the material should be defined as
R0 = 18.5 , c1 = 0.925 , and c 2 = 0.15 . The parameters necessary to create the
longitudinal reinforcement bar material are summarized in Table 4.13.
Elastic bond slip springs with a stiffness defined by equation (4.7) have been shown to
accurately capture the additional flexibility of bond slip and yield penetration in RC
columns. This model assumes that the reinforcement bars are sufficiently developed and
will not experience pull out. The elastic springs are represented by the rotSpring2D
function, which can be downloaded from the OpenSees website
(http://opensees.berkeley.edu/).
4.5.2 Composite Beam Summary
The Steel02 material model can be used to define the steel beams, with the yield strength
and hardening ratio either determined experimentally or assumed as R y Fy and 2% of the
initial stiffness, respectively. The concrete slab can be modeled with the Concrete02
material with a peak stress ( f cc' ) of 1.3 f c' and an effective slab width equal to the width
of the column that the beam frames into. The strain corresponding to the maximum
concrete stress ( cc ) should be taken as -0.002 and the strain marking the beginning of
208
the residual strength ( c 2 ) as -0.05. The post-peak residual stress of the slab ( fc'2 ) should
be taken as 20% of the peak stress, f cc' . These modeling assumptions assume that the
shear studs have been designed to provide a fully composite section according the
requirements in the AISC Seismic Provisions (2002). The remaining concrete
parameters, including the tensile strength and softening, can be defined similar to that of
the core concrete described in Section 4.5.1. These parameters needed to define the steel
beam and concrete slab are summarized in Tables 4.12 and 4.13, respectively.
In addition to the specific recommendations provided for the RC columns and the
composite beams, general recommendations were also provided to help in the overall
convergence of flexibility-based beam-column elements. These issues include: (1)
providing a finer discretization of the fiber section in regions where the material is non-
ductile (e.g. concrete cover) in order to capture the softening response, (2) providing
seven gauss integration points along the length of the member, and (3) enforcing the
mitigation of error at the element level by setting the maximum number of iterations to
10 and the minimum tolerance to 1x10-18.
4.5.3 Composite Joint Summary
The Joint2D model is recommended to represent the composite joint panels in OpenSees.
Joint dimensions should correspond to the depth of the steel beam and the width of the
concrete column. The two different deformation mechanisms, panel shear and vertical
bearing, can be represented by two nonlinear springs in series. The panel shear spring
can be best represented by the Steel02 model with a strength and stiffness defined by
equations (2.33) and (4.9), respectively. The remaining Steel02 parameters determined in
the calibration study are recommended as follows: R0 = 18.5 , c1 = 0.96 , and c 2 = 0.15 .
The vertical bearing spring can be defined using the Hysteretic material model with the
backbone defined in Fig. 4.31. The strength and stiffness of this spring can be defined by
equation 2.34 and (4.8), respectively. The recommended values for the pinching factors,
x and y , are 0.2 and 0.5, respectively, and the damage factors, d and d E , are 0.015
209
and 0.0, respectively. Again, the values needed to model the vertical bearing spring are
summarized in Fig
4.6 Conclusions
Calibration studies have the shown that the analytical models used in this study are able
to consistently and accurately capture the behavior of subassembly tests. The intention of
this study is not to overly tune the analytical models to the point where they perfectly fit
the specific tests that were used in this calibration study, but rather the approach is to
provide simple recommendations that are grounded in mechanics that could be used to
accurately model a variety of different types of structures.
The pseudo-dynamic test of a 3-story composite RCS moment frame described in

Chapter 3 has provided a unique opportunity to extend the validation of the analytical
models from the typical subassembly tests to a full-scale building subjected to realistic
excitations. Using the recommendations derived from the subassembly tests, the
analytical model of the test frame is able to capture the experimental response quite well
during the 50/50 and 10/50 events. Interstory drifts and story shear forces, both critical
engineering demand parameters necessary for design and assessment, are captured
relatively well throughout the entire time history. While there are some occurrences of
minor local buckling in the steel beams during the 10/50 event, the frame behavior is still
largely dominated by flexural hinging of the RC columns and composite beams.
Therefore, the fiber models are able to accurately capture the deformation mechanisms
that occur in the test frame during these events.
Comparison of the maximum considered event (2/50) highlighted some of the limitations
of the fiber beam-column elements. During the first 18-seconds of the event, the
analytical model represents the behavior of the test frame fairly well, but beyond this
point, large local buckling in the steel beam hinges began to dominate the response of the
test frame. It is acknowledged that the fiber beam-column models are not able to capture
this phenomenon and only model the flexural yielding behavior of these hinges. This is
210
also compounded by the fact that this event is now the third major earthquake that the test
frame was subjected to and the damage in the RC column base hinges and the composite
beam hinges continue to accrue over the repeated cycles. Based on the results within this
study, it is suggested that the reliability of the fiber beam-column models to accurately
model the realistic inelastic behavior be questioned once interstory drift ratios exceed 3-
4%.
There is a noticeable elongation of the natural period that occurs in the test frame that is
not accurately captured by the analytical model. This difference stems from two primary
issues: (1) the rate of deterioration of the components stiffness in the test frame is not
accurately simulated by the fiber beam-column elements and joint models and (2) strain
rate effects and the force relaxation that occurs due to the pseudo-dynamic algorithm
soften the global response of the frame. The latter issue is inherent to the testing method
and could only be corrected if the test were conducted in real time or using a shake table.
As far as the analytical models, this difference in stiffness degradation is partly due to the
local buckling that is not being captured in the simulation, but also a result of stiffness
degradation of the RC columns and bond slip deterioration that occurred at the base of
the 1st floor. These differences are readily apparent when comparing the stiffness of the
frame during the final pushover.
4.6.1 Future Work
An improved hysteretic model has recently been implemented into OpenSees that gives
the user more control over the strength degradation of the backbone (Ibarra 2004). The
Hysteretic model used in this study does not directly allow the degradation of the
strength, so therefore this was accomplished by imposing a cap on the strength at 5%
rotation, beyond which there is a negative stiffness which brings it down to a residual
strength of 65% of Mvb. For future studies, it would be better suited to recalibrate the
joint bearing model using the recently developed model proposed by Ibarra (2004).
Despite this recommendation, the model used in this study will prove to be sufficient
211
given that the joints in RCS frames are inherently strong and do not experience large
deformations that would push the negative backbone of the current model.
The large differences in the residual drifts between the simulated and measured response
highlight the sensitivity of this important engineering demand parameter to the precise
details of the inelastic hysteretic model. These details include parameters such as yield
strength, hardening stiffness, unloading and reloading stiffness, pinching, and the strength
and stiffness degradation. Each of these parameters can be assigned a mean and
coefficient of variation based on the calibration study of the component. This suggests
that a probabilistic approach could be adopted when defining these hysteretic models, and
the residual drifts could be evaluated considering the distribution of each of the
parameters.
Local buckling of the steel beams proved to be an important behavior effect that
ultimately had a large impact on the response of the frame when subjected to larger
excursions. The next step in modeling these composite RCS frames is to calibrate hinge
models, such as that proposed by Ibarra (2004), to capture this loss of strength and
stiffness in these composite beam hinges (which primarily occurs when subjected to
negative bending). These updated composite beam models will provide the capability to
model these composite frames out to very large drifts and perform more detailed analysis
that are able to capture the onset of structural collapse.
The deformation-based damage model used in this study has been shown to work well
with correlating the analytical response to the physical damage, but there are some
problems stemming from the assumptions used to compute the plastic deformation in the
fiber beam-column elements. In addition, the definition of plastic rotation becomes less
clear when dealing with a highly pinched component and would likely lead to
overestimations of damage. Switching to an energy-based damage model could avoid
some of the problems observed in this study. On the other hand, these energy models
require the force as well as the hinge rotation, which may also introduce another problem
in the evaluation of measured hinge response from the test frame given that the forces are
212
interpreted from strain gauges and are not exact. Nevertheless, these energy-based
models provide another path to investigate the data from this study and could prove to be
an important validation exercise for these models.
213
Table 4.1 Summary of RC column specimens used in calibration study.
Test Test Rebar f c' Fy,rebar Fy,trans
Config Section H (mm) B (mm) L (mm) P f c' Ag s
Specimen Details (MPa) (MPa) (MPa)
Tsai 2002
FFH08 12- 650 650 3000 0.086 0.017
FFL08 Cant- 36mm 650 650 3000 0.086 0.017
41.3 414 414
FRL08 ilever 12- 340 340 1200 0.084 0.025
FRL60 19mm 340 340 1200 0.586 0.025
Tanaka &
Park 1990
8 bars,
Double
#2, 3, 4 dbar= 25.6 474 333 400 400 1600 0.2 0.026
Ended
20mm
214
Table 4.2 - Composite beam validation study
Experiment Best Fit Parameters Description of Test and Failure
'
1.3 f c Fully composite beam with two steel columns part of a 1-story, 1.5-bay substructure.
Tagawa (1989) C 2 0.05 Local buckling occurred in the lower beam flange (pt. B) at the 4th cycle. They
b =b assume bslab = bcol and f c',max =1.8 f c' . Details of the test are shown in Fig. XXX.
slab col
1.3 f
c
' Fully composite beam with rigid column and panel zone. Inelastic behavior is
Bursi and Ballerini governed by steel beam yielding and concrete crushing. Web and flange buckling
C 2 = 0.05 0.07
(1996) occurred at the first negative cycle (/y=4). Final failure mechanism is governed by
bslab = bcol crushing and uplifting of composite slab. Details of the test are shown in Fig. XXX.
Small scale cantilever composite beam that is fixed to a rigid steel column which is
1.3 f c' attached to a strong wall. The entire width of the concrete slab also runs into the
Uang (1985) CG3 C 2 0.07 strong wall, which may help provide a larger effective width for the plastic moment.
bslab = 2.5bcol Severe local buckling occurred at every cycle of the loading protocol. Details of the
test are shown in Fig. XXX.
Fully composite beam with a majority of the inelasticity occurring in the beam,
although it was predicted that the steel columns were 40% weaker than the beam.
Early local buckling occurred in the steel beam at cycle 19- and continued to grow as
Lee (1987) EJ-WC Failure in columns the test progressed. Final failure was due to fracture in the lower beam flange near
the middle of the local buckle. Lee (1987) generated analytical models to simulate
this test with bslab = b AISC and an effective concrete stress of 1.3 f c' . Details of the
test are shown in Fig. XXX.
0.85 f c'
Partially composite beams with 75% of the AISC-LRFD required number of shear
Cheng (2002) ICLCS C 2 = 0.012
studs. There were very little shear studs placed within the region of higher moment.
bslab = bcol The composite beams and to a lesser extent the composite joints governed the
0.85 f c' inelastic action of the specimen. Shear studs fractured at about 3% drift and local
buckling and slab separation occurred at 4% drift. Fracture of the bottom beam
Cheng (2002) INUCS C 2 = 0.012
flange occurred at the 6% drift cycle. Details of the test are shown in Fig. XXX.
bslab = bcol
215
Table 4.2 (cont.) - Composite beam validation study
'
0.85 fc The only difference from the first two specimens is that the imposed displacements
Cheng (2002) ICLPS C 2 = 0.012 followed near fault earthquake response (Krawinkler 2000). Local buckling occurred
bslab = bcol during the large 8% drift excursions. Details of the test are shown in Fig. XXX.
Partially composite beam (25-35%) with shear studs designed to develop 1.3 f c' over
bslab = bcol . Failure of the shear studs at their welds often occurred. Overall
deterioration due to local and lateral buckling of the steel beams. Attained strengths
Civjan et al (2001) Not analyzed for the composite beams showed that the effective concrete stress was less than or
equal to 0.85 f c' . The slab compressive zone was found to be wider than the column
face, initiating at the far column flange. Slip of composite slab was witnessed as
well as shortening of the beams.
Partially composite beam (55% and 35%) assuming composite strength based on
AISC-LRFD specifications (0.85 f c' , bslab = beff ). The maximum composite strengths
Leon, Hajjar et.al (1998) Not analyzed attained were 0.73-0.80 M p+,calc (specimen 2) and 0.90-0.92 M p+,calc (specimen 3). In
specimen 2, the bottom flange of the beam fractured during the second cycle of 1.5%
drift. Specimen 3 experienced local buckling at 2% drift and ultimately failed in low
cycle fatigue type ruptures of the bottom girder flange.
Fully composite beam with shear studs designed to meet the requirements of AISC
design specifications (1994). They found that the attained strength of the composite
Bugeja et al (2000) Not analyzed beams suggest that bslab = b AISC with an effective concrete stress of 0.85 f c' . If the
plastic moment is computed using 1.3 f c' and bslab = bcol , the ultimate moment is only
9% lower than by assuming 0.85 f c' and bslab = b AISC .
216
Table 4.3 Material properties for composite beam test specimens.
Composite Beam
'
Fy , web Fy , flange Fy ,rebar Column
Test fc
Properties
(MPa) (MPa) (MPa) (MPa)
Tagawa et Fy , web =377MPa
24.5 283.8 329.8 355.7
al.1989 Fy , flange =286MPa
Bursi and Fy , web =300MPa
Ballerini 39.0 299.7 299.7 481.6
1996
Fy , flange =300MPa
Uang 1985
29.4 285.9 254.9 544.3 -
(CG3)
Lee 1987 Fy , web =270MPa
35.1 260.4 252.5 413.4
(EJ-WC) Fy , flange =251MPa
f c' =50MPa
Cheng 2002 Fyr=443MPa
22.5 478.5 444.2 -
(ICLCS) Fyh=431MPa
s=0.0175
f c' =54MPa
21.0 478.5 444.2 -
(ICLPS) Fyh=431MPa
s=0.0175
f c' =50MPa
24.3 478.5 444.2 -
(INUCS) Fyh=431MPa
s=0.0175
217
Table 4.4 Plastic rotation capacity of test frame components.
pu (rad)
Floor
RC Columns Steel Beams Comp. Joints
Inner 0.15
1st 0.072 0.087
Outer 0.31
Inner 0.16
2nd 0.093 0.093
Outer 0.41
Inner 0.16
3rd 0.098 0.10
Outer 0.42
Table 4.5 Correlation between the Mehanny damage index and the expected damage in
the component.
Original Proposed
Anticipated Damage State
D Range D Range
Little to no damage in element and corresponds to immediate
0.00-0.30 0.00-0.50
occupancy damage level
Structural element experiences noticeable damage, such as
spalling of cover concrete and minor shear cracking in RC
columns, and hinging and some local buckles in steel beams.
0.30-0.60 0.50-0.70
In terms of the component damage and necessary repairs, this
region of the damage index roughly corresponds to a life safety
limit state.
Structural element is assumed to be at a near collapse state,
0.60-0.95 0.70-0.95 with extensive cracking and hinge formation in RC columns
and significant hinging and local buckles for the steel beams.
The damage is so extensive that the capacity of the element is

>0.95 >0.95
assumed to be exhausted.
Table 4.6 Comparison of the simulated to measured maximum plastic rotation in 1st
floor interior column.
st
1 Floor Int. Column Hinge (1C3) Max. Plastic Rotation (rad)
OpenSees Measured Error
Event
Lower Upper Lower Upper Lower Upper
Hinge Hinge Hinge Hinge Hinge Hinge
50/50 0.007 0.001 0.008 0.005 14% 88%
10/50-1a 0.022 0.002 0.026 0.008 13% 81%
10/50-1b 0.020 0.002 0.016 0.006 -22% 74%
2/50 0.025 0.002 0.057 0.015 56% 89%
10/50-2 0.017 0.001 0.017 0.010 3% 88%
218
Table 4.7 Comparison of the simulated to measured maximum plastic rotation in 1st
floor exterior column
1st Floor Ext. Column Hinge (1C4) Max. Plastic Rotation (rad)
Event
50/50 0.008 0.001 0.008 0.004 -9% 67%
10/50-1a 0.024 0.002 0.021 0.004 -13% 64%
10/50-1b 0.021 0.001 0.015 0.004 -36% 70%
2/50 0.027 0.002 0.055 0.016 51% 90%
10/50-2 0.018 0.001 0.016 0.010 -9% 87%
Table 4.8 Comparison of the simulated to measured maximum plastic rotation in 2nd
floor interior column
2nd Floor Ext. Column Hinge (2C3) Max. Plastic Rotation (rad)
Event
50/50 0.003 0.002 0.001 0.004 -142% 43%
10/50-1a 0.008 0.012 0.001 0.004 -600% -176%
10/50-1b 0.008 0.013 0.003 0.005 -172% -182%
2/50 0.015 0.024 0.006 0.016 -158% -51%
10/50-2 0.009 0.014 0.003 0.007 -178% -94%
Table 4.9 Comparison of the simulated to measured maximum plastic rotation in 2nd
floor exterior column
2nd Floor Ext. Column Hinge (2C4) Max. Plastic Rotation (rad)
Event
50/50 0.007 0.005 0.002 0.005 -294% 0%
10/50-1a 0.013 0.013 0.003 0.008 -374% -68%
10/50-1b 0.012 0.013 0.003 0.007 -333% -77%
2/50 0.014 0.021 0.004 0.029 -268% 28%
10/50-2 0.013 0.013 0.002 0.017 -706% 26%
219
Table 4.10 Comparison of the simulated to measured maximum plastic rotation in
1st floor beam.
1st Floor Beam Hinge (1B1) Max. Plastic Rotation (rad)
Event
South North South North Lower Upper
50/50 0.008 0.004 0.003 0.001 -204% -322%
10/50-1a 0.014 0.014 0.012 0.008 -14% -78%
10/50-1b 0.013 0.012 0.007 0.005 -86% -176%
2/50 0.015 0.017 0.063 0.017 76% 3%
10/50-2 0.012 0.010 0.010 0.001 -24% -818%
Table 4.11 Comparison of the simulated to measured maximum plastic rotation in

2nd floor beam.
2nd Floor Beam Hinge (2B3) Max. Plastic Rotation (rad)
Event
South North South North Lower Upper
50/50 0.005 0.007 0.003 0.003 -73% -132%
10/50-1a 0.005 0.018 0.003 0.004 -66% -349%
10/50-1b 0.004 0.015 0.002 0.003 -91% -347%
2/50 0.005 0.016 0.003 0.004 -57% -261%
10/50-2 0.004 0.013 0.002 0.002 -54% -641%
Table 4.12 Summary of OpenSees input parameters for definitions of Concrete02

materials (uniaxialMaterial Concrete02)
Variable Name Concrete Core Concrete Cover Concrete Slab
Equation 4.1
f cc' 0.85 f c',nom 1.3 f c',nom
using 0.85 f c',nom
cc Equation 4.2 -0.002 -0.002
Equation 4.3
f c'2 0.0 0.2*1.3 f c',nom
using 0.85 f c',nom
c2 Equation 4.4-4.5 -0.010 -0.050
= Eu Ec 0.1 0.1 0.1
ft Equation 4.4 0.0 Equation 4.4
Stevens (1991) Stevens (1991)
Ets 0.0
or Linear Approx. or Linear Approx.
220
Table 4.13 Summary of OpenSees input parameters for definitions of Steel02 materials
(uniaxialMaterial Steel02)
Variable Name Longitudinal Rebar Steel Beam Joint Shear Spring
Mps
Fy 1.25Fy,rebar 1.1Fy,beam
Equation 2.33
Kes
E 200,000MPa 200,000MPa
Equation 4.9
b 0.02 0.02 0.02
Ro 18.5 18.5 18.5
cR1 0.925 0.925 0.96
cR2 0.15 0.15 0.15
a1 0 0 0
a2 1 1 1
a3 0 0 0
a4 1 1 1
Table 4.14 Summary of OpenSees input parameters for definitions of Hysteretic

materials (uniaxialMaterial Hysteretic)
Variable Name Vertical Bearing
0.9Mvb
+M1
Equation 2.34
0.9Mvb/Keb
+1
Equation 2.34, 4.8
Mvb
+M2
Equation 2.34
+2 0.05
0.65Mvb
+M3
Equation 2.34
+3 0.10
-M1,-1,
-M2,-2, Negative of above values
-M3,-3,
x 0.2
y 0.5
d 0.015
dE 0.0
221
600
400
200
Stress (MPa)
-200
-400
-600
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
Strain
Figure 4.1 Hysteretic response for the OpenSees Steel02 material model.
10
(f t)
0
-10
-20 (f' , )
Stress (MPa)
c2 c2
-30
-40
-50
-60
Core Concrete
(f'cc , cc ) Cover Concrete
-70
-0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04
Strain
Figure 4.2 Hysteretic response for the OpenSees Concrete02 material model.
222
4
x 10
1
Hysteretic Model (d2 , f 2 )
0.8 x = y = 0.5
d = 0.01
0.6 dE = 0.0
(d1 , f 1 ) (d3 , f 3 )
0.4
Moment (kN-mm)
0.2
-0.2
-0.4
-0.6
-0.8
-1
-0.1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1
Rotation (rad)
Figure 4.3 Backbone and cyclic response of Hysteretic model in OpenSees.
8
(cr, f cr) Carrasquillo/Stevens
Linear Approx.
7
6
Concrete Stress (MPa)
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Tensile Strain x 10
-3
Figure 4.4 Typical tensile softening response of reinforced concrete.
223
slip
s>y

y
y = yield stress of rebar Gap opens (slip) due Fiber section element
s = stress in rebar including to yield penetration is used in series with
strain hardening and bond slip bond slip spring
Figure 4.5 Representation of deformations from bond slip and yield penetration.
20
P
18
16
Confined
14
Bond Stress, q (MPa)
Concrete
12
Elastic Rebar
10
4 Bar Pull Out
0
0 1 2 3 4 5 6 7 8 9 10
Slip, u (mm)
Figure 4.6 Idealized backbone response of elastic reinforcing bar pull out. (Fillipou et
al. 1983)
224
6
x 10
2.5
Tsai 2002
FFH08
1.5
Moment(kN-mm)
0.5
-0.5
-1.5
OpenSees
Test
-2.5
-250 -150 -50 0 50 150 250
Drift (mm)
Figure 4.7 RC Column calibration against subassembly test: Tsai 2002-FFH08, base
springs included.
6
x 10
2.5
Tsai 2002
FFH08
No Base Springs
1.5
Moment(kN-mm)
0.5
-0.5
-1.5
OpenSees
Test
-2.5
-250 -150 -50 0 50 150 250
Drift (mm)
Figure 4.8 RC Column calibration against subassembly test: Tsai 2002-FFH08, no base
springs.
225
6
x 10
2.5
Tsai 2002
FFL08
1.5
Moment(kN-mm)
0.5
-0.5
-1.5
OpenSees
Test
-2.5
-250 -150 -50 0 50 150 250
Drift (mm)
Figure 4.9 RC Column calibration against subassembly test: Tsai 2002-FFL08, grouted
splice within hinge zone.
5
x 10
4
Tsai 2002
3 FRL08
1
Moment(kN-mm)
-1
-2
-3 OpenSees
Test
OS-Backbone w/ Couplers
-4
-150 -100 -50 0 50 100 150
Drift (mm)
Figure 4.10 RC Column calibration against subassembly test: Tsai 2002-FRL08,
grouted splice within hinge zone (OS model with coupler influence shown as backbone).
226
5
x 10
6
Tsai 2002
FRL60
4
2
Moment(kN-mm)
-2
-4
OpenSees
Test
-6
-150 -100 -50 0 50 100 150
Drift (mm)
Figure 4.11 RC Column calibration against subassembly test: Tsai 2002-FRL60,
grouted splice within hinge zone, high axial load.
Tension stress
d1 Tension stress
goes to zero
Bond-Slip
Depending on d1, Spring
the bond slip in
one column may d2
be affected by the
other column
Cantilever OpenSees Model
Tsai 2002
Double-Ended
Tanaka and Park 1990
Figure 4.12 Schematics of the cantilever and double-ended test setup with respect to
bond slip and OpenSees modeling.
227
200
Tanaka and Park 1990, No. 4

150
100
50
Force (kN)
-50
-100
-150
OpenSees
Test
-200
-100 -50 0 50 100 150
Displacement (mm)
(a) Bond slip spring included.
200

150
100
50
Force (kN)
-50
-100
-150
OpenSees
Test
-200
-100 -50 0 50 100 150
Displacement (mm)
(b) No bond slip spring.
Figure 4.13 RC Column calibration against subassembly test: Tanaka et al. (1990) test
#4.
228
200

150
100
50
Force (kN)
-50
-100
-150
OpenSees
Test
-200
-80 -60 -40 -20 0 20 40 60 80
Displacement (mm)
#3.
200

150
100
50
Force (kN)
-50
-100
-150
OpenSees
Test
-200
-100 -50 0 50 100 150
Displacement (mm)
#2.
229
RC Column
bcol
bAISC Concrete Slab
tslab
Figure 4.16 Definition of effective slab width showing both AISC-LRFD and the
column width.
147.6
Loading Beam P, 3.54
41
1.18 2.95
Concret e slab
W 1 4x3 0 Wire mesh
Point B Point A 0.236 @ 3.94
134
Reaction
W 1 6x5 7 Fra me W 14x30
295 118
(a) Test setup and specimen Tagawa et al.1989 (1in=25.4mm)
47.24
1.97
P, Concret e slab
0.79 2.75
IP E 330
8 bars - 0.47
55.12
HE 360B
IP E 330
157.48
(b) Test setup and specimen Bursi and Ballerini 1996 (1in=25.4mm)
30
1
0.56 1
P,
Concret e slab
Wire mesh
0.0625 @ 1
M 6x4.4
M 6x4.4
45
(c) Test setup and specimen Uang 1985 (1in=25.4mm)
230
47.24
W 12x65 3.5
66.93
(weak axis) 1.2 3
P,
Concret e slab
Wire mesh
0.214 @ 4
66.93
W 18x35
90.55 W 18x35
(d) Test setup and specimen Lee 1987 (1in=25.4mm)

2000
P,
1600
P, 75
20
75
concrete slab
Wire mesh
596x199x10x15 7mm @ 100mm
1600
650mm square RC 596x199x10x15

12#36 bars
2700 2700
All dimensions are in millimeters
(e) Test setup and specimen (ICLCS, INUCS, ICLPS) Cheng 2002
Figure 4.17 Composite beam subassembly dimensions and cross-section details.
Uang 1985
6 f'c=1.3f'c & bslab =bcol
C2=0.05
4
Tip Force (kips)
-2
-4
-6
-3 -2 -1 0 1 2 3
Tip Displacement (in)
Figure 4.18 Composite beam calibration against subassembly test: Uang (1985) test.
(1kip = 4.448kN, 1in = 25.4mm)
231
80
Bursi & Ballerini 1996

C2=0.05
40
20
Tip Force (kips)
-20
-40
-60
-80
-5 -4 -3 -2 -1 0 1 2 3 4 5
Figure 4.19 Composite beam calibration against subassembly test: Bursi and Ballerini
(1996) test. (1kip = 4.448kN, 1in = 25.4mm)
500
400 Tagawa 1989

f'c=1.3f'c & bslab =bcol
300 C2=0.05
200
Horizontal Force (kN)
100
-100
-200
-300
-400
-500
-80 -60 -40 -20 0 20 40 60 80 100
Displacement at pt. A (mm)
Figure 4.20 Composite beam calibration against subassembly test: Tagawa (1989) test
horizontal force versus displacement at A.
232
5
x 10
5
Tagawa 1989
C2=0.05
3
Beam Moment (kN-mm)
-1
-2
-3
-4
-0.04 -0.03 -0.02 -0.01 0 0.01 0.02
Beam Rotation (rad)
Figure 4.21 Composite beam calibration against subassembly test: Tagawa (1989) test
beam moment versus rotation.
50
40 Lee 1987
30 C2=0.012
20
Tip Force (kips)
10
-10
-20
-30
-40
-50
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Figure 4.22 Composite beam calibration against subassembly test: Lee (1987) test.
(1kip = 4.448kN, 1in = 25.4mm)
233
6
x 10
2 Cheng 2002
INUCS - East Beam
1.5
C2=0.05
1
Moment (kN-mm)
0.5
-0.5
-1
-1.5
-2
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
Beam Rotation (rad)
Figure 4.23 Composite beam calibration against subassembly test: Cheng (2002),
specimen INUCS-East Beam.
6
x 10
2 Cheng 2002
INUCS - West Beam
1.5
C2=0.05
1
Moment (kN-mm)
0.5
-0.5
-1
-1.5
-2
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
Beam Rotation (rad)
specimen INUCS-West Beam.
234
6
x 10
2 Cheng 2002
ICLCS - East Beam
1.5
C2=0.05
1
Moment (kN-mm)
0.5
-0.5
-1
-1.5
-2
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
Beam Rotation (rad)
specimen ICLCS-East Beam.
6
x 10
2 Cheng 2002
ICLCS - West Beam
1.5
C2=0.05
1
Moment (kN-mm)
0.5
-0.5
-1
-1.5
-2
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
Beam Rotation (rad)
specimen ICLCS-West Beam.
235
6
x 10
2 Cheng 2002
ICLPS - East Beam
1.5
C2=0.05
1
Moment (kN-mm)
0.5
-0.5
-1
-1.5
-2
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
Beam Rotation (rad)
specimen ICLPS-East Beam.
6
x 10
2 Cheng 2002
ICLPS - West Beam
1.5
C2=0.05
1
Moment (kN-mm)
0.5
-0.5
-1
-1.5
-2
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
Beam Rotation (rad)
specimen ICLPS-West Beam.
236
Frame
Element
Fixed-end
Multipoint Node 2
Constraint
Frame
Node 3 Element
Frame Node 1
Element
Center Node
Node 4 Rotational Spring
Frame
Element
Figure 4.29 Schematic of OpenSees joint element used in this study.
400
Joint Panel Shear Rotation 0.02Kes
300
ps)ps
(Mps/Kes, MM
200
100
M PanelShear
-100
-200
-300 K initial = 2Mps/0.01

K hardening = 0.02Kinitial
-400
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation
Figure 4.30 Panel shear hysteretic model (backbone and cyclic model).
237
400 Joint Bearing Rotation
300 (0.05, M vb )
(4.5e-3, 0.9M vb )
200 (0.10, 0.65M vb )
100
M vb
-100
-200
-300
-400
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
Figure 4.31 Vertical bearing hysteretic model (backbone and cyclic model).
400 Total Joint Rotation

OJB1-0
300 M ps = 10106 kN-mm
M vb = 7327 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
Figure 4.32 Calibration results for joint specimen OJB1-0, which primarily fails in
vertical bearing.
238
OJB4-0
300 M ps = 10098 kN-mm
M vb = 7301 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
vertical bearing.

OJB5-0
300 M ps = 10102 kN-mm
M vb = 8326 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
vertical bearing.
239
OJB6-1
300 M ps = 9988 kN-mm
M vb = 6976 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
vertical bearing.

OJS1-1
300 M ps = 4436 kN-mm
M vb = 4801 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
Figure 4.36 Calibration results for joint specimen OJS1-1, which primarily fails in panel
shear.
240
OJS2-0
300 M ps = 3309 kN-mm
M vb = 4463 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
shear.

OJS3-0
300 M ps = 6789 kN-mm
M vb = 9453 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
shear.
241
OJS4-1
300 M ps = 6789 kN-mm
M vb = 9453 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
shear.

OJS5-0
300 M ps = 6276 kN-mm
M vb = 10834 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
shear.
242
OJS6-0
300 M ps = 7040 kN-mm
M vb = 10279 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
shear.

OJS7-0
300 M ps = 7040 kN-mm
M vb = 10279 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
shear.
243
HJS1-0
300 M ps = 9561 kN-mm
M vb = 15848 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
Figure 4.43 Calibration results for joint specimen HJS1-0, which primarily fails in panel
shear.

HJS2-0
300 M ps = 9561 kN-mm
M vb = 15848 kN-mm
200
Beam Shear (kN)
100
-100
-200
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
Figure 4.44 Calibration results for joint specimen HJS2-0, which primarily fails in panel
shear.
244
400 Joint Panel Shear Rotation 400 Joint Bearing Rotation
OJS3-0 OJS3-0
300 300
200 200
Beam Shear (kN)
Beam Shear (kN)

100 100
0 0
-100 -100
-200 -200
-300 -300
Test Test
-400 OpenSees -400 OpenSees
-0.1 0 0.1 -0.1 0 0.1

Rotation (rad) Rotation (rad)

OJS4-1 OJS4-1
300 300
200 200
Beam Shear (kN)
Beam Shear (kN)

100 100
0 0
-100 -100
-200 -200
-300 -300
Test Test
-0.1 0 0.1 -0.1 0 0.1


OJS5-0 OJS5-0
300 300
200 200
Beam Shear (kN)
Beam Shear (kN)
100 100
0 0
-100 -100
-200 -200
-300 -300
Test Test
-0.1 0 0.1 -0.1 0 0.1

Figure 4.45 Contributions from panel shear and vertical bearing spring for joint
specimen OJS3-0, OJS4-1, and OJS5-0.
245
4m Composite RCS
joint element
4m RC column Composite beam

fiber element Rigid
fiber element
link
4m Released end Leaning

Node column
Bond slip
base springs Zero length
7m 7m 7m
Figure 4.46 Schematic of the analytical model of the test frame.
Disp. Force (n+2)T

(n+2)T
dn+2
(n+1)T
(n+1)T Force
dn+1 Relaxation
dn
nT
nT
Ramp Hold Ramp
Real Time dn dn+1 dn+2 Disp.
Figure 4.47 Ramp and hold phases of pseudo-dynamic testing and the concept of force
relaxation.
246
200
Test
150 TCU082 50/50 OpenSees
100
Roof Disp. (mm) 50
-50
-100
-150
-200
0 5 10 15 20 25 30 35 40 45
Time (sec)
4000
Test
2000
Base Shear (kN)
1000
-1000
-2000
-3000
-4000
0 5 10 15 20 25 30 35 40 45
Time (sec)
3 3
2 2
Floor
Floor
1 1
0 0
-6 -4 -2 0 2 4 6 -4 -3 -2 -1 0 1 2 3 4
IDR Story Shear (1000kN)
Figure 4.48 OpenSees versus test frame response for 50/50 event: (a) roof
displacement, (b) base shear, (c) peak IDR, and (d) peak story shear.
247
400
LP89G04 10/50 1a
300
200
Roof Disp. (mm)
100
-100
-200 Test
OpenSees
-300
0 1 2 3 4 5 6 7
Time (sec)
4000
3000 LP89G04 10/50 1a
2000
Base Shear (kN)
1000
-1000
-2000
-3000 Test
OpenSees
-4000
0 1 2 3 4 5 6 7
Time (sec)
3 3
2 2
Floor
Floor
1 1
0 0
-6 -4 -2 0 2 4 6 -4 -3 -2 -1 0 1 2 3 4
Figure 4.49 OpenSees versus test frame response for 10/50-1a event: (a) roof
248
300
Test
LP89G04 10/50 1b OpenSees
200
100
Roof Disp. (mm)
-100
-200
-300
0 5 10 15 20 25 30 35 40
Time (sec)
4000
Test
3000 LP89G04 10/50 1b OpenSees
2000
Base Shear (kN)
1000
-1000
-2000
-3000
-4000
0 5 10 15 20 25 30 35 40
Time (sec)
3 3
2 2
Floor
Floor
1 1
0 0
-6 -4 -2 0 2 4 6 -4 -3 -2 -1 0 1 2 3 4
Figure 4.50 OpenSees versus test frame response for 10/50-1b event: (a) roof
249
400
Test
TCU082 2/50 OpenSees
200

0
-200
-400
Frame test stopped
at 28 seconds
-600
0 5 10 15 20 25 30 35 40 45 50
Time (sec)
4000
Test
2000
Base Shear (kN)
1000
-1000
-2000
-3000
-4000
0 5 10 15 20 25 30 35 40 45 50
Time (sec)
3 3
2 2
Floor
Floor
1 1
0 0
-6 -4 -2 0 2 4 6 -4 -3 -2 -1 0 1 2 3 4
Figure 4.51 OpenSees versus test frame response for 2/50 event: (a) roof displacement,
(b) base shear, (c) peak IDR, and (d) peak story shear.
250
300
Test
LP89G04 10/50 2 OpenSees
200
100
Roof Disp. (mm)
-100
-200
-300
0 5 10 15 20 25 30 35 40
Time (sec)
4000
Test
3000 LP89G04 10/50 2 OpenSees
2000
Base Shear (kN)
1000
-1000
-2000
-3000
-4000
0 5 10 15 20 25 30 35 40
Time (sec)
3 3
2 2
Floor
Floor
1 1
0 0
-6 -4 -2 0 2 4 6 -4 -3 -2 -1 0 1 2 3 4
Figure 4.52 OpenSees versus test frame response for 10/50-2 event: (a) roof
251
2
TCU082 - 50%in50yr
1.8 Analytical Results
1.6
1.4
T = 1.05sec
freq (Hz)
1.2
0.8
0.6 T = 1.15sec
0.4
0.2
10 15 20 25 30 35
time (sec)
Figure 4.53 Contours of the power spectral density for the frequency of the analytical
response throughout the time history (10/50-1) using a 10-second sliding window.
2
TCU082 - 50%in50yr
1.8 Experimental Results
1.6
1.4
freq (Hz)
1.2 T = 1.15sec
0.8
0.6
T = 1.3sec
0.4
0.2
10 15 20 25 30 35
time (sec)
Figure 4.54 Contours of the power spectral density for the frequency of the measured
252
2 LP89G04 - 10%in50yr - 1
Analytical Results
1.8
1.6
1.4
freq (Hz)
1.2
0.8
0.6 T = 1.3sec
0.4
0.2
5 10 15 20 25 30 35
time (sec)
Figure 4.55 Contours of the power spectral density for the frequency of the analytical
2
LP89G04 - 10%in50yr - 1
1.8 Experimental Results
1.6
1.4
freq (Hz)
1.2
1 T = 1.4sec
0.8
0.6
0.4
0.2
10 15 20 25 30
time (sec)
Figure 4.56 Contours of the power spectral density for the frequency of the measured
253
2
Analytical
1.9 Experimental
LP89G04 - 0.8*10%in50yr - 1
1.8
1.7
Predominate Period (sec)
1.6
1.5
1.4
1.3
1.2
1.1
1
0 5 10 15 20 25 30 35 40
time (sec)
Figure 4.57 Plot of the predominate period of a sliding 10-second window over the
displacement time history of the first 10/50 event.
3
Analytical
2.8 TCU082 - 2%in50yr Experimental
2.6
2.4
Predominate Period (sec)
2.2
1.8
1.6
1.4
1.2
1
0 5 10 15 20 25 30
time (sec)
displacement time history of the 2/50 event.
254
2.6
2.4 LP89G04 - 0.8*10%in50yr - 2
Predominate Period (sec) 2.2
1.8
1.6
1.4
1.2
0.8
0 5 10 15 20 25 30 35 40
time (sec)
displacement time history of the second 10/50 event.
3500
kos/ktest = 1.9 7%
3000
2500
Base Shear (kN)
2000
1500
kos
1000
500 ktest
Test
OpenSees
0
0 1 2 3 4 5 6 7 8
Roof Drift Ratio (%)
Figure 4.60 Analytical versus experimental comparison of the roof drift versus base
shear during the final static pushover of the test frame.
255
300
1 st Event
2 nd Event
200
100
-100
-200
Test Frame Results

-300
0 5 10 15 20 25 30 35 40
Time (sec)
Figure 4.61 Comparison of the roof displacement response measured from the test frame
for the first (1b) and second 10/50 (2) event.
300
1 st Event
2 nd Event
200
100
-100
-200
OpenSees Results
-300
0 5 10 15 20 25 30 35 40
Time (sec)
Figure 4.62 Comparison of the roof displacement response predicted by OpenSees for
the first (1b) and second 10/50 (2) event.
256
3 3
2 2
Moment (10 3kN-m)

Moment (10 kN-m)
1B1S050 2B3N050
1 1
3
0 0
-1 -1
-2 -2
-3 -3
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
Rotation (rad/1000) Rotation (rad/1000)
Figure 4.63 Analytical versus measured response of beams during 50/50 event.
3 3
2 2
Moment (10 kN-m)
Moment (10 kN-m)

1B1N050 2B3S050
1 1
3
3
0 0
-1 -1
-2 -2
-3 -3
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
Figure 4.64 Analytical versus measured response of beams during 10/50-1a event.
3 3
2 2
Moment (10 3kN-m)
Moment (10 kN-m)
1B1S050 2B3N050
1 1
3
0 0
-1 -1
-2 -2
-3 -3
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
Figure 4.65 Analytical versus measured response of beams during 10/50-1b event.
257
3 3
2 2
Moment (10 kN-m)
Moment (10 kN-m)

1B1S050 2B2N050
1 1
3
3
0 0
-1 -1
-2 -2
-3 -3
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
Figure 4.66 Analytical versus measured response of beams during 2/50 event.
3 3
2 2
Moment (10 kN-m)

Moment (10 3kN-m)
1B1S050 2B3N050
1 1
0 3 0
-1 -1
-2 -2
-3 -3
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
Figure 4.67 Analytical versus measured response of beams during 10/50-2 event.
3 3
2 2
Moment (10 kN-m)
Moment (10 kN-m)
1C2d050 1C4d050
1 1
3
0 0
-1 -1
-2 -2
-3 -3
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
Figure 4.68 Analytical versus measured response of columns during 50/50 event.
258
3 3
2 2
Moment (10 kN-m)
Moment (10 kN-m)

1C2d050 1C4d050
1 1
3
3
0 0
-1 -1
-2 -2
-3 -3
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
Figure 4.69 Analytical versus measured response of columns during 10/50-1a event.
3 3
2 2
Moment (10 kN-m)
Moment (10 kN-m)

1C2d050 1C4d050
1 1
3
0 3 0
-1 -1
-2 -2
-3 -3
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
Figure 4.70 Analytical versus measured response of columns during 10/50-1b event.
3 3
2 2
Moment (10 kN-m)
Moment (10 kN-m)
1C1d050 1C4d050
1 1
3
0 0
-1 -1
-2 -2
-3 -3
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
Figure 4.71 Analytical versus measured response of columns during 2/50 event.
259
3 3
2 2
Moment (10 kN-m)
Moment (10 kN-m)

1C2d050 1C4d050
1 1
3
3
0 0
-1 -1
-2 -2
-3 -3
-80 -60 -40 -20 0 20 40 60 80 -80 -60 -40 -20 0 20 40 60 80
Figure 4.72 Analytical versus measured response of columns during 10/50-2 event.
6 6
4 4
Joint Shear (10 kN)
Joint Shear (10 kN)

1J3 2J3
2
3
3
0 0
-2 -2
-4 -4
-6 -6
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
Total Rot. (rad/1000) Total Rot. (rad/1000)
Figure 4.73 Analytical versus measured response of joints during 50/50 event.
6 6
4 4
Joint Shear (10 kN)
Joint Shear (10 kN)
1J3 2J3
2
3
2
3
0 0
-2 -2
-4 -4
-6 -6
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
Figure 4.74 Analytical versus measured response of joints during 10/50-1a event.
260
6 6
4 4
Joint Shear (10 kN)
Joint Shear (10 kN)

1J3 2J3
2
3
2
3
0 0
-2 -2
-4 -4
-6 -6
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
Figure 4.75 Analytical versus measured response of joints during 10/50-1b event.
6 6
4 4
Joint Shear (10 kN)
Joint Shear (10 kN)

1J3 2J3
2
3
3
0 0
-2 -2
-4 -4
-6 -6
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
Figure 4.76 Analytical versus measured response of joints during 2/50 event.
6 6
4 4
Joint Shear (10 kN)
Joint Shear (10 kN)
1J3 2J3
2
3
2
3
0 0
-2 -2
-4 -4
-6 -6
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30
Figure 4.77 Analytical versus measured response of joints during 10/50-2 event.
261
3
2
50/50 Event
Floor 1 st Fl: 199% err.
1
2 nd Fl: -1% err.
3 rd Fl: 40% err.
0
-60 -40 -20 0 20 40 60
3
2
10/50-1a Event
Floor
1 st Fl: 33% err.

1
2 nd Fl: -22% err.
3 rd Fl: -19% err.
0
-60 -40 -20 0 20 40 60
3
2
10/50-1b Event
Floor
1 st Fl: 77% err.

1
2 nd Fl: 60% err.
3 rd Fl: 45% err.
0
-60 -40 -20 0 20 40 60
3 3 = -329
2 2 = -244
2/50 Event
Floor
1 st Fl: 96% err.

1 1 = -135
2 nd Fl: 95% err.
3 rd Fl: 97% err.
0
-60 -40 -20 0 20 40 60
3
2
10/50-2 Event
Floor
1 st Fl: 107% err.

1
2 nd Fl: 122% err.
3 rd Fl: 134% err.
0
-60 -40 -20 0 20 40 60
Residual Displacement (mm)
Figure 4.78 Comparison of the residual displacements from the analytical (thin line) and
experimental (thick line) model after each pseudo-dynamic event.
262
400 Total Joint Rotation diff. in peak
OJB1-0 strength
300 M ps = 10106 kN-mm
M vb = 7327 kN-mm
200 negative
backbone
Beam Shear (kN)
100
kunloading
0 zero load
-100 difference in
resid. drift
difference in
-200
strength
degradation
-300
Test
-400 OpenSees
-0.15 -0.1 -0.05 0 0.05 0.1 0.15

Rotation (rad)
vertical bearing.
67 52 66 68 49 67
50 36 43 44 36 49
44 36 36 36 35 43
30 44 44 30
TCU082-EW: 50%in50year
Figure 4.80 Damage index values (>30%) after 50% in 50 year event.
263
73 58 72 74 56 76
56 42 46 50 52 45 42 60
31 31
56 51 49 51 53 50 51 66
39 57 57 39
LP89G04-NS: 10%in50year-1a
Figure 4.81 Damage index values (>30%) after 10% in 50 year 1a event.
93 81 30 91 93 30 79 94
76 67 50 71 73 50 66 82
32 50 50 31
39 54 54 37
76 73 50 73 73 50 74 84
60 78 78 60
LP89G04-NS: 0.8*10%in50year-1b
Figure 4.82 Damage index values (>30%) after 10% in 50 year -1b event.
264
101 93 48 99 101 49 91 102
30
43 89 82 73 84 86 73 81 93 41
54 76 76 53
61 77 77 59
41 90 89 72 88 89 71 90 97 39
31 31
77 92 92 77
TCU082-EW: 2%in50year
Figure 4.83 Damage index values (>30%) after 2% in 50 year event.
39 104 98 63 103 104 64 97 105 40

33 33
30 38 37
57 95 90 87 92 93 87 89 98 55
64 84 84 63
70 86 85 69
53 96 95 87 95 95 86 96 101 53
37 37
85 97 97 85
LP89G04-NS: 0.8*10%in50year-2
Figure 4.84 Damage index values (>30%) after final 10% in 50 year event.
265
(a) 50/50 (b) 10/50-1b (c) 2/50 (d) 10/50-2
TCU082 LP89G04 TCU082 LP89G04
1 I II
50/50 10/50-1b 2/50 10/50-2
0.8
1C3
Damage Index
0.6
0.4
0.2
DI : Lower Hinge
DI : Upper Hinge
0
0 20 40 60 80 100 120 140 160 180
(e) Time (sec)
0.4
DI maxExc: Lower Hinge
Damage Index
0.3 DI maxExc: Upper Hinge

0.2
0.1
0
0 20 40 60 80 100 120 140 160 180
(f) Time (sec)

1 I
50/50 10/50-1b 2/50 10/50-2
0.8
1C3
Damage Index
0.6
0.4
0.2
0
0 20 40 60 80 100 120 140 160
(g) Time (sec)
Figure 4.85 (a-d) Photos of damage progression in the 1st floor int. column after each
main event. Evolution of damage indices using (e,f) OpenSees and (g) measured data.
266
(a) 50/50 (b) 10/50-1b (c) 2/50 (d) 10/50-2

1 I II
50/50 10/50-1b 2/50 10/50-2
0.8
1C4
Damage Index
0.6
0.4
0.2
DI
DI maxExc
0
0 20 40 60 80 100 120 140 160 180
Time (sec)
(e)

1 I
50/50 10/50-1b 2/50 10/50-2
0.8
1C4
Damage Index
0.6
0.4
0.2
0
0 20 40 60 80 100 120 140 160
Time (sec)
(f)
Figure 4.86 (a-d) Photos of damage progression in the 1st floor exterior column after
each main event. Evolution of damage index using (e) OpenSees and (f) measured data.
267
(a) 50/50 (b) 10/50-1b (c) 2/50 (d) 10/50-2

1 I II
50/50 10/50-1b 2/50 10/50-2
0.8
2C3
Damage Index
0.6
0.4
0.2
DI
DI maxExc
0
0 20 40 60 80 100 120 140 160 180
Time (sec)
(e)

1 I
50/50 10/50-1b 2/50 10/50-2
0.8
2C3
Damage Index
0.6
0.4
0.2
0
0 20 40 60 80 100 120 140 160
Time (sec)
(f)
Figure 4.87 (a-d) Photos of damage progression in the 2nd floor interior column after
268
(a) 50/50 (b) 10/50-1b (c) 2/50 (d) 10/50-2

1 I II
50/50 10/50-1b 2/50 10/50-2
0.8
2C4
Damage Index
0.6
0.4
0.2
DI
DI maxExc
0
0 20 40 60 80 100 120 140 160 180
Time (sec)
(e)

1 I
50/50 10/50-1b 2/50 10/50-2
0.8
2C4
Damage Index
0.6
0.4
0.2
0
0 20 40 60 80 100 120 140 160
Time (sec)
(f)
Figure 4.88 (a-d) Photos of damage progression in the 2nd floor exterior column after
269
(a) 50/50 (b) 10/50-1b (c) 2/50 (d) 10/50-2

1 I II
50/50 10/50-1b 2/50 10/50-2
0.8
1B1
Damage Index
0.6
0.4
0.2
DI
DI maxExc
0
0 20 40 60 80 100 120 140 160 180
Time (sec)
(e)

1 I
50/50 10/50-1b 2/50 10/50-2
0.8
1B1
Damage Index
0.6
0.4
0.2
0
0 20 40 60 80 100 120 140 160
Time (sec)
(f)
Figure 4.89 (a-d) Photos of the damage progression in the 1st floor beam after each main
event. Evolution of damage index using (e) OpenSees and (f) measured data.
270
(a) 50/50 (b) 10/50-1b (c) 2/50 (d) 10/50-2

1 I II
50/50 10/50-1b 2/50 10/50-2
0.8
2B3
Damage Index
0.6
0.4
0.2
DI
DI maxExc
0
0 20 40 60 80 100 120 140 160 180
Time (sec)
(e)

1 I
50/50 10/50-1b 2/50 10/50-2
0.8
2B3
Damage Index
0.6
0.4
0.2
0
0 20 40 60 80 100 120 140 160
Time (sec)
(f)
Figure 4.90 (a-d) Photos of the damage progression in the 1st floor interior column after
271
Chapter 5: Applications
5.1 Introduction
In this chapter, the design recommendations proposed in Chapter 2 are implemented and
evaluated in the designs of three case study buildings. These include a 3, 6, and 20-story
buildings with perimeter frame systems. Each of these buildings are modeled using the
recommendations outlined in Chapter 4 and then subjected to multiple ground motions
representing a range of earthquake intensities up through the 2% in 50 year hazard level.
The results from this study are used to evaluate the performance of composite RCS
moment frames as influenced by the current and proposed design recommendations. The
performance of the 3-story building also provides further insights into the performance of
the test frame discussed in Chapter 3.
5.2 Case Study Buildings
The moment frame of the 3-story case-study building is identical to the test frame
discussed in Chapter 3. Pertinent details of this frame design are briefly reviewed in this
chapter; for complete details the reader is referred to Chapter 3. The 6 and 20-story
designs are more representative of the building heights where RCS systems are most
competitive with conventional steel or RC systems.
The layout of the 6-story building is presented in Fig. 5.1 and follows the general layout
of the theme structure proposed as part of the US-Japan program on hybrid structures.
For the perimeter frame configuration, two additional columns (columns B1/B7 and
E1/E7 on Fig. 5.1) have been added to the original theme structure plan. The layout of
the 20-story case study building (Fig. 5.2) is well known to the research community as
one of three buildings that has been exhaustively studied as part of the SAC Joint Venture
investigation on Steel Framed Buildings (Gupta and Krawinkler, 1999). In the 20-story
building, the beams framing into the corner columns are released in one direction to
272
avoid complicated three-dimensional joint details and biaxial bending in the corner
columns.
All three building systems have been designed as perimeter moment resisting RCS frame
systems according to the recommended design provisions presented in Chapter 2. The
unit floor loads used in the design and modeling of each of the case study buildings are
summarized in Table 5.1. The live load is based on typical office floor load requirements
from the ASCE 7-02 standards. An additional concentrated wall weight of 1.0kPa is also
considered along the perimeter of the building (1.2kPa was assumed for the 3-story
building). Seismic design forces are based on a highly seismic site representative of Los
Angeles or San Francisco with mapped spectral accelerations of S s = 1.5 g and
S1 = 0.72 g , using IBC 2003. The soil condition at the building location is assumed to be
that of the site class D. The buildings are assigned a Seismic Use Group I, which
together with the seismic hazard level classifies the building into Seismic Design
Category D (as per IBC 2003).
The calculated seismic mass (including the weight of the structural members) and the
natural periods for each of the case study buildings are summarized in Table 5.1. The
natural periods are computed from the OpenSees model with gravity loads applied and
zero tensile strength in the concrete to represent cracked conditions. These natural
periods are consistent with those calculated using elastic models with averaged positive
and negative composite beam stiffness, 50% rigid panel zones, and effective cracked
column stiffness according to equation 2.5. The periods calculated according to the
simplified equation in the IBC 2003 (equation 2.3) (Ta) for the 3, 6, and 20-story frames
are 0.44, 0.68, and 2.43 seconds, respectively. Given that the computed periods from the
analytical models are greater than the code defined values, the IBC 2003 permits one to
use an effective period of 1.2Ta for calculating the design base shear. Thus, the base
shears are determined using periods of 0.53, 0.82, and 2.91 seconds for the 3, 6, and 20-
story building, respectively. Each of these periods is marked on the IBC (2003) design
hazard spectra in Fig. 5.3 with the corresponding design base shear coefficient (not
including the provision for accidental torsion). Note that the resulting base shear for the
273
20-story frame falls below the lower limit of the design base shear (represented by the
dashed line in Fig. 5.3). Therefore, the 20-story frame is designed based on the minimum
base shear coefficient of V W = 0.044 . The calculated building weights and governing
design base shear coefficient, including the accidental torsion effect, are presented in
Table 5.1.
Member dimensions and properties for the 6 and 20-story frames are shown in Tables 5.2
and 5.3, respectively; properties of the 3-story frame are reported in Chapter 3. Rolled W
shapes of grade 50 steel ( Fy = 345MPa ) are used for the beams and f c' = 41MPa (6ksi )
normal strength concrete is used for the reinforced concrete columns. The longitudinal
and transverse steel reinforcement in the columns is designed according to the seismic
details and recommendations as given in ACI-318 Chapter 21 with a nominal yield
strength of Fy = 414 MPa (60ksi ) . The general layout of the transverse reinforcement is
shown in Fig. 5.4 and 5.5 for the 6 and 20 story frames, respectively. The restraining
bars are provided to meet the clear span distance requirements of crossties (356mm or
14in) from ACI-318 and may vary depending on whether a steel erection column is
needed in the construction process. The slab is normal weight concrete with a minimum
specified compression strength of f c' = 27.6 MPa (4ksi ) . The controlling aspects of the
design of the buildings are presented in the following sections.
5.2.1 Beam Design
The beam sizes in all frames are controlled by the negative bending strength requirements
for the 1.2D+0.5L+1.0E load combination (IBC 2003). As recommended in Chapter 2,
the shear studs are designed to develop the full composite strength according to the
AISC-LRFD (2002) and the AISC Seismic Provisions (2002). The latter reference
applies a 25% reduction factor on the strength of the shear studs defined in the first
reference in order to account for the cyclic demands in seismic design.
274
5.2.2 Column Design
The design of the RC columns in each of the case study frames is controlled by the
strong-column weak-beam criteria. Table 5.4 lists the computed nominal strengths of the
interior and exterior columns, the plastic strength of the composite beams (according to
Section 2.3.2.2.1), and the corresponding strong-column weak-beam ratios according to
equation 2.46 for the 3-story frame. These ratios are slightly different than those
presented in Chapter 3 due to the fact that tributary gravity loads are now considered in
the columns and, therefore, additional strength is provided based on the column
interaction diagram. The two exterior columns are handled separately due to overturning
effects, which vary the axial loads and leads to large differences in strength. The beam
that frames into the exterior column with the larger strength (compressive overturning
force) will be in negative bending (i.e. steel only), while the opposite exterior column that
is weaker must resist composite action associated with positive beam bending. This is
the reason for the two SCWB ratios for the exterior columns in Table 5.4. The average
SCWB ratios over the floor considering composite beam action are 1.2, 1.1, and 1.0,
indicating that the spirit of the code is met by satisfying the SCWB criterion over the
entire floor. Considering only the bare steel beam strength, the SCWB criterion is met at
every joint with the average ratios increasing by about 20%. The alternative SCWB
criteria proposed by the SEAOC Blue Book are also specified in Table 5.4. This reveals
that both the 1st and 2nd floor are approximately 40% under-designed based on the
requirements of the alternative (proposed SEAOC) SCWB check (considering composite
beam strength).
Table 5.5 presents the SCWB information for the 6-story building for the frame lines 1
and 7. This direction of framing, as opposed to frame lines A and F, is modeled and
simulated in OpenSees later in this chapter. The SCWB design for the 6-story frame
pushes the minimum limits of the specified criteria, similar to the design for the 3-story
frame. The average ratios over each floor are 1.10, 0.96, 1.09, 1.02, 1.14, and 0.57 for
the first through roof level, in order. Again, if only the steel beam is considered, the
average SCWB ratios increase by about 17%, implying that there is some conservatism in
275
the design of the columns. The alternative SEAOC Blue Book SCWB ratios are also
reported considering composite beam action, which reveals that the strength of the
columns would have to be increased by almost 50% in order to satisfy this proposed
criterion.
Table 5.6 presents the same SCWB information for the 20-story building for the frame
lines A and F. Again, this represents this framing direction is modeled and simulated in
OpenSees later in this chapter. The SCWB ratios for this frame are much more
conservative than the two previous designs, with the average ratios over the each floor
equal approximately 1.60 (compared to ~1.0 for the other two building designs). Despite
this conservatism, this design would still require about a 20% additional strength in the
columns in order to satisfy the proposed alternative SEAOC Blue Book SCWB criterion.
While this conservative SCWB design was originally a result of miscalculation of the
plastic strength of the composite beams, this resulting frame design provides some insight
into the influence on the SCWB criteria.
5.2.3 Composite Joint Design
The composite beam-column joints in these frames are designed with the following
standard details: (1) face bearing plates, (2) band plates, and (3) joint ties within the beam
depth. The strength of the joints in the 3, 6, and 20-story frames, calculated according to
the updated strength model discussed in Chapter 2, are presented in Tables 5.7, 5.8, and
5.9, respectively. The ratio of the nominal joint strength to the nominal strengths of the
beams (i.e. strong-joint weak-beam ratio) is also reported. The design of the joints for
these case study buildings shows that the standard joint details will usually provide
sufficient strength to automatically satisfy the strong-joint weak-beam criterion.
A plan view of the column and a typical beam column joint in the 20-story frame is
shown in Fig. 5.6. Notice that in this detail the joint ties are allowed to pass directly
through the web of the main girder and the gravity beams. This is clearly seen in Fig.
5.7, which represents cross section A-A from Fig. 5.6. The band plate detail is utilized
276
both above and below the beam along with the face bearing plates (FBP) for both the
main girder and the gravity beam (Fig. 5.8). This detail is representative of the details in
the 6-story frame as well. The joint details in the 3-story frame are discussed in Chapter
3 and include a transverse beam framing into both sides of the column and without
through-web joint ties.
5.2.4 Drift Limitations
When checking the drift limitations for each of the frames, it should be noted that IBC
(2003) and ASCE 7 (2002) allow the designer to compute the base shear based on the
calculated period of the frame rather than the code-defined value. In addition, the lower
limit of the design base shear, represented by the dashed line in Fig. 5.3 does not apply to
the drift check. For these case-study buildings, this resulted in a reduction of the base
shear by approximately 25, 40, and 50% for the 3, 6, and 20-story frames. Using these
reduced forces, all of the frames automatically satisfy the 2% interstory drift limits set by
the IBC (2003) and ASCE 7 (2002) and no further resizing of the members is required.
While this is not always the case in the design of RCS frames, it does highlight the
efficiency of composite beams and RC columns to satisfy both strength and stiffness
requirements concurrently. This is in contrast to conventional RC or steel moment
resisting frames, where the sizes of the beams are typically controlled by the drift
criterion.
5.3 Nonlinear Dynamic Time History Analyses
Second-order inelastic dynamic analyses are performed on the three case study buildings
using the OpenSees analysis platform. Three analytical models are generated using the
recommendations presented in Chapter 4. A stripe analysis technique, described in
Section 5.3.1, is implemented in this study to investigate the performance of the three
case study buildings at a variety of hazard levels.
277
5.3.1 Ground Motion Scaling Techniques
The approach taken in the nonlinear time history analyses is to subject the analytical
model to measured ground motions, each of which represents a particular hazard level for
the building under investigation, and then report the engineering demand parameters of
interest, such as interstory drift ratio (IDR) and floor accelerations. The result of these
analyses can be represented as a scatter plot between the earthquake ground motion
intensity and the engineering demand parameter of interest. This strategy is shown in
Figure 5.9a, where each point represents the building response (maximum IDR) from a
single earthquake ground motion. One problem with this approach is that there is often a
lack of substantial data in the less frequent hazard levels (i.e. 2%in50years), because
there are very few recorded events that can provide these high intensities. This lack of
high intensity ground motions is why other approaches, such as the incremental dynamic
analysis (IDA) method (Vamvatsikos 2002), were developed.
A key assumption of most techniques that utilize recorded ground motions as input for
nonlinear response analyses is that it is reasonable to scale existing records to represent a
range of earthquake intensities. Shome (1999) has shown that scaling ground motion
records by the spectral acceleration at the first natural period of a structure is appropriate
to represent various earthquake intensity levels and does not introduce a bias into the
results. This conclusion is limited to first-mode dominated structures and does not
necessarily apply to frames that have significant higher mode effects. The higher-mode
limitation poses a problem for the 20-story frame and discussion on an alternate scaling
technique for this building is provided in Section 5.3.4.
If one accepts the legitimacy of scaling ground motions, one approach to organize the
analyses is to incrementally scale a single earthquake to a variety of ground motion
intensities resulting in what is often called an IDA curve. This can be thought of as the
dynamic equivalent of the static pushover curve for a single ground motion. This process
can be repeated for a number of different earthquake events and results in a series of IDA
278
curves, as is represented in Figure 5.9b. These curves are often taken up to collapse of
the structure and can exhibit nonlinear hardening and softening prior to collapse.
An alternate approach to the IDA concept is to select and scale different records, which
are more suitable to represent different hazard levels for a particular building and site.
This is in contrast to scaling one record to a wide range of intensities. This method,
which can be referred to as the stripe analysis technique, involves selecting a specific
number of appropriate ground motions and scaling each of them to the same hazard level,
thereby creating a stripe of response points. This is represented in Figure 5.9c for three
hazard levels. The appropriateness of a record depends on several factors, all of which
attempt to gauge how suitable the record is to represent the particular hazard level. This
topic will be discussed further in Section 5.3.3 when records are selected for each of the
case study frames. The process results in a series of stripes at various hazard levels, from
which a statistical analysis can be performed to find the mean and coefficient of variation
at each level. It is not appropriate nor is it necessary to connect the dots between each
stripe since it is not likely that the same set of earthquakes is repeated for each hazard
level. What is often reported are the curves that connects the means and a plus and minus
standard deviations for each stripe level. This stripe analysis approach is adopted in this
study.
5.3.2 Limitations of Analytical Models
As discussed in Chapter 4, the fiber beam-column element models can accurately capture
the flexural hinging response in composite steel beams and RC columns when it is
limited to steel yielding and concrete crushing and cracking. When pushed to larger
inelastic rotations, the response of these hinges begin to be dominated more by other
effects such as local buckling in the steel beams and reinforcing bar bond deterioration
and buckling in the RC columns. Given the fundamental assumptions of these fiber
models, these types of degrading behaviors are not captured in the response. Therefore,
when the response exceeds about 4% interstory drift, it is unlikely that the fiber models
are accurately capturing the complete behavior. It is for this reason that the analyses
279
presented in this chapter will not push these buildings out to very large drifts by scaling
the records up to simulate global collapse. Rather the limits of the model will be
acknowledged and the results presented in this chapter will be limited to earthquake
record intensities slightly above the 2% in 50 year event (or the maximum considered
event), where drifts will likely reach the limits of the validity of the fiber elements.
5.3.3 Ground Motion Selection
The earthquake records used in these analyses were selected from a suite of 80
earthquake ground motions. This suite of records was originally compiled by Medina
(2001) from the Pacific Earthquake Engineering Research (PEER) Center strong motion
database, where all records are consistently processed. All ground motions were
recorded on free-field sites that can be classified as site class D according to the IBC
2000 provisions. Figure 5.10 shows the distribution of the magnitude (Mw) and distance
(R) pairs for each of the 80 records.
In this study, 15 unique earthquake records are chosen for each stripe level, where each
record is scaled to a common spectral acceleration ( S a (T1 ) ) to represent a specific hazard
level for the building. Each of these records has been selected from the suite of 80
records using the following criteria:
1. The spectral acceleration of the unscaled record should be as close as possible to
the desired Sa-level, such that the scaling factor applied to the record is as near as
possible to unity. Adhering to this will avoid scaling records by very large or
very small factors. Unfortunately, this criterion cannot always be satisfied since
there is only a limited range of ground motions available for use. For this reason,
when larger scaling is required, a second criterion (#2) is also considered.
2. The epsilon of the earthquake record, which is defined by Baker (et al. 2005) as
the number of standard deviations by which an observed logarithmic spectral
acceleration differs from the mean logarithmic standard deviation, is also
considered in the selection of ground motions. Epsilon is computed by
subtracting the mean predicted ln S a (T1 ) from the records unscaled ln S a (T1 ) and
280
dividing by the logarithmic standard deviation (Baker et al. 2005). Both the mean
predicted ln S a (T1 ) and the logarithmic standard deviation are calculated from
attenuation relationships given the magnitude and distance to the rupture zone for
the record of concern. Baker et al. points out that at low mean annual frequency
of exceedance the ground motions are generally associated with positive-epsilon
events. Baker et al. argue that the scaling of zero or negative epsilon events to
achieve larger intensity records may not be appropriate and will likely over
estimate the demand on the structure. Therefore, they recommended to select
records with appropriate epsilon values for the type of scaling that one wishes to
perform. This means that when selecting records to scale up to represent lower
frequency events, records with positive epsilons should be used.
3. The records from the PEER strong motion database have been filtered for noise
using causal Butterworth filters with high-pass (HP) and low-pass (LP)
frequencies. The useable bandwidth of the records for the purpose of engineering
analysis is within 0.8 of the LP frequency and 1.25 of the HP frequency. The
limit on the low pass frequency is not of concern for the frames under
investigation given their long natural periods. On the other hand, the high pass
frequency of the record should be sufficiently lower than the natural frequency of
the structure, or rather the high pass period should be larger than the natural
period of the structure. This is particularly important considering that the natural
period of the frames will elongate as the structure becomes more nonlinear at
lower hazard levels. This criterion was most critical for the 20-story frame,
resulting in the elimination of 40 of the records in the ground motion suite, which
have high pass frequencies of 0.2Hz (5 seconds) that were considered too close to
the natural frequency of the frame (0.25 Hs, 4 seconds).
Figures 5.11 through 5.13 provide a graphical representation of the first two steps of the
ground motion selection process by plotting the spectral acceleration versus the epsilon of
each of the records for a period of 1.0, 1.5, and 4.0 seconds 1 . These figures shows that a
1
Note that the records for the 6-story building were chosen and scaled based on a natural period of 1.5
seconds, whereas Table 5.1 reports the period as 1.4 seconds. This difference reflects a change in the
281
large number of the earthquakes fall within the low spectral acceleration and low epsilon
region, which is expected since these are the more common events. The more intense
events (i.e. lower frequency) are represented by the data points that have both a large
spectral acceleration and large epsilon. Using the information in these graphs and the
selection criteria presented in steps 1-3, records are selected for each frame and stripe
level. The final scaled events for the 3-story frame through the 0.6g stripe level are
shown in Figs. 5.14a-g. For all stripes above 0.6g, the records used for the 0.6g stripe are
repeated given the lack of recorded ground motions in these higher intensity regions. The
results for the 6-story frame up through the 0.4g stripe level are shown in Figs. 5.15a-f,
where all stripes above 0.4g were created by re-scaling the 0.4g stripe. The records for
the 20-story frame were selected using the same technique presented here, however,
scaling of these records was performed in slightly a different manor, as described in the
next section.
5.3.4 Weighted 20-story Scaling Technique
One of the fundamental assumptions of scaling records based on S a (T1 ) is that the
structural response is largely dominated by the first-mode of vibration. This is clearly not
the case for a 20-story building, where the influence of higher modes is likely to be
significant. This suggests that an alternative scaling technique should be implemented to
account for the effects of the higher modes. This issue has been investigated by Shome
(1999) who recommends that a weighted-average scaling technique be used in the cases
where more than one mode has a large impact on the final response of a structure. The
number of modes that one should consider and their corresponding weighted values can
be determined from the modal-mass participation factors of the elastic structure. Table
5.10 lists the modal properties of the 20-story frame and their corresponding modal-mass
participation factors. Based on these participation factors, it seems reasonable to consider
the first three modes of the frame in the earthquake scaling procedure. The weighted
model between selecting the records and the final analysis model and is presumed to be small enough not to
have a large impact on the final results.
282
factors (WFi) are interpreted from these mass participation ratios and are presented in the
last column of Table 5.10.
# modes
considered
Sa (Ti )Target
Scale Factor = i =1 S a (Ti ) Record
WFi (0.1)
The solid line in Fig. 5.16 is an example of an unscaled response spectrum (IV79e12).
The white circles are points on the IBC equal hazard curve for the first three modes of the
20-story frame, each of which represent a Sa (Ti )Target value. The corresponding points
on the unscaled record at these three periods represent the S a (Ti )record values. Using
these data points and the weight values proposed in Table 5.10, a weighted scale factor of
4.14 can be obtained for this record from equation (0.1). The results of the record
selection (presented in Section 5.3.3) and the weighted average scaling for the 20-story
frame are shown in Figs. 5.17a-f for stripe levels up through Sa (T1 )Target = 0.20 g .
Beyond Sa (T1 )Target = 0.20 g , the records are repeated with only a difference in scaling
factors.
Whereas the scaled records for the 3 and 6-story frames are all scaled to a common
spectral acceleration value, S a (T1 ) ; using this weighted average scaling technique, the
scaled records for the 20-story building do not share a common spectral value. Rather
there is some variability in the hazard level that corresponds to the weight factor used at
each specific period. As seen in Fig. 5.17f, the variability is small around 4 seconds
where the weight factor corresponds to 75%. This variability increases dramatically at
the 1.5 and 0.8 seconds where the weight factors are only 15 and 10%, respectively. In
order to organize the analysis results for the 20-story frame, the engineering demand
parameters used to evaluate the performance of the 20-story frame will be plotted against
Sa (T1 )Target , despite the fact that there exists a small amount of variation at this period.
283
5.4 3-Story Perimeter Frame Results
This section will first begin by subjecting the analytical model of the test frame,
developed in Chapter 4, to the stripe analysis at four different hazard levels
corresponding to the levels tested in the laboratory. These results are investigated to give
us some insight to the response of the test frame and how this compares to other events.
The OpenSees model will then be adjusted from the laboratory conditions to represent the
case-study building with appropriate gravity loads and damping. Using this updated
model, a complete stripe analysis is simulated up to hazard levels just beyond the 2% in
50year event, defined by the IBC 2003 ( S a (T1 ) =1.08g). Important engineering demand
parameters are then calculated at each stripe level to evaluate the frame performance.
5.4.1 Interpreting the Test Frame Results
The measured maximum interstory drift ratios from the test frame are represented in Fig.
5.18 with the inverted triangle for the first four pseudo-dynamic events (discussed in
Chapter 3). These records are as follows:
1. 1999 Chi-Chi record scaled to 50/50 hazard level ( S a (T1 ) =0.41g)
2. 1989 Loma Prieta record scaled to 10/50 hazard level ( S a (T1 ) = 0.68g, 1a)
3. 1989 Loma Prieta record scaled to 80% of the 10/50 hazard level ( S a (T1 ) = 0.54g,
1b)
4. 1999 Chi-Chi record scaled to the Taiwanese 2/50 hazard level ( S a (T1 ) = 0.92g)
The simulated results from the analytical model of the test frame developed in Chapter 4
are also shown in Fig. 5.18 with the diamond marker. Recall that this model simulates
the laboratory conditions of the test frame, with zero viscous damping and where the full
gravity load is applied to the leaning column. This figure shows that the measured drift
for the first three loading events is captured very well by the analytical simulations. In
the fourth event (2/50), there is a large difference between the predicted and measured
maximum drifts, which can be largely attributed to the limitations in the fiber model
284
discussed in Section 5.3.2. This figure is one of the primary motivations for questioning
the validity of the fiber models beyond about 4% interstory drift.
Using this same analytical model, nonlinear earthquake time history analyses are
simulated using the ground motions selected in Section 5.3.3 corresponding to the
following four hazard levels: (1) S a (T1 ) = 0.40g, (2) S a (T1 ) = 0.60g, (3) S a (T1 ) = 0.72g,
and (4) S a (T1 ) = 0.9g. These levels represent the range of the Taiwanese spectral hazard
levels of 50, 10, and 2% in 50 years that were simulated in the composite RCS test frame
described in Chapter 3. In Fig. 5.19, the maximum interstory drifts for each scaled
ground motion are plotted against the intensity of the record, which is represented by the
intensity measure S a (T1 ) . The median and the 16th and 84th percentile response values
are also represented in this figure by the dashed lines. Note that the response points
described in Fig. 5.18 are repeated in Fig. 5.19. These stripe analyses of the test frame
provides the opportunity to (1) consider the performance of the test frame under different
input ground motions, and (2) evaluate the severity of the records that were used in the
test.
Before comparing the response of the test frame at each stripe level with that of the
measured and simulated test response, it is important to make a couple of key distinctions
between the two. Recall that the test frame was not repaired after each earthquake event,
so therefore the damage accumulates throughout all the loading events, making the frame
more flexible and susceptible to further damage. Contrast this to stripe analysis response
points, where each of the simulations represents a single event that is subjected to an
undamaged frame. In addition, there are several aspects of the pseudo-dynamic testing
methodology that could lead to larger drifts than would be experienced under actual
earthquake loading conditions (discussed in Chapter 4).
Now focusing on Fig. 5.19, it is apparent that the first three loading events fall within the
median and 84th percentile of the predicted response points. This implies that the records
selected for use in the test were representative of ground motions that were slightly more
285
severe than the median at their respective stripe levels and produced larger deformation
demands. At the S a (T1 ) = 0.9g stripe level, it is interesting to note that the predicted test
response point falls on the median of the stripe analysis response points. This was
intentional in the selection of this event, given that this record was carefully selected to
produce a moderate amount of damage and remain within the capabilities of the
laboratory. Given the limitations of the fiber beam-column, it is likely that the response
points in Fig. 5.19 which are equal to or larger than 4% interstory drift are beginning to
diverge from the actual response. This accounts for 6 of the 15 total records that were
analyzed at this hazard level. Despite these modeling limitations, there are important
implications that can be deduced from the analyses at the S a (T1 ) = 0.9g hazard level.
These are discussed in a subsequent section after the local response comparisons.
Figure 5.20 plots the simulated maximum response for selected columns, beams, and
joints as well as the corresponding measured values from the test frame. For the column
plastic rotations (Figs. 5.20a-b), the measured response of the test frame falls within the
range of simulated results for the first three events, but differences occur during the 2/50
event when measured response points fall in the tail of the predicted distribution in Figs.
5.20a. Similar trends are apparent in the beam plastic rotation plots in Figs. 5.20c-d,
where maximum measured plastic rotations for an interior and exterior beam hinge are
represented on these plots with an inverted triangle. Again, the maximum values from
the 2/50 event correspond to the excursion where severe local buckling had increased the
flexibility of the test frame. In addition to this, there is some measurement error when
attempting to track the large plastic rotations that occurred in these hinges in the test
frame, particularly in the measured point in Fig. 5.20c that shows plastic rotations up to
6% in the beam. While the 2nd floor joint response in Fig. 5.20f is also captured
relatively well, there are some noticeable differences in the simulated versus measured
response of the 1st floor joint (Fig. 5.20e). While some of this may be due to deficiencies
in the models, these particular joints measurements are prone to error because the data is
inferred from beam and column tiltmeters that likely includes additional rotation from
these members.
286
There are a few important conclusions to draw from the results shown in Figs. 5.19 and
5.20:
1. While the full-scale test presented in Chapter 3 provided a unique opportunity to
evaluate the performance of the frame under a series of earthquakes, Fig. 5.19
reveals that the performance exhibited in this test is not a definitive estimate of
performance. Rather it is but one instance within a larger distribution of possible
random earthquake ground motions.
2. The 1999 Chi-Chi record (TCU082) was initially chosen because the OpenSees
models predicted that this motion would produce a moderate amount of damage to
the test frame and stay within the limitations of the laboratory. This is confirmed
in Fig. 5.19, which shows that the analytical prediction of the test response lies on
the median of the 15 records analyzed at this level. One might consider what
would have happened if instead of selecting the TCU082 event, the LP89hda
record (1989 Loma Prieta), which is labeled on Fig. 5.19, was selected for the test
at the 2/50 hazard level? The peak drift in this event (2.2%) is well within the
capabilities of the analytical models, so it is reasonable to suspect that the test
frame would have matched the predicted response rather well. It also reveals that
there would have been much less damage under this LP89hda event than what
was observed in the test frame (under the TCU082 event). This would have likely
led to much different conclusions regarding the performance and modeling of
composite RCS frames. On the other hand, what if the LP89g03 record (marked
on Fig. 5.19) was selected to represent the 2/50 hazard level event on the test
frame? Predicted drifts for this event are very large (6.4%) and beyond the
limitations of the fiber models. In this case, it is likely that if local buckling and
general stiffness degradation was incorporated into the frame performance, this
record has caused enough damage and drift to lead to global collapse of the frame.
3. What would the response of the frame to the 2/50 event have been if it was
subjected to an undamaged frame. In other words, how significant to the final
response of the test frame was the fact that the frame had already experienced the
equivalent of two major earthquakes (50/50 and 10/50 event)? Note that a
moderate amount of local buckling had already occurred in most of the steel beam
287
hinges during the first two events, which leads local buckling at much earlier
stages in the 2/50 event. This would increase the flexibility and reduce the
capacity of the beam hinges in negative bending as compared to an undamaged
beam hinge and ultimately lead to larger interstory drifts.
5.4.2 Modeling Variations for Realistic Building Case Study
The following adjustments are made to the 3-story model of the test frame in order to
account for more realistic conditions in an actual building:
1. In the pseudo-dynamic frame test, the only gravity load on the frame test was that
of the frame itself. All other gravity loads that were tributary to the frame and to
gravity columns supported by the frame were modeled in the pseudo-dynamic
algorithm with a leaning column. This same distribution of loading was
implemented in the analytical model of the test frame. For the model results
described later in this chapter, which are intended to represent the real building,
the gravity loads are distributed between the perimeter frame and gravity column
according to the actual floor framing plan. Roughly one-third of the gravity (full
dead and 25% live load) is applied to the moment frame and two-thirds is applied
to the leaning columns. This does not change the effective P-delta effects on the
structure, but the shift of gravity load to the RC columns will increase their
moment capacity by approximately 10% and increase the effective stiffness by
approximately 15%.
2. As discussed in Chapter 3, there was no viscous damping included in the pseudo-
dynamic algorithm for the test frame. Accordingly, viscous damping was set to
zero in the analytical model presented in Chapter 4. The building model in this
chapter does, on the other hand, include 2% viscous damping at the 1st and 3rd
modes of the structure. Given that the natural period of this structure (Tn = 1sec)
falls within the velocity-sensitive region of response spectrum, damping is
expected to result in a noticeable decrease in the response of system as compared
to structures that fall within the acceleration or displacement-sensitive regions of
the spectrum (Chopra 1995).
288
In order to highlight the differences of the laboratory versus the realistic building
analytical model, the test frame loading protocol is analyzed using the revised model and
compared to the original model. Results for the two analysis models, as well as the
measured results, are summarized in Table 5.11. The results show that the adjustments
to the mold reduce the maximum interstory drifts by about 20% in the larger earthquakes,
which is primarily due to the viscous damping added to the realistic model. This effect
will be investigated further in Section 5.4.2.2.
Both models incorporate the measured material properties of the test frame, as described
in Chapter 3. The measured steel and concrete strengths are presented in Tables 5.12 and
5.13, respectively. These tables show that the measured values all exceed the expected
strength values. While there is a very large material overstrength of the concrete in the
RC columns, this only leads to approximately a 10% increase in the moment capacity of
these columns given that they are largely controlled by the yielding of the longitudinal
reinforcement. The measured yield strength of the steel in the 2nd floor beams is
approximately 30% higher than the expected yield strength (1.1Fy), which has the
potential to have a large impact on the balance of strength between the beams and the
columns. Fortunately, given that the measured strength of the concrete and the rebars
were also found to be higher than the expected values, these differences do not severely
impact the capacity design criteria (e.g. SCWB provisions) and the performance of the
frame. One could imagine a case where these measured properties could adversely affect
the system performance and shift the hinging from the beams into the columns, perhaps
leading to an undesirable story mechanism. While these results are based on a single test
frame, this does highlight the potential problems that one could face in the design,
analysis, and performance prediction of real building systems. One way to approach the
analysis of this type of problem is implement a sensitivity analysis of several of the key
properties of the frame that could affect the system performance.
289
5.4.2.1 Static Pushover Response
Figure 5.21 shows the roof drift ratio versus the base shear from a static pushover
analysis on the 3-story building using the IBC 2003 force distribution. This figure shows
that the building has a maximum strength of approximately 2.4 times the design base
shear. This overstrength is within the range expected (2-3) of buildings designed per
current standards. In Fig. 5.22, the interstory drift profile of the building is plotted at
various stages of the pushover analysis. The expected levels of deformation can be
computed using the so-called displacement-coefficient method from FEMA 273. This
method determines the target displacement, t, to which the structure is expected to be
pushed during a design level earthquake. For the 3-story frame, the 10/50 target
displacement level is approximated as 1.9% roof drift. From Figs. 5.21 and 5.22, at this
level of drift, the building is expected to reach its maximum strength with deformations
tending to concentrate in the 2nd floor. As the frame is pushed further, deformations
continue to concentrate in the first two floors, indicating that the 2-story mechanism
observed in the test frame is also captured in the analytical model.
5.4.2.2 Derivation of Damping and Effects on Maximum Response
Initially, the analytical models used in this study contained no damping so as to mimic
conditions in the test frame. In order to more accurately model the conditions of a real
building, 2% damping is included in this updated model. In OpenSees, damping is
assumed to be linearly proportional to the mass and stiffness matrix, according to the
Rayleigh equation (Chopra, 1995).
C =M + K (0.2)
where:
M, K = Mass and stiffness matrix of the MDOF structure
, = mass and stiffness parameters of proportionality determined by to the
following equations:
290
2 ( j j ii )
=
2j i2
= 2ii i2
where damping ratios, i and j, are assumed at two frequencies, i
and j, of the structure.
In modeling moment resisting frames, the mass matrix is typically defined as a diagonal
matrix with the mass of each floor along the diagonal. In OpenSees, the stiffness matrix
used to calculate the damping matrix per (0.2) can be defined as one of three variations:
(1) the current iteration of the model ( K current ), (2), the last committed state of the model
( K lastCommitted ), or (3) the initial state of the model ( K initial ). The first and second options
are roughly the same in a time history analysis since the load steps are small, although the
later seems to make more sense given that the stiffness is based on a converged step
rather than the current Newton iteration. Both of these variations of the Rayleigh
damping equation will cause the damping matrix to continually change as nonlinearities
in the system occur, which maintains an effective equivalent damping, but would result in
equilibrium imbalances that could ultimately be detrimental to the reliability of the
computed results (Bernal, 1994; Kannan and Powell, 1973). The third option is to use
K initial , which implies a constant damping matrix over the entire time history analysis
according to equation (0.2). This approach leads to the potential for the development of
spurious forces at degrees of freedom with small (or zero) inertias, which is a result of the
loss of orthogonality of the damping matrix when yielding occurs in the structure (Bernal,
1994). This effect only occurs as the frame experiences inelasticity, and has a larger
effect on the computed forces than the maximum displacements, given that the plastic
excursions typically occur over short periods of time. This approach would also lead to a
change in effective damping according to the amount of nonlinearity that occurs during
the time history. There are ways to condition the damping matrix such that these
spurious forces are not introduced into the system (Bernal, 1994), but these methods are
currently not implemented in OpenSees.
291
The effects of the different formulations of the damping matrix, as well as zero damping,
are compared in Fig. 5.23. This figure shows the median response of the model at the
three stripe levels ( S a (T1 ) = 0.40, 0.72, and 0.90g) with four different assumptions for
damping: (1) zero damping, (2) K current , (3) K lastCommit , and (4) K initial . The results show
that the constant damping matrix based on the initial stiffness causes the biggest
reduction of the maximum response (approximately 20%), which is likely a combination
of the increase in relative damping as the structure gets more nonlinear and the effect of
the spurious forces. The K lastCommit damping shows about 5% larger response than the
K initial , but also leads to convergence problem as 5 of the records were unable to converge
(~10% of events). This problem gets worse when the damping is based on K current
causing a total of 10 records to fail to converge, thereby making the median response
unreliable.
In this study, it was decided to base the damping matrix on K initial . This study accepts the
spurious nodal forces at the massless degrees of freedom, in exchange for better global
convergence of the model. It also is based on the assumption that the inherent damping
in the building is associated with physical properties that would be present throughout an
earthquake. As discussed later in Chapter 6, this problem with damping is something that
needs to be resolved in OpenSees. For now, the differences shown in the median drift
response are small enough to accept this problem.
The alpha and beta coefficients (equation (0.2)) for the 3-story frame are computed
according to the 1st and 3rd modal frequencies of the frame assuming a cracked stiffness
of the columns and beams, anticipating some elongation of the period during the
earthquake excitation. Figure 5.24 depicts the damping ratio as a function of the
frequency by assuming 2% damping at 6.11 and 40.65 rad/sec, which correspond to the
square root of the 1st and 3rd eigenvalues of the cracked stiffness of the 3-story RCS
frame. The frequency corresponding to the initial uncracked stiffness of the frame is also
marked on this figure at approximately 8 rad/sec, which corresponds to a damping ratio
of approximately 1.7%.
292
5.4.2.3 Global Response
The stripe analysis results are plotted in Fig. 5.25 with the EDP as the maximum
interstory drift ratio (the maximum IDR that occurred in any floor of the frame). Each
stripe contains the results for 15 ground motions corresponding to the events chosen in
Section 5.3.3. The median, 16th, and 84th percentiles for each stripe level are connected
by the solid and dotted lines, respectively. This figure reveals that the response of the
structure with respect to increasing earthquake intensities remains relatively proportional
to the intensity up to the design hazard level ( S a (T1 ) = 0.72g). At this hazard level, larger
nonlinearities in the frame begin to increase the variability of the response, with the
predicted median response just above 2.5% interstory drift and a dispersion ( ln idr |Sa ) of
17.5%. Note that this median exceeds the design drift limit of 2% set by the IBC 2003
and the ASCE-7 (2002). With an implicit lognormal assumption, these results indicate
that there is a probability of 78% that the frame will exceed the design limit of 2% drift in
a design level event. As the frame is pushed to the 2% in 50 year level ( S a (T1 ) = 1.08g),
the median response increases out to 3.5% interstory drift with a dispersion of 28.9%.
Note that 6 of the 15 records at this hazard level exceed 4% interstory drift, implying that
the assumptions in the fiber models are beginning to diverge from what would be the true
response (e.g., due to local buckling in the steel beams, bond slip in the RC columns,
etc.).
Figure 5.26 contains six plots corresponding to the first 6 IDA stripe levels ( S a (T1 ) =
0.045-0.5g) each depicting the instantaneous drift profile of the frame at the time of the
maximum interstory drift. The dark circle on each of the drift profiles indicates the floor
at which the maximum interstory drift occurred. Figure 5.27 depicts the same
information for the remaining 6 IDA stripes ( S a (T1 ) =0.6-1.15g). These plots show that
the frame responds primarily in a first mode of vibration during the lower level IDA
stripes. As the intensity is increased to the design level ( S a (T1 ) =0.72g) and higher, the
293
maximum drifts begin to occur in the 2nd story, implying that the nonlinearities are
beginning to concentrate in this floor. This behavior is a bit different from the test frame,
where the drift concentrated in the 1st and 2nd story, but was largest in the 1st. Figures 5.28
and 5.29 reevaluate this interstory drift information by plotting the absolute maximum
value for each floor given each event at each IDA stripe level. These maximum IDR
values do not necessarily correspond to the same time in the record. The median, 16th,
and 84th percentiles are also plotted in these figures. The maximum median drift at the
design level ( S a (T1 ) =0.72g) is 2.3%, indicating that the drift limitations of 2% defined
by the ASCE 7 (2002) has been exceeded. These figures also confirm that deformations
are starting to concentrate in the 2nd floor of the frame at the higher intensity levels. This
is evident at the 2/50 level ( S a (T1 ) =1.08g), where the median drift at the 2nd floor is
3.6% interstory drift (84th percentile of 4.9%). The concentration of deformation in the
2nd floor was also captured in the static pushover analysis described in Section 5.4.2.1.
5.4.2.4 Member Plastic Rotations
Figure 5.30 contains six plots that show the maximum plastic rotations in the lower and
upper hinges of the interior columns of each floor as a function of increasing earthquake
intensity. As expected, the base hinges in the 1st story columns experience the most
significant plastic rotations in all column hinges, with the median values of 0.013 and
0.020 rad of plastic rotation for the 10/50 and 2/50 hazard events. These figures also
show that the plastic rotation in the 2nd story column reach 0.007 and 0.011rad in the
lower hinge and 0.004 and 0.007rad in the upper hinge for the 10/50 and 2/50 hazard
levels, respectively. Similar trends exist for the exterior columns in Fig. 5.31. This
reinforces the trends seen in the maximum IDR plots (Figs. 5.28-5.29) and confirms that
hinging occurs in the upper and lower hinges of the 2nd story columns.
The absolute maximum plastic rotations in the composite beams of each floor are shown
in Fig. 5.32 as a function of increasing earthquake intensity. This figure shows that
plastic rotations are distributed fairly evenly throughout all floors with median values of
0.010, 0.011, and 0.012rad in each of the three floors for the 10/50 hazard level. These
294
values increase to 0.018, 0.020, and 0.020 for the 2/50 hazard level. This shows that
despite the yielding that is occurring in the columns, the beams at each floor are also
experiencing a significant amount of hinging.
Rather than plotting the absolute maximum plastic rotation in the beams, consider the
maximum positive (i.e., composite beam) and negative (i.e., steel only) plastic rotations
separately, as shown in Fig. 5.33. This figure reveals that there is very little positive
plastic rotation (corresponding to composite action) in the events less than the 10/50
hazard level, and the median response at higher earthquake intensities (2/50) reaches
approximately one-half of the negative plastic rotation (i.e., steel only). This confirms
that the plasticity of the beams is dominated by the response in negative bending and that
the slab prevents large plastic rotations in positive bending. This reveals that more of the
rotation demand is placed on the columns when the beams are in positive bending, which
further substantiates the consideration of the composite strength of the beams in the
strong-column weak-beam design of the RC columns.
The maximum joint rotations for the interior and exterior beam-column joints are shown
in Figs. 5.34 and 5.35, respectively. A line is marked at 2% joint rotation on each of
these plots because this is recognized as the deformation level at which the joint attains
its maximum strength. These figures further substantiate other data which show that the
composite RCS joints are inherently strong and capable of forcing hinging to occur in the
surrounding beams and columns.
In order to get a better appreciation of the relative distribution of these plastic

deformations, the median and the 16th and 84th percentile for the response at the 10/50
hazard level in Figs. 5.30 through 5.35 are re-plotted in a more compact way in Fig. 5.36.
This new figure shows side-by-side comparisons of the median plastic rotations
(represented by the heavy black bars) for the interior and exterior columns (upper and
lower hinge), the beams, and the interior and exterior joints. In addition, the 16th and 84th
percentile values are represented by the white circles connected by a horizontal line. This
shows that for the 10/50 hazard level, the most severe plastic rotations occur in the 1st
295
floor interior and exterior base hinges. The lower hinge in the 2nd floor is experiencing
about half the plastic rotation seen at the base hinges. Again, the distribution of plastic
hinging is relatively even in the beams for each floor and approximately the same
magnitude as the 1st floor column base hinges. The joints all remain relatively elastic.
These same damage trends extend to the 2/50 hazard level, as shown in Fig. 5.37.
5.4.2.5 Damage Indices
As discussed in Chapter 4, the Mehanny damage index (2001) has been shown to
correlate the history of plastic rotations in a component with the severity of damage and
the repairability. These damage indices provide a better measure of damage since they
account for both the maximum plastic excursions as well as the cumulative damage
caused by repeated excursions. The monotonic plastic rotation capacities of the
components of the frame are computed using the methods described in Section 4.4.1 and
are summarized in Table 5.14. Note that the predicted plastic rotation capacities for the
RC columns are rather large, particularly in the exterior columns. Recall the discussion
in Section 4.4.1, which, based on work by Fardis et al. (2001), argued that the rotation
capacities for the interior columns are reasonable, but acknowledged that those for the
exterior columns are excessively large given that the model used in this study does not
consider all possible failure modes in the column (e.g., reinforcing bar buckling or
fracture).
The plastic rotation history of each of the components in the frame has been processed
with the Mehanny damage index for each of the time history analyses. These results are
plotted in a similar manner as the maximum plastic rotations plots that have been already
presented in Section 5.4.2.4. The computed damage indices for in the interior and
exterior columns are presented in Figs. 5.38 and 5.39, respectively. The damage index
plots for the composite beams are shown in Fig. 5.40 and the interior and exterior joints
are shown in Fig. 5.41 and 5.42, respectively. On each of these plots are three vertical
lines that separate the negligible to moderate (0.5), moderate to significant (0.7), and
significant to loss of capacity (0.95) damage states. Each of these damage states are
296
described in Table 5.15. Also, two horizontal lines are drawn on these figures to indicate
the 10/50 and 2/50 hazard levels.
The first thing that stands out about these plots is the extreme outliers that first appear at
the S a (T1 ) = 0.3g stripe and continue up through the maximum S a (T1 ) = 1.15g stripe.
These response points belong to a single event, a record from the 1979 Imperial Valley
event (IV79dlt), which is scaled throughout the range of this stripe levels. This record is
particularly damaging because the event lasts about 100 seconds with approximately 50
seconds of strong motion, which leads to a significant number of plastic excursions.
Compare this to the more modest events that last about 40 seconds and have strong
motion durations that range from 10-20 seconds. This shows that these damage indices
are able to capture the duration effect of a longer event, something which is not reflected
in the maximum plastic rotations.
Shifting the focus to the median values, the damage indices indicate that the 1st story
column base hinges experience the most damage of all the column hinges (Fig. 5.38).
The 84th percentile damage index of these hinges cross into the moderate damage state
just below the 10/50 hazard level while the median crosses into this region at the 2/50
hazard level. From the distribution at 10/50 hazard level, it is estimated that there is an
18% probability that these hinges fall within the moderate damage state and only a 5%
chance of exceeding this damage state. These probabilities increase to 32 and 18%,
respectively, at the 2/50 hazard level. The only other column hinges that experiences
noteworthy damage is the lower and upper hinges of the 2nd story columns, which have
approximately a 15 and 7% probability of exceeding the negligible damage level at the
2/50 level, respectively. The damage in the lower hinge of these columns is counter to
the assumption of the SEAOC SCWB provision that presumes that columns will tend to
incur damage in the upper column hinge of the story.
The damage indices for the composite beams (Fig. 5.40) predict much earlier and more
significant damage than seen in the columns. Table 5.16 breaks down the probability of
the beam hinges in each floor being within moderate, significant, and loss of capacity
297
damage state levels for the 10/50 and 2/50 hazard levels. This table shows that (1) there
are much higher probabilities of damage that requires some sort of repair than what was
seen for the RC column hinges and (2) despite the damage seen in some of the upper
story columns (suggesting the formation of a story mechanism), the beam hinges are still
experiencing high inelastic rotations.
The damage indices for the interior and exterior beam-column joints are shown in Figs.
5.41 and 5.42. Other than the outlier data points from the IV79dlt record, the predicted
damage indices for the joints all remain within the negligible damage state.
Figure 5.43 summarizes the median and 16th and 84th percentile values for the damage
indices in the columns, beam, and joints at the 10/50 hazard level. This figure shows that
the median damage is indicative of a side sway frame mechanism where the column base
hinges and the beam hinges in each floor experience the most damage, which is
representative of the damage observed in the test frame after the design level event. The
level of damage in these hinges is indicative of the performance expected by FEMA 356,
with the 84th percentile levels entering the moderate damage region.
Figure 5.44 summarizes this same data for the 2/50 hazard level, which exhibits a similar
trend as the 10/50 hazard level. This shows that the beams are much more likely to
experience moderate to significant damage, which would imply that local buckling is
becoming more prevalent in these hinges. These damage states correlate to those
observed in the test frame and should be expected given the higher level of drifts at this
hazard level (median interstory drift of 3.5%). The median damage in the base hinges of
the 1st floor column is within the moderate level while the 84th percentile falls within the
significant level. This figure also shows that the predictions imply that the lower hinge of
the second floor column is beginning to reach moderate levels of damage (84th
percentile). Recall that the test frame showed more damage in the upper hinge of the 2nd
floor than in the lower hinge, while the analytical model is predicting the opposite. This
may be attributed to the fact that the predicted drifts of the frame are exceeding the
limitations set on the fiber models, with 6 of the 15 events surpassing 4% maximum drift.
298
Given that local buckling is not being adequately captured in the 1st floor beam at these
drift levels, the analytical models of the beams are not capturing the strength degradation
in these hinges. The result may be that the beams are imposing unrealistically high forces
into the columns, leads to hinging in the lower hinges of the 2nd story columns.
Another interesting difference between the test behavior and the response of the
analytical models is that the 2nd floor beams are experiencing roughly the same amount of
damage (moderate to significant) as the 1st and 3rd floor beams, which was not observed
in the test. The damage in the 2nd floor beam was limited to minor yielding in the test
frame. This difference is likely attributed to the fact that the axial loads (from the
tributary gravity loads) have increased the capacity of the RC columns, thereby
increasing the average SCWB ratio by approximately 15% from the original test frame
model. This change has the potential to shift more of the damage into the 2nd floor beams
than what was observed in the test frame.
The performance assessment of the 6-story building will focus on lateral resisting frames
corresponding to frame lines 1 and 7 from the building plan presented in Fig. 5.1. This
frame is modeled in OpenSees according to the recommendations in Chapter 4 and
subjected to a complete stripe analysis up through the hazard level associated with the
maximum considered event, which is defined by IBC 2003 as S a (T1 ) =0.72g and can be
assumed as the intensity of a 2% in 50 years ground motion. The performance of the

frame is evaluated in a similar manner as the 3-story frame presented in the previous
section. Damping is defined as 2% at the first and third modes of the frame, which
corresponds to an alpha and beta coefficients of 0.16 and 1.36x10-3 (equation (0.2),
shown graphically in Fig. 5.45). Tributary gravity loads (full dead and 25% live load) are
imposed on the moment frame and the remaining loads are imposed upon a leaning
column. The ground motion records used at each hazard level have been discussed in
Section 5.3.3.
299
5.5.1 Static Pushover Response
The results from a static pushover of the 6-story building model are shown in Fig. 5.46.
This figure shows that the maximum strength of the building is approximately 2.9 times
the design base shear. The drift profiles of the frame at the several stages of the pushover
are shown in Fig. 5.47. At the design base shear, interstory drifts remain below 0.5%,
with a relatively even level of deformation demand throughout all stories. The maximum
strength of the building occurs at approximately 2.0% drift. The target displacement, t,
from FEMA 273 is computed as 1.6% for a design level earthquake. At this drift level,
deformations are beginning to concentrate in the middles stories, with the most
pronounced drifts in 3rd and 4th stories. This localization of damage continues in these
floors as the frame is pushed to higher drift demands.
5.5.2 Global Response
Maximum interstory drifts versus the hazard level, as defined by S a (T1 ) , are plotted for
the 6-story frame in Fig. 5.48. The median response of the 6-story frame exhibits a slight
hardening effect with increasing earthquake intensities. At this design hazard level, the
median value is approximately 1.9% interstory drift with a dispersion ( ln idr|Sa ) of about
13%. Note that the median response is just below the 2% drift limit set by the IBC 2003
and the ASCE-7 (2002). Assuming a lognormal distribution, this corresponds to a
probability of 32% that the frame will exceed the 2% drift limit in a design level event.
At the 2/50 hazard level ( S a (T1 ) = 0.72g), the frame is pushed to a median drift of 2.5%
with a dispersion of approximately 20%. Note that the drift ratios for all of the 2/50
records are less than 4%, which suggests that the fiber beam-column elements can
accurately simulate the response. In fact, only two events exceed this threshold as the
earthquake intensity is increased to the largest intensity analyzed in this stripe analysis,
S a (T1 ) = 0.8g.
300
Figures 5.49 and 5.50 shows the instantaneous drift profiles of each analysis at each
stripe hazard level. These plots show that the maximum drift (marked by the dark circle)
typically occurs within the 3rd, 4th, or 5th story for all hazard levels. As the earthquake
intensities are increased, it is apparent that there is a bulging effect of the middle stories
indicating that a localization of response in this region is beginning to control the
response of the building. Figures 5.51 and 5.52 show the absolute maximum interstory
drifts for each floor over each of the earthquakes at each hazard level. These plots show
similar trends in that the middle 4 stories display the largest displacement demands. This
concentration of demands is also captured in the static pushover response described in
Section 5.5.1. The 1st-story only reaches a median drift of 1.8%, indicating that rotation
demand at the base hinges of the 1st-story column is likely much less than what was seen
in the 3-story building.
5.5.3 Member Plastic Rotations
For brevity, only the median, 16th and 84th percentile summary plots are examined for
general trends and insights into the performance of the 6-story frame. Figure 5.53 shows
the summary plot for the 10/50 stripe level ( S a (T1 ) = 0.48g). This figure shows that the
bulging effect of the middle stories is apparent in the maximum beam plastic rotations,
with median values reaching as high as 1% in the 3rd floor beams. The base column
hinges are subjected to median plastic rotations of only 0.6% and 0.8% in the interior and
exterior columns, respectively. This is approximately half of the demand that was seen in
the 10/50 level for the 3-story frame. This difference is likely attributed to the
localization of the response in the middle stories, which alleviates the demands on the
lower story. This figure also shows that the median joint rotations are generally larger
than what was seen in the 3-story frame, but are still well within their maximum capacity.
A summary of the plastic rotation demands for the 2/50 hazard level is shown in Fig.
5.54. The 3rd-floor beam is subjected to the highest deformation demand, with the
median plastic rotation reaching approximately 1.5%. The bulging effect of the middle
floors is still apparent in the maximum beam plastic rotations. The demand in the interior
301
and exterior base column hinges are approximately 1.0 and 1.2%, respectively, which
again is low compared to the demands seen in the 3-story frame. There are also
comparable plastic rotations occurring in the lower hinges of the 3rd story columns, and to
a lesser degree in the upper hinges of the 4th and 5th story columns. While damage is
occurring in the upper story columns, there are no signs of development of story
mechanisms at these hazard levels. The deformation demands in the composite joints
remain within the same limits that were seen in the 10/50 hazard level (0.5-1.0%).
5.5.4 Damage Indices
The damage index developed by Mehanny et al. (2001) is used to process the plastic
rotation histories from each of frame components, which can then be correlated to a
specific physical damage state level according to the relationships in Table 5.15. The
plastic rotation capacities (pu) for each of the components in the 6-floor frame are
computed according to the methods outlined in Section 4.4.1 and are summarized in
Table 5.17. As with the hinge rotation demands, only the overall trends are reviewed
through the median, 16th and 84th percentile summary plots.
Figure 5.55 shows the summary of the damage indices in each of the components of the
frame for the 10/50 hazard level. The median damage in the 3rd and 4th floor beams
remain within the negligible damage state but the 84th percentiles are pushed to the
moderate damage state, which is comparable to what was seen in the results of the 3-story
frame. The 1st story column base hinges remain within the negligible damage state and is
less severe than what was seen in the 3-story frame. This again can be attributed to the
higher mode effects on the 6-story frame that tend to push the middle stories to larger
deformations than the lower stories. The composite joints all remain within the
negligible damage state, but do tend to experience more damage than the joints in the 3-
story frame.
Summary of the damage indices for the 2/50 hazard level are shown in Fig. 5.56. This
figure shows that the 3rd and 4th floor beams experience the most damage of all the
302
components in the frame, with medians of moderate damage and the 84th percentiles
approaching significant damage. Damage in the 1st story base hinges remains limited
with only the 84th percentile of the exterior columns experiencing moderate levels of
damage. Similar to the 10/50 hazard level, the lower hinge of the 3rd story exterior
column is also showing comparable damage to the 1st story base hinges. The 84th
percentile of the 5th story columns are also on the verge of moderate damage. The
damage in the composite joints has not changed much from what was predicted in the
10/50 hazard level.
The performance assessment of the 20-story building will focus on lateral resisting
frames corresponding to frame lines A and F from the building plan presented in Fig. 5.2.
The analytical modeling will follow the recommendations presented in Chapter 4 and
subjected to a complete stripe analysis up through the 2% in 50year hazard level, which is
defined by IBC 2003 as Sa (T1 )Target =0.18g. Recall that in Section 5.3.4, an alternative
weighted average scaling technique was implemented in the scaling of the ground
motions for the 20-story building. The performance of the frame is evaluated in a similar
condensed manner as the 6-story frame presented in the previous section. Damping is
defined as 2% at the at the first and fourth mode of the building, which corresponds to an
alpha and beta coefficients of 0.049 and 2.88x10-3 (equation (0.2), shown graphically in
Fig. 5.57). Note that the damping at the first mode is approximately 1.8%, which is due
to a change in the model that was not accounted for in the Rayleigh coefficients. This is
only a small error and the impact on the final results should be insignificant. Tributary
gravity loads (full dead and 25% live load) are imposed on the moment frame and the
remaining loads are imposed upon a leaning column. The ground motion records used at
each hazard level have been discussed in Section 5.3.3.
303
5.6.1 Static Pushover Response
It should be acknowledged that the static pushover of the 20-story building is less likely
to capture the realistic deformation and damage pattern of the dynamic response of the
building. Nevertheless, the results of the static pushover are shown in Fig. 5.58 for
comparison to the trends observed later in the dynamic analyses. This figure shows that
the building reaches a maximum strength of approximately 1.9 times the design base
shear, which is substantially lower than the overstrengths in the 3 and 6-story frame. At
the design base shear, the deformation demands push the middle stories more that the
upper and lower stories (Fig. 5.59). As the frame is pushed further, deformations begin
to concentrate in the lower stories, particularly in the 1st through 5th floors. These types
of results highlight the limitations of the equivalent static lateral load approach used in
design to capture higher mode effects often found in taller structures. Comparison to the
actual dynamic response will be made in the following section.
5.6.2 Global Response
Maximum interstory drifts are plotted against the hazard level, represented by
Sa (T1 )Target , for the 20-story frame in Fig. 5.60. The median, 16th and 84th percentile
response are also shown with the solid and dashed lines, respectively. The median shows
a slight hardening response with increasing earthquake intensities, which is similar to the
results of the 6-story building. At the design hazard level, the median value is 2.2% with
a dispersion ( ln idr|Sa ) of 19%. Assuming a lognormal distribution, this corresponds to a
67% probability that the ASCE-7 (2002) drift limit of 2% will be exceeded in this frame
during a design level event. At the 2/50 hazard level ( Sa (T1 )Target = 0.27g), the frame
experiences a median drift of 2.9% with a dispersion of 15%. The drift limitations (4%
interstory drift) set on the fiber beam-column models is exceeded only under one ground
motion at Sa (T1 )Target = 0.3g, suggesting that the analysis results are fairly accurate..
304
From the onset of this study, it was recognized that higher mode effects would influence
the response of this building. These effects are highlighted in Figs. 5.61 and 5.62, which
show the maximum instantaneous drift profiles of each time history analysis at each
hazard level. The floors that experience the maximum interstory drift are marked by the
dark circle. Under very low excitations ( Sa (T1 )Target = 0.01 and 0.02g in Fig. 5.61), the
building experiences an upper story whiplash effect, with maximum drifts occurring in
the 17th and 18th floors. As the earthquake intensities are increased, the shape of the drift
profile changes from record to record, with no real distinct pattern. At the design level
hazard, the maximum drifts occur throughout the 9th through 13th floor and the 3rd floor.
This trend continues at the two longest intensities ( Sa (T1 )Target = 0.27 and 0.3g), where six
of the fifteen events produce maximum drifts in the 3rd floor, with the rest distributed
throughout the 9th through 18th floor.
Figures 5.63 and 5.64 show the absolute maximum drifts of each floor for each hazard
level in a non-synchronous manner. These data show similar trends as the instantaneous
drift profiles. At higher intensities, the maximum drifts occur in the 1st through 4th floors
and the 10th through the 17th floors. Contrast this to the predicted deformation demands
in the static pushover analysis, where only high demands in the lower floors were
captured. This shows that evaluating the response of a taller building solely on the static
pushover results will likely underestimate the demand in the upper stories associated with
higher mode effects.
5.6.3 Member Plastic Rotations
The median summary plots of the maximum plastic rotations in the columns, beams, and
joints are shown in Figs. 5.65 and 5.66 for the 10/50 and 2/50 hazard levels, respectively.
Focusing on the 10/50 level (Fig. 5.65), it is interesting to observe that the plastic
deformations in the upper story columns are very small. Even at the 84th percentile
response, the columns in the upper floors remain below 0.4% plastic rotation. Similarly,
at the 2/50 hazard level, the median plastic rotations in the upper story columns remain
305
below 0.5%. Contrast this to the to the upper story columns in the 3 and 6-story frames
that experienced moderate levels plastic rotation and damage, even in the design level
event. This behavior is likely attributed to relatively stronger columns in this frame
compared to the two previously discussed frames. The SCWB ratios in this 20-story
frame were, on average, 60% larger than the code minimum, considering the composite
strength of the beams. Recall that the alternate SEAOC Blue Book SCWB criterion
would still require approximately 20% more strength in the columns.
The 1st story column base hinges do experience moderate plastic rotation demands,
although not quite as high as those predicted in the 3-story building. Recall that the 6-
story building experienced very low inelastic demands at the base column hinges and that
the largest demands were in the middle stories. While the 20-story building has
significant higher mode effects, they tend to produce larger deformation demands in the
lower stories of the frame (as described in Section 5.6.2). At the both the 10/50 (Fig.
5.65) and 2/50 hazard levels (Fig. 5.66), the exterior column base hinges experience
much larger plastic rotation demands than the interior columns. The primary reason for
this is that during the large excursions where the building is being pushed in one
direction, the exterior columns experience large compressive and tensile forces in order
to resist the overturning moment. The tension force dramatically reduces the moment
capacity of the column thereby leading to very large plastic rotation demands. On the
opposite column, the compressive force increases the moment capacity of the columns,
thereby reducing the plastic rotation demand. Given that both of these columns are
considered together in the summary plots, this is also the reason for the large amount of
scatter in the exterior column plastic rotation demands. It should also be recognized that
the rotation capacity of the column would also change given the large change in axial
loads (e.g. increasing the axial load beyond the balance point of the column interaction
diagram would reduce its deformation capacity). This effect is rather interesting
considering that it could not be captured using nonlinear hinge models that must be
calibrated at a particular axial load.
306
The maximum plastic rotations in the composite beams follow the drift profiles seen in
Section 5.6.2, where larger demands occur in the lower four stories and between the 10th
and 16th floor. The median plastic rotations remain below 1% in the 10/50 hazard level
(Fig. 5.65) and 2% in the 2/50 hazard level (Fig. 5.66), both of which are well within the
rotation capacity of these composite beam hinges.
Joint deformations are approaching 1% distortion in the 9th through 14th floors at both the
10/50 and 2/50 hazard level. This is comparable to the demand in the joints for the 6-
story frame and again larger than the demands in the 3-story frame. Recall that the
maximum joint strength is expected to occur at approximately 2% joint distortion, which
again demonstrates the large relative strength of composite joints to the surrounding
members.
5.6.4 Damage Indices
Similar to the two previous frames, damage indices are used to interpret the plastic
rotation histories from each of the analyses. The monotonic plastic rotation capacities
(pu) for each of the components in the 20-floor frame are summarized in Table 5.18.
Similar to the interior columns of the 3-story frame, the predicted puvalues for the RC
columns tend be on the higher side with an average of approximately 0.15 radians. While
these are high, they are not unreasonable considering the work of Fardis et al. (Fardis et
al. 2003; Panagiotakos et al. 2001). Again, in future work, alternative definitions of RC
column capacities will be considered (see Section 5.4.2.5). Nevertheless, there is a very
limited amount of plastic rotations in the columns in this building, so this should not have
a large impact on the results. The general trends from the median, 16th and 84th percentile
summary plots for the 10/50 and 2/50 hazard levels are discussed herein.
Figure 5.67 and 5.68 shows the median summary plot for the damage indices in each of
the components of the frame at the 10/50 and 2/50 hazard level, respectively. These
figures follow similar trends that were found in the plastic rotation summary plots.
Damage in the upper story columns is predicted as negligible in both hazard levels. The
307
median response of the 1st story base column hinges fall within the moderate damage
state, with slightly more damage occurring in the exterior columns due to the large
changes in axial load and moment capacity. Despite attaining moderate levels of plastic
rotations (1-2%) at the 2/50 hazard level, the damage in these hinges remain relatively
low. This implies that the number of inelastic cycles that these hinges experience during
these events is relatively small.
These figures show that the beams experience the greatest amount of damage, which is
more concentrated in the 2nd through 5th floors and the 15th through 20th floors. At the
10/50 level, the median response remains just below the threshold of moderate damage.
At the 2/50 level, the level of predicted damage is increased, particularly in the 2nd, 3rd
and 19th floors, which have a 84th percentile response that falls in the significant damage
state (implying extensive hinge formation and a high probability of local buckling).
Under both hazard levels, the damage in the joints remains negligible.
5.7 Conclusions
5.7.1 Seismic Design
The case study buildings highlight the efficiency of design that is possible with
composite RCS frames. For all three case study buildings, the composite beam designs
are controlled by strength requirements and the RC columns according to the strong-
column weak-beam criterion. Drift is automatically satisfied in each of the frames, which
highlights the inherent stiffness that composite RCS systems possess over conventional
steel or concrete frames. Using fairly standard details, the composite joints in all three
case study buildings were able to satisfy the strong-joint weak-beam criterion. This is in
contrast to conventional RC frames where the column sizes are often controlled by the
beam-column joint design requirements. Alternatively, seismically designed steel frames
typically require the use of fully welded connections with web doubler plates.
308
5.7.2 General Seismic Performance
All frames exhibit excellent seismic performance up through the 2/50 hazard level. The
predicted damage is comparable to what was observed in the test frame described in
Chapter 3 and meets the spirit of the life-safety (10/50) and collapse-prevention (2/50)
performance states as defined by FEMA 356. Predicted plastic rotations in the beams are
well within the rotation capacities observed in the test frame and RCS subassembly tests.
Damage in the RC columns is within the limits of the damage states observed in the test
frame; moreover, the column deformation demands and damage appear to be less in the
taller buildings, as compared to the 3-story test frame.
5.7.3 Drift Criterion
Figure 5.69 plots the maximum interstory drifts from each frame at their corresponding
design level hazard (10/50). The IBC (2003) and ASCE-7 (2002) design drift limit of 2%
is also marked on this figure for comparison with the predicted inelastic drifts. As
mentioned earlier, the probability that the 3, 6, and 20-story buildings exceed this design
level drift is approximately 78, 32, and 67%, respectively. One could argue that the
intention of the drift criterion is to ensure that the median response falls approximately on
or below 2% interstory drift. Given the empirical nature of the structural modification
factors (R and Cd) that are used in the design process, it is interesting to see that the
inelastic deformations are actually not that far off the drift limit. In fact, if more
conservatism was implemented in the design of these frames, the medians of the 3 and
20-story frame may shift closer to the 2% drift limit.
An important point regarding these results is that the frames are all designed (for strength
and drift) according to a design base shear that is distributed up the height of the frame
according to the IBC equivalent static loading pattern. This loading pattern can vary
from linear to parabolic up the height of the structure, but ultimately, it cannot capture the
type of higher mode effects seen in this study for the 6 and 20-story buildings. This
309
highlights the inherent limitations of the equivalent static lateral load approach used in
design.
5.7.4 Composite Joint Performance
The composite joints experienced negligible amounts of damage in each of the frames.
In the 3-story frame, the joints remained relatively elastic up through the 2/50 hazard
level. While this is also true for the 6 and 20-story frames up to the 10/50 hazard level,
the joints in these frames begin to show some minor nonlinearity at the 2/50 level. This
implies that the joint demand in the taller buildings is higher than that in low-rise
buildings.
Based on comparisons to the composite strength of the beams, RC columns in the 3 and
6-story buildings were designed to the minimum limits of the strong-column weak-beam
(SCWB) criteria considering the composite beam strength. This design approach resulted
in some hinging in the upper story columns, but it did not lead to any significant story
mechanisms in the range of hazard levels investigated in the stripe analyses. This implies
that while the current SCWB ratio does not prevent damage in the upper story column
hinges, it does seem to provide sufficient protection from the formation of a story
mechanism at the 10/50 and 2/50 hazard levels. However, the uncertainties introduced
by variability in the measured material strengths, unaccounted changes in member sizes,
and other construction issues could potentially alter the column and beam strength
enough to alter this observation. In addition to this, the static pushover results revealed
that deformations in both the 3 and 6-story frames began to concentrate in a couple of
floors at higher demand levels, implying the likelihood of the formation of a story
mechanism at even higher intensity ground motions.
The 20-story building adopted a different (albeit inadvertent) design approach where the
columns were conservatively sized to provide, on average, 60% more strength than what
310
is required by the minimum limits of the traditional SCWB. Recall that this design still
does not meet the more stringent requirements of the proposed SEOAC Blue Book
provision, which essentially increases the SCWB ratio to 2.0. The predicted damage in
the upper story columns in this building is negligible, with the columns remaining
relatively elastic throughout all the considered hazard levels. These results imply that for
this case the SEOAC Blue Book provision may in fact be too conservative and that a
more reasonable SCWB ratio may be somewhere between the two cases considered in
this study. On the other hand, if the composite strength of these beams were ignored in
the SCWB calculations, the strength of the columns would be close to satisfying a
strength ratio of 2.0.
5.7.6 Damage Distribution
The 3-story building experienced damage that is consistent with a first-mode sway
motion, with damage concentrating in the 1st story column base hinges and the beam
hinges in each floor. This is also fairly representative of the damage seen in the test
frame response described in Chapter 3. There were some differences including some
predicted damage in the lower hinge of the of the 2nd story columns that was not seen in
the test. Recall that six of the fifteen 2/50 ground motions push this building beyond the
threshold of the fiber models, implying that local buckling would occur in these events.
This lack of strength deterioration in the 1st floor beams is likely the reason that damage
is predicted in these 2nd story lower column hinges. In addition, the predicted damage in
the 2nd floor beams was found to be generally larger than what was observed in the test.
The distribution of damage in the 6 and 20-story buildings are controlled by localization
of inelastic demands in the mid-region stories and associated higher mode effects. For
the 6-story frame, this results in much less demand on the 1st story column base hinges
and more damage in the middle story columns and beams. The 20-story frame has two
regions of higher amounts of damage; one in the lower stories (1st-4th floor) and the
second in the upper floors (10th-16th floors).
311
In the 20-story building, the drastic change in axial load in the exterior columns due to
the overturning moment proved to have a large impact on the plastic rotation demand and
damage that occurred in the exterior column hinges. This is something that was not seen
in the 3 and 6-story frames, presumably because the overturning moment is not as
significant as in taller buildings. This effect shows the importance of being able to
capture the interaction between moment capacity and axial loads. This can be accurately
modeled with elements using either fiber sections or P-M interaction yield surfaces.
Traditional moment versus rotation hinge models are not able to capture this effect.
The influence of higher modes in taller buildings and the resulting distribution of damage
provide some perspective to what was observed in the 3-story test frame:
1. Damage in the base column hinges for taller structures may not be as significant
as what was observed in the 3-story test.
2. It is much more difficult in taller structures to achieve an even distribution of
damage over all of the floors.
3. Axial loads due to overturning in the RC columns can lead to more damage in the
exterior columns. As most RC frame columns are designed below the balance
point, induced tension forces can dramatically reduce the bending resistance and
lead to more significant column hinging.
312
Table 5.1 Summary of design values for each of the case study buildings
3-Story RCS 6-Story RCS 20-Story RCS
4.41kPa 3.64kPa 4.12kPa
Floor
92psf 76psf 86psf
Dead Load 3.64kPa, (5.22kPa)
4.26kPa 3.21kPa
Roof 76psf, (109psf)
89psf 67psf
(w/penthouse)
2.40kPa 2.40kPa 2.40kPa
Live Load
50psf 50psf 50psf
2 2
585 kN-s /m 701 kN-s /m 686 kN-s2/m
Typical Floor Mass 2 2
40 k-s /ft 48 k-s /ft 47 k-s2/ft
585 kN-s2/m 556 kN-s2/m 628 kN-s2/m
Roof Mass 2 2
40 k-s /ft 38 k-s /ft 43 k-s2/ft
17,241 kN 39,836 kN 133,990 kN
Total Weight
3,876 kips 8,956 kips 30,120 kips
Vdesign/W (+torsion) 0.134 0.097 0.046
Calculated
1.0sec 1.4sec* 4.0sec
Natural Period, Tn
*Note that the earthquake hazard level and ground motion selection and scaling are based on a period of
1.5seconds, which represents a slightly older model. This difference is presumed to be negligible in the
final results.
Table 5.2 Member design schedule of 6-story case study building.

PERIMETER COLUMNS PERIMETER BEAMS
Floor
A1,B1,C1,D1,E1, A2,A3,A4,A5,A6 Frame Line Frame Line
# A1,F1,A7,F7
A7,B7,C7,D7,E7 F2,F3,F4,F5,F6 1&7 A&F
650x650 mm 650x750 mm 650x750 mm
12#32 bars 8#43,4#36 bars 8#32,4#29 bars W 690x125 W 610x101
1-2
25.6"x25.6" 25.6"x29.5" 25.6"x29.5" W 27x84 W 24x68
12#10 bars 8#14,4#11 bars 8#10,4#9 bars
650x650 mm 650x750 mm 650x750 mm
12#29 bars 8#36,4#32 bars 12#29 bars W 610x101 W 530x92
3-4
25.6"x25.6" 25.6"x29.5" 25.6"x29.5" W 24x68 W 21x62
12#9 bars 8#11,4#10 bars 12#9 bars
600x600 mm 600x700 mm 600x700 mm
12#25 bars 12#36 bars 12#25 bars W 530x92 W 460x89
5-6
23.6"x23.6" 23.6"x27.6" 23.6"x27.6" W 21x62 W 18x60
12#8 bars 12#11 bars 12#8 bars
Notes:
(1) Column reinforcement: Fy = 414MPa (60ksi)
(2) Column concrete: fc = 41.4MPa (6ksi)
(3) Slab concrete: fc = 27.6MPa (4ksi)
(4) Steel beams: Fy = 345MPa (50ksi)
313
Table 5.3 Member design schedule of 20-story case study building.
PERIMETER COLUMNS PERIMETER BEAMS
Floor # Frame
B1,C1,D1,E1; A2,A3,A4,A5,A6; Frame
A1,F1,A7,F7 Line A &
B7,C7,D7,E7 F2,F3,F4,F5,F6; Line 1 & 7
F
B2 1016x762mm
12#36 bars
B1
40"x30"
1 12#11 bars
W 760x147
2 762x762mm 1016x762mm W 30x99
12#43 bars 12#36 bars W
3
30"x30" 40"x30" 690x125
889x762mm
4 12#14 bars 12#11 bars W 27x84
12#32 bars
5 35"x30"
12#10 bars
6
7
8
W 760x134
9 W 30x90
889x762mm
10 12#36 bars 762x762mm
W
35"x30" 12#36 bars
11 762x762mm 610x113
12#11 bars 30"x30"
12#32 bars W 24x76
12 12#11 bars
30"x30"
13 12#10 bars
14 W 530x101
762x762mm W 21x68
15 12#32 bars W 530x92
30"x30" 762x762mm W 21x62
16 12#32 bars
12#10 bars
17 30"x30"
W 460x74
12#10 bars
18 762x762mm W 18x50
12#25 bars 762x762mm
19 30"x30" 12#25 bars 762x762mm W 460x52
12#8 bars 30"x30" 12#25 bars W 460x52 W 18x35
20 12#8 bars 30"x30" W 18x35
12#8 bars
Notes:
(1) Column reinforcement: Fy = 414MPa (60ksi)
(2) Column concrete: fc = 41.4MPa (6ksi)
(3) Slab concrete: fc = 27.6MPa (4ksi)
(4) Steel beams: Fy = 345MPa (50ksi)
314
Table 5.4 Summary of column and beam strengths with the corresponding SCWB ratios
for the 3-story perimeter frame. (units: kN,mm)
Mc,col Mp,beam
Mc,col/Mp,beam Mc,colFloor/Mp,beamFloor
(1x106) (1x106)
Floor
Traditional SCWB
Interior Ext (t) Ext (c) Comp. Steel SEAOC SCWB
Interior Ext (t) Ext (c)
1 1.76 1.12 1.33 1.91 1.28 1.09 0.89 1.59 0.62
2 1.72 0.59 0.70 1.65 1.11 1.23 0.72 1.2 0.57
1.67
3 0.60 0.63 0.84 0.49 0.9 0.71 1.3 0.91
1.20
for the 6-story perimeter frame (frame line 1 and 7). (units: kN,mm)
Mc,col Mp,beam
6 Mc,col/Mp,beam Mc,colFloor/Mp,beamFloor
(1x10 ) (1x106)
Floor
Traditional SCWB
1 2.01 0.86 1.39 2.19 1.64 1.05 0.81 1.66 0.54
2 2.01 0.91 1.33 2.19 1.64 0.91 0.75 1.34 0.54
3 1.48 0.74 0.88 1.67 1.19 1.03 0.91 1.48 0.53
4 1.48 0.78 0.88 1.67 1.19 0.93 0.91 1.45 0.53
5 1.19 0.75 0.84 1.37 0.97 1.01 1.1 1.7 0.54
6 1.18 0.77 0.80 1.37 0.97 0.5 0.56 0.83 0.54
315
for the 20-story perimeter frame (frame line A and F). (units: kN,mm)
Mc,col Mp,beam
Mc,col/Mp,beam Mc,colFloor/Mp,beamFloor
(1x106) (1x106)
Floor
Traditional SCWB
1 3.44 1.89 3.46 2.27 1.64 1.53 1.69 4.24 1.03
2 2.54 1.95 3.51 2.27 1.64 1.28 1.75 4.31 0.83
3 2.48 2.01 3.57 2.27 1.64 1.26 1.8 4.35 0.82
4 2.43 2.07 3.59 2.27 1.64 1.23 1.85 4.33 0.82
5 2.37 2.13 3.54 2.27 1.64 1.2 1.9 4.27 0.80
6 2.32 2.18 3.49 2.27 1.64 1.18 1.94 4.21 0.79
7 2.28 2.23 3.44 2.27 1.64 1.12 1.58 3.67 0.78
8 2.10 1.34 2.60 2.27 1.64 1.06 1.2 3.11 0.66
9 2.06 1.37 2.51 1.90 1.35 1.25 1.44 3.64 0.78
10 2.02 1.38 2.41 1.90 1.35 1.23 1.46 3.5 0.77
11 1.99 1.39 2.31 1.90 1.35 1.21 1.47 3.34 0.75
12 1.95 1.40 2.21 1.90 1.35 1.19 1.47 3.18 0.73
13 1.91 1.40 2.09 1.90 1.35 1.16 1.23 2.7 0.72
14 1.88 0.94 1.55 1.45 0.97 1.43 1.3 3.07 0.88
15 1.57 0.94 1.44 1.45 0.97 1.28 1.29 2.85 0.76
16 1.52 0.93 1.33 1.45 0.97 1.24 1.27 2.61 0.73
17 1.48 0.91 1.21 1.45 0.97 1.01 1.25 2.35 0.70
18 0.96 0.90 1.08 0.80 0.45 1.5 2.22 4.6 0.97
19 0.90 0.87 0.98 0.80 0.45 1.41 2.14 4.15 0.91
20 0.85 0.83 0.88 0.80 0.45 0.68 1.05 1.96 0.85
Table 5.7 Strength of composite joints and the strong-joint weak-beam ratios for the 3-
story case study frame. (units: kN,mm)
Mp,beamNom Mjoint,Nom
SJWB Ratios
(1x106) (1x106)
Floor
Exterior Exterior
Comp. Steel Interior Exterior Interior
Comp. Steel
1 1.77 1.16 2.95 2.05 1.00 1.16 1.76
2 1.53 1.01 2.46 1.72 0.97 1.12 1.70
3 0.79 0.44 1.68 1.12 1.36 1.42 2.53
316
Mp,beamNom Mjoint,Nom
6 SJWB Ratios
(1x10 ) (1x106)
Floor
Exterior Exterior
Comp. Steel
1 1.96 1.36 4.09 2.34 1.23 1.19 1.72
2 1.96 1.36 4.09 2.34 1.23 1.19 1.72
3 1.49 0.98 3.42 1.94 1.38 1.30 1.97
4 1.49 0.98 3.42 1.94 1.38 1.30 1.97
5 1.22 0.80 2.62 1.47 1.30 1.21 1.83
6 1.22 0.80 2.62 1.47 1.30 1.21 1.83
Mp,BeamNom Mjoint,Nom
SJWB Ratios
(1x106) (1x106)
Floor
Exterior Exterior
Comp. Steel
1 1.95 1.36 6.04 4.04 1.83 2.07 2.97
2 1.95 1.36 5.15 3.45 1.56 1.77 2.54
3 1.95 1.36 5.15 3.45 1.56 1.77 2.54
4 1.95 1.36 5.15 3.45 1.56 1.77 2.54
5 1.95 1.36 5.15 3.45 1.56 1.77 2.54
6 1.95 1.36 5.15 3.45 1.56 1.77 2.54
7 1.95 1.36 5.15 3.45 1.56 1.77 2.54
8 1.95 1.36 4.28 2.87 1.29 1.48 2.12
9 1.64 1.12 3.64 2.44 1.32 1.49 2.19
10 1.64 1.12 3.64 2.44 1.32 1.49 2.19
11 1.64 1.12 3.64 2.44 1.32 1.49 2.19
12 1.64 1.12 3.64 2.44 1.32 1.49 2.19
13 1.64 1.12 3.64 2.44 1.32 1.49 2.19
14 1.24 0.80 3.03 2.03 1.48 1.64 2.52
15 1.24 0.80 3.03 2.03 1.48 1.64 2.52
16 1.24 0.80 3.03 2.03 1.48 1.64 2.52
17 1.24 0.80 3.03 2.03 1.48 1.64 2.52
18 0.69 0.37 2.22 1.46 2.09 2.11 3.97
19 0.69 0.37 2.22 1.46 2.09 2.11 3.97
20 0.69 0.37 2.22 1.46 2.09 2.11 3.97
317
Table 5.10 Modal properties of 20-story frame and corresponding weights for record
scaling.
Mass Weighted
Mode i (rad/s) Ti (sec)
Participation Values
1 1.55 4.04 71.8% 75%
2 4.32 1.45 12.8% 15%
3 7.45 0.84 4.2% 10%
4 11.17 0.56 3.1% 0%
5 15.15 0.42 2.3% 0%
Table 5.11 Comparison of IDRMAX of two OpenSees models with test frame.
IDRMAX
OS: Difference
Earthquake Event Measured OS:
Lab in Models
Response Realistic Building
Conditions
50%in50yr
1999 ChiChi 1.91% 1.52% 1.35% 11%
S a (T1 ) =0.408g
10%in50yr-1a
1989 Loma Prieta 3.10% 3.18% 2.37% 25%
S a (T1 ) =0.68g
10%in50yr-1b
1989 Loma Prieta 2.67% 2.48% 1.98% 20%
S a (T1 ) =0.544g
2%in50yr
1999 ChiChi 5.78% 3.79% 2.93 % 23%
S a (T1 ) =0.92g
10%in50yr-2
1989 Loma Prieta 2.75% 2.60% 2.14% 18%
S a (T1 ) =0.544g
Table 5.12 Measured strengths of steel tension coupons.

Fy Percent Difference
Steel
(MPa) from Fy,exp
Flange 409 8%
1st Floor
Web 442 16%
Flange 484 28%
2nd Floor
Web 517 36%
Flange 407 7%
3rd Floor
Web 431 14%
#11 bars 527 6%
318
Table 5.13 Measured crushing strength of concrete cylinders.
Percent Difference from
Concrete f c' (MPa)
Nominal Strength
89.0 (lower) 115%
1st Floor Columns
70.8 (upper) 71%
2nd Floor Columns 68.2 65%
3rd Floor Columns 68.4 65%
Slab 31.0 12%
Table 5.14 Plastic rotation capacity of 3-story case study frame components.
pu (rad)
Floor
Inner 0.150
1st 0.072 0.087
Outer 0.306
Inner 0.161
2nd 0.093 0.093
Outer 0.415
Inner 0.162
3rd 0.098 0.102
Outer 0.418
Table 5.15 Correlation between the Mehanny damage index and the expected damage
in the component.
D Range Anticipated Damage State
Negligible: Little to no damage in element and corresponds
0.00-0.50
to immediate occupancy damage level
Moderate: Structural element experiences noticeable
damage, such as spalling of cover concrete and minor shear
cracking in RC columns, and hinging and some local buckles
0.50-0.70
in steel beams. In terms of the component damage and
necessary repairs, this region of the damage index roughly
corresponds to a life safety limit state.
Significant: Structural element is assumed to be at a near
collapse state, with extensive cracking and hinge formation
0.70-0.95
in RC columns and significant hinging and local buckles for
the steel beams.
Loss of Capacity: The damage is so extensive that the
>0.95
capacity of the element is assumed to be compromised.
319
Table 5.16 Probability of beam hinges being in a specific damage state given a 10/50
and 2/50 hazard level.
Pr( x | 10/50 hazard) Pr( x | 2/50 hazard)
x= x=
Beams
Moderate Significant Loss Moderate Significant Loss
0.5<DI<0.7 0.7<DI<0.95 DI<0.95 0.5<DI<0.7 0.7<DI<0.95 DI<0.95
1st
20% 6% 1% 32% 21% 11%
Floor
2nd
22% 3% 0.2% 49% 21% 3%
Floor
3rd
25% 7% 1% 44% 24% 5%
Floor
pu (rad)
Floor
Inner 0.089 0.070
1st, 2nd 0.052
Outer 0.088 0.077
rd th Inner 0.096 0.073
3 ,4 0.060
Outer 0.082 0.080
Inner 0.089 0.074
5th, 6th 0.094
Outer 0.079 0.081
320
pu (rad)
Floor RC Columns Comp. Joints
Steel Beams
Inner Outer Inner Outer
1st 0.126
0.079 0.089
2nd
3rd
4th 0.154
0.125
5th 0.054
0.075 0.086
6th
7th
8th
9th
0.070 0.081
10th
11th 0.145
12th 0.172 0.077
th
13 0.073 0.083
th
14
15th
16th
0.166 0.090
17th
0.076 0.087
18th
0.187
19th
0.188 0.077
20th 0.079 0.089
321
F
2 @ 6.40m (2 @ 21)
83mm Normal Weight
Concrete over 51mm Metal Deck
D
35.20m (115.5)
9.60m (31.5)
C
2 @ 6.40m (2 @ 21)
A
1 2 3 4 5 6 7
6 @ 6.40m = 38.40m (6 @ 21 = 126)
Figure 5.1 Typical floor plan of 6-story case study building.
F
63.5mm Normal Weight
Concrete over 76mm Metal Deck
E
5 @ 6.1m = 30.5m (5 @ 20 = 100)
Released End
A
1 2 3 4 5 6 7
6 @ 6.10m = 35.60m (6 @ 20 = 120)
Figure 5.2 Typical floor plan of 20-story case study building.
322
0.14
(T = 0.53 sec, V/W = 0.125)

0.12
(T = 0.82 sec, V/W = 0.111)
V/W - Base Shear Design Coefficient
0.1
0.08
0.06
(T = 2.91 sec, V/W = 0.044)

0.04
(T = 2.91 sec, V/W = 0.031)
0.02
IBC 2003/ASCE 7-02 Design Spectra

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Period (sec)
Figure 5.3 IBC 2003 design hazard spectra with the code-defined periods (1.2Ta) for
the 3, 6, and 20-story frames labeled on the curve.
323
#5-75mm o.c.
#4-75mm o.c. #4-75mm o.c.
600mm
600mm 700mm
Restraining #5-75mm o.c.
Bars #4-75mm o.c.
650mm
750mm
Figure 5.4 Schematic of typical transverse reinforcement in RC columns for 6-story
perimeter frame.
324
Restraining Restraining #5-100mm o.c.
#4-100mm o.c. Bars #4-100mm o.c.
Bars
762mm
762mm 889mm
Restraining
Bars #5-100mm o.c.
762mm
1016mm
Figure 5.5 Schematic of typical transverse reinforcement in RC columns for 20-story
perimeter frame.
325
889mm
Steel Erection
Column
(Optional)
762mm
Restraining
Bars
Column Section
(Below Beam)
889mm
A 38mm
B 13mm
85mm
W760x147
48mm
B A
Gravity Beam
Column Section
(Beam Level)
Figure 5.6 Typical plan view of the column section just below the beam and in the
beam-column joint. (section A-A and B-B in Figs. 5.7 and 5.8)
326
Band Plate
265mm
FBP
W760x147
Gravity Beam
Ties within
Joint
Holes in web
Section A-A from Fig. 5.6
Figure 5.7 Cross-section of typical beam-column joint depicting the location of the
joints ties.
762mm
Band Plate
16mm thick
FBP: 16mm thick

W760x147
Section B-B from Fig. 5.6
Figure 5.8 Cross-section of typical beam-column joint with band plate and face bearing
plate details.
327
EQ EQ
Intensity Intensity
IDRmax IDRmax
(a) (b)
EQ
Intensity
IDRmax
(c)
Figure 5.9 (a) The traditional cloud approach of nonlinear time history analyses and
two alternative concepts for scaling ground motions using (b) incremental scaling of
single ground motions and (c) the stripe analysis technique.
7
6.8
6.6
6.4
Mw
6.2
5.8
5.6
10 15 20 25 30 35 40 45 50 55 60 65
R (km)
Figure 5.10 Magnitude and distance to the rupture pairs for ground motion records used
in this study.
328
4
2
Epsilon (1sec)
-1
-2
R. Medina's EQ Database
-3
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
S a(1sec) (g)
Figure 5.11 Epsilon versus spectral acceleration at a period of 1 second for the 80
ground motions considered in this study.
2.5
1.5
1
Epsilon (1.5sec)
0.5
-0.5
-1
-1.5
-2
-2.5
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
S a(1.5sec) (g)
Figure 5.12 Epsilon versus spectral acceleration at a period of 1.5 seconds for the 80
329
2.5
1.5
1
Epsilon (4.04sec)
0.5
-0.5
-1
-1.5
-2
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
S a(4.04sec) (g)
Figure 5.13 Epsilon versus spectral acceleration at a period of 4 seconds for the 80
2
NR94glp - 0.75
1.8 S a,Target = 0.05g IV79e13 - 0.63
IV79pls - 0.76
1.6 BO42elc - 0.95
NR94bad - 0.59
1.4 IV79vct - 0.65
WN87cat - 1.04
WN87flo - 0.68
1.2
PS86h06 - 0.98
S a (g)
MH84cap - 0.67
1 CO83c08 - 0.58
WN87cts - 0.93
0.8 PS86ino - 0.63
WN87sse - 0.57
0.6 LV80stp - 0.55
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.14a Response spectrum for selected ground motions at 0.05g stripe hazard
level for the 3-story building.
330
2
IV79wsm - 0.94
1.8 S a,Target = 0.09g NR94nya - 1
LP89sjw - 1.06
1.6 IV79e01 - 0.87
LP89fms - 0.96
1.4 IV79cmp - 0.95
NR94cas - 0.89
MH84sjb - 1.04
1.2
NR94sor - 0.89
S a (g)
LV80srm - 0.99
1 IV79cc4 - 0.86
LV80kod - 0.96
0.8 WN87har - 1.03
LV80stp - 1.1
0.6 WN87stc - 1.09
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.14b Response spectrum for selected ground motions at 0.09g stripe hazard
2
SH87bra - 1.13
1.8 S a,Target = 0.2g IV79e12 - 1.27
SF71pel - 1.29
1.6 LP89hvr - 1.36
SH87wsm - 0.82
1.4 NR94fle - 1.26
NR94pic - 1.35
MH84g03 - 1.21
1.2
NR94php - 1.34
S a (g)
NR94hol - 0.86
1 PS86psa - 1.29
NR94jab - 1.29
0.8 NR94dwn - 1.3
NR94lh1 - 0.85
0.6 PM73phn - 0.91
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.14c Response spectrum for selected ground motions at 0.2g stripe hazard level
for the 3-story building.
331
2
LP89cap - 1.08
1.8 S a,Target = 0.3g NR94stc - 1.02
SH87icc - 0.97
1.6 LP89g04 - 0.88
LP89g03 - 0.79
1.4 NR94hol - 1.29
LP89svl - 1.16
NR94lh1 - 1.28
1.2
IV79chi - 1.07
S a (g)
IV79qkp - 0.89
1 NR94cen - 0.88
LP89a2e - 1.17
0.8 IV79dlt - 1.15
WN87cas - 0.77
0.6 WN87bir - 0.71
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.14d Response spectrum for selected ground motions at 0.3g stripe hazard level
2
NR94stc - 1.36
1.8 S a,Target = 0.4g SH87icc - 1.29
LP89g04 - 1.18
1.6 LP89g03 - 1.06
LP89svl - 1.54
1.4 NR94cnp - 0.8
IV79chi - 1.43
IV79qkp - 1.19
1.2
NR94cen - 1.17
S a (g)
IV79dlt - 1.54
1 LP89hch - 0.79
LP89hda - 0.73
0.8 LP89slc - 0.72
WN87cas - 1.03
0.6 WN87bir - 0.95
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.14e Response spectrum for selected ground motions at 0.4g stripe hazard level
332
2
LP89g04 - 1.47
1.8 S a,Target = 0.5g LP89g03 - 1.32
LP89svl - 1.93
1.6 NR94lh1 - 2.14
NR94cnp - 1
1.4 IV79chi - 1.79
IV79qkp - 1.49
NR94cen - 1.47
1.2
LP89a2e - 1.95
S a (g)
IV79dlt - 1.92
1 LP89hch - 0.99
LP89hda - 0.91
0.8 LP89slc - 0.9
WN87cas - 1.29
0.6 WN87bir - 1.19
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.14f Response spectrum for selected ground motions at 0. 5g stripe hazard level
2
SH87icc - 1.94
1.8 S a,Target = 0.6g LP89g04 - 1.77
LP89g03 - 1.59
1.6 LP89svl - 2.31
NR94cnp - 1.2
1.4 IV79chi - 2.14
IV79qkp - 1.79
NR94cen - 1.76
1.2
LP89a2e - 2.34
S a (g)
IV79dlt - 2.31
1 LP89hch - 1.19
LP89hda - 1.1
0.8 LP89slc - 1.09
WN87cas - 1.55
0.6 WN87bir - 1.42
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.14g Response spectrum for selected ground motions at 0.6g stripe hazard level
333
2
LV80kod - 0.82
1.8 S a,Target = 0.03g MH84gmr - 0.81
WN87cat - 1.41
1.6 WN87flo - 1.44
NR94glp - 1.48
1.4 NR94lv2 - 0.75
NR94sor - 0.79
NR94sse - 0.73
1.2
CO83c05 - 0.73
S a (g)
IV79vct - 0.93
1 MH84cap - 0.95
MH84sjb - 0.87
0.8 PS86h06 - 0.98
WN87cts - 1.24
0.6 WN87har - 1.18
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
2
MH84agw - 1.08
1.8 S a,Target = 0.06g MH84g02 - 0.9
WN87w70 - 1.03
1.6 WN87wat - 1.11
SF71pel - 0.97
1.4 SH87bra - 1
LP89hvr - 0.93
NR94bad - 1.15
1.2
NR94del - 0.85
S a (g)
NR94lh1 - 0.98
1 BO42elc - 0.86
IV79cmp - 1.01
0.8 IV79nil - 0.88
LV80stp - 0.92
0.6 PS86ino - 0.86
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
334
2
IV79e13 - 0.92
1.8 S a,Target = 0.1g PM73phn - 1.01
PS86psa - 1.14
1.6 WN87cas - 0.88
LP89agw - 0.94
1.4 NR94far - 1.18
NR94fle - 0.94
LP89fms - 1.04
1.2
LP89sjw - 1.17
S a (g)
NR94dwn - 0.9
1 NR94loa - 1.12
NR94php - 0.97
0.8 NR94pic - 1.15
NR94ver - 1.08
0.6 MH84hch - 0.98
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.15c Response spectrum for selected ground motions at 0.1g stripe hazard level
2
IV79chi - 1.01
1.8 S a,Target = 0.2g IV79qkp - 0.82
WN87cas - 1.77
1.6 LP89cap - 0.72
LP89hch - 0.81
1.4 LP89hda - 1.14
LP89svl - 0.91
NR94hol - 0.95
1.2
SH87pls - 0.87
S a (g)
BM68elc - 1.46
1 LP89a2e - 1.19
LP89slc - 0.85
0.8 NR94cen - 0.93
IV79dlt - 0.94
0.6 WN87bir - 1.62
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
335
2
IV79chi - 1.51
1.8 S a,Target = 0.3g IV79qkp - 1.23
LP89cap - 1.08
1.6 LP89g03 - 0.85
LP89g04 - 0.81
1.4 LP89hch - 1.21
LP89svl - 1.36
NR94hol - 1.43
1.2
NR94stc - 0.8
S a (g)
SH87icc - 1.03
1 SH87pls - 1.31
LP89a2e - 1.79
0.8 LP89slc - 1.28
NR94cen - 1.4
0.6 IV79dlt - 1.41
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.15e Response spectrum for selected ground motions at 0. 3g stripe hazard
2
IV79chi - 2.02
1.8 S a,Target = 0.4g IV79qkp - 1.65
LP89cap - 1.44
1.6 LP89g03 - 1.13
LP89g04 - 1.07
1.4 LP89hch - 1.62
LP89svl - 1.81
NR94cnp - 0.82
1.2
NR94hol - 1.9
S a (g)
NR94stc - 1.07
1 SH87icc - 1.37
SH87pls - 1.75
0.8 LP89slc - 1.71
NR94cen - 1.86
0.6 IV79dlt - 1.88
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.15f Response spectrum for selected ground motions at 0.4g stripe hazard level
336
T3 T2 T1
2
IV79e12 - Unscaled
1.8 IV79e12 - 4.14
IBC Hazard Curve
1.6
1.4
1.2
S a (g)
0.8
0.6
Scale factor = 4.14
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.16 Example of weighted average scaling technique.
T3 T2 T1
2
WN87bir - 0.47
1.8 S a,Target = 0.01g WN87cas - 0.55
NR94loa - 0.57
1.6 MH84g03 - 0.62
NR94del - 0.64
1.4 NR94sse - 0.69
SH87bra - 0.71
NR94far - 0.72
1.2
NR94jab - 0.78
S a (g)
IV79pls - 0.83
1 NR94nya - 0.87
NR94lh1 - 0.9
0.8 NR94ver - 0.92
PS86ino - 1.05
0.6 LV80stp - 1.32
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
337
T3 T2 T1
2
IV79chi - 0.49
1.8 S a,Target = 0.02g IV79wsm - 0.61
LP89sjw - 0.62
1.6 LP89fms - 0.62
IV79cal - 0.71
1.4 NR94fle - 0.87
IV79nil - 0.91
WN87bir - 0.95
1.2
WN87cas - 1.09
S a (g)
NR94loa - 1.13
1 MH84g03 - 1.25
NR94del - 1.27
0.8 NR94sse - 1.38
SH87bra - 1.42
0.6 NR94far - 1.44
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
T3 T2 T1
2
IV79cal - 1.59
1.8 S a,Target = 0.05g IV79chi - 1.1
IV79e01 - 1.34
1.6 IV79e12 - 1.16
IV79qkp - 0.85
1.4 IV79wsm - 1.38
LP89g03 - 0.75
LP89hda - 0.71
1.2
NR94cnp - 0.62
S a (g)
NR94stc - 0.88
1 SH87wsm - 0.73
LP89fms - 1.39
0.8 LP89sjw - 1.39
IV79dlt - 0.77
0.6 IV79nil - 2.04
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.17c Response spectrum for selected ground motions at 0.05g stripe hazard
338
T3 T2 T1
2
IV79chi - 2.19
1.8 S a,Target = 0.1g IV79e01 - 2.67
IV79e12 - 2.33
1.6 IV79qkp - 1.71
IV79wsm - 2.76
1.4 LP89g03 - 1.5
LP89hch - 1
LP89hda - 1.41
1.2
LP89svl - 1.18
S a (g)
NR94cnp - 1.23
1 NR94stc - 1.76
SH87icc - 1.09
0.8 SH87wsm - 1.45
LP89sjw - 2.78
0.6 IV79dlt - 1.54
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
T3 T2 T1
2
IV79chi - 3.9
1.8 S a,Target = 0.18g IV79e01 - 4.75
IV79e12 - 4.14
1.6 IV79qkp - 3.04
IV79wsm - 4.91
1.4 LP89g03 - 2.68
LP89hch - 1.78
LP89hda - 2.52
1.2
LP89svl - 2.09
S a (g)
NR94cnp - 2.19
1 NR94stc - 3.13
SH87icc - 1.94
0.8 SH87wsm - 2.59
LP89sjw - 4.95
0.6 IV79dlt - 2.75
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.17e Response spectrum for selected ground motions at 0.18g stripe hazard
339
T3 T2 T1
2
LP89hch - 2
1.8 S a,Target = 0.2g SH87icc - 2.18
LP89svl - 2.35
1.6 NR94cnp - 2.46
LP89hda - 2.83
1.4 SH87wsm - 2.91
LP89g03 - 3.01
IV79dlt - 3.09
1.2
IV79qkp - 3.41
S a (g)
NR94stc - 3.52
1 IV79chi - 4.39
IV79e12 - 4.65
0.8 IV79e01 - 5.34
IV79wsm - 5.52
0.6 LP89sjw - 5.57
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Period (s)
Figure 5.17f Response spectrum for selected ground motions at 0.2g stripe hazard level
1.2
3-Story RCS
Perimeter Frame
1
TCU082-2/50
0.8
LP89G04-10/50 1a
S a (g)
0.6
LP89G04-10/50 1b
0.4 TCU082-50/50
0.2
OS: Test Conditions
Test Results
0
0.01 0.02 0.03 0.04 0.05 0.06 0.07
IDRMAX
Figure 5.18 Plot of the measured and simulated maximum IDR from the first four
events of the pseudo-dynamic loading protocol.
340
1.2
3-Story RCS LP89hda
Perimeter Frame IDRmax=2.2%
1
TCU082-2/50
0.8
LP89G04-10/50 1a
S a (g)
0.6
LP89G04-10/50 1b LP89g03
IDRmax=6.4%
0.4 TCU082-50/50
0.2
OS: Test Conditions
Test Results
0
0.01 0.02 0.03 0.04 0.05 0.06 0.07
IDRMA X
Figure 5.19 Comparison of stripe analysis study and the measured and simulated drift
from the test frame.
341
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
0.2 Column Int. Lower Hinge 0.2 Column Int. Upper Hinge
Floor: 1 Floor: 2
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
(a) (b)
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
0.2 Composite Beams 0.2 Composite Beams

Floor: 1 Floor: 2
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
(c) (d)
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
0.2 Interior Joints 0.2 Interior Joints

Floor 1 Floor 2
0 0
0 0.01 0.02 0.03 0 0.01 0.02 0.03
Maximum Joint Rotation Maximum Joint Rotation
(e) (f)
Figure 5.20 Simulate IDA stripe response for selected columns (a,b), beams (c,d), and
joints (e,f) compared to the measured response from the frame test.
342
2.5
2
Design
1.5
/V
BaseShear
1
V
0.5
3-Story Perimeter Frame

0
0 0.005 0.01 0.015 0.02 0.025 0.03
Roof Drift Ratio
Figure 5.21 Static pushover curve for 3-story frame using IBC 2003 force distribution.
3
V
design
RDR = 0.01
2.5 RDR = 0.02
RDR = 0.029
2
Floor
1.5
0.5
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
IDR
Figure 5.22 IDR profile of 3-story frame during the pushover at the design base shear
and selected roof drift ratios.
343
1.2 1.2
1 1
**
0.8 0.8
*
Sa (g)
0.6 0.6
0.4 0.4
- Last Committed K
0.2 - Initial K 0.2
Non-converged Events
- Current K - Last Com. K: *3, **2 events
Zero Damping - Current K: *4, **6 events
0 0
0 1 2 3 4 0.7 0.8 0.9 1
IDRMAX Ratio of Medians to Zero Damping
Figure 5.23 Comparison of the median response of 3-story RCS frame with zero
damping and 2% damping based on initial and last committed stiffness matrix.
0.04
= 0.2126
0.035 = 0.0008554
0.03
, Percent Critical Damping
0.025
1 3
0.02
1,initial
0.015 2
0.01
0.005
0
0 10 20 30 40 50 60
, Frequency (rad/s)
Figure 5.24 Relationship between damping ratio and frequency for the 3-story RCS
frame as defined by the Rayleigh equation.
344
1.2
3-Story RCS
IBC 2%in50yr
Perimeter Frame
1 1.08g
0.8
IBC 10%in50yr
0.72g
S a (g)
0.6
0.4
0.2
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
IDRMA X
Figure 5.25 Stripe analysis plot of maximum interstory drift versus hazard level for the
3-story RCS frame.
345
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
IDR (S a =0.045g) IDR (S a =0.09g)
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
IDR (S a =0.2g) IDR (S a =0.3g)
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
IDR (S a =0.4g) IDR (S a =0.5g)
Figure 5.26 Drift profile of 3-story frame at the time of maximum drift during each
event scaled to the common hazard level labeled in the x-axis. (Sa=0.045g-0.5g)
346
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
-0.06 -0.03 0 0.03 0.06 -0.06 -0.03 0 0.03 0.06
IDR (S a =0.6g) IDR (S a =0.72g)
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
-0.06 -0.03 0 0.03 0.06 -0.06 -0.03 0 0.03 0.06
IDR (S a =0.8g) IDR (S a =0.9g)
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
-0.06 -0.03 0 0.03 0.06 -0.06 -0.03 0 0.03 0.06
IDR (S a =1.08g) IDR (S a =1.15g)
347
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
IDR (S a =0.045g) IDR (S a =0.09g)
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
IDR (S a =0.2g) IDR (S a =0.3g)
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
IDR (S a =0.4g) IDR (S a =0.5g)
Figure 5.28 Maximum drift at each floor of 3-story frame during each event in the
corresponding stripe level (bold lines: median, 16th, and 84th percentile).
348
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
0 0.02 0.04 0.06 0 0.02 0.04 0.06
IDR (S a =0.6g) IDR (S a =0.72g)
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
0 0.02 0.04 0.06 0 0.02 0.04 0.06
IDR (S a =0.8g) IDR (S a =0.9g)
3 3
2.5 2.5
2 2
Floor
Floor
1.5 1.5
1 1
0.5 0.5
0 0
0 0.02 0.04 0.06 0 0.02 0.04 0.06
IDR (S a =1.08g) IDR (S a =1.15g)
349
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
Floor: 1 Floor: 1
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
Floor: 2 Floor: 2
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
Floor: 3 Floor: 3
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
Figure 5.30 Relationship between the maximum plastic rotation in the interior columns
and the scaled spectral acceleration.
350
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
0.2 Column Ext. Lower Hinge 0.2 Column Ext. Upper Hinge
Floor: 1 Floor: 1
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
Floor: 2 Floor: 2
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
Floor: 3 Floor: 3
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
Figure 5.31 Relationship between the maximum plastic rotation in the exterior columns
and the scaled spectral acceleration.
351
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4

Floor: 1 Floor: 2
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
1.2
0.8
S a (g)
0.6
0.4
0.2 Composite Beams

Floor: 3
0
0 0.02 0.04 0.06 0.08
p (rad)
Figure 5.32 Relationship between the maximum plastic rotation in the beams and the
scaled spectral acceleration.
352
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
Composite Beams, +p Composite Beams, -p
0.2 0.2
Floor: 1 Floor: 1
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
0.2 0.2
Floor: 2 Floor: 2
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
0.2 0.2
Floor: 3 Floor: 3
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
p (rad) p (rad)
Figure 5.33 Relationship between the maximum positive (left column) and negative
(right column) plastic rotation in the beams and the scaled spectral acceleration.
353
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4

Floor 1 Floor 2
0 0
0 0.01 0.02 0.03 0 0.01 0.02 0.03
1.2
0.8
S a (g)
0.6
0.4
0.2 Interior Joints

Floor 3
0
0 0.01 0.02 0.03
Maximum Joint Rotation
Figure 5.34 Relationship between the maximum rotation for the interior joints and the
354
1.2 1.2
1 1
0.8 0.8
S a (g)
S a (g)
0.6 0.6
0.4 0.4
0.2 Exterior Joints 0.2 Exterior Joints

Floor 1 Floor 2
0 0
0 0.01 0.02 0.03 0 0.01 0.02 0.03
1.2
0.8
S a (g)
0.6
0.4
0.2 Exterior Joints

Floor 3
0
0 0.01 0.02 0.03
Maximum Joint Rotation
Figure 5.35 Relationship between the maximum rotation for the exterior joints and the
355
Int. Cols . Ext. Cols . Joints
Beam s
(Low/Upp) (Low/Upp) (Int/Ext)
3 3 3 3
2 2 2 2
Floor
1 1 1 1
S =0.72g
a
0 0.02 0.04 0 0.02 0.04 0 0.02 0.04 0 0.02 0.04

p (rad) p (rad) p (rad) total (rad)
Figure 5.36 Summary of the median and standard deviation of plastic rotations of 3-
story frame members at the 10%in50year level (Sa = 0.72g).

Beam s
3 3 3 3
2 2 2 2
Floor
1 1 1 1
S =1.08g
a
0 0.02 0.04 0 0.02 0.04 0 0.02 0.04 0 0.02 0.04

356
1.2 1.2
2/50 2/50
1 1
0.8 10/50 0.8 10/50

S a (g)
S a (g)
0.6 0.6
0.4 0.4
Floor: 1 Floor: 1
M S LC M S LC
0 0
0 0.5 1 0 0.5 1
DI DI
1.2 1.2
2/50 2/50
1 1
0.8 10/50 0.8 10/50

S a (g)
S a (g)
0.6 0.6
0.4 0.4
Floor: 2 Floor: 2
M S LC M S LC
0 0
0 0.5 1 0 0.5 1
DI DI
1.2 1.2
2/50 2/50
1 1
0.8 10/50 0.8 10/50

S a (g)
S a (g)
0.6 0.6
0.4 0.4
Floor: 3 Floor: 3
M S LC M S LC
0 0
0 0.5 1 0 0.5 1
DI DI
Figure 5.38 Relationship between the final value of the damage index for the interior
columns and the scaled spectral acceleration.
357
1.2 1.2
2/50 2/50
1 1
0.8 10/50 0.8 10/50

S a (g)
S a (g)
0.6 0.6
0.4 0.4
Floor: 1 Floor: 1
M S LC M S LC
0 0
0 0.5 1 0 0.5 1
DI DI
1.2 1.2
2/50 2/50
1 1
0.8 10/50 0.8 10/50

S a (g)
S a (g)
0.6 0.6
0.4 0.4
Floor: 2 Floor: 2
M S LC M S LC
0 0
0 0.5 1 0 0.5 1
DI DI
1.2 1.2
2/50 2/50
1 1
0.8 10/50 0.8 10/50

S a (g)
S a (g)
0.6 0.6
0.4 0.4
Floor: 3 Floor: 3
M S LC M S LC
0 0
0 0.5 1 0 0.5 1
DI DI
Figure 5.39 Relationship between the final value of the damage index for the exterior
columns and the scaled spectral acceleration.
358
1.2 1.2
2/50 2/50
1 1
0.8 10/50 0.8 10/50

S a (g)
S a (g)
0.6 0.6
0.4 0.4

Floor: 1 Floor: 2
M S LC M S LC
0 0
0 0.5 1 0 0.5 1
DI DI
1.2
2/50
1
0.8 10/50
S a (g)
0.6
0.4
0.2 Composite Beams

Floor: 3
M S LC
0
0 0.5 1
DI
Figure 5.40 Relationship between the final value of the damage index for the beams and
the scaled spectral acceleration.
359
1.2 1.2
2/50 2/50
1 1
0.8 10/50 0.8 10/50

S a (g)
S a (g)
0.6 0.6
0.4 0.4

Floor 1 Floor 2
M S LC M S LC
0 0
0 0.5 1 0 0.5 1
DI DI
1.2
2/50
1
0.8 10/50
S a (g)
0.6
0.4
0.2 Interior Joints

Floor 3
M S LC
0
0 0.5 1
DI
Figure 5.41 Relationship between the final value of the damage index for the interior
joints and the scaled spectral acceleration.
360
1.2 1.2
2/50 2/50
1 1
0.8 10/50 0.8 10/50

S a (g)
S a (g)
0.6 0.6
0.4 0.4
0.2 Exterior Joints 0.2 Exterior Joints

Floor 1 Floor 2
M S LC M S LC
0 0
0 0.5 1 0 0.5 1
DI DI
1.2
2/50
1
0.8 10/50
S a (g)
0.6
0.4
0.2 Exterior Joints

Floor 3
M S LC
0
0 0.5 1
DI
Figure 5.42 Relationship between the final value of the damage index for the exterior
joints and the scaled spectral acceleration.
361
Beam s
3 3 3 3
2 2 2 2
Floor
1 1 1 1
S =0.72g
a M S LC M S LC M S LC M S LC
0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0
DI DI DI DI
Figure 5.43 Summary of the median and standard deviation of damage indices of
frame members in 3-story frame at the 10%in50year level (Sa = 0.72g).

Beam s
3 3 3 3
2 2 2 2
Floor
1 1 1 1
S =1.08g
0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0
DI DI DI DI
Figure 5.44 Summary of the median and standard deviation of damage indices of
frame members in 3-story frame at the 2%in50year level (Sa = 1.08g).
362
0.04 6
= 0.16129
0.035 = 0.00136
5
0.03
4
0.025
1 3
0.02
2
0.015
0.01
0.005
0
0 10 20 30 40 50 60
, Frequency (rad/s)
2.5
2
Design
/V
1.5
BaseShear
1
V
0.5

0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Roof Drift Ratio
363
6
V
design
RDR = 0.01
5 RDR = 0.02
RDR = 0.03
4
Floor
0
0 0.01 0.02 0.03 0.04 0.05 0.06
IDR
0.9
6-Story RCS Perimeter Frame
0.8
0.7 IBC 2%in50yr

0.72g
0.6
0.5
IBC 10%in50yr
S a (g)
0.48g
0.4
0.3
0.2
0.1
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
IDRMA X
6-story RCS frame.
364
6 6
5 5
4 4
Floor
Floor
3 3
2 2
1 1
0 0
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
IDR (S a =0.03g) IDR (S a =0.06g)
6 6
5 5
4 4
Floor
Floor
3 3
2 2
1 1
0 0
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
IDR (S a =0.1g) IDR (S a =0.2g)
6 6
5 5
4 4
Floor
Floor
3 3
2 2
1 1
0 0
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
IDR (S a =0.3g) IDR (S a =0.4g)
365
6 6
5 5
4 4
Floor
Floor
3 3
2 2
1 1
0 0
-0.1 -0.05 0 0.05 0.1 -0.1 -0.05 0 0.05 0.1
IDR (S a =0.48g) IDR (S a =0.6g)
6 6
5 5
4 4
Floor
Floor
3 3
2 2
1 1
0 0
-0.1 -0.05 0 0.05 0.1 -0.1 -0.05 0 0.05 0.1
IDR (S a =0.72g) IDR (S a =0.8g)
366
6 6
5 5
4 4
Floor
Floor
3 3
2 2
1 1
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
IDR (S a =0.03g) IDR (S a =0.06g)
6 6
5 5
4 4
Floor
Floor
3 3
2 2
1 1
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
IDR (S a =0.1g) IDR (S a =0.2g)
6 6
5 5
4 4
Floor
Floor
3 3
2 2
1 1
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
IDR (S a =0.3g) IDR (S a =0.4g)
367
6 6
5 5
4 4
Floor
Floor
3 3
2 2
1 1
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
IDR (S a =0.48g) IDR (S a =0.6g)
6 6
5 5
4 4
Floor
Floor
3 3
2 2
1 1
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
IDR (S a =0.72g) IDR (S a =0.8g)
368
Beam s
6 6 6 6
5 5 5 5
4 4 4 4
Floor
3 3 3 3
2 2 2 2
1 1 1 1
S =0.48g
a
0 0.02 0.04 0 0.02 0.04 0 0.02 0.04 0 0.02 0.04


Beam s
6 6 6 6
5 5 5 5
4 4 4 4
Floor
3 3 3 3
2 2 2 2
1 1 1 1
S =0.72g
a
0 0.02 0.04 0 0.02 0.04 0 0.02 0.04 0 0.02 0.04

369
Beam s
6 6 6 6
5 5 5 5
4 4 4 4
Floor
3 3 3 3
2 2 2 2
1 1 1 1
S =0.48g
0 0.50 1.0 0 0.50 1.0 0 0.50 1.0 0 0.50 1.0
DI DI DI DI
Figure 5.55 Summary of the median and standard deviation of damage indices of 6-

Beam s
6 6 6 6
5 5 5 5
4 4 4 4
Floor
3 3 3 3
2 2 2 2
1 1 1 1
S =0.72g
0 0.50 1.0 0 0.50 1.0 0 0.50 1.0 0 0.50 1.0
DI DI DI DI
370
0.04
= 0.04889
0.035 = 0.00288
0.03
0.025 5
0.02 1 4
0.015 3
2
0.01
0.005
0
0 5 10 15 20 25
, Frequency (rad/s)
1.8
1.6
1.4
Design
1.2
/V
1
BaseShear
0.8
V
0.6
0.4
0.2
0
0 0.005 0.01 0.015 0.02 0.025 0.03
Roof Drift Ratio
371
20
V
design
18 RDR = 0.01
RDR = 0.02
16 RDR = 0.025
14
12
Floor
10
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
IDR
0.35
20-Story RCS Perimeter Frame
0.3
IBC 2%in50yr
0.25 0.27g
0.2
S a (g)
IBC 10%in50yr
0.18g
0.15
0.1
0.05
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
IDRMA X
20-story RCS frame.
372
20 20
Floor 15 15
Floor
10 10
5 5
0 0
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
IDR (S a =0.011125g) IDR (S a =0.02225g)
20 20
15 15
Floor
Floor
10 10
5 5
0 0
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
IDR (S a =0.05g) IDR (S a =0.1g)
20 20
15 15
Floor
Floor
10 10
5 5
0 0
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
IDR (S a =0.178g) IDR (S a =0.2g)
373
20 20
Floor 15 15
10 10
5 5
0 0
-0.05 -0.025 0 0.025 0.05 -0.05 -0.025 0 0.025 0.05
IDR (S a =0.267g) IDR (S a =0.3g)
374
20 20
Floor 15 15
Floor
10 10
5 5
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
IDR (S a =0.011125g) IDR (S a =0.02225g)
20 20
15 15
Floor
Floor
10 10
5 5
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
IDR (S a =0.05g) IDR (S a =0.1g)
20 20
15 15
Floor
Floor
10 10
5 5
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
IDR (S a =0.178g) IDR (S a =0.2g)
375
20 20
Floor 15 15
Floor
10 10
5 5
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
IDR (S a =0.267g) IDR (S a =0.3g)
376
22 22 22 22
Beam s
20 20 20 20
18 18 18 18
16 16 16 16
14 14 14 14
12 12 12 12
Floor
10 10 10 10
8 8 8 8
6 6 6 6
4 4 4 4
2 2 2 2
S =0.178g
a
0 0 0 0
0 0.02 0.04 0 0.02 0.04 0 0.02 0.04 0 0.02 0.04
377
22 22 22 22
Beam s
20 20 20 20
18 18 18 18
16 16 16 16
14 14 14 14
12 12 12 12
Floor
10 10 10 10
8 8 8 8
6 6 6 6
4 4 4 4
2 2 2 2
S =0.267g
a
0 0 0 0
0 0.02 0.04 0 0.02 0.04 0 0.02 0.04 0 0.02 0.04
378
22 22 22 22
Int. Cols.
Ext. Cols. Joints
(Low/Upp) Beams
S =0.178g (Low/Upp) (Int/Ext)
a
20 20 20 20
18 18 18 18
16 16 16 16
14 14 14 14
12 12 12 12
Floor
10 10 10 10
8 8 8 8
6 6 6 6
4 4 4 4
2 2 2 2
M S LC M S LC M S LC M S LC
0 0 0 0
0 0.50 1.0 0 0.50 1.0 0 0.50 1.0 0 0.50 1.0
DI DI DI DI
379
22 22 22 22
Int. Cols.
Ext. Cols. Joints
(Low/Upp) Beams
S =0.267g (Low/Upp) (Int/Ext)
a
20 20 20 20
18 18 18 18
16 16 16 16
14 14 14 14
12 12 12 12
Floor
10 10 10 10
8 8 8 8
6 6 6 6
4 4 4 4
2 2 2 2
M S LC M S LC M S LC M S LC
0 0 0 0
0 0.50 1.0 0 0.50 1.0 0 0.50 1.0 0 0.50 1.0
DI DI DI DI
380
10% in 50 year Hazard Level IDRMAX |EQ Record
Median,16th ,84th perc.
IBC/ASCE7 Drift Limit
20-story
P(IDRMAX0.02) = 67%
6-story
P(IDRMAX0.02) = 32%
3-story
P(IDRMAX0.02) = 78%
0.01 0.015 0.02 0.025 0.03 0.035

IDRMAX
Figure 5.69 Maximum interstory drift response for each case study building at the
10/50 hazard level.
381
Chapter 6: Conclusions
The main objective of this research is to assess the seismic performance of composite
RCS frames and, thereby, validate provisions for the seismic design of these systems. An
underlying goal is to fill the knowledge gaps in design and construction of composite
RCS systems and to facilitate their acceptance as a viable alternative to traditional steel or
concrete construction. This research employs complementary full-scale experimental
testing and comprehensive analytical studies to evaluate seismic design criteria and the
overall system response. In addition to providing insight into composite RCS systems,
this work has broader implications toward the development of a general methodology and
enabling tools for performance based earthquake engineering.
In this chapter, a summary of the work done throughout this research highlighting the
main contributions is presented, followed by conclusions and recommendations.
Suggestions for future research are also included.
6.1 Summary
Design Provisions: A detailed review and interpretation of the seismic design criteria
for composite RCS frames is provided in Chapter 2. Given that these systems combine
reinforced concrete (RC) columns and steel (or composite) beams, the design criteria for
these moment frames are compiled from several different sources. In addition, a
literature search on the latest research on the design and performance of RC columns,
composite beams, and composite joints is compiled. Final recommendations for the
design of composite RCS frames are proposed and later exercised and validated in the
design of a full-scale test (Chapter 3) and subsequent case-study building designs
(Chapter 5).
Updated Joint Design Model: A proposed update to the 1994 ASCE beam-column joint
design guidelines (ASCE 2004) is developed, which incorporates information from
several of the latest studies on composite joints as well as results from the full-scale
382
testing program (Chapter 3). These updated guidelines extend the 1994 ASCE model to
include a wider variety of joints details, reduce the requirements for transverse ties within
the joint height, differentiate the strength of interior and exterior joints, and incorporate
performance-based requirements to limit deformation and demand in joints. Based on
observations from subassembly tests, the updated model identifies the strength of the
inner and outer panel separately, considering both shear and vertical bearing deformation
mechanisms. Using a database of RCS composite joint tests, predicted strengths from
both of these models are compared to the measured values from subassembly tests. The
updated guidelines are implemented in the design of the three case-study buildings and
are used to define the backbone of the rotational springs in the analytical models of the
composite joints.
Full-Scale Testing: A full-scale 3-story composite RCS frame is designed, constructed,

and tested in collaboration with researchers at the National Center for Research on
Earthquake Engineering in Taipei, Taiwan. Designed with the intention of pushing the
minimum limits of the recommendations presented in Chapter 2, the test frame is pseudo-
dynamically subjected to a series of four earthquake ground motions, ranging in
intensities from what is considered a frequent event (50%in50year) up through a rare
event (2%in50year probability of exceedance). Peak transient drifts during the design
level and maximum considered earthquake loading events reached as high as 3.0% and
5.5%, respectively. Upon completing the pseudo-dynamic tests, the frame is then quasi-
statically loaded to interstory drift ratios as high as 10%. Global and local behavior
results are examined for each test and performance and design implications are discussed.
Results of the frame test provide the basis for assessing the effectiveness of design
recommendations from Chapter 2 and to evaluate the validity of analytical models from
Chapter 4. In addition, differences in observed behavior between connection
subassembly tests and the frame test provide evidence that the subassembly tests tend to
exaggerate the amount of damage that will occur in real buildings subjected to large
earthquakes.
383
Analytical Modeling Recommendations: Modeling recommendations are provided to
accurately simulate to the performance of composite RCS frames within the structural
analysis software, OpenSees. Using fiber beam-column elements, RC columns and
composite beams are calibrated against a series subassembly tests and recommendations
for material models are presented, considering the effects such as confined and
unconfined concrete and effective stress and width of composite slabs. Composite joints
are represented by a finite joint model with two nonlinear springs to simulate the vertical
bearing and panel shear deformation mechanisms. Using the updated strength model in
Chapter 2 to define the backbone of the joint behavior, the simulation models are
calibrated against a series of subassembly tests. General recommendations on improving
convergence are also provided.
Damage Indices: A damage index developed by Mehanny et al. (2001) is used to

process the plastic rotation histories from the analytical models and the measured results
to obtain information on the predicted damage states and necessary levels of repair.
These predictions are compared to the observed damage in the test frame, which provides
implications on the validity of the damage model and also how well the analytical models
capture the distribution and extent of the damage observed in the test frame.
Case Study Building Design and Performance: The different aspects of this study
regarding design, testing, and analysis are brought together and applied in the seismic
design and performance assessment of three case study buildings. Standing 3, 6, and 20
stories tall, these buildings are designed with the intent to examine the key design aspects
of these frames as well as to assess the performance of RCS systems in the range of
heights that they are expected to be competitive with other structural systems. The
building designs highlight the efficiency of composite RCS systems to concurrently
satisfy strength, stiffness, and strong-column and strong-joint weak-beam requirements.
These frames are simulated in OpenSees using static pushover and nonlinear time history
analyses under multiple hazard levels up through the maximum considered event (2% in
50 year). The results of these simulations help provide insight and perspective into the
384
performance of the test frame as to inform conclusions regarding the design
recommendations and general performance of composite RCS frames.
6.2 Major Findings and Conclusions
The main finding and general conclusions from this work are summarized in the
following sections.
6.2.1 Seismic Performance of Composite RCS Frames
Designed to interrogate the minimum limits of current building code requirements, the
full-scale test frame exhibited excellent seismic performance up through the maximum
considered earthquake level. The damage patterns after each pseudo-dynamic earthquake
event were representative of the performance expected in moment resisting frames
designed by current building codes. The columns, beams, and joints all performed in a
ductile manner and did not experience any sudden or unexpected failures. The fact that
the test frame performed well through four major earthquake loadings and still
maintained most of its strength (2.8 times the design base shear) through large ductilities
during the final pushover test demonstrates the robustness of RCS moment frames. The
composite joints maintained their strength and stiffness throughout the entire loading
protocol and experienced very limited damage. This sort of behavior is expected by these
composite joints given their inherent strength compared to the surrounding beams and
columns.
The performance assessment of the case study buildings reinforce the observations from
the test frame, with all frames demonstrating excellent seismic performance up through
the 2/50 hazard level. Predicted median plastic rotations and damage indices are within
the limits observed in the test and meet the spirit of the life-safety (10/50) and collapse-
prevention (2/50) performance states as defined by FEMA 356.
385
6.2.2 Performance of Precast Splices
The grouted RC column splices and the bolted steel beam splices exhibited adequate
strength and stiffness to provide equivalent behavior to cast-in-place concrete or welded
steel systems throughout all of the pseudo-dynamic loading events. The flange plates on
the 1st floor steel beam did experience ductile rupture in the final stages of the quasi-static
pushover test (at a drift ratio of 10%); however, this behavior is considered as acceptable
under the design intent. Moreover, with little additional cost, the splice could be
designed to either eliminate or further postpone this failure mode.
6.2.3 Structural Period Elongation
The stiffness deterioration of the frame resulted in a significant amount of period

elongation during each of the pseudo-dynamic earthquakes. After the first earthquake
loading (corresponding to the 50% in 50 year event), the effective stiffness of the frame
reduced to 60% of its initial value, as indicated by a shift in period from 1.0 second to 1.3
seconds. After the design level (10% in 50 year) event, the period further lengthened to
1.5 seconds, indicating that the stiffness had reduced to 45% of its original value. After
the maximum considered event (2% in 50 year), the period of the frame was 1.7 seconds
or 35% of its original stiffness. This elongation of the period was shown to have a large
influence in changing the spectral demand on frame under subsequent earthquake events.
This type of behavior has implications in both scaling and selection of ground motions
for future testing and also reinforces the idea of including spectral shape information into
the intensity measures anticipating the elongation of the period (Cordova et al. 2001,
Baker and Cornell, 2005).
6.2.4 SCWB
Despite satisfying the strong-column weak-beam criteria, a two-story mechanism

between the 1st and 2nd-floor of the test frame began to develop during the maximum
considered event and became even more pronounced in the final static pushover. Based
386
on this response, the current SCWB criteria appear unable to prevent this mechanism
from occurring in the test frame. An alternative SCWB provision proposed for the
SEAOC Blue Book (Maffei et al. 2004) shows promise in that it was able to identify the
weakness of the columns in the 2nd-floor. However, the frame studies also suggested that
the proposed SEAOC provisions are probably more conservative than necessary to
provide reasonably good performance.
6.2.5 Top Floor Joints
The proposed reinforcing bar plate detail for the roof beam-column joint of a composite
RCS frame has been shown in both subassembly tests and the full-scale testing program
to possess adequate strength and stiffness to force hinging to occur in the surrounding
beams. This joint performed well in both tests and is recommended as a practical detail
for use in top floor (roof) joints.
6.2.6 IBC 2003/ASCE 7-2002 Drift Criterion
Despite meeting the stiffness criterion set by the IBC 2003 and ASCE 7 (2002), which
limits the interstory drift ratios under the design earthquake loads to 2%, the test frame
reached drift ratios of 3% during the design level event. In the case study buildings, the
3, 6, and 20-story frame experience a median drift at the design level earthquake of
approximately 2.5, 1.9, and 2.2%, respectively. If one interprets the drift criterion to
imply that the median response of the building falls approximately on or below 2%
interstory drift, then these results show that the case study buildings, which pushed the
minimum limits of the design, are relatively close to meeting this expectation.
6.2.7 Full-scale System versus Subassembly Test Behavior
The boundary conditions enforced in typical beam-column subassembly tests exaggerate

the amount of damage and deterioration that occurs in the hinges as compared to the
response of full-scale structural systems. The severity of steel beam flange and web local
387
buckling was limited in the frame test by the continuity of the steel beam and the
continuous top flange support of the composite slab. Local buckling of the steel beam
tends to be exaggerated in subassembly tests where the steel beam is allowed to
physically shorten, leading to an accordion effect where local buckles build up over
cycles.
Composite action of the slab was maintained throughout the entire loading protocol with
no occurrence of shear stud fracture in the test frame. This can be attributed to (1) the
continuity of the slab and beams and (2) the realistic introduction of load through the
floor system that alleviates the tension stresses in the slab due to flexural bending. The
typical beam-column subassembly setup is not able to capture this effect and tends to
induce higher stresses in the slab and impose excessive slip on the shear studs than what
may be present in real buildings. This effect leads to the prediction of an excessive
amount of damage in slab and therefore more strength and stiffness degradation in the
beam.
6.2.8 Validity of Fiber Beam-Column Models
The fiber beam-column elements used to model the RC columns and composite steel
beams were shown to capture the test frame behavior quite well up through
approximately 3% drifts. As the frame was pushed into larger excursions during the 2%
in 50 year event, the analytical models could not accurately capture the response where
local buckling in the hinge zone of the steel beams began to dominate the frame behavior.
Given the fundamental principals behind the fiber element models (i.e., plane sections
remain plane and uniaxial material behavior), one must recognize that there are some
limitations to the type of behavior that they can accurately model. While these models
can accurately capture the response of flexural hinging, problems occur as the hinges are
pushed to large plastic rotations when their behavior begins to deviate from the plane
sections remain plane assumption with the occurrence of local buckling in the beam
flange, followed by web buckling in the beam and deterioration of the concrete, bond
slip, and possibly rebar buckling in the RC column. Therefore, the validity of the
388
standard beam-column fiber models is questionable as the drift exceeds values of 4%,
which implies that they are inaccurate in simulating excessively large drifts or collapse.
6.2.9 Mehanny Damage Index
A damage index proposed by Mehanny et al. (2001) was validated using interpreted
plastic rotations from the hinge response measured in the frame. These results showed
that the index correlates well with the observed damage in the test frame. While there are
some differences between predicted and observed damage in some of the lightly damaged
hinges, these can be attributed to the estimation process used to interpret the plastic
rotations in these components.
The fiber beam-column models combined with the Mehanny damage index accurately
predicted the regions of the test frame that experienced severe damage over the duration
of the pseudo-dynamic loading events. However, there were instances where the model
tended to overestimate the amount of damage compared to what was observed in the test
frame. This is likely due to the inability of the fiber elements to capture the softening
behavior (i.e. local buckling) that would lead to damage concentration in highly loaded
regions, which will shield damage in other regions. What this shows is that while these
damage models are useful to detect levels of damage, their accuracy is limited by the
accuracy of the fiber beam-column element.
6.2.10 Perspective on the Performance of the Test Frame
The full-scale test presented in Chapter 3 provided a unique opportunity to evaluate the
performance of a composite RCS frame under a series of earthquakes representing a
range of hazard levels. It is important to recognize that this is not the definitive estimate
of performance, but rather, only one instance within a larger distribution of possible
random earthquake ground motions. Subsequent analytical simulations of the test frame
(Chapter 5) emphasize the inherent variability in the structural response when subjected
to different ground motions. These results demonstrate that the response of the frame
389
could have been quite different than what was observed in the test. More specifically,
given a sample of 15 ground motions representative of the maximum considered
earthquake hazard, the response of the test frame could have ranged from relatively minor
damage (IDRMAX 2%), up to, conceivably, global collapse of the frame.
6.3 Design and Analytical Modeling Recommendations
The full-scale testing program and the complementary analytical studies provided the
opportunity to investigate several key design features of current building codes. Based
on the results from these studies, the recommendations in the following sections are
made.
The case study building simulations reinforce the results observed in the test frame
(Section 6.2.4), showing that when the SCWB design was pushed to its minimum limits,
damage in the upper floor columns occurred during the more intense loading events
(around the 2/50 hazard level). Despite incurring some damage, there were no signs of
pronounced story mechanism in the time history analyses at the hazard levels
investigated. The static pushover, on the other hand, was able to pick up some of the
weakness of the 3 and 6-story frame, with deformation demands beginning to localize in
just a couple of the stories as the frame was pushed to higher drifts. This suggests that at
higher intensity events, a story mechanism may develop and begin to dominate the
response of these frames. This implies that while the current SCWB ratio does not
prevent damage in the upper story column hinges, it does seem to provide a sufficient
amount protection from the formation of a story mechanism at the 10/50 and 2/50 hazard
levels. However, the uncertainties introduced during the design and construction stages
(i.e. measured versus expected strengths, unaccounted changes in member sizes, etc.)
could potentially shift the balance of column and beam strengths and lead to excessive
column hinging in the upper stories.
390
The more conservatively designed 20-story frame, which was approximately 60%
stronger than the minimum SCWB limit, did not experience any sort of damage in the
upper floor columns. Despite this good behavior, the columns are still 20% under-
designed based on the proposed SEAOC provisions, implying that this alternative
approach is overly conservative for this building.
6.3.2 Bolted Beam Splice Design
Based on the performance of the bolted beam splices in the test frame, it is recommended
that the location of the edge of the splice be at least two times the depth of the steel beam
away from the column face to avoid interaction of the splice and the hinging zone of the
beam. This beam splice can be designed as a simple bolted connection with flange plates
and a shear tab designed with a strength to develop the expected plastic moment of the
steel beam ( 1.1Ry M p ), which is described in detail in Chapter 2. While the bolts in this
splice can be designed for bearing resistance, it should be recognized that there is a high
likelihood of bolt slip even in very frequent earthquake events (50% in 5 year event).
While the occurrence of slip is not detrimental to the performance of the splice or the
frame, it does produce a very loud and sharp noise which has been referred to as bolt
banging. With multiple splices throughout the building, this phenomenon can prove to
be a frightening experience for the building occupants and would likely lead to required
post-earthquake inspection. To postpone this phenomenon, the bolts could be designed
as slip critical according to the AISC-LRFD (2002) using the slip critical force of a bolt
(Equation J3.1,1.13Tb N s ). In the test frame, this would have required approximately
50% more bolts than a typical bearing design.
6.3.3 Column Grouted Splice Design
Precast RC columns are spliced using grouted connections, which should be designed to
develop the full plastic moment of the section. It is recommended that the column splices
be located within the middle third of the column length for both the structural integrity of
the hinge and ease of construction. Placement of the splice within the column hinge zone
391
was investigated and found to provide sufficient strength but with more pronounced
stiffness and strength degradation at larger drifts.
6.3.4 Composite beams
Based on the results in the test frame (Chapter 3) and the calibration studies (Chapter 4),
it is recommended that the composite strength and stiffness of the beam be considered in
both in the design (i.e. strength, stiffness, and SCWB) and the analytical modeling stages.
It was shown that the plastic strength of a beam could easily be 30-40% stronger than a
bare steel section when considering the composite strength of the slab. Not accounting
for this strength could shift the hinging from the beams into the columns, leading to an
undesirable story mechanism. The stiffness of the composite beams can be handled
directly with the fiber beam-column element or by taking an average stiffness of the steel
and composite beam assuming that the member is in double curvature. This effect can
increase the stiffness of the bare steel beams between 50 to 100%. It is also
recommended that beams are designed as fully composite beams according to the AISC-
LRFD (2002) and shear studs are designed according to the AISC Seismic Provisions
(2002). For plastic strength purposes, the effective slab width can be taken as equal to
the width of the column with an effective stress of 1.3 f c' .
6.3.5 Updated Joint Guidelines
Updated joint design guidelines (Chapter 2) are recommended for use in place of the
earlier 1994 ASCE guidelines. The updated model increased the accuracy of the original
joint bearing and shear strength models from a mean predicted-to-measured value of 0.76
and 0.80 to 0.92 and 0.96, respectively. This improvement is also reflected in the
consistency of the strength predictions as the coefficient of variation on these joint
bearing and shear mean values have decreased from the original (15% and 16%) to the
updated model (8% and 14%). In addition to the strength models, the resistance factors
(phi-factors) were also re-evaluated by processing the results of the calibration study with
392
the beta-reliability method (Ravindra and Galambos, 1978) and are recommended as 0.75
and 0.85 for the joint bearing and shear strengths, respectively.
6.3.6 Bond-Slip in RC Columns
A simple elastic spring was proposed in Chapter 4 to model the bond slip deformations
that occur in reinforced concrete columns. Bond slip flexibility is not included in the
effective RC column stiffness calculations proposed in Equation 2.5. This effect was
shown to decrease the elastic stiffness of the frame by approximately 10%. It is
recommended that these springs are incorporated into analytical models of moment
frames with RC columns to account for the flexibility that bond slip adds to the response.
6.3.7 Analytical Modeling of Composite RCS Frames
This study has utilized the fiber beam-column elements and a 2-dimensional joint model
from the OpenSees simulation platform. A succinct set of modeling guidelines have been
proposed in Chapter 4 for these elements, which have been shown to accurately capture
the behavior of RC column, composite beam, and composite joint subassembly tests. As
described in Section 6.2.8, the simulated response of these frames begin to diverge from
the true response at drifts greater than 4% when the hinges in the steel beams begin to be
dominated by local buckling.
6.4 Future Work
Based on discussions with practicing structural engineers and contractors, there is clear
interest in use of composite RCS moment resisting frames for seismic design. The
advantages that these systems offer for seismic design been demonstrated in this research,
particularly, the inherent ductility and robustness of the systems. With many of the
questions of their seismic performance answered, a remaining practical challenge to
greater utilization of composite RCS systems lies with convincing the construction
industry of their economical advantages over conventional seismic resisting systems.
393
Discussions with construction firms and industry engineers have pointed to the
development of the precast system as being the key to making composite RCS frames
competitive in the current market. While some constructability issues were addressed in
this research, a more thorough handling of this topic is required to highlight the benefits
of composite RCS frames compared to the more traditional moment frames and other
systems. This could consist of construction engineering and cost analysis that would
compare the construction of RCS frames versus other competitive systems.
6.4.1 Calibration of Deteriorating Models
Limitations in the fiber beam-column elements prevents the ability to simulate large drifts
(>4%) and perform collapse studies. It is for these reasons that models that include more
significant strength and stiffness deterioration capabilities, such as the nonlinear (moment
versus rotation) hinge models proposed by Ibarra (2003), should be calibrated to use with
the RC columns and composite steel beams. These models are most important for the
composite steel beams since they will be able to capture the strength and stiffness
deterioration that occurs in these hinges due to local buckling in the flange and web of the
steel beam. This effect was shown to be an important effect in the response of the test
frame to the large earthquake intensities when interstory drifts exceeded 3-4%. These
models will provide the tools to more accurately simulate large drifts in these composite
RCS frames and allow the study of the collapse capacity of the system.
6.4.2 Investigation of Subassembly Boundary Conditions
Both of these boundary condition effects have proven to be a key difference between the
performance of steel beams and composite slabs in continuous moment frame systems
compared to a typical subassembly test. While no recommendations are made directly in
this study, this does leave some work to determine how these subassembly tests should be
interpreted given that these boundary conditions are not accurately captured. One idea is
to create a subassembly setup that will be able to accurately capture the realistic boundary
conditions of a continuous moment frame. While this would be a more complicated
394
setup than what is done for typical beam-column tests, the main intention would be
obtain a direct comparison between this and the traditional setup. This would give some
insight as to how much of the strength and stiffness deterioration in the composite beam
is due to the boundary conditions and perhaps provide a way to reinterpret the traditional
subassembly tests.
6.4.3 Alternative Energy-Based Damage Models
While the deformation-based damage model used in this study was shown to work in
capturing the physical damage of the test frame, there are some problems stemming from
the general assumptions used to compute plastic deformation. As described in Chapter 4,
these are due to (1) the overestimation of plastic rotations in OpenSees for unsymmetrical
sections (i.e. composite beams) and (2) the general problem of defining plastic rotations
for a highly pinched response. Alternatively, the use of an energy-based damage model
(i.e., Kratzig et al. 1989, Mehanny 1999) could help avoid some of the problems
observed in this study given that the calculation of hysteretic energy is less sensitive to
the issues described here. Measured and simulated results from this study could be
reinterpreted using these alternative damage measures and validated against the
performance of the test frame.
395
Bibliography
ACI (2002). Building Code Requirements for Structural Concrete, ACI-318-02,

American Concrete Institute, Farmington Hills, MI.
ACI-ASCE Committee 352 (1985). Building Code Requirements for Structural

Concrete, Report No. ACI 318-99, American Concrete Institute, Farmington
Hills, Michigan.
Altoontash, A. (2004). Simulation and Damage Models for Performance Assessment of

Reinforced Concrete Beam-Column Joints, PhD. Thesis, Department of Civil and
Environmental Engineering, Stanford University, CA.
AISC (1997). American Institute of Steel Construction, Inc., Seismic Provisions for
Structural Steel Buildings, Load and Resistance Factor Design, 2nd Edition,
Chicago, Illinois, 1997.
AISC (1999). Load and Resistance Design Specification for Structural Steel Buildings,
2nd Ed., American Institute of Steel Construction, Chicago, IL, 1999, 2001 and
2005.
AISC (2002). Seismic Provisions for Structural Steel Buildings, American Institute of
Steel Construction, Chicago, IL, 2002 and 2005.
ASCE Guidelines (1994). Guidelines for Design of Joints Between Steel Beams and
Reinforced Concrete Columns, Journal of Structural Division, ASCE, Vol.
120(8), pp. 2330-2357.
ASCE (2002). Minimum Design Loads for Buildings and Other Structures, SEI/ASCE
7-02, ASCE, Reston, VA, 2002 and 2005.
Aslani, H. (2005). Probabilistic Earthquake Loss Estimation and Loss Disaggregation in

Buildings, Ph.D. Dissertation, Department of Civil and Environmental
Engineering, Stanford University.
Aktan, A.E., M. ASCE, and Bertero, V.V. (1987). Evaluation of Seismic Response of
RC Buildings Loaded to Failure, Journal of Structural Engineering, Vol. 113,
No. 5, May 1987.
Baba, N., Nishimura, Y. (2000). Seismic Behavior of RC Column to S Beam Moment

Frames, Proc.12WCEE.
Baker, J. and Cornell, A.C.(2005) A Vector-Valued Ground Motion Intensity Measure

Consisting of Spectral Acceleration and Epsilon, Earthquake Engineering and
Structural Dynamics, Vol. 34, No. 10, pp. 1193-1217.
396
Bernal, D. (1994). Viscous Damping in Inelastic Structural Response, Journal of
Structural Engineering, Vol 120, No. 4. pp 1241-1254.
Building Seismic Safety Council (BSSC), 1995, 1994 Edition NEHRP Recommended
Provisions for the Development of Seismic Regulations for New Buildings, Part 1:
Provisions, part II: Commentary, developed for Federal Emergency Management
Agency, FEMA 222 and 223, Washington DC.
Bugeja, M., Bracci, J.M., and Moore, W.P. (1999). Seismic Behavior of Composite
Moment Resisting Frame Systems, Technical Report CBDC-99-01, Dept. of
Civil Engrg., Texas A & M University.
Bugeja, M., Bracci, J.M., and Moore, W.P. (2000). Seismic Behavior of Composite RCS
Frame Systems. Journal of Structural Engineering, 126(4), 429-436.
Bursi, O.S., and Ballerini, M. (1996). Behavior of Steel-Concrete Composite

Substructure with Full and Partial Shear Connection, Proceedings of 11th World
Conference on Earthquake Engineering, Paper No. 771, Acapulco, Mexico, 1996.
Carrasquillo, R.L., Nilson, A.H., and Slate, F.O. (1981). Properties of High Strength
Concrete Subjected to Short Term Loads, ACI Structural Journal, V.78, No. 3,
May-June 1981, pp. 171-178.
Chen, C.C. and Lin, N.J. (2002), Seismic Behavior of Steel Beam-to-RC Column Joint,
Technical Report, National Center for Research on Earthquake Engineering.
(in Chinese)
Cheng, C.T. and Cian, P.H. (2002), Composite Behavior of Slab and Steel Beam-to-RC
Column, Technical Report, National Center for Research on Earthquake
Engineering. (in Chinese)
Civjan, S.A., Engelhardt, M.D., and Gross, J.L. (2000). Retrofit of Pre-Northridge
Moment-Resisting Connections, Journal of Structural Engineering, V. 126,
No.4, April 2000.
Civjan, S.A., M.ASCE, and Singh, P. (2003). Behavior of Shear Studs Subjected to
Fully Reversed Cyclic Loading, Journal of Structural Engineering, V.129, No.
11, November 2003.
Chopra, A.K. (1995). Dynamics of Structures. Theory and Applications to Earthquake

Engineering, Prentice-Hall.
Committee on Design of Steel Building Structures of the Committee on Metals,

Structural Division. Compendium of Design Office Problems, Journal of
Structural Engineering, Vol 118, No. 12, December, 1992.
397
Cordova, P.P., Deierlein, G. G., Mehanny, S.S.F. and Cornell, C. A. (2001).
Development of a Two-Parameter Seismic Intensity Measure and Probabilistic
Assessment Procedure, The Second U.S.-Japan Workshop on Performance-Based
Earthquake Engineering Methodology for Reinforced Concrete Building
Structures, Sapporo, Hokkaido, March 2001.
Cordova, P.P., Chen, C.H., Lai, W.C., Deierlein, G.G., and Tsai, K.C. (2006). Pseudo-
Dynamic Test of Full-Scale Composite RCS Frame, John A. Blume Earthquake
Engineering Center, Department of Civil Engineering, Stanford University, CA
(in press).
Deierlein, G.G., Yura, J.A., and Jirsa, J.O. (1988). Design of Moment Connections
forcomposite Framed Structures, PMFSEL Report No. 88-1, University of Texas
at Austin, Texas.
Deierlein, G.G., Sheikh, T.M., Yura, J.A., and Jirsa, J.O. (1989). Beam-Column
Moment Connections for Composite Frames: Part 2, Journal of Structural
Engineering ASCE, Vol. 115, November 1989, pp. 2877-2896.
Deierlein, G.G. (2000). New Provisions for the Seismic Design of Composite and
Hybrid Structures Earthquake Spectra, EERI, 16(1), pp. 163-178.
Deierlein, G.G. and Noguchi, H. (2004). Overview of US-Japan Research on the

Seismic Design of Composite Reinforced Concrete and Steel Moment Frame
Structures, Journal of Structural Engineering, ASCE, 130(2), pp. 361-367.
Dooley, K.L. and Bracci, J.M. (2001). Seismic Evaluation of Column-to-Beam Strength
Ratios in Reinforced Concrete Frames, ACI Structural Journal, V.98, No.6,
Nov.-Dec. 2001.
Du Plessis, D.P. and Daniels, J.H. (1972). Strength of Composite Beam to Column
Connections. Report No. 374.3, Fritz Engineering Lab., Lehigh University,
Bethlehem, Pa.
Fardis, M. N. and Biskinis, D. E. (2003). Deformation Capacity of RC Members, as

Controlled by Flexure or Shear, Otani Symposium, 2003, pp. 511-530.
FEMA-273 (1997). NEHRP Commentary on the Guidelines for the Seismic

Rehabilitation of Buildings, American Society of Civil Engineers (funded by
Federal Emergency Management Agency), 1997.
FEMA-350 (2000). Recommended Seismic Design Criteria for New Steel Moment-
Frame Buildings, SAC Joint Venture (funded by Federal Emergency
Management Agency), June 2000
398
FEMA-351 (2000). Recommended Seismic Evaluation and Upgrade Criteria for
Existing Welded Steel Moment-Frame Buildings, SAC Joint Venture (funded by
Federal Emergency Management Agency), July 2000.
FEMA-356 (2000). Prestandard and Commentary for the Seismic Rehabilitation of

Buildings, ASCE (funded by Federal Emergency Management Agency),
November 2000.
Filippou, F.C. (1986). A Simple Model for Reinforcing Bar Anchorages Under Cyclic
Excitations, ASCE, Journal of Structural Engineering, Vol. 112, No. 7, July
1986.
Filippou, F.C., Popov E.P., and Bertero V.V. (1986). "Modeling of R/C joints under
cyclic excitations," Journal of Structural Engineering, Vol. 109, No. 11, pp.
2666-2684.
Foutch, D.A., Goel, S.C., Roeder, C.W. (1987). Seismic Testing of Full-Scale Steel
Building Part 1, Journal of Structural Engineering, Vol. 113, No. 11,
November 1987.
Goel, S.C. (2004). United States-Japan Cooperative Earthquake Engineering Research

Program on Composite and Hybrid Structures, Journal of Structural
Engineering, Vol. 130, No. 2, February 2004.
Griffis, L.G. (1986). Some Design Considerations for Composite-Frame Structures,

AISC Engineering Journal, Second Quarter, pp. 59-64.
Griffis, L.G. (1992). Composite Frame Construction, Constructional Steel Design, Ed.
P.J. Dowling, J.E. Harding, R. Bjorhovde, Elsevier Applied Sciences, NY, pp.
523-554.
Gupta, A. and Krawinkler, H. (1999). Seismic Demands for Performance Evaluation of

Steel Moment Resisting Frame Structures, John A. Blume Earthquake
Engineering Center, Report No. 132, Department of Civil Engineering, Stanford
University.
Ibarra, L. (2003). Global Collapse of Frame Structures Under Seismic Excitations, PhD.
Thesis, Department of Civil and Environmental Engineering, Stanford University,
CA.
International Code Council (2003). International Code Council, Falls Church, VA, 2000,
2003, and 2005.
Kannan, A.E., and Powell, G.H. (1973). Drain-2D, a general purpose computer program
for dynamic analysis of inelastic plane structures, Report No. UCB/EERC 73-6,
Earthquake Engineering Research Center, Univ. of California, Berkeley, CA.
399
Kanno, R. (1993). Strength, Deformation, and Seismic Resistance of Joints Between
Steel Beams and Reinforced Concrete Columns, PhD. Thesis, Cornell
University, Ithaca, NY
Kanno, R. and Deierlein, G.G. (1994). Cyclic Behavior of Joints Between Steel Beams
and Reinforced Concrete Columns, ASCE Structures Congress Proceedings.
Kanno, R. and Deierlein, G.G. (1997). Seismic Behavior of Composite (RCS) Beam-
Column Joint Subassemblies, Composite Construction in Steel and Concrete III,
American Society of Civil Engineers, Reston, VA, 1997, pp. 236-249.
Kanno, R. and Deierlein, G.G. (2000). Design Model of Joints for RCS Frames
Composite Construction IV, ASCE, (in press).
Kanno, R. and Deierlein, G.G. (2002). Design Model of Joints for RCS Frames.
Composite Construction in Steel and Concrete IV, ASCE Reston, VA., 947-958.
Kaul, R. (2004). Object Oriented Development of Strength and Stiffness Degrading

Models for Reinforced Concrete Structures, Ph.D. Thesis, Department of Civil
and Environmental Engineering, Stanford University, CA.
Kemp, A.R. and Dekker, N.W. (1991). Available Rotation Capacity in Steel and
Composite Beams, The Structural Engineer, Vol. 69, No. 5, March, 1991, pp.
88-97.
Kratzig, W.B., Meyer, I.F., and Meskouris, K. (1989). Damage Evolution in

ReinforcedConcrete Members under Cyclic Loading, 5th International
Conference on Structural Safety and Reliability, San Francisco, CA, Vol. II, pp.
795-802.
Krawinkler, H., Miranda E., Bozorgnia, Y. and Bertero, V. V. (2004). Chapter 9:

Performance Based Earthquake Engineering,, Earthquake Engineering: From
Engineering Seismology to Performance-Based Engineering, CRC Press, Florida.
Kuramoto, H. (1996). Seismic Resistance of Through Column Type Connections for

Composite RCS Systems. Proc., 11 WCEE, Paper No. 1755, Elsevier Science.
Kuramoto, H. and Nishiyama, I. (2004). Seismic Performance and Stress Transferring

Mechanism of Through-Column-Type Joints for Composite Reinforced concrete
and Steel Frames. Journal of Structural Engineering, 130(2), 352-360.
Lee, K. and Foutch, D.A. (2002). Seismic Performance Evaluation of Pre-Northridge

Steel Frame Buildings with Brittle Connections, Journal of Structural
Engineering, Vol. 128. No. 4, April 2002.
400
Lee, S.J. (1987). Seismic Behavior of Steel Building Structures with Composite Slabs,
Ph.D. Thesis, Department of Civil Engineering, Lehigh University, Bethlehem,
1987
Lee, S.J. and Lu, L.W. (1989). Cyclic Tests of Full-Scale Composite Joint
Subassemblages, Journal of Structural Division, ASCE, Vol. 115, No. 8, Aug.
1989, pp. 1977-1998.
Leon, R.T. and Deierlein, G.G. (1995). An Overview of Codes, Standards, and
Guidelines for Composite Construction, ASCE Conference Proceedings, Boston,
May 1995, pp. 1297-1300
Leon, R.T. and Flemming, D.J. (1997). The Shear Resistance of Headed Studs Used
with Profiled Steel Sheeting, Composite construction in Steel and Concrete III,
325-338.
Leon, R.T., Hajjar, J.F., and Gustaafson, M.A., (1998). Seismic Response of Composite
Moment-Resisting Connections. I: Performance, Journal of Structural
Engineering, ASCE 124(8), 868-876.
Liang, S. and Ding, D. (1994). Comparative Experimental lStudies of Models of Self-

Controlled and Ordinary Frames on the Shaking Table, Earthquake Engineering
and Structural Dynamics, Vol. 24, 533-547 (1995).
Liang, X., Parra-Montesinos, G, and Wisght, J.K. (2003). Seismic Behavior of RCS
Beam-column Subassemblies and Frame Systems Designed Following a Joint
Deformation-Based Capacity Design Approach, Report Number UMCEE 03-10,
University of Michigan, MI.
Liang, X. and Parra-Montesinos, G.J. (2004). Seismic behavior of reinforced concrete

column-steel beam subassemblies and frame systems. Journal of Structural
Engineering, 130(2), 310-319.
Maffei, J., Stanton, J., Nigel, P., and Park, R. (2004). "Design Approaches,"Seismic
Design of Precast Building Structures, State of the Art Report [Robert Park
editor], Commission 7, Federation International du Beton, Lausanne, Switzerland,
Chapter 4, January 2004.
Mahin, S.A. and Shing, P.B. (1985). "Pseudodynamic Method for Seismic Testing,"
Journal of Structural Engineering, Vol. 111, No. 7, pp. 1482-1503.
Mann, A.P. and Morris, L.J. (1984). Lack of Fit in High Strength Connections, Journal
of Structural Engineering, ASCE, Vol. 110, No. 6, June, 1984, pp 1235-1252.
401
Mazzoni, S. and Moehle, J.P. (2001). Seismic Response of Beam-Column Joints in
Double-Deck Reinforced Concrete Bridge Frames. ACI Structural Journal 98
(3): 259-269.
McKenna, F. and G. L. Fenves (1999). G3 Class Interface Specification Version 0.1

Preliminary Draft, Pacific Earthquake Engineering Research Center, University
of California at Berkeley, CA
Medina, R. and Krawinkler, H. (2003), Seismic Demands for Non-Deteriorating Frame

Structures and Their Dependence on Ground Motions, Report PEER 2003/13,
Pacific Earthquake Engineering Center, October 2003.
Mehanny, S.S., Cordova, P.P., and Deierlein, G.G. (2000). Seismic Design of Composite
Moment Frame Buildings Case Studies and Code Implications, Composite
Construction IV, ASCE, in press.
Mehanny, S.S.F. and Deierlein, G.G. (2001). Seismic Damage and Collapse Assessment
of Composite Moment Frames. Journal of Structural Engineering, ASCE,
127(9), 1045-1053.
Noguchi, H. and Kim, K. (1998). Shear Strength of Beam-to-Column Connections in

RCS System, Proceedings of Structural Engineers World Congress, Paper
No.T177-3, Elsevier Science, Ltd.
Noguchi, H. and Uchida, K. (2004). Finite Element Method Analysis of Hybrid

Structural Frames with Reinforced concrete Columns and Steel Beams. Journal
of Structural Engineering, 130(2), 328-335.
Neuenhofer, A. and Filippou, F.C. (1997). "Evaluation of Nonlinear Frame Finite

Element Models", Journal of Structural Engineering, American Society of Civil
Engineers, Vol. 123, No. 7, July 1997, pp. 958-966.
Ollgaard, Slutter, R. G., and Fisher, J. W. (1971). "Shear strength of stud connectors in
lightweight and normal-weight concrete." Engineering Journal, 8(2), pp. 5564.
Panagiotakos, T. B. and Fardis, M. N. (2001). Deformations of Reinforced Concrete at

Yielding and Ultimate, ACI Structural Journal, Vol. 98, No. 2, March-April
2001, pp. 135-147.
Parra-Montesinos, G. and Wight, J.K., (2000). Seismic Behavior, Strength and Retrofit
of Exterior RC Column-to-Steel Beam Connections, Report No. UMCEE 00-09,
University of Michigan, MI.
Parra-Montesinos, G. and Wight, J.K. (2001). Modeling Shear Behavior of Hybrid RCS
Beam-Column Connections, Journal of Structural Engineering, ASCE, 127(1),
pp. 3-11.
402
Parra-Montesinos, G., Liang, X., and Wight, J.K., (2003). Towards Deformation-Based
Capacity Design of RCS Beam-Column Connections, Engineering Structures,
25(5), 681-690.
Paulay, T. and Priestly, M.J.N. (1992). Seismic Design of Reinforced Concrete and
Masonry Buildings, John Wiley & Sons.
PEER (2005). Pacific Earthquake Engineering Research Center: PEER Strong Motion
Database, University of California, Berkeley, http://peer.berkeley.edu/smcat/
(September 15, 2005).
Ravindra, M. K. and Galambos, T. V. (1978). Load and Resistance Factor Design for
Steel. ASCE Journal of the Structural Division, 104, ST9, 1337-1353.
RCSC Committee 15 (2000). Specification for Structural Joints Using ASTM A325 or
A490 Bolts. Research Council on Structural Connections, June 2000.
Roeder, C.W., Foutch, D.A., and Goel, S.C. (1987). Seismic Testing of Full-Scale Steel
Building Part II, Journal of Structural Engineering, Vol. 113, No. 11,
November 1987.
Ruiz-Garcia, J. (2004). Performance-Based Assessment of Existing Structures

Accounting for Residual Displacements, PhD. Thesis, Department of Civil and
Environmental Engineering, Stanford University, CA.
Saatcioglu, M. and Ozcebe, G. (1989). Response of Reinforced Concrete columns to

Simulated Seismic Loading, American Concrete Institute, ACI Structural
Journal, January-February 1989, pp. 3-12.
Saatcioglu, M. and Grira, M. (1999). Confinement of Reinforced Concrete Columns

with Welded Reinforcement Grids, ACI Structural Journal, January-February
1999, pp 29-39.
Schwein, R.L. (1999). The Banging Bolt Syndrome, Modern Steel Construction,
November 1999.
Scott, B. D., Park R., and Priestley, M. J. N. (1982). Stress-Strain Behavior of Concrete
Confined by Overlapping Hoops and Low and High Strain Rates, ACI Journal,
January-February 1982, No. 1, pp. 13-27.
Sheikh, T.M., Yura, J.A., and Jirsa, J.O. (1987). Moment Connections between
SteelBeams and Concrete Columns, PMFSEL Report No. 87-4, University of
Texas at Austin, Texas.
403
Sheikh, T.M., Deierlein, G.G., Yura, J.A., and Jirsa, J.O. (1989). Beam-Column
Moment Connections for Composite Frames: Part 1, Journal of the Structural
Division, ASCE, Vol. 115, No. 11, Nov. 1989, pp. 2858-2876.
Shome, N., (1999). Probabilistic Seismic Demand Analysis of Nonlinear Structures,

Ph.D. Thesis, Stanford University.
Stevens, N. J., Uzumeri, S.M., Collins, M.P., and Will, G.T. (1991). Constitutive Model
for Reinforced Concrete Finite Element Analysis, ACI Structural Journal, V.88,
No. 1, January-February 1991.
Tagawa, Y., Kato, B., and Aoki, H. (1989). Behavior of Composite Beams in Steel
Frame Under Hysteretic Loading, Journal of Structural Division, ASCE, Vol.
115, No. 8, Aug. 1989, pp. 2029-2045.
Tanaka, H., and Park, R. (1990). Effect of Lateral Confining Reinforcement on Ductile
Behavior of Reinforced Concrete Columns, Report No. 90-2, Department of
Civil Engineering, University of Cantebury, Christchurch, New Zealand.
Tide, R.H.R. (1999). Banging Bolts Another Perspective, Modern Steel Construction,
November 1999.
Tsai, K.C. and Chen, P.C. (2002). A Study of RC Column-to-Foundation and Steel
Beam-to-RC Column Joints for an RCS Frame Specimen, Technical
Report, National Center for Research on Earthquake Engineering. (in Chinese)
Uang C.M., and S. Kiggins (2003). Reducing Residual Drift of Buckling-Restrained

Braced Frames as a Dual System, Proceedings of the International Workshop on
Steel and Concrete Composite Construction (IWSCCC), October 8-9, Taipei,
Taiwan.
Vamvatsikos, D. and Cornell, C.A. (2002). Incremental Dynamic Analysis. Earthquake

Engineering and Structural Dynamics 31, 491-514.
404

TR155 Cordova

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

TR155 Cordova

Загружено:

Авторское право:

Доступные форматы

Department of Civil and Environmental Engineering

VALIDATION OF THE SEISMIC PERFORMANCE OF

Paul Phillip Cordova

Report No. 155

The John A. Blume Earthquake Engineering Center

2007 The John A. Blume Earthquake Engineering Center

All Rights Reserved

To the rest of the professors in the Structural Engineering and Geomechanics

I would also like to acknowledge the contribution of Professor Gustavo Parra-Montesinos

And of course, my unending thanks to my beautiful wife, Mrs. Sylvie Cordova.

Chapter 1: Introduction ............................................................................................... 1

1.1 Motivation for this Study .................................................................................. 3

Chapter 2: RCS System Design ................................................................................... 9

2.1 Background of RCS Composite Moment Frames............................................. 9

Chapter 3: Full-Scale Composite RCS Test Frame ................................................. 76

3.1 Background ..................................................................................................... 76

Chapter 4: Analytical Modeling and Validation .................................................... 162

4.1 Introduction ................................................................................................... 162

Chapter 5: Applications ........................................................................................... 272

5.1 Introduction ................................................................................................... 272

Chapter 6: Conclusions ............................................................................................ 382

6.1 Summary ....................................................................................................... 382

Bibliography ................................................................................................................. 396

2.1 Joint width comparisons between ASCE and Updated Guidelines............... 58

3.1 Rationale for large-scale test ......................................................................... 120

4.1 Summary of RC column specimens used in calibration study...................... 214

2.1 Typical construction sequence of composite RCS moment frames............... 63

3.1 Plan View of Building.................................................................................... 123

During 1990s, considerable research on composite construction was conducted through a

An integral element in the performance assessment of structures is the ability to

1.3 Scope and Organization of Report

In Chapter 2, a brief summary of the background on the rationale and development of

2.1 Background of RCS Composite Moment Frames

In Japan, RCS composite systems were developed as an alternative to low-rise reinforced

Another important advantage of these composite systems is the ability to accommodate

An alternative precast construction method can also be implemented in the erection of

2.2.1 Beam-Column Composite Joints

During the 1990s, the United States-Japan Cooperative Earthquake Engineering

2.2.2 Small-Scale Frame Tests

2.2.3 Trial Design Studies of System Performance

2.3 System and Member Design Guidelines

but the result should not be less than:

2.3.2 Member Design Requirements: Part II of the AISC Seismic Provisions

2.3.2.1 Reinforced Concrete Columns

2.3.2.1.1 Strong-Column Weak-Beam

2.3.2.1.1.1 SEAOC Blue Book Provisions

M c = sum of nominal moment strength in all columns framing into the

M b = sum of nominal moment strengths at each end of each beam, taken at

2.3.2.1.2 Precast RC Column Splice Design

2.3.2.2 Composite Steel Beams

2.3.2.2.1 Plastic Strength of Composite Beams

Qn , stat = 0.5 Asc f c' Ec Asc Fu (2.11)

Civjan (2003) performed a series of experimental and analytical tests on modified

2.3.2.2.3 Steel Beam Splices

2.3.2.2.3.1 Bearing versus Slip Design: Bolt Banging

2.4 Composite Joint Design Guidelines

2.4.1 Joint Deformations

2.4.2 General Detailing Requirements

2.4.3 Effective Joint Width

bm = ( b f + b ) 2 b f + h 1.75b f (2.17)b band plates w/no joint hoops

C = ( x h) ( y bf ) (2.17)c bo = 12tbp b b f (2.18)b

2.4.4 Joint Strength

ASCE Guidelines Updated Guidelines

2.4.5 Inner Panel Shear Strength