Blind Prediction of a Full-Scale Steel Frame Test

BLIND PREDICTION OF A FULL-SCALE
3D STEEL FRAME TESTED UNDER

DYNAMIC CONDITIONS

A Dissertation Submitted in Partial Fulfilment of the Requirements
for the Master Degree in

Earthquake Engineering

By

ANNA PAVAN

Supervisor: Dr RUI PINHO

May, 2008

Istituto Universitario di Studi Superiori di Pavia
Universit degli Studi di Pavia

The dissertation entitled Blind prediction of a full-scale 3D steel frame tested under dynamic
conditions, by Anna Pavan, has been approved in partial fulfilment of the requirements for
the Master Degree in Earthquake Engineering.

Rui Pinho -------

Stelios Antoniou

Abstract
i

ABSTRACT

A blind analysis contest for a full-scale four-story building was announced in 2007 by the
executive committee of the E-Defense steel building project, sponsored by the National
Research Institute for Earth Science and Disaster Prevention in Japan.
In this contest, each participant should predict the structural response before and after the test
was performed using the three-dimensions shaking table located in Miki City, Hyogo
Prefecture, Japan.

This work presents the results obtained from dynamic analyses performed on the building
using SeismoStruct, a fibre element-based program. Through comparing the results given by
the program with experimental ones, the aim of the work was to demonstrate that nonlinear
dynamic response of steel buildings is possible.

The influence that some modelling choices (hysteretic rules, mass discretization, damping
parameter) have on the prediction of the global structural response were investigated in order
to obtain the best representation of the real building.

The final model and analysis have a negligible difference from experimental results validating
computer capabilities in predicting nonlinear dynamic response also of steel buildings.

Keywords: steel structure; blind prediction; fibre element; dynamic non-linear analysis.

Acknowledgements
ii

ACKNOWLEDGEMENTS

I would like to express my gratitude to Dr. Rui Pinho for everything he taught me, for his patience, his
dedication towards the competition of this work. Many thanks also to all the people that work in
ROSE School for all the enthusiasm they offer to the realization of this important and unique project.
Special thanks to all my friends in ROSE School for the intensity of each moment spent together.

Index
iii

TABLE OF CONTENTS

ABSTRACT........................................................................................................................................... i
ACKNOWLEDGEMENTS.................................................................................................................. ii
1 INTRODUCTION ........................................................................................................................ 1
1.1 Outline of the Contest ........................................................................................................... 1
1.2 Objectives ............................................................................................................................. 1
1.3 Organization of the work ...................................................................................................... 2
2 OUTILINE OF THE SPECIMEN................................................................................................ 3
2.1 Geometry............................................................................................................................... 3
2.2 Slabs...................................................................................................................................... 5
2.3 Connections........................................................................................................................... 6
2.4 Non structural Elements........................................................................................................ 7
2.4.1 External Walls................................................................................................................... 7
2.4.2 Internal Partitions, openings and ceiling.......................................................................... 7
2.4.3 Exterior Stairs ................................................................................................................... 7
2.4.4 Safety System.................................................................................................................... 7
2.5 Weights ................................................................................................................................. 8
3 STRUCTURAL MODELLING.................................................................................................. 10
Index
iv
3.1 Gravity load on beams ........................................................................................................ 10
3.2 Materials ............................................................................................................................. 13
3.2.1 Concrete .......................................................................................................................... 13
3.2.2 Steel................................................................................................................................. 14
3.3 Sections............................................................................................................................... 23
3.4 Connections......................................................................................................................... 27
3.4.1 Cycling loading test of composite beam......................................................................... 27
3.4.2 Cycling loading test of column....................................................................................... 29
3.4.3 Modelling attempt ........................................................................................................... 31
3.5 Element Classes .................................................................................................................. 33
3.6 Damping.............................................................................................................................. 33
3.7 Constraints .......................................................................................................................... 34
3.8 Integration Scheme ............................................................................................................. 34
4 THE LABORATORY TEST...................................................................................................... 35
4.1 Ground Motion.................................................................................................................... 35
4.2 Measurement during the experiment .................................................................................. 38
4.3 Concrete Characteristics Measured during the test............................................................. 39
5 BLIND PREDICTION AND EXPERIMENTAL RESULTS.................................................... 41
5.1 General remarks.................................................................................................................. 41
5.2 Comparison of the results ................................................................................................... 45
5.2.1 Maximum value of relative displacement from the base at each floor ........................... 46
5.2.2 Maximum value of drift angle and residual drift in each story....................................... 48
5.2.3 Maximum value of absolute acceleration at each floor .................................................. 49
5.2.4 Maximum value of story shear ay each story ................................................................. 50
5.2.5 Maximum value of overturning moment at each floor ................................................... 50
5.2.6 Maximum strain at a specified point in elastic range ..................................................... 51
5.2.7 Time of building collapse ............................................................................................... 52
6 MODEL CALIBRATION.......................................................................................................... 53
7 CONCLUSIONS......................................................................................................................... 61
REFERENCES ................................................................................................................................... 62
APPENDIX A: Structural Drawings ..................................................................................................A1
APPENDIX B: Material Test Results................................................................................................. B1
Index
v

LIST OF FIGURES

Figure 1_Organization of the work..................................................................................................... 2
Figure 2_Steel building specimen. Main structure framing elevations. (unit:m) ............................... 3
Figure 3_Steel building specimen. Main structure framing plan. (unit m)......................................... 4
Figure 4_Structure building specimen. Secondary beams plans. (unit:m) ......................................... 4
Figure 5_Overall view of the specimen. ............................................................................................. 5
Figure 6_2FL, 3FL, 4FL Floor section detail. (unit: mm) .................................................................. 5
Figure 7_RFL Floor section detail. (unit: mm)................................................................................... 6
Figure 8_Beam to Column connection (right) and Base column connection (left). (unit: mm)......... 6
Figure 9_Column trees factory inspection (right). Beam-to-column connection detail (left). ........... 6
Figure 10_Internal LGS partition panels (right), steel angles reinforcement for aluminium sashes
installation (centre) and lightgauge steel for hanging ceiling. ............................................................ 7
Figure 11_Safeguard system scheme and specimen almost completed for the test. .......................... 8
Figure 12_Typical floor beams tributary area. (unit: m) ................................................................. 11
Figure 13_Procedure scheme to determine beams additional distributed mass. ............................. 11
Figure 14_Typical stress-block for the two kind of steel used for structural elements. ................... 15
Figure 15_3D model used for eigenvalue analysis ........................................................................... 18
Index
vi
Figure 16_North-South component acceleration spectrum. ............................................................. 18
Figure 17_Bi-linear constitutive model for column steel. ................................................................ 21
Figure 18_Bi-linear constitutive model for different steel used for beams. ..................................... 23
Figure 19_Deck slab thickness geometry. ........................................................................................ 24
Figure 20_Applied displacement time history. ................................................................................. 25
Figure 21_Model of the beam with the applied displacement. The same analysis was performed for
both slab width. ................................................................................................................................. 26
Figure 22_Hysteretic curves for the two beams with different slab effective width. ....................... 26
Figure 23_Composite beam cycling loading testing set-up. ............................................................. 27
Figure 24_Drawings of the specimen. .............................................................................................. 28
Figure 25_Testing results scheme..................................................................................................... 28
Figure 26_Rotation angle b imposed on the beam.......................................................................... 28
Figure 27_Testing results.................................................................................................................. 29
Figure 28_ Column cycling loading testing set-up. .......................................................................... 29
Figure 29_Explanatory scheme spread by the committee. ............................................................... 30
Figure 30_Hypothesis of new testing set-up..................................................................................... 30
Figure 31_Springs calibration results. .............................................................................................. 32
Figure 32_60% scaled NS ground acceleration time history........................................................... 36
Figure 33_60% scaled EW ground acceleration time history........................................................... 36
Figure 34_60% scaled vertical ground acceleration time history. .................................................... 36
Figure 35_3D model with dynamic loads applied. ........................................................................... 37
Figure 36_Three consecutive NS components. ................................................................................ 37
Figure 37_Three consecutive EW components. ............................................................................... 37
Figure 38_Three consecutive vertical components........................................................................... 38
Figure 39_Transducers position at 1-st floor(right), 2-nd and 3-rd floor (left). Plan. (unit: mm) .... 39
Figure 40_4-th floor and roof transducers position. Plan. (unit: mm) .............................................. 39
Figure 41_Transducers position. Sections. (unit: mm)..................................................................... 39
Figure 42_Concrete stress-block for different floor slab. ................................................................. 40
Figure 43_Definition of floor and story (unit:mm)........................................................................... 41
Figure 44_Gloobal coordinates......................................................................................................... 42
Figure 45_Definition of residual drift............................................................................................... 43
Figure 46_Position of the strain gage of column 1A. ....................................................................... 44
Figure 47_Strain gage position Figure 48_Strain gauges on the specimen. .................... 45
Index
vii
Figure 49_Maximum relative displacements at every level in two directions. ................................ 46
Figure 50_Deformed shape of the pre-test analysis model (out of scale). From left hand, in North-
South direction, East-West, perspective view and detail of the first floor beam-to-column
connection. ........................................................................................................................................ 46
Figure 51_Specimen collapse obtained in the laboratory with a table motion equal to 100% level of
Takatory motion................................................................................................................................ 46
Figure 52_Plastic hinges formation. From left, in y=0m frame, y=6m, x=0m, x=5m and x=10m. . 47
Figure 53_Maximum drift angle at every floor in two directions. ................................................... 48
Figure 54_Maximum residual drift at every floor in two directions. ............................................... 48
Figure 55_Maxima absolute acceleration at every level in two directions....................................... 49
Figure 56_Maximum story shear at very floor in two directions. .................................................... 50
Figure 57_Maximum overturning moment at every floor in two directions. ................................... 50
Figure 58_Maximum strain at column face. ..................................................................................... 51
Figure 59_Column 1A axial force time-history from pre-test analysis. ........................................... 52
Figure 60_Column 1A stress-block. ................................................................................................. 52
Figure 61_Stress-strain curve for column steel. Detail..................................................................... 54
Figure 62_Maximum floor relative displacement after model calibration. ...................................... 55
Figure 63_Maximum story drift angle after model calibration. ....................................................... 56
Figure 64_Maximum story residual drift angle after model calibration........................................... 56
Figure 65_Maximum floor absolute acceleration after model calibration........................................ 57
Figure 66_Maximum story shear after model calibration................................................................. 57
Figure 67_Maximum story overturning moment after model calibration. ....................................... 58
Figure 68_Column maximum axial strain after model calibration. .................................................. 59
Figure 69_Acceleration and Displacement spectra. Highlighted the two building periods. ............ 59
Index
viii

LIST OF TABLES

Table 1_Members schedule (mm) ...................................................................................................... 4
Table 2_Secondary beams schedule (mm) ......................................................................................... 5
Table 3_Table of calculated weights (kN).......................................................................................... 8
Table 4_Floor weights and storey height............................................................................................ 9
Table 5_ Primary Steel elements self weight.................................................................................... 11
Table 6_ Secondary beams and Slab self weight.............................................................................. 12
Table 7_Weight of slab portion that collaborates in the composite beam. ....................................... 12
Table 8_Floor net self weight. .......................................................................................................... 13
Table 9_Additional mass distributed on girders. .............................................................................. 13
Table 10_Concrete parameters used in the pre-test model ............................................................... 14
Table 11_Steel parameters derived from experimental data............................................................. 15
Table 12_ Moment of inertia of structural elements......................................................................... 16
Table 13_Elastic frame elements properties. .................................................................................... 17
Table 14_Columns yielding rotation. ............................................................................................... 20
Table 15_Column plastic strain. ....................................................................................................... 20
Table 16_Linear distribution of displacement. ................................................................................. 21
Index
ix
Table 17_Plastic strain requested by different girders...................................................................... 22
Table 18_Steel bi-linear constitutive model characteristic parameters. ........................................... 23
Table 19_Effective slab width and thickness for different beams. ................................................... 27
Table 20_Input parameters used in the hysteretic model.................................................................. 32
Table 21_Peak ground acceleration for 60% scaled ground motion. ............................................... 36
Table 22_PGA for different ground motion intensities. ................................................................... 38
Table 23_Characteristic parameters for concrete. ............................................................................ 40
Table 24_Error evaluation in predicting relative displacements in different analyses. .................... 47
Table 25_Error evaluation in predicting interstory drift. Residual drift values for different analyses.
........................................................................................................................................................... 48
Table 26_ Error evaluation in predicting absolute acceleration in different analyses. ..................... 49
Table 27_ Error evaluation in predicting shear forces in different analyses. ................................... 50
Table 28_ Error evaluation in predicting overturning moment in different analyses....................... 51
Table 29_Axial Strain at column face values. .................................................................................. 51
Table 30_Column maximum axial strain comparison. ..................................................................... 54
Table 31_Error evaluation for floor relative displacement after model calibration. ........................ 55
Table 32_ Error evaluation for floor story drift and residual drift after model calibration. ............. 56
Table 33_Error evaluation for floor absolute acceleration after model calibration.......................... 57
Table 34_Error evaluation for story shear after model calibration................................................... 58
Table 35_Error evaluation for overturning moment after model calibration. .................................. 58

Charter 1 . Introduction
1

1 INTRODUCTION
1.1 Outline of the Contest

In May 2007 the executive committee of the E-Defense steel building project, sponsored by
the National Research Institute for earth Science and Disaster Prevention in Japan, announced
the 2007 Blind Analysis Contest for a full-scale four-story steel building, which was tested to
collapse in September 2007 on the world's largest three-dimensional shaking table located at
Miki City, Hyogo Prefecture, Japan. The test was conducted by applying a scaled version of
near-fault motion recorded during the 1995 Kobe earthquake.
Two analysis method were categorized as 2D and 3D and, inside these two, participants were
divided between researchers and practicing engineers. Each participant had to predict the
response before and after the test. Because the actual loadings were determined during the
course of the testing based on observed response, the contest was organized in two parts: pre-
test analysis based on anticipated earthquake loadings, and post-test analysis using the actual
loadings. The requirement was that analytical model for the post-test analysis had to be
identical to that for pre-test analysis.
Under executive committee, two working groups were organized: the Analysis Method and
Verification WG was responsible for the announcement, distribution of data, answering
questions, and determination of contest winners and the "Building Collapse Simulation WG
that produced the experimental data for the collapse of the building.
The work presented in following pages analyzed a 3D model of the building and was part of
the researcher category.
1.2 Objectives

The aim of this work was to demonstrate the that a good prediction of nonlinear dynamic
response of a steel structure is possible through appropriate modelling choices.
Charter 1 . Introduction
2
In the past, the precision in foreboding the response of reinforce concrete buildings was
largely verified; the same started to occur for steel frames more recently and with a smaller
amount of example.
The contest at E-Defense represented an opportunity to compare computer analysis results
with experimental ones, adding one more prove to their validation in predicting also steel
frames behaviour.
1.3 Organization of the work

The first part of the work was focused in collecting all the information from the Analysis
Method and Verification Working Group. Structural geometry was provided, together with
details of loading conditions, material test results on steel, ideal acceleration time-history and
response spectrum of seismic motion.
On the base of collected data, a 3D model was constructed to be analyzed in SeismoStruct.
Different modelling choices (mass discretization, hysteretic rules, damping) were studied in
order to obtain the best representation of the real building. A pre-test nonlinear dynamic
analysis was performed for and incipient collapse level.
After the test was carried out, some supplemental data were provided (material test results on
concrete and measured acceleration). Analytical model was updated with the new data and
analyzed once more for three consecutive seismic levels: elastic, incipient collapse and
collapse level. This stage of the process was called post-test analysis and it was still a blind
prediction.
Results obtained through the blind prediction were compared to experimental ones and
several other investigations regarding modelling choices were made to improve the response
prediction validity.

Figure 1_Organization of the work.
Charter 2 . Outline of the Specimen
3

2 OUTILINE OF THE SPECIMEN
2.1 Geometry

The building is made by four-storey steel moment resisting frames. Along x-direction there
are two frames composed by two bays 5m long, while in y-direction the frames are three, with
one bay 6m long. Interstorey height is 3.5m and at the top of the building there is a parapet
0.9m height from the net height of the roof slab, for a total height of the building equal to
15.275m.

Figure 2_Steel building specimen. Main structure framing elevations. (unit:m)
4

Figure 3_Steel building specimen. Main structure framing plan. (unit m)
The frames consist in square tube columns and composite beams with wide flanges girders.
The geometry of steel elements is presented in the following table.
Table 1_Members schedule (mm)
Beam Column
Floor G1 G11 G12 Story
R H- 346x174x6x9 H- 346x174x6x9 H- 346x174x6x9 4 RHS- 300x300x9
4 H- 350x175x7x11 H- 350x175x7x11 H- 340x175x9x14 3 RHS- 300x300x9
H- height x width x web thickness x flange thickness, RHS- height x width x thickness

At each floor there are also secondary beams, the geometry of which depends from the level.

1_GF+800 Level 2_ 2FL, 3FL

3_ 4FL 4_RFL
Figure 4_Structure building specimen. Secondary beams plans. (unit:m)
5
Table 2_Secondary beams schedule (mm)
B20 H- 200x100x5.5x8
B29 H- 294x200x8x12
B34 H- 346x174xx9
B35 H- 350x175x7x11

Figure 5_Overall view of the specimen.
2.2 Slabs

Slabs at second, third and forth level consist in composite deck floor, 175mm height. Instead,
roof floor is a reinforced concrete slab, with a flat steel deck at its bottom, for an overall
thickness of 150mm.

Figure 6_2FL, 3FL, 4FL Floor section detail. (unit: mm)

6

Figure 7_RFL Floor section detail. (unit: mm)
2.3 Connections

All connections are made using details and fabrication practice developed following the 1995
Hyogoken-Nanbu earthquake. The figure below shows a typical beam-to-column connection.
First a column tree is built by welding together the column and the starting part of the
beam. After, the rest of the beam is connected to the tree using a bolted connection. This type
of connection forces the eventual formation of the plastic hinge away from the weaker joint
between the column and the beam.

Figure 8_Beam to Column connection (right) and Base column connection (left). (unit: mm)
Column bases are connected to concrete blocks (1.5m height) with steel plates, anchored to
them through steel bolts; these concrete blocks create the connection with the shaking table.

Figure 9_Column trees factory inspection (right). Beam-to-column connection detail (left).
7
2.4 Non structural Elements
2.4.1 External Walls
External walls consist in ALC (autoclaved, aerated concrete) panels, 0.125m thick, fixed on to
the support beams at the top and the bottom of their edges. These panels are connected
between them using the HDR (Hebel Dry Rocking) method, which does not use mortar but
rabbet type panel that are allowed to rock in case of earthquake.
Panels are fixed to the girder at every level through ruler angles and plates to maintain
equilibrium against overturning but are effectively supported by wide flanges beams H-
300x150x6.5x9 (see drawing S03.1 in Appendix A), positioned at 500mm (axis position)
from the bottom end of the building. In this way, all the weight coming from the weight of the
ALC panel is unloaded directly at the foundations, without affecting any other structural
element above.
2.4.2 Internal Partitions, openings and ceiling
Internal partitions are made using LGS (light gauge steel) backing board installed on
aluminium frames. Glass windows with aluminium sash were installed within external ALC
panels openings and reinforced by steel angles. A lightgauge steel was used to hanging ceiling
at each floor.

Figure 10_Internal LGS partition panels (right), steel angles reinforcement for aluminium sashes
installation (centre) and lightgauge steel for hanging ceiling.
2.4.3 Exterior Stairs
During construction phases, steel stairs were put up to facilitate the building process. After the
specimen was completed, they were removed. This specification was made because all
elevation drawings in Appendix A show the presence of these stairs but effectively their
presence did not need to be taken in consideration for modelling.
2.4.4 Safety System
An anti-collapse safeguard system was built to protect shaking table from collapse of the
structure. The system has four components: steel blocks inside the building at each floor (see
drawing S07.1 in Appendix A), outside fence at first story, diagonal cables at first two stories
(see drawing S17.1 in Appendix A) and jumbo tray lay directly of the shaking table. The
system prevent interstory drift beyond 1/3.5 rad (equal to 0.29 rad circa).

8

Figure 11_Safeguard system scheme and specimen almost completed for the test.

2.5 Weights

The following table presents the weights of each part spread by the committee.
Table 3_Table of calculated weights (kN)
Floor
Steel
Frame
Exterior
Wall
Interior
Wall Ceiling Parapet
Safeguard
System
Cantilevel
Floor Total
Roof Floor 459 20 12 71 2 565
4-th Story 19 79 35 133
4-th Floor 270 24 3 47 4 349
3-rd Story 18 73 30 122
3-rd Floor 260 32 3 47 4 347
2-nd Story 18 73 30 8 130
2-nd Floor 260 41 47 4 352
1-st story 27 76 12 115
Total 1248 200 302 95 19 71 162 15 2113

9
Each part includes:
Floor: slab, steel beams, deck plates, handrails
Steel Frame (Story): columns, diaphragms, connection panels
Steel Frame (Floor): girders, stud bolts, high tension bolts, slice plates, fire resistive covering
materials
Exterior Wall: ALC panels, reinforcement material for openings, ALC support beam on 1-st
floor, glass, sash
Interior Wall: plaster boards, light-gauge steel backing, doors
Ceiling: plaster boards, rock wool sound absorbing boards, light-gauge steel backing
Parapet: RC parapet on roof floor
Safeguard System: steel tables on floor slab, diagonal wires
Cantilever floor: cantilever floor for temporary stairs, handrails
The following table is a summarizes which is the load applied at every floor. It has to be
reminded that the weight of external ALC panels was not considered to affect any structural
element a part from the beams at 1-st floor. For this reason it was not taken in consideration in
the calculation of the floor weight.

Table 4_Floor weights and storey height
Floor number
j
Floor weight
W
j
(kN)
Story number
i
Story height
h
i
(m)
5 631.5 4 3.5
4 476.5 3 3.5
3 473.0 2 3.5
2 474.5 1 3.875
total 2055.5

Charter 3 . Structural Modelling
10

3 STRUCTURAL MODELLING
All analysis have been carried out using SeismoStruct, a finite element analysis program, used
for seismic analysis of framed building. The software is fibre element-based, able to predict
accurately the distribution of damage because it spreads material inelasticity both along the
element length and across its section depth.
In what follows, both eingenvalue and dynamic time-history analysis performed are
presented. The first was used to evaluate the natural period of vibration of the building while
the second one was use to predict the nonlinear inelastic response of the structure when
subjected to earthquake loading.
In this chapter, all the choices made to model the real building in the most accurate way are
explained. Some of them were made on the base of data spread by the committee; other ones
needed to derive from self-made assumption, based on engineering judgement and software
capabilities.
3.1 Gravity load on beams

To determine the amount of mass to apply on each beam, its tributary area was defined on the
geometry of the problem, assuming a distributed mass over the diaphragm. Mass was
internally converted in vertical load by the software. As a consequence, the vertical load on
the beam was independent of the span direction of the slab. This had a little consequence on
the results since building global response (i.e. floor shear, floor acceleration, etc.) depends on
the mass per floor and not really from how the weight is transferred to the beams.
11

Figure 12_Typical floor beams tributary area. (unit: m)
Starting from the given data, the additional distributed mass to apply on each beam was
calculated, subtracting the self weight of the structure (composite beams and columns). The
following scheme explains the process followed to arrive to the additional mass value.

Figure 13_Procedure scheme to determine beams additional distributed mass.
Table 5_ Primary Steel elements self weight.
Element Quantity
Cross Section
Area(m
2
)
Length (m) Self Weight (kN)
2G1 4 0.0083 5 13.006
3G1 4 0.0071 5 11.140
4G1 4 0.0063 5 9.814
RG1 4 0.0052 5 8.182
2G11 2 0.0083 6 7.803
3G11 2 0.0083 6 7.803
4G11 2 0.0063 6 5.888
RG11 2 0.0052 6 4.909
2G12 1 0.0101 6 4.736
3G12 1 0.0083 6 3.902
4G12 1 0.0079 6 3.675
RG12 1 0.0052 6 2.455
Column 6 0.0102 14.052 67.079
Total 150.393

12
The following table presents secondary beams self weight that, together with primary steel
elements self weight, was subtracted to find slab self weight.
Table 6_ Secondary beams and Slab self weight.
Floor Element Quantity
Length
(m)
Cross section
area (m
2
)
Beams Self
Weight
(kN)
Slab Self
Weight
(kN)
Slab Self
Weight
(kN/m
2
)
B35 2 6 0.006 5.753
B29 8 2.5 0.007 10.858
B20 4 2 0.003 1.630
2-3
CB20 4 1.25 0.003 1.019
240.741 4.012
B35 2 6 0.006 5.753
B29 16 2.5 0.007 21.715
B20 4 0.75 0.003 0.611
4
CB20 4 1.25 0.003 1.019
240.902 4.015
B34 2 6 0.0051 4.774
B20 2 1.5 0.003 0.611
CB20 2 1.25 0.003 0.509
RF
B25 2 2.5 0.004 1.420
343.685 5.728

In SeismoStruct, beams were defined as Composite I sections. In this way the program
considers also the weight of the strip of slab that collaborates with the girder in the composite
action as self weight. For this reason, this portion of load does have to be considered in the
definition of additional mass. In paragraph 3.2.2 the procedure followed to determine beams
effective width is presented. Here, just the final results is used to calculate beams effective
self weight.
Table 7_Weight of slab portion that collaborates in the composite beam.
Girder b
eff
(m)
Length
(m) A
eff
(m
2
)
Slab Self
Weight (kN)
2G1 0.575 5 2.875 11.536
2G11 0.575 4.85 2.78875 11.189
2G12 1.15 4.85 5.5775 22.379
3G1 0.571 5 2.855 11.455
3G11 0.575 4.858 2.79335 11.208
3G12 1.15 4.858 5.5867 22.416
4G1 0.525 5 2.625 10.539
4G11 0.525 4.95 2.59875 10.434
4G12 1.05 4.95 5.1975 20.868
RG1 0.496 5 2.48 14.206
RG11 0.496 5.008 2.483968 14.228
RG12 0.496 5.008 2.483968 14.228

Hence, subtracting from initial data the weights of steel elements and collaborating slab, it
was possible to determine the net weight of the slab.

13
Table 8_Floor net self weight.
Floor number
j
Floor weight
W
j
(kN)
Net Floor weight
W
j net
(kN)
5 631.5 508.028
4 476.5 356.401
3 473 342.675
2 474.5 340.870
tot. 2055.5 1547.97

Finally, the following table presents values of additional mass per unit length applied to the
beams in the model.
Table 9_Additional mass distributed on girders.
Floor number
j
Net Floor
weight
W
j net
(kN)
Element
Tributary
Area
(m
2
)
Load on the
element
(kN)

Length
(m)
Load
(kN/m)
Additional
Mass
(ton/m)
RG1 6.2511 45.532 5 11.326 1.155
RG11 8.7489 63.725 6 12.841 1.309
5 437.028
1

RG12 17.4978 127.450 6 21.242 2.165
4G1 6.2511 37.132 5 7.426 0.757
4G11 8.7489 51.969 6 8.661 0.883
4 356.401
4G12 17.4978 103.937 6 17.323 1.766
3G1 6.2511 35.702 5 7.140 0.728
3G11 8.7489 49.967 6 8.328 0.849
3 342.675
3G12 17.4978 99.934 6 16.656 1.698
2G1 6.2511 35.514 5 7.103 0.724
2G11 8.7489 49.704 6 8.284 0.844
2 340.870
2G12 17.4978 99.408 6 16.568 1.689
3.2 Materials
3.2.1 Concrete
For pre-test analysis, precise data regarding properties of concrete were not distributed by the
organizing committee. Actually, they had the intent to determine the effective stress-strain
relation during the shacking table test (experimental results regarding concrete constitutive
model can be consulted in paragraph 4.3).
The only specification made from them during pre test preparation says that (see drawing
S04.1 in Appendix A):
the concrete shall be as follow

Position Concrete
type
Design Strength
(N/mm
2
)
Quality Strength
(n/mm
2
)
Slump (cm) Max.size of coarse
aggregate (mm)
Floor Slab Plain 21 24 15
20(Crushed stone)
20 (Ballast)

1
The value presented here is the results of the net weight of 5-th floor minus parapet self weight: 508.028-
71=437.028 kN. Actually, the weight of the parapet was then added as additional distributed mass only on RG1
and RG11 beams because the central beam RG12 is not affected by it.
14

The strength of structural concrete can be the compression strength of the specimen
sampled at the factory. The difference between the strength of structural concrete and
the specimen can be considered to be F(=3N/mm
2
). The curing method of specimen is
as shown below and shall refer to the administrated age.
1) 28 days: standard water curing or site water curing
2) Over 28 days and less that 91 days: site can sealed curing

In the computer model, a nonlinear constant confinement concrete model was adopted for the
concrete of slabs. It is a uniaxial model in which a constant confining pressure is assumed
through the entire stress-strain range.
Five parameters that characterized the model were calibrated using values able to generalize
as much as possible the properties of any type of concrete, to overcome the lack of
information. For the compressive strength, the suggested value of 24MPa was assumed. The
tensile strength was considered to be equal to 1/10 of the compressive one. The strain at peak
stress, for normal strength plain concrete, usually varies between 0.002 and 0.0022mm/mm. A
default value equal to 0.002 was assumed. The constant confinement factor k
c
is used to scale
up the stress-strain relationship throughout all the strain range and is defined as the ratio
between the confined and the unconfined compressive stress of the concrete. Also in his case,
the default value of 1. was assumed. Finally, the concrete specific weight was specified.
Table 10_Concrete parameters used in the pre-test model
f
cu
cylinder compressive strength 24 MPa
f
t
tensile strength 2.4 MPa
c
strain at peak stress 0.002 mm/mm
1.0 if unconfined
k
c
confinement factor
1.2 if confined
specific weight 24 kN/m
3

3.2.2 Steel

For steel elements all experimental data regarding constitutive law were released by the
committee. For beam elements, both data about flanges and web were given. Moreover, data
regarding base plates, diaphragms and anchor bolts were given. This last set of data, regarding
detailing specifications and not really primary structural elements, was not utilized in the
computer model. The choice was to consider column bases as fully fixed to the ground,
without modelling base-plates and anchor bolts. Furthermore, the panel zone was not treated
in any detailed way: beams were connected directly to columns edge points. This choice was
taken considering that the lack of contribution to global deformation coming from panel zone
yielding could be compensated by the absence in the model of non structural elements (as for
example internal partitions) that add stiffness and reduce structural deflection
Two different kinds of steel were utilized in the specimen: SN400B for beams and BCR295
for column. The probable yield stress for SN400B is 300 MPa while for BCR295 is 380
15
MPa. Following two graphs present and example of the stress strain curve for the two kind of
steel. On the left hand side, it is possible to see the BCR295 relation, which seems to have a
more constant post-yielding behaviour, compared to the SN400B, in which hardening begins
suddenly, after an accumulation of strain at a more or less constant level of stress.
Column
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

Beam
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

Figure 14_Typical stress-block for the two kind of steel used for structural elements.
More precisely, the committee specified that for first floor columns a different kind of steel
was utilized with respect to the steel used for columns at all the other levels.
The committee specified the yielding stress
y
of for every element; the relative value of
yielding strain was red from the graph and the Elastic Modulus was calculated as
y
y
E
=
The following table is a summary of the yielding stress and strain values experimentally
determined for different type of steel and the calculated Elastic Modulus..
2

Table 11_Steel parameters derived from experimental data.
Element E (MPa)
y
(MPa)
Column 1 89351 330 0.0037
Column 2 92222 332 0.0036
Flanges
203750 326 0.0016
2G1-2G11-3G11-3G12
Web
181951 373 0.0021
Flanges
212827 308.6 0.0016
4G12
Web
221562 354.5 0.0016
Flanges
201818 333 0.0017
RG1-RG11-RG12
Web
206703 382.4 0.0019
Flanges
223482 301.7 0.0014
4G1-4G11
Web
216364 349.9 0.0017
Flanges
174765 297.1 0.0017
2G12
Web
204258 316.6 0.0016
Flanges
199551 311.3 0.0016
3G1
Web
194211 369 0.0019

From the table, it can be noticed is that yielding stress for columns is quite smaller than the
expected value, while beam flanges
3
follow more often the expected behaviour (around

2
All data regarding experimental results about material properties can be consulted in Appenix B.
16
300MPa). Furthermore, columns yielding strain is significantly bigger than beams one
(sometimes even the double). As a consequence, also columns Young Modulus E results to be
considerably far from the usual value that is 200000MPa, assuming an unexpected value
around 90000MPa.
As a conclusion, steel used in columns resulted to be softer than the one used for beams.
Another argument needs to be underline at this point. As known, the geometry of a structural
element determines its moments of inertia I
i
around the i-axis, and its stiffness is a function of
the product EI
i
.
In the following table it is possible to read the second moment of inertia of the several
elements used in the structure.
Table 12_ Moment of inertia of structural elements.
Element Ix(mm
4
) Iy(mm
4
)
COL 142,000,000 142,000,000
2G1-2G11-3G11-3G12 235,000,000 17,400,000
4G12 156,000,000 12,500,000
RG1-RG11-RG12 110,000,000 7,910,000
4G1-4G11 135,000,000 9,840,000
2G12 267,000,000 21,300,000
3G1 198,000,000 14,500,000

As can be noticed, I
x
value for columns is generally smaller than the one for beams. This
point, together with the one above regarding elements Young Modulus, started to give some
suggestions about building expected response, in particular its deformation. Beams resulted to
be stiffer than columns, letting to think that probably the majority of the deformation will
occur in the columns.
These considerations could be confirmed or denied only by the performed analysis and by the
experimental test. However, at this point, they needed to be noticed to have an idea about
what the model should be able and was expected to reproduce.
In the model, a bi-linear curve was adopted to represent the constitutive law for the steel, on
the base of the experimental data given by the committee. The aim of the bi-linearization was
to offer the best linear regression of the experimental data.
Every element, columns and girders, was characterized using its specific stress-strain curve.
Steel models used for girders were calibrated using flanges constitutive law. In fact, flanges
were considered to better represent the post-yielding behaviour, being the first part that
plasticizes in the section.
The procedure followed to obtain the bi-linear model was quite simple and based on equal
energy dissipation principle. Since the yielding point was determined by experimental results
worked on specimens, the main issue was to determine the hardening parameter, defined as

3
Flanges constitutive rule is of particular interest for the modelling (see what follows).
17
the ratio between the initial elastic modulus and the post-yielding modulus. The starting point
was to determine for which amount of strain the parameter needed to be defined because this
could generate different values of post yielding modulus. The assumption was to determined
the plastic strain required by girders when the building equivalent single-degree of freedom is
subjected to spectral acceleration.
An Eingenvalue Analysis was performed to determine the building elastic natural period of
frequency. For this type of analysis, the program considers that the elastic properties of
elements remain constant during all the procedure and hence it requires just the specification
of sectional mechanical properties, not of material and section type.
The following table presents the data utilized to characterize the elastic frame elements of the
model.
Table 13_Elastic frame elements properties.
Element
Cross
Section area
(m
2
)
E
(kN/m
2
)
Ix (m
4
) Iy (m
4
)
G
(kN/m
2
)
J (m
4
)
Own
mass
(tonnes)
EA (kN)

EIx
(kNm
2
)
EIy
(kNm
2
)
GJ
(kNm
2
)
Applied
Mass
4

(tonnes/m)
COL 0.01020 8.94E+07 0.00014 0.00014 3.44E+07 2.83E-07 0.7956 911380 12688 12688 9.7204 0.7956
2G1 0.00834 2.04E+08 0.00024 0.00002 7.84E+07 3.57E-07 0.6503 1698664 47881 3545 27.9578 1.3743
3G1 0.00714 2.00E+08 0.00020 0.00001 7.68E+07 2.19E-07 0.5570 1424994 39511 2893 16.8344 1.2850
4G1 0.00629 2.23E+08 0.00014 0.00001 8.60E+07 1.93E-07 0.4907 1405919 30170 2199 16.5707 1.2477
RG1 0.00525 2.02E+08 0.00011 0.00001 7.76E+07 1.08E-07 0.4091 1058535 22200 1596 8.3972 1.5641
2G11 0.00834 2.04E+08 0.00024 0.00002 7.84E+07 3.57E-07 0.6503 1698664 47881 3545 27.9578 1.4943
3G11 0.00834 2.04E+08 0.00024 0.00002 7.84E+07 3.57E-07 0.6503 1698664 47881 3545 27.9578 1.4993
4G11 0.00629 2.23E+08 0.00014 0.00001 8.60E+07 1.93E-07 0.4907 1405919 30170 2199 16.5707 1.3737
RG11 0.00525 2.02E+08 0.00011 0.00001 7.76E+07 1.08E-07 0.4091 1058535 22200 1596 8.3972 1.7181
2G12 0.01012 1.75E+08 0.00027 0.00002 6.72E+07 6.65E-07 0.7894 1768622 46662 3722 44.7309 2.4784
3G12 0.00834 2.04E+08 0.00024 0.00002 7.84E+07 3.57E-07 0.6503 1698664 47881 3545 27.9578 2.3483
4G12 0.00785 2.13E+08 0.00016 0.00001 8.19E+07 4.42E-07 0.6125 1671338 33201 2660 36.1548 2.3785
RG12 0.00525 2.02E+08 0.00011 0.00001 7.76E+07 1.08E-07 0.4091 1058535 22200 1596 8.3972 2.5741

Where:
E = elastic modulus
I
x
= moment of inertia around the major axis
I
y
= moment of inertia around the minor axis
G = shear modulus
J = modulus of rigidity

4
It was obtained as the sum of the own mass and the applied one.
18

Figure 15_3D model used for eigenvalue analysis
Form the analysis, the first mode period of vibration is equal to sec 30 . 1 = T .
The spectral acceleration of the equivalent single degree of freedom was then found from the
North-South (x-direction in the model) response spectrum
5
spread by the committee.

NS Component Acceleration Spectrum
0
5
10
15
20
25
30
35
0.1 1 10
T sec
S
a

m
/
s
2

Figure 16_North-South component acceleration spectrum.
From the spectrum,
2
/ 06 . 25 s m S
a
= .
Since the pre-test analysis is conducted using a load whose intensity is scaled at 60% of its
original intensity, the scaled spectral acceleration was determined.
2
60
/ 034 . 15 s m S
a
=
The equivalent SDOF own circular frequency of vibration was defined as

5
Only the North-South spectrum was considered for simplicity of the procedure.
19
sec
8489 . 4
2 rad
T
= =

and its height was assumed to be
m H H
top eff
13 . 10 475 . 14 7 . 0 7 . 0 = = = .
At this point, spectral displacement could be find: m
S
S
a
d
639 . 0
2
60
60
= =
.
From the displacement, it was possible to determine the SDOF-base rotation, defined as
rad
H
S
eff
d
TOT
06312 . 0
13 . 10
639 . 0
60
= = =
At this point, it was possible to determine the plastic deformation required to the several
structural elements by the applied acceleration.
TOT
is the column rotation, and it is given by the sum of an elastic (
y
) and a plastic (
p
)
component:
p y TOT
+ =
As known,
y
is the column rotation at yielding point. Before it was underlined as beams are
quite stiffer than columns. This allows (for the purpose of this calculation) to consider
columns as fully fixed at both edges, and hence their deformed shaped under lateral loads
could look like in the figure below.

From elasticity relations, it was possible to know
- the displacement at column mid-height
12 3
2
2
2 /
2
H
H
H
y y

=
=
- the yielding curvature in the column section
2 /
c
y
y
h
=
- the yielding rotation in the column

6 2 /
H
H
y
y

= =

Hence, yielding rotation in the column resulted to be:
6
2 /
H
h
c
y
y

20
Where
= displacement at column mid-height
H = column height
y
= yielding curvature of column section
y
= yielding strain of column steel
h
c
= column cross section width
y
= column yielding rotation
H and h
c
were already determined by the geometry of the problem and
y

was known from
column stress-block. Columns yielding rotation was then calculated.
Table 14_Columns yielding rotation.
H (m) h
c
(m)
y

y
(rad)
col 1 3.875 0.30 0.0037 0.01593
col 2 3.5 0.30 00036 0.014

Colum plastic strain could be calculated from its plastic rotation.
y TOT p
= and
p p p
L =
2 /
c
p
p
h
=
2
c
p
y TOT
p
h
L

=

Where
p
= column plastic rotation
p
= plastic curvature of column section
L
p
= plastic hinge length (an average value of 0.5 x column width was assumed)
p
= column max plastic strain
Table 15_Column plastic strain.

p
(rad) L
p
(m)
p

col 1 0.04719 0.15 0.04719
col 2 0.04912 0.15 0.04912
Once the level of strain was known, the second branch of the bi-linear curve could be
determined. The slope was determined such that there was equal area under both the
experimental and bi-linear relations between the yield point and the strain request.
21
Column
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

Figure 17_Bi-linear constitutive model for column steel.
Considering that the height of the building is quite small, a linear distribution of the
displacements with height was adopted. The table below presents displacements calculated in
North-South direction for every level, starting from the equivalent SDOF system
displacement
60 d
S , through the relation:
eff
d i
H
hi
S u
60
=
Where
u
i
= lateral displacement at i-th level
h
i
= i-th level height
H
eff
= equivalent SDOF system height
S
d60
= equivalent SDOF lateral displacement
Table 16_Linear distribution of displacement.
Level height (m) displ(m)
1 3.500 0.221
2 7.000 0.441
3 10.525 0.664
4 14.052 0.886

The target rotation at the column base is the one explained before, while the rotation at the
beam ends can be estimated from this through the geometry.
p c b
b c
b
l h l
l

=

Where:
b
= is the rotation at the beam end
c
= is the rotation at the column base
l
b
= is the length of the beam
22
h
c
= is the column width
l
p
= is the beam plastic hinge length (an average value of 0.5 x the depth of the beam was
assumed).
Hence, through the relations:
p
b
p
l
= and
2 /
b
p
p
h
=
Where:
p
= is the strain request
p
= is the curvature of the beam section
h
b
= is the width of the beam
it was then possible to estimate the strain request for every girder.
Table 17_Plastic strain requested by different girders.
Girder Span(m) h
b
(m) l
p
(m)
b
(rad) b
(m
-1
)
p

2G1 5 0.4 0.18 0.0698 0.3879 0.078
2G11 6 0.4 0.18 0.0686 0.3812 0.076
2G12 6 0.39 0.1755 0.0686 0.3906 0.076
3G1 5 0.396 0.1782 0.0698 0.3917 0.078
3G11 6 0.4 0.18 0.0686 0.3812 0.076
3G12 6 0.4 0.18 0.0686 0.3812 0.076
4G1 5 0.35 0.1575 0.0695 0.4411 0.077
4G11 6 0.35 0.1575 0.0683 0.4338 0.076
4G12 6 0.34 0.153 0.0683 0.4462 0.076
RG1 5 0.346 0.1557 0.0694 0.4460 0.077
RG11 6 0.346 0.1557 0.0683 0.4387 0.076
RG12 6 0.346 0.1557 0.0683 0.4387 0.076

2G1- 2G11 - 3G11 - 3G12 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

4G12 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

23
RG1 - RG11 - RG12 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

4G1 - 4G11 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

2G12 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

3G1 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

Figure 18_Bi-linear constitutive model for different steel used for beams.
The following table presents characteristic parameters (Elastic Modulus, yielding stress and
hardening parameter) obtained after bi-linearization for each set of elements and used to
model different steel types constitutive rule.
Table 18_Steel bi-linear constitutive model characteristic parameters.
Element E (MPa)
y
(MPa) r
Column 1 89351 330 0.05
Column 2 92222 332 0.05
2G1-2G11-3G11-3G12 203750 326 0.12
4G12 212830 308.6 0.1
RG1-RG11-RG12 201820 333 0.15
4G1-4G11 223480 301.7 0.13
2G12 174770 279.1 0.2
3G1 199550 311.1 0.18

3.3 Sections

Columns were modelled using Rectangular Hollow section, characterized by steel modelled
as explained before. Girders were modelled using Composite I section for which three
materials had to be defined: the steel for the profile, the concrete for the cover and the
confined concrete
6
. Moreover, to use this type of section some information about slab
thickness and its effective width needed to be input. SeismoStruct does not allow the
modelling of a slab with a shape different from the flat one. For this reason, for girders at
second, third and forth level, it was necessary to define an equivalent slab thickness to model

6
Definition of material was explained in paragraph 3.2.
24
a flat slab able to give the same contribution to the compression strength given by the real
slab.
Taking in consideration that the slab spans in x-direction, girders were characterized with
different values of equivalent slab thickness, depending from their spanning direction.

Figure 19_Deck slab thickness geometry.
For beam spanning parallel to the slab, the contribution to compression strength given from
this last was considered to derive from and average thickness obtained as:
m
h
h H t
s
1375 . 0
2
1
= + =
While, for girders spanning in y-direction, only the thinner part of slab was considered as
effectively able to contribute to beam compression strength.
m h H t 1 . 0
2
= =

To use this type of section it was also necessary to define the effective width of slab that
collaborates with the steel girder in steel-concrete composite action.
Different Codes present different formulae to calculate slab width. To decide which one to
follow for the modelling, a comparison between two of them was performed to demonstrate
that results are not really conditioned by this parameter. The internal girder 2G11, which
spans in y-direction for 6 meters, was taken as example. The slab effective width in this case
resulted to be equal to 0.1m.
Two codes were then compared: LRFD (Load and Resistance Factor Design) from AISC
(American Institute of Steel Construction), Second Edition 1994, and the New Zealand Code
3101.
The LRFD requires that:
+ =
ei eff
b b b
0
and
2
i
ei
b
b <
Where
b
0
= distance between the centres of the outstand shear connectors (0 in this case because shear
connectors lie in one line. See figure 6)
25
b
ei
= value of the effective width of the concrete flange on each side of the web, taken as L
e
/8
but not greater than the geometric width b
i
(5m in this case). L
e
should be taken as the
approximate distance between points of zero bending moment. The code specifies that, for
multi-span beams, L
e
can be considered to be equal to 0.7L = 4.2m, where L in the span
length. In this case, the beam is a single-span, so L
e
has a greater value. However, L
e

cannot be considered equal to L because at beam edges there is a moment resistance given
by the connection with the column. To mediate between these two limit conditions L
e

was considered to be equal to 5m.
For LRFD provisions, slab effective width results to be equal to m
L
e
625 . 0
8
5
8
= =
New Zealand Code, for fully composite, completely connected beams, provides the effective
width to be:
4
L
t b
w eff
+ = and
) ( ) (
) (
'
) (
8
2 1
1
s b s b
s b
eff
s b eff
s eff
t d t d
t d
L b
t d b
t b
+ + +
+
+ <
+ <
<

Where
L = span of the beam, 6m.
t
w
= web thickness, 8mm.
t
s
= slab thickness, 0.1m.
d
b
= beam width, 400mm.
L = distance between two adjacent beams, 5m.
In this case, the most restrictive condition resulted to be ) (
s b eff
t d b + < which forced the
effective width to be equal to 0.5m.
It can be noticed that, in this case, slab effective width results to be 20% smaller than the one
given by American Code.
A Static time-history analysis was performed to verify if this difference could condition the
results. The same system was modelled using both slab widths, maintaining constant all other
characteristics (steel beam and column properties, span, etc) and subjected to the same
generic displacement time-history, applied in vertical direction to the beam free edge.
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
0 2 4 6 8 10 12 14 16 18 20
time (sec)
d
i
s
p
l
a
c
e
m
e
n
t
(
m
)

Figure 20_Applied displacement time history.
26

Figure 21_Model of the beam with the applied displacement. The same analysis was performed for both
slab width.

The following plot compares the same hysteretic curve resulting from the two cases. On the
horizontal axis there is the displacement of the free end while on the vertical axis it is possible
to read the total base shear.
Displacement-Total Base Shear
-250
-200
-150
-100
-50
0
50
100
150
200
250
300
-0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04
displacement (m)
S
h
e
a
r

(
k
N
)
NZ Code
LRFD

Figure 22_Hysteretic curves for the two beams with different slab effective width.
It can be noticed as the difference between the two results is negligible, as expected, and
hence every code provision could be followed without changing significantly the results.
It was decided to follow the New Zealand Code.
The following table summarizes the input data used for slab width and thickness in the model
for every beam.

27
Table 19_Effective slab width and thickness for different beams.
Element b
eff
(m) t
s
(m)
2G1 0.575 0.1
2G11 0.575 0.1375
2G12 1.15 0.1375
3G1 0.571 0.1
3G11 0.575 0.1375
3G12 1.15 0.1375
4G1 0.525 0.1
4G11 0.525 0.1375
4G12 1.05 0.1375
RG1 0.496 0.15
RG11 0.496 0.15
RG12 0.992 0.15

3.4 Connections

Preliminary cyclic loading tests of composite beam and column were performed to determine
joints deformation. The testing set-up is presented in what follows.
3.4.1 Cycling loading test of composite beam

Figure 23_Composite beam cycling loading testing set-up.
28

Figure 24_Drawings of the specimen.
A 2.5m long cantilever beam was tested by applying a cyclic load at its free end. As a
consequence, deformation of the beam itself and of the panel zone was produced and
measured. The following scheme represent parameters involved in the experiment.

Figure 25_Testing results scheme.
The organizing committee spread the rotation angle of the beam b time-history.

Figure 26_Rotation angle b imposed on the beam.
Q: applied force
bM=Q x bL: moment at the beam end
bL: length of the beam
b=[t-(fxbL+c)]/bL: rotation angle of the beam
c: vertical displacement of the beam end
pM=bM-cHxbd: panel moment
cH=Qx(bL+L)cL: shear force of column
L: horizontal distance between beam end and pin support
cL: vertical distance of pin support
bd: distance between centre of flange
p: deformation angle of the panel zone
Rotation Angle
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
0 5 10 15 20
time (sec)
a
n
g
l
e

(
r
a
d
)
29
Together with this, also the hysteretic curves representing the moment at the end of the beam
bM versus the rotation angle of the beam b and the panel zone moment pM versus the
deformation angle of the panel zone p were given.

Figure 27_Testing results.
It results quite clear how these data do not allow the construction of a computer model able to
reproduce the testing set up. Actually, b is just one of the total rotation components that
interest in the system. There are also p, whose values could eventually be derived from the
hysteretic curve, and f, that is the rotation angle at the beam end, whose values are unknown
and cannot be derived from the given data. As a consequence, a rotation-time history cannot
be applied in the model. For the same reason, a the free-end displacement time history cannot
be derived and imposed on the beam. Actually, imposing the effective force time-history at
the free end of the beam could represent the easiest and most precise trial of reproducing what
was done in the laboratory. Without knowing it, an adequate computer model could not be
worked out.
3.4.2 Cycling loading test of column
As for the beam, the set-up for the test conducted on the steel column was spread.

Figure 28_ Column cycling loading testing set-up.
-1000
-800
-600
-400
-200
0
200
400
600
800
1000
-0.1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06
bQ(rad)
b
M

(
k
N
m
)

-1,000
-800
-600
-400
-200
0
200
400
600
800
1,000
-0.06 -0.04 -0.02 0 0.02 0.04 0.06
pg (rad)
p
M

(
k
N
m
)

30
Together with this, a scheme of what was realized was given.

Figure 29_Explanatory scheme spread by the committee.
At this point, it is clear that some discrepancies happened between the test set-up and the
scheme that aimed to reproduce it. In the scheme, the point indicated as A is supported in the
horizontal plane. This appears to be in clear contrast with what is represented in the test set-
up, where A is the point at which the horizontal force is applied and should be able to move
laterally because there are no other elements that restrain it. It could be assumed that figure 28
presents a mistake and that in reality A is the upper end of the whole vertical element.

Figure 30_Hypothesis of new testing set-up.
However, also this interpretation results to be meaningless because the column specimen RHS
300x9.0 is 1468mm long and ends clearly where the lateral force is applied. The rest of the
length is part of the laboratory instrumentation. For this reason, figure 29 should not represent
an unique element, with the same characteristics, spanning between A and B. The deduction is
that point A is correctly indicated in the figure that represents the test set-up. One more issue
becomes clear: since A is the point of application of the vertical and the lateral forces and is
supposed to move in the horizontal plain, figure 29 should not show a rotation angle at its
location.

These fundamental inconsistencies generates the impossibility of a clear and unique
interpretation of the given data and hence their reproduction through a computer model.
31
3.4.3 Modelling attempt
Despite the consciousness that analytical models for the whole testing set-up could not be
created because of several problems related to the interpretation of the given data,
experimental test results related just to the specific part of panel zone and column base
connection were reproduced, considering them as isolated from the rest of the testing system.
To do this, two rotational spring elements, one for each part, were implemented within the
software. The figure below shows the model set-up, in which two coincident nodes are
connected by a spring that is subjected to an imposed rotation. Rotation time-histories for the
panel zone and the base column deformation were derived from given hysteretic curves.
Each spring was characterized using a Modified Richard-Abbot model, an hysteretic loop able
to model all sorts of steel and composite connections. The model is characterized by thirty
parameters, fifteen for positive branches and fifteen for negative ones. Anyway, since a
symmetrical behaviour was assumed, the second set of parameters was assumed identical to
the first one.
Input parameters:
- Ka=Kd: initial stiffness for the upper bound curve. Typical value may range from
15000 to 50000kNm/rad.
- Ma=Md: strength for the upper bound curve, normally equal to 1.1-1.3 of the yielding
moment capacity of the connection. Its values may typically range from 75 to 250
kNm.
- Kpa=Kpd: post elastic stiffness for the upper bound curve. Typical value ranges from
0.02Ka (Kd) to 0.05Ka (Kd).
- Na=Nd: shape parameter for the upper bound curve. Its value is normally assumed as
equal to 2.
- Kap=Kdp: initil stiffness for the lower bound curve. Its value is normally assumed
equal to Ka (Kd).
- Map=Mdp: strength for the lower bound curve. Its value may typically range from
0.45Ma (Md) to 0.65 Ma (Md).
- Kpap=Kpdp: post-elastic stiffness for the lower bund curve. Its value is normally
assumed as equal to Kpa (Kpd).
- Nap=Ndp: shape parameter for the lower bound curve. Its value is normally assumed
as equal to 1.
- t1a=t1d: empirical parameter related to the pinching. Its value may typically range
from 5 to 20.
- t2a=t2d: empirical parameter related to the pinching. Its value may typically range
from 0.15 to 0.5.
- Ca=Cd: empirical parameter related to the pinching. Its value is normally assumed
equal to 1.
- iKa=iKd: empirical coefficient related to the stiffness damage rate. Its value may
typically range from 3 to 25.
- iMa=iMd: empirical coefficient related to the strength damage rate. Its value may
typically range from 0.01 to 0.1.
- Ha=Hd: empirical coefficient defining the level of isotropic hardening. Its value may
typically range from 0.01 to 0.04.
- Emaxa=Emaxd: maximum value of deformation (rotation) reached in the loading
history.
32
.
Table 20_Input parameters used in the hysteretic model.
Positive Branches Negative Branches
Panel Zone Column Base Panel Zone Column Base
Ka 258337 49740 Kd 258337 49740
Ma 600 450 Md 360 450
Kpa 7750 0 Kpd 7750 0
Na 2 2 Nd 2 2
Kap 258337 49740 Kdp 258337 49740
Map 416 150 Mdp 279 150
Kpap 7750 0 Kpdp 7750 0
Nap 1 1 Ndp 1 1
t1a 5 5 t1d 2 5
t2a 0.15 0.15 t2d 0.15 0.15
Ca 1 0.5 Cd 1 0.5
ika 15 0 ikd 0.01 0
ima 0.2 0.25 imd -0.035 0.25
Ha -0.04 0 Hd 0 0
Emax 0.0472 0.05 Emax 0.009 0.05

Figure 31_Springs calibration results.
These parameters should be calibrated when considered to be part of a specific system, where
physical elements (beams/columns) are subjected to loads and show their own cyclic
response. Spring elements should be defined dependently from this. Unfortunately, because of
all the reasons that were explained in previous two paragraphs, the real situation of the
experiment could not be reproduced. As a consequence, all these parameters were not
calibrated on the base of engineering knowledge and judgement but merely trying to emulate
what was represented on the given curves.
Results obtained from springs responses were quite close to the experimental ones.
Nevertheless, at this point it was considered if this difficult calibration of an advanced joint
model could be necessary for a reasonable accurate overall modelling of the structure. The
decision was to try to maintain the model simple, neglecting the deformation given by panel
Panel Zone Cyclic Response
-1,000
-800
-600
-400
-200
0
200
400
600
800
1,000
-0.06 -0.04 -0.02 0 0.02 0.04 0.06
pY (rad)
p
M

(
k
N
m
)
Experimental
Modelling

Base Column Cyclic Response
-500
-400
-300
-200
-100
0
100
200
300
400
500
-0.06 -0.04 -0.02 0 0.02 0.04 0.06
rotB (rad)
M
B

(
k
N
m
)
Experimental
Modelling
33
zone yielding and column base buckling, assuming that they could be somehow
compensated by the equal absence of modelling of certain sources of stiffness (i.e. non-
structural walls).
3.5 Element Classes

Beam and columns were modelled as Inelastic frame elements infrm-, that are 3-
dimensional elements characterized both by geometric and material inelastcity. Bi-linear
material uniaxial response, defined in 3.2.2, is the particular response of every individual
fibre in which elements cross section is divided. In this model, a number of 200 fibres were
used for every cross section, because quite an high level of inelasticity is expected to be
required to the elements and hence a more accurate analysis should to be performed.
The program integrates the uniaxial material response to obtain the sectional stress-strain state
of the element, accounting for the spread of inelasticity along the element length and across
the section depth.
In infrm elements also local geometric non-linearity, deriving from the interaction between
axial force and transverse deformation of the frame, is included (beam-column effect).
3.6 Damping

In infrm elements formulation, hysteretic damping is already taken in consideration by
material inelasticity that characterizes fibres. As known, in dynamic non-linear analysis, this
kind of damping represents the dissipation of the greatest part of energy induced in the
building by earthquake force. Another kind of damping was considered in the analysis: this is
non-hysteretic damping, able to represent the dissipation of a small amount of energy
mobilized during dynamic response of structures through phenomena such friction between
structural and non-structural members, friction in opened concrete cracks, energy radiation
through foundation.
This kind of damping was defined globally for the whole structure and was assumed to be
tangent stiffness-proportional. The choice of not considering the proportionality between the
damping and the mass was taken on the base of the fact that, for building that do not follow a
rigid body motion, it could generate unrealistic energy dissipation. Moreover, the fact of
considering a tangent stiffness proportionality came from several considerations. Initial
stiffness is related to the cracking of the structure and hence it can be considered to give a
realistic response only until the crack really happens. After this point, the hysteretic behaviour
of the building given by the initial stiffness would be overestimated, in the sense that the
actual value of stiffness is decreased and, as a consequence, the amount of energy dissipated
in every cycle is smaller. Tangent stiffness is linked to the point of yielding in a structure and
hence has a smaller value with respect of initial stiffness. This means that, until cracking does
not occur in the structure, and hence at the very beginning of inelastic response, this stiffness
under-esteems energy dissipation but after this it can give a better representation of the same.
In addition to this, it was considered that this structure in mainly in steel. The concrete is
34
present only on slabs and hence the point of cracking would be still less representative for the
whole structure with respect of the point of yielding.
Considering that the analysis that had to be performed was expected to be highly inelastic, for
its numerical stability a small amount of equivalent viscous damping, set equal to 0.5%, was
introduced in the model.
Stiffness matrix multiplier was typically calculated as
T
k
=
Where
T =1.35sec is the first-mode period of vibration of the structure (see 3.2.2)
= equivalent viscous damping= 0.05
In the software, the tangent-stiffness proportional damping matrix is updated at every load
increment.
3.7 Constraints

Slabs were modelled as rigid diaphragms with no bending stiffness out of plane. At every
levels, nodes were linked between them with rigid links in XY plane.
3.8 Integration Scheme

The Hilber-Hughes-Taylor algorithm was used was used as method of time integration. It is
an implicit single step integration algorithm, which requires only the previous step solution to
get the new solution, solving simultaneous equations to get the new step. It calls for the
characterisation of three parameters: , and . The parameter alpha () is used to control the
level of numerical dissipation. It can play a beneficial role in dynamic analysis, mainly
through the reduction of higher spurious modes' contribution to the solution (which typically
manifest themselves in the form of very high short-duration peaks in the solution), thus
increasing both the accuracy of the results as well numerical stability of the analysis.
According to its authors [Hilber et al., 1977], optimal solutions, in terms of solution accuracy,
analytical stability and numerical damping are obtained for values of
2
) 1 ( 25 . 0 = , for
which the integration scheme is unconditionally stable, and = 5 . 0 , with 0 3 / 1 ,
avoiding problems related to artificial damping that can be built in to the method.
Charter 4 . The Laboratory Test
35

4 THE LABORATORY TEST
4.1 Ground Motion

The building was shaken and collapsed by applying a scaled version of the near-fault motion
recorded in Takatori during the 1995 Kobe earthquake.
More 3D shaking table tests were performed consecutively with increasing levels of seismic
motion to evaluate the effect of plastic deformation:
- Takatori scaled to 40% (elastic level)
- Takatori scaled to 60% (incipient collapse seismic level, elasto-plastic)
- Takatori in full scale (collapse seismic level)
All result data (e.g. interstory drift, floor displacement) have to be submitted for incipient
collapse level while the collapse level was used to evaluate the time of collapse (defined as
the time at which the interstory drift angle, both in x or y-direction, reach 0.13 or -0.13 rad).
It was required to predict the response before and after the tests.
Pre-test analysis was conducted for an incipient collapse level, immediately prior to the
building collapse. For this seismic level, the recoded ground motion intensity was scaled to
60%. Graphs below present the three components of ground motion time history used for pre-
test analysis.
36
NS-Component
-8
-6
-4
-2
0
2
4
6
8
0 10 20 30 40
t ime (sec)
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
e
c
2
)

Figure 32_60% scaled NS ground acceleration time history.
EW-Component
-8
-6
-4
-2
0
2
4
6
8
0 10 20 30 40
t ime (sec)
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
e
c
2
)

Figure 33_60% scaled EW ground acceleration time history.
Vertical Component
-8
-6
-4
-2
0
2
4
6
8
0 10 20 30 40
time (sec)
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
e
c
2
)

Figure 34_60% scaled vertical ground acceleration time history.
Table 21_Peak ground acceleration for 60% scaled ground motion.
Component PGA (m/sec2)
North-South 3.633
East-West 3.942
Vertical 1.676

It was required to apply the North-South component parallel to the longest side of the
building and East-West component parallel to the shortest one.
37

Figure 35_3D model with dynamic loads applied.
For post- test response acceleration records of the shaking table were used. Time history
analysis was carried out for all three ground motion intensities consecutively. In the model,
ten seconds of delay were left between one motion and the one immediately after. In this
lapses of time the building was expected to move in free vibration.
NS Component
-10
-8
-6
-4
-2
0
2
4
6
8
0 20 40 60 80 100 120
t ime (sec)
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
e
c
2
)

Figure 36_Three consecutive NS components.
EW Component
-10
-8
-6
-4
-2
0
2
4
6
8
0 20 40 60 80 100 120
time (sec)
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
e
c
2
)

Figure 37_Three consecutive EW components.
38
VerticalComponent
-10
-8
-6
-4
-2
0
2
4
6
8
0 20 40 60 80 100 120
t ime (sec)
A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
e
c
2
)

Figure 38_Three consecutive vertical components.
Table 22_PGA for different ground motion intensities.
Intensity 40% 60% 100%
Component PGA (m/sec2) PGA (m/sec2) PGA (m/sec2)
North-South
3.028
4.6576
8.5657
East-West
3.327
5.4506
8.4916
Vertical
1.4285
2.3481
3.8481

If the 60% intensity PGA in different directions are compared with the respective one used for
pre-test analysis (see table 20), it is possible to notice that value experimentally obtained with
shaking table are considerably bigger.
4.2 Measurement during the experiment

Following pictures show the way in which floor accelerations and displacements were
measured during the laboratory test. As can be noticed, different transducers were positioned
inside the specimen. For acceleration recording, the transducer was positioned exactly in the
middle point of the top surface of every slab and ground floor. For displacements, two laser
transducer were positioned at each level to measure x-direction (north-south) displacement
and one transducer for y-direction measurement. Their positions have different eccentricity
from the geometrical centre of the building level.

39
Figure 39_Transducers position at 1-st floor(right), 2-nd and 3-rd floor (left). Plan. (unit: mm)

Figure 40_4-th floor and roof transducers position. Plan. (unit: mm)

Figure 41_Transducers position. Sections. (unit: mm)
4.3 Concrete Characteristics Measured during the test

As explained in paragraph 3.2, the committee did not spread any stress-strain relation for
concrete used for slabs because they had the intent to measure it during the experimental test.
After this happened, precise concrete characteristics were known and used for post-test blind
prediction.
40
Concrete at 2-nd Floor
0
10
20
30
40
0 1000 2000 3000 4000 5000 6000 7000
Strain (10-6)
S
t
r
e
s
s
(
M
P
a
)

Concrete at 3-rd Floor
0
10
20
30
40
0 1000 2000 3000 4000 5000 6000 7000
Strain (10-6)
S
t
r
e
s
s
(
M
P
a
)

Concrete at 4-th Floor
0
10
20
30
40
0 1000 2000 3000 4000 5000 6000 7000
Strain (10-6)
S
tr
e
s
s
(
M
P
a
)

Conctere at Roof Floor
0
10
20
30
40
0 1000 2000 3000 4000 5000 6000 7000
Strain (10-6)
S
t
r
e
s
s
(
M
P
a
)

Figure 42_Concrete stress-block for different floor slab.
Table 23_Characteristic parameters for concrete.
Floor f
cu
(MPa) f
t
(MPa)
c

2-nd 33.53 3.35 0.00154
3-rd 28.8 2.88 0.0012
4-th 27.54 2.754 0.00127
Roof 35.9 3.59 0.002467

These new parameters were used in post-test analysis performed. What need to be noticed is
that all of them have a greater value with respect of the ones used in pre-test analysis. In
particular, strength values are even 67% (roof concrete) bigger than the value assumed before
(24 MPa). Taking this in consideration, post-test blind prediction is expected to give smaller
value of building deformation, due to concrete strength that in pre-test was underestimated
7
.

7
In the post-test model, values of confinement factor k
c
and concrete specific weight were kept as in the pre-
test model: k
c
= 1.0 for unconfined concrete, k
c
=1.2 for confined concrete, =24 kN/m
3
.
Charter 5 . Blind Prediction and Experimental Results
41

5 BLIND PREDICTION AND EXPERIMENTAL RESULTS
5.1 General remarks

For 3D analysis the response in both x- and y-direction had to be obtained.
The committee decided that the absolute displacement u
j
(t) and the absolute acceleration
j

(t) of the j-th floor had to be evaluated on the upper surface at the centre (x ; y)= (3000;5000)
of the floor slab (see figure 35). For absolute they meant the value measured with respect to
the global coordinate , as shown in figure 36.

Figure 43_Definition of floor and story (unit:mm)
42

Figure 44_Gloobal coordinates.
The values corresponding to j = 5 are the responses on the upper face of roof and the values
corresponding to j = 0 are the responses on the face of the shaking table.
8

The time t is measured from the beginning of the analysis at each seismic level and the base
acceleration records were considered to be absolute acceleration on the face of the shaking
table.
Both for pre-test and for post-test analysis, data to be submitted for incipient collapse level
were:
Maximum value of relative displacement from the base at each floor
The relative displacement from the base at j-th floor was obtained by
) ( ) ( ) (
0
t u t u t d
j j
=
The maximum value in time history was defined as ) ( max t d
j
j
, j = 2,,5.
Maximum value of drift angle in each story
The drift angle in the i-th story was obtained by
i
i i
i
h
t u t u
t R
) ( ) (
) (
1

=
+

Where
h
i
= story height
The maximum value in time history was defined as ) ( max t R
i
i
, i = 1,,4.
Residual drift angle in each story

8
In the model as level j = 0 was considered the base of the building, on the to of 1500mm high concrete block.
Actually, these blocks were considered to be so stiff that their presence could not create any different interaction
between the building and the shaking table/ground. For this reason their presence was neglected.
43
The residual drift angle in the i-th story was obtained as the average value between the
absolute values at local maximum and local minimum at the end of the base acceleration
record.

Figure 45_Definition of residual drift.
Maximum value of absolute acceleration at each floor
The maximum value of absolute acceleration in time history was given by
) (
..
max t u
j
j
, j = 2,5.

Maximum value of story shear ay each story
The story shear from the experiment was calculated as follows. The inertia force at the j-
th floor was obtained as
) ( ) (
..
t
j
g
W
t F
u
j
j
=
Where
g = acceleration of gravity (9.81 m/sec
2
)
W
j
= floor weight (kN)
The story shear of the i-th story was obtained as
+ =
=
5
1
) ( ) (
i j
j i
t F t Q
For which the maximum value in time history was give by ) ( max t Q
i
i
, i = 1,,4.
44
Maximum value of overturning moment at each floor
The overturning moment at j-th floor from the experiment was calculated as follows

+ = =
=
= =
5
1
4 1
) ) ( ( ) ) ( ( ) (
j k j i
i i
k
j i
i k OVTj
h t Q h t F t M
for which the maximum value in time history is given as ) ( max t M
j OVT
j
, j = 2,,5.

Maximum strain at a specified point in elastic range
The maximum of the absolute value of the strain in time history had to be evaluated at a
specific point (centre of a certain column in the first story) that has to be assumed to be
in the elastic range. The stain to be reported had to be the one due to the dynamic
response, that is, it does not contain the initial strain due to dead loads.
In the shaking-table test, the strain was measured using 588 strain gauges (gauge length
was 5mm) positioned in the whole specimen and in particular in the column surface. For
evaluation of pre-test analysis, strain gage was located on the surface seen from the
direction of the arrow in figure 38 in column 1A, in its lower section.

Figure 46_Position of the strain gage of column 1A.
The location of the gage for evaluation of post-test analysis was changed. It was still
maintained in column 1A but translated from internal to external surface.
45

Figure 47_Strain gage position Figure 48_Strain gauges on the specimen.
for post-test analysis.

For collapse level, and hence only for post-test analysis, the time of building collapse needed
to be computed. This time was defined as the time, from the beginning of the collapse-level
motion, at which the drift angle at any story in x- or y-direction reaches 0.13 or -0.13rad.
5.2 Comparison of the results

Following graphs present comparison of measured results with blind prediction results both
from pre-test and for post-test analysis (see figure1), for incipient collapse level (60% level of
Takatori record). Graphs present also results obtained by the best three teams in the two
categories of the contest (researchers and practicing engineers) to better visualize gaps
between different predictions.
The method of judgement adopted by the Verification Working Group to established the best
predictions was based on the RMS (Root Mean Squared) error, computed from
=
j
j i j i i
F F E
2 *
, ,
) (
Where
F
i,j
= analysis result of i-th response quantity at j-th floor
*
, j i
F = test result of i-th response quantity at j-th floor
E
i
= RMS error of i-th response quantity
In this context, the main purpose was to demonstrate the accuracy of computer analysis in
predicting nonlinear dynamic response of steel buildings. For this reason the error in the
prediction was evaluated as a percentage of the same measured quantity. After each couple of
graphs, that present the comparison for the same parameter in two directions, there is a table
which shows the percentage of error. A result was considered to be acceptable if its
percentage of error was smaller than 20%.
46
This limit value was fixed taking in consideration values of safety coefficients to be applied in
the design. In fact, with an error smaller than 30% from the dynamic analysis and safety
coefficients applied, the design can be considered to be conservative.
5.2.1 Maximum value of relative displacement from the base at each floor

MaximumRelative Displacement NS-Direction (mm)
1
2
3
4
5
0 50 100 150 200 250 300 350 400 450
F
l
o
o
r
measured
pre-test
post-test
best results
6

MaximumRelative Displacement EW-Direction (mm)
1
2
3
4
5
0 50 100 150 200 250 300 350 400 450
F
l
o
o
r
measured
pre-test
post-test
best results

Figure 49_Maximum relative displacements at every level in two directions.

Figure 50_Deformed shape of the pre-test analysis model (out of scale). From left hand, in North-South
direction, East-West, perspective view and detail of the first floor beam-to-column connection.

Figure 51_Specimen collapse obtained in the laboratory with a table motion equal to 100% level of
Takatory motion.

47

Figure 52_Plastic hinges formation. From left, in y=0m frame, y=6m, x=0m, x=5m and x=10m.
Table 24_Error evaluation in predicting relative displacements in different analyses.
TEST LEVEL Displ. NS (mm) error % Displ. EW (mm) error %
1 (GL) 0 0
2 75 47
3 135 95
4 177 134
MEASURED
5 (Roof) 200 157
1 0 0
2 128 71% 105 123%
3 218 61% 185 95%
4 282 59% 247 84%
PRE TEST
5 321 61% 292 86%
1 0 0
2 171 128% 114 143%
3 259 92% 175 84%
4 329 86% 232 73%
POST TEST
5 373 87% 272 73%

Looking at the results obtained, the first aspect that was clear was that the prediction of
relative displacements was completely overestimated in both analyses. None of the obtained
results could be considered acceptable. Pre-test displacements in North-South direction
resulted to be smaller than post-test ones probably because of different values of input
accelerations that could have required a different response to the building. Blind prediction
building deformed shapes were however in line with the real one. Actually, both analyses
were able to reproduce a soft-story collapse mode, that was actually expected to happen
because of the characteristics of structural elements (see paragraph 3.2.2) and was confirmed
by experimental test, as can be seen in figure 43.
The reason why these results were so far from the measured quantities will be discussed in the
following chapter where a model calibration made on the base of experimental data will be
presented.

48
5.2.2 Maximum value of drift angle and residual drift in each story

Maximum Story Drift Angle NS-direction
1
2
3
4
0.000 0.010 0.020 0.030 0.040 0.050 0.060
F
l
o
o
r
measured
pre-test
post-test
best results
2

Maximum Story Drift Angle EW-direction
1
2
3
4
0.000 0.010 0.020 0.030 0.040 0.050
F
l
o
o
r
measured
pre-test
post-test
best results

Figure 53_Maximum drift angle at every floor in two directions.

MaximumResidual Story Drift Angle NS-direction
1
2
3
4
-0.012 -0.008 -0.004 0.000 0.004 0.008 0.012 0.016 0.020 0.024 0.028
F
l
o
o
r
measured
pre-test
post-test
best results

MaximumResidual Story Drift Angle EW-direction
1
2
3
4
-0.012 -0.008 -0.004 0.000 0.004 0.008 0.012 0.016 0.020
F
l
o
o
r
measured
pre-test
post-test
best results

Figure 54_Maximum residual drift at every floor in two directions.
Table 25_Error evaluation in predicting interstory drift. Residual drift values for different analyses.
TEST LEVEL Drift NS error % Drift EW error % Residual Drift NS Residual Drift EW
2 0.0193 0.012145 0.0024 -0.0004
3 0.0173 0.01404 0.0020 -0.0003
4 0.0124 0.01348 0.0012 -0.0003
MEASURED
5 (Roof) 0.0073 0.0067183 0.0002 -0.0001
2 0.03651 89% 0.030016 147% 0.009125 0.005198
3 0.02614 52% 0.023616 68% 0.00021 0.000708
4 0.01830 48% 0.019231 43% 0.000159 0.000702
PRE TEST
5 0.01148 58% 0.013472 101% 0.000049 0.000639
2 0.0488 153% 0.0326 168% 0.027 0.018
3 0.0250 45% 0.0215 53% 0.000874 0.0019
4 0.0200 61% 0.0178 32% 0.000626 0.0015
POST TEST
5 0.0126 73% 0.013 94% 0.000437 0.001046

Considering comments made for maximum relative displacement at every floor, maximum
drift angle prediction resulted to be far from real results. This was true in particular in North-
South direction, where the maximum relative displacement happened to be. The prediction
49
was wrong especially at first floor that was the one characterized by the excess of
deformation.
For what concerns residual drifts, the percentage of error was not presented because, being the
values very small, it would be huge even where the difference was not so much, giving a
wrong understanding of the validity of the prediction. Looking at the graphs in figure 46,
results seemed to be totally wrong only at first floor. In fact, at other levels, values are quite
close to the best six predictions of other teams. This could be explained by the fact that the
first floor was required to work at a very high level of plasticity, given as result a considerably
big value of irrecoverable deformation.
5.2.3 Maximum value of absolute acceleration at each floor

MaximumAbsolute Acceleration NS-direction (mm/sec2)
1
2
3
4
5
0 2000 4000 6000 8000 10000 12000 14000 16000
F
l
o
o
r
measured
pre-test
post-test
best results

MaximumAbsolute Acceleration EW-direction (mm/sec2)
1
2
3
4
5
0 2000 4000 6000 8000 10000 12000 14000 16000
F
l
o
o
r
measured
pre-test
post-test
best results

Figure 55_Maxima absolute acceleration at every level in two directions.
Table 26_ Error evaluation in predicting absolute acceleration in different analyses.
TEST LEVEL Acc.NS (mm/sec2) error % Acc.NS (mm/sec2) error %
1 (GL) 4663 5469
2 6421 7672
3 6982 7797
4 8618 6984
MEASURED
5 (Roof) 9382 8687
1 4663 5469
2 9504 48% 7208 6%
3 8484 22% 9461 21%
4 8295 4% 9495 36%
PRE TEST
5 9071 3% 10005 15%
1 4663 5469
2 8803 37% 10701 39%
3 9096 30% 11349 46%
4 11111 29% 10707 53%
POST TEST
5 10020 7% 10285 18%

Even for absolute floor accelerations the prediction resulted to be not really precise. Looking
at the graphs, however, it could be noticed how the post-test analysis reproduced the
measured shape of accelerations distribution at different floors. Probably, pre-test analysis
50
was not able to do the same because not all the data used in the model were the same as in the
test. Actually, concrete characteristics were not the real ones and input ground motion was
based on ideal acceleration time-history.
5.2.4 Maximum value of story shear ay each story

MaximumStory Shear NS-direction (kN)
1
2
3
4
0 400 800 1200 1600 2000
F
l
o
o
r
measured
pre-test
post-test
best results

MaximumStory Shear EW-direction (kN)
1
2
3
4
0 400 800 1200 1600 2000
F
l
o
o
r
measured
pre-test
post-test
best results
2

Figure 56_Maximum story shear at very floor in two directions.
Table 27_ Error evaluation in predicting shear forces in different analyses.
TEST LEVEL Shear NS (kN) error % Shear EW (kN) error %
2 1420 1165
3 1256 1086
4 997 843
MEASURED
5 (Roof) 608 562
2 1443 2% 1316 13%
3 1142 9% 1101 1%
4 791 21% 827 2%
PRE TEST
5 439 28% 484 14%
2 1519 7% 1324 14%
3 1358 8% 1110 2%
4 858 14% 788 7%
POST TEST
5 485 20% 497 12%

5.2.5 Maximum value of overturning moment at each floor

MaximumOverturning Moment NS-direction (kNmm)
1
2
3
4
5
0.00E+00 4.00E+06 8.00E+06 1.20E+07 1.60E+07
F
l
o
o
r
measured
pre-test
post-test
best results

MaximumOverturning Moment EW-direction (kNmm)
1
2
3
4
5
0.00E+00 4.00E+06 8.00E+06 1.20E+07 1.60E+07
F
l
o
o
r
measured
pre-test
post-test
best results

Figure 57_Maximum overturning moment at every floor in two directions.
51
Table 28_ Error evaluation in predicting overturning moment in different analyses.
TEST LEVEL MovT NS (Nmm) error % MovT EW (Nmm) error %
1 (GL) 1.52E+07 1.30E+07
2 9.94E+06 8.60E+06
3 5.55E+06 4.75E+06
4 2.11E+06 1.94E+06
MEASURED
5 (Roof) 0 0
1 13067560 14% 11903415 8%
2 8150316 18% 7931047 8%
3 4278632 23% 4507888 5%
4 1547475 27% 1706861 12%
PRE TEST
5 0 0
1 14044393 8% 11283042 13%
2 9053616 9% 7597458 12%
3 4646348 16% 4268428 10%
4 1709353 19% 1754595 10%
POST TEST
5 0 0

From previous graphs, it could be noticed how predictions for shear forces and floor
overturning moment were quite precise. This could be explained by the fact that these two
quantities depend from equilibrium conditions that are easier to respect. Displacements, for
example, require also some compatibility equations that could be more difficult to verify.
5.2.6 Maximum strain at a specified point in elastic range

Axial Strin at the Column
1
2
3
4
0.0000 0.0005 0.0010 0.0015

Figure 58_Maximum strain at column face.
Table 29_Axial Strain at column face values.
Axial Strai at Column
MEASURED
0.001176
PRE TEST
0.000457
POST TEST
0.000476

To evaluate column axial strain, its axial force time-history was extrapolate from output data
given by the software. From this, axial force time history was found subtracting gravity load
value (i.e. axial force in the column at time = 0sec).
52
The graph below presents, as example, results from pre-test analysis.
Column 1A_Axial Force time-history
-1400
-1200
-1000
-800
-600
-400
-200
0
200
400
600
800
1000
1200
1400
0 5 10 15 20 25 30
Time (sec)
F
o
r
c
e
(
k
N
)

Figure 59_Column 1A axial force time-history from pre-test analysis.
The maximum value of axial force due to dynamic action was then found and, dividing this
by column area, maximum axial stress, which resulted to be 96 . 92
_ max,
=
PRE axial
MPa for
pre-test and 29 . 97
_ max,
=
POST axial
MPa for post-test. Finally, from column stress-block,
maximum value of axial strain was determined.
Column 1
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

Column 1
0
50
100
150
200
0.0000 0.0005 0.0010 0.0015 0.0020
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)
stress-block
pre-test
post test

Figure 60_Column 1A stress-block.
From the comparison of the results, axial strain predicted values resulted to be more or less
the half of the measured one, but however in the range of the best predictions.
5.2.7 Time of building collapse

Time of collapse
1
2
3
4
5.50 5.70 5.90 6.10 6.30 6.50
measured
post-test
best results

Charter 6. Model Calibration
53

6 MODEL CALIBRATION
After the comparison between blind prediction results and experimental one, it was clear that
something was wrong in the performed analyses. The first issue was to understand if the
distance between two sets of values was due to numerical model limitations or to mistakes
made during the pre-processing.
Researching the causes of this discrepancy, the attention was focused on the model used for
pos-test blind prediction. This because it was built using all available data from organizing
committee, resulting closer to the real specimen, and could offer a better point of view of
what happened during the analysis.
Amongst all the post-processing data collected to submit the required results, only floor
relative displacements and floor maximum accelerations were directly extrapolated from
output data given by the software (the rest of the data were derived from these through
relations presented in 5.1). Observing graphs that summarize these two quantities, it was quite
clear that the shape of predicted quantities was similar to the measure one. Predicted values
were significantly bigger, but the general behaviour of the building was reproduced quite well
at every floor. This was the first reason that helped to understand that the mistakes had not to
be related to the software but to wrong input data. In fact, results coming from SeismoStruct
were able to reproduce the building global response and soft-story collapse mode, but they
were for some reason multiplied by an incremental factor. The point now was to understand
from where, inside the input data, that incremental factor could come.
Looking at relative displacement results, it was clear that the modelled building was
considerably softer that the real specimen. The lack of stiffness could not be related to the
geometry of structural element, clearly identified by structural drawings since the beginning.
Instead, it could be quite easily connected to material property, in particular to the steel used
for the columns, because its stress-strain curve was modelled as a bi-linear regression whose
properties were derived from experimental results.
54
As explained in paragraph 3.2.2. columns were realized with steel BCR295, which expected
yielding stress is 380MPa. Instead, the spread stress-strain curve gave a yielding stress equal
to 330MPa and a Young Modulus E around 90000Mpa. This small value of E could be the
cause of the excess in elements deformation and, as a consequence, in building deflection.
Another thing was noticed studying blind prediction results for column axial strain. As
explained before, once the maximum axial stress was determined, from the given constitutive
curve, maximum axial strain was read. Since the stress due to dynamic axial action was
smaller that the yielding one, elastic relations between stress-strain could be applied.
E
=
The table below present, for two performed analysis, the value of strain calculated graphically
from the stress block and the value of the same strain calculated using the elastic relation
above.
Table 30_Column maximum axial strain comparison.
Prediction E (MPa)
y
(MPa)
max,axial
(MPa) from stress block from elastic relation
pre-test 89351 330 92.69 0.000457 0.00104
post-test 89351 332 97.29 0.000476 0.00109

It was evident that some kind of contradiction happened because the results should be the
same in both cases. Instead, the experimental test conducted on the column specimen
provided a value of strain more or less the half of the value calculated through elastic relation.
This suggested that column Young Modulus had to be the double of the one used in the
analysis.
At this point, material data provided by the committee were checked again. Stress-strain curve
for the column was analyzed more in detail. Graph below presents a zoom in the initial part of
the curve.

Column
0
100
200
300
400
500
0.0000 0.0025 0.0050 0.0075 0.0100
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

Figure 61_Stress-strain curve for column steel. Detail.
Column 1
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
tr
e
s
s
(
M
P
a
)
55
It resulted quite clear now that the value of 330MPa, specified by the committee, was
probably related to an effective yield of the material an not to the initial one, that occurs
around 230MPa. This was the reason that caused a small value of E modulus and, as a
consequence, the very soft behaviour of the material. The effective value of the Elastic
Modulus resulted now to be around 204000MPa.
Another analysis was then performed for both models maintaining the same input data a part
from column steel for which typical characteristic of steel BCR295 were assumed:
yielding stress MPa
y
380 =
Elastic Young Modulus MPa E 200000 =

All the results are presented in what follows.
Maximum Relative Displacement NS-Direction (mm)
1
2
3
4
5
0 50 100 150 200 250 300 350 400 450
F
l
o
o
r
pre-t est blind predict ion
post -t est blind predict ion
measured
pre-t est calibrated
post -t est calbrat ed

MaximumRelative Displacement EW-Direction (mm)
1
2
3
4
5
0 50 100 150 200 250 300 350 400 450
F
l
o
o
r
measured
pre-t est calibrat ed
post -t est calibrat ed

Figure 62_Maximum floor relative displacement after model calibration.
Table 31_Error evaluation for floor relative displacement after model calibration.
TEST LEVEL Displ. NS (mm) error % Displ. EW (mm) error %
1 (GL) 0 0
2 75 47
3 135 95
4 177 134
MEASURED
5 (Roof) 200 157
1 0 0
2 86 15% 82 74%
3 153 13% 125 32%
4 196 11% 160 19%
PRE TEST
5 222 11% 191 22%
1 0 0
2 80 7% 52 11%
3 146 8% 96 1%
4 191 8% 135 1%
POST TEST
5 219 10% 163 4%

56

Maximum Story Drift Angle NS-direction
1
2
3
4
0.000 0.010 0.020 0.030 0.040 0.050 0.060
F
l
o
o
r
measured

Maximum Story Drift Angle EW-direction
1
2
3
4
0.000 0.010 0.020 0.030 0.040 0.050 0.060
F
l
o
o
r
post -t est blind prediction
measured

Figure 63_Maximum story drift angle after model calibration.
MaximumResidual Story Drift Angle NS-direction
1
2
3
4
-0.012 -0.008 -0.004 0.000 0.004 0.008 0.012 0.016 0.020 0.024 0.028
F
l
o
o
r
pre-t est blind prediction
post -test blind prediction
measured
post -test calibrat ed

MaximumResidual Story Drift Angle EW-direction
1
2
3
4
-0.012 -0.008 -0.004 0.000 0.004 0.008 0.012 0.016 0.020
F
l
o
o
r
measured

Figure 64_Maximum story residual drift angle after model calibration.
Table 32_ Error evaluation for floor story drift and residual drift after model calibration.
TEST LEVEL Drift NS error % Drift EW error % Residual Drift NS Residual Drift EW
2 0.0193 0.0121 0.00236 0.00041
3 0.0173 0.0140 0.00199 0.00030
4 0.0124 0.0135 0.00115 0.00026
MEASURED
5 (Roof) 0.0073 0.0067 0.00019 0.00013
2 0.0246 28% 0.0236 94% 0.00783 0.01192
3 0.0192 12% 0.0142 1% 0.00184 0.00126
4 0.0124 0% 0.0134 1% 0.00066 0.00110
PRE TEST
5 0.0075 4% 0.0116 72% 0.00025 0.00080
2 0.0230 19% 0.0149 23% 0.00761 0.00455
3 0.0189 10% 0.0137 2% 0.00199 0.00068
4 0.0130 5% 0.0110 18% 0.00058 0.00062
POST TEST
5 0.0082 13% 0.0083 23% 0.00016 0.00042

57

MaximumAbsolute Acceleration NS-direction (mm/sec2)
1
2
3
4
5
0 2000 4000 6000 8000 10000 12000 14000 16000
F
l
o
o
r
pre-test blind prediction
post -test blind prediction
measured
pre-test calibrat ed
post -test calibrat ed

MaximumAbsolute Acceleration EW-direction (mm/sec2)
1
2
3
4
5
0 2000 4000 6000 8000 10000 12000 14000 16000
F
l
o
o
r
pre-test blind prediction
post-test blind prediction
measured
post-test calibrated

Figure 65_Maximum floor absolute acceleration after model calibration.
Table 33_Error evaluation for floor absolute acceleration after model calibration.
TEST LEVEL Acc.NS (mm/sec2) error % Acc.NS (mm/sec2) error %
1 (GL) 4663 5469
2 6421 7672
3 6982 7797
4 8618 6984
MEASURED
5 (Roof) 9382 8687
1 4663 5469
2 7288 14% 8419 10%
3 8338 19% 8303 6%
4 8608 0% 9196 32%
PRE TEST
5 8987 4% 9733 12%
1 4663 5469
2 6942 8% 6858 11%
3 7879 13% 7896 1%
4 9325 8% 8067 16%
POST TEST
5 10006 7% 9311 7%

MaximumStory Shear NS-direction (kN)
1
2
3
4
0 400 800 1200 1600 2000
F
l
o
o
r
pre-test blind predict ion
post-t est blind prediction
measured
post-t est calibrated

MaximumStory Shear EW-direction (kN)
1
2
3
4
0 400 800 1200 1600 2000
F
l
o
o
r
post-t est blind prediction
measured
pre-test calibrated
post-t est calibrat ed

Figure 66_Maximum story shear after model calibration.

58
Table 34_Error evaluation for story shear after model calibration.
TEST LEVEL Shear NS (kN) error % Shear EW (kN) error %
2 1420 1165
3 1256 1086
4 997 843
MEASURED
5 (Roof) 608 562
2 1670 18% 1263 8%
3 1231 2% 980 10%
4 839 16% 767 9%
PRE TEST
5 435 28% 471 16%
2 1621 14% 1318 13%
3 1257 0% 1106 2%
4 906 9% 763 9%
POST TEST
5 484 20% 450 20%

MaximumOverturning Moment NS-direction (kNmm)
1
2
3
4
5
0.00E+00 4.00E+06 8.00E+06 1.20E+07 1.60E+07
F
l
o
o
r
post-t est blind predict ion
measured
post-t est calibrated

MaximumOverturning Moment EW-direction (kNmm)
1
2
3
4
5
0.00E+00 4.00E+06 8.00E+06 1.20E+07 1.60E+07
F
l
o
o
r
pre-t est blind prediction
post -t est blind prediction
measured
post -t est calibrated

Figure 67_Maximum story overturning moment after model calibration.
Table 35_Error evaluation for overturning moment after model calibration.
TEST LEVEL MovT NS (Nmm) error % MovT EW (Nmm) error %
1 (GL) 15200000 13000000
2 9940000 8600000
3 5550000 4750000
4 2110000 1940000
MEASURED
5 (Roof) 0 0
1 14419506 -5% 11346693 -13%
2 8691618 13% 7566256 12%
3 4467984 19% 4198407 12%
4 1533321 27% 1660421 14%
PRE TEST
5 0 0
1 14415918 5% 12258018 6%
2 9128409 8% 7721403 10%
3 4827377 13% 4103968 14%
4 1707047 19% 1588465 18%
POST TEST
5 0 0

59
Axial Strin at the Column
1
2
3
4
0.00025 0.00075 0.00125
post-test blind prediction
measured
pre-test calibrated
post-test calibrated

Figure 68_Column maximum axial strain after model calibration.
It can be noticed how all the results are significantly better than the ones obtained with the
blind predictions. Using the calibrated post-test model (the closest to the real specimen)
displacements were evaluated with a maximum error or 11%, which becomes around 6% in
average. For what concerns accelerations, the error is significantly reduced from previous
results and it is around 10% in average. Floor shear and overturning moment are still closely
matched.
It could be concluded that the prediction closely reproduced the real building dynamic
response.
As expected, a significant increment in columns E value increased building overall stiffness,
reducing floor displacements.
The first natural period of the structure was re-calculated through an eingenvalue analysis of
the model. It was found to be equal to T=1.16sec while building first circular frequency
resulted equal to =5.389rad/sec.
NS Component Acceleration Spectrum
0
5
10
15
20
25
30
35
0.1 1 10
T sec
S
a
m
/s
2
T=1.3sec
T=1.16sec

NS Component Displacement Spectrum
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0.1 1 1.9 2.8 3.7 4.6 5.5 6.4 7.3 8.2 9.1 10
T sec
S
a

m
/s
2
T=1.3sec
T=1.16sec

Figure 69_Acceleration and Displacement spectra. Highlighted the two building periods.
Looking at Displacement Spectrum, it resulted clear how a decreased building period required
a smaller spectral displacement. Spectral acceleration was also changed and this required a
different response to the building.
The overall building strength did not really changed: increasing column yielding stress from
330 to 380MPa made a negligible difference and this was the reason why floor shear force
60
and overturning moment did not really changed from blind prediction results. Moreover
strength influence on displacement demand is negligible according to the equal displacement
principle.
Charter 7. Conclusions
61

7 CONCLUSIONS

For E-Defense contest, two analyses were ran: one for the pre-test an the other for post-test
blind prediction. The difference between the two was related to the actual characteristic of
concrete used for slabs and to the value of acceleration time-histories used during the test.
After the comparison between predicted and experimental results, it was clear that blind
prediction was not satisfactory.
Therefore a revised model calibration was performed in order to improve analysis results. It
was found then that changing just one value, i.e. column steel Young Modulus, analyses
measured parameters closely matched experimental ones. However, it has also to be clear that
in a non-linear time history the scatter in evaluated results is unavoidable.
From comparison of results coming from final analysis and the experimental test, it was clear
that predicting non-linear dynamic response of steel buildings (at least, of this one) is possible
by using of appropriate modelling choices (hysteretic rules, mass discretization, damping
parameter, etc).
The choice of this work was to neglect the contribution to building deformation given by
panel zone yielding, considering this to be somehow compensated by the equal absence in
the model of other sources of stiffness, as for example non-structural walls.
Nevertheless, further investigations should be performed to evaluate any eventual influence of
panel zone deformation modelling on the accuracy of the prediction. In fact it would be
interesting to numerically evaluate the difference between results obtained from the a fibre
model, as the one here analyzed, characterized by a spread of inelasticity along the elements,
and the results given by an hybrid lumped-plasticity/fibre model, in which part of the
deformation is also concentrated in the panel zone.
References
62

REFERENCES
Broderick, B.M., Elnashai A.S. and Izzuddin B.A. [1994], Observations on the effect of numerical
dissipation on the nonlinear dynamic response of structural systems, Engineering Structures, Vol.
16, No. 1, pp. 51-62.
Casarotti, C. (2005), Adaptive Pushover-based Methods for Seismic Assessment and Design of Bridge
Structures, PhD Thesis, European School for Advances Studies in Reduction of Seismic Risk
(ROSE School), University of Pavia, Italy.
Chopra, A.K. [1980], Dynamic of Structures A Primer, Earthquake Engineering Research Institute,
Berkley, California.
Hilber, H.M., Hughes, T.J.R. and Taylor R.L. [1977], Improved numerical dissipation for time
integration algorithms in structural dynamics, Earthquake Engineering and Structural Dynamics,
Vol. 5, No. 3, pp. 283-292.
Izzuddin, B.A., [1991]. Nonlinear Dynamic Analysis of Framed Structures. PhD Thesis, Imperial
College, University of London (UK).
Leon, R. [2007], Seismic Design of Steel and Composite Structures, ROSE School Course Notes,
Pavia, Italy.
Lpez-Menjivar, M.A. [2004], Verification of a displacement-based Adaptive Pushover method for
assessment of 2-D Reinforced Concrete Buildings, PhD Thesis, European School for Advances
Studies in Reduction of Seismic Risk (ROSE School), University of Pavia, Italy.
Manual of Steel Construction [1994], Load & Resistance Factor Design, Volume 1 Structural
Members, Specifications & Codes, American Institute of Steel Constructions, Inc., Chicago,
Illinois.
Miranda, E. [2006], Structural Dynamics, ROSE School Course Notes, Pavia, Italy.
References
63
National Research Institute for Earth Science and Disaster Prevention, Hyogo Earthquake Engineering
Research Center. Blind Analysis Contest web site: http://www.blindanalysis.jp/index_e.html
Nogueiro, P., Simnes da Silva, L., Bento, R. [2003] , Influence of joint slippage on teh cyclic
response of steel frame, Proceeding of 9
th
International Conference on Civil and Structural
Engineering Computing, paper n66, Civil-Comp Press, Stirling, United Kingdom.
Nogueiro, P., Simnes da Silva, L., Bento, R., Simnes R., Numerical implementation and calibration
of a hysteretic model with pinching for the cyclic response of steel and composite joints.
NZS3404, [1997], Steel Structures Standard, New Zealand Standards, Wellington.
NZS3101, [2006], Concrete Structures Standard, New Zealand Standards, Wellington.
Petrini, L., Pinho, R., Calvi, G.M. [2004] Criteri di progettazione antisismica degli edifici, IUSS
Press, Pavia, Italy.
Pietra, D., Pinho, R. and Antoniou, S. [2006], Verification of displacement-based adaptive pushover
for seismic assessment of high-rise steel buildings, Proceeding of First European Conference on
Earthquake Engineering and Seismology, paper n. 956, Geneva, Switzerland.
SeismoSoft [2005], SeismoStruct A Computer program for static and dynamic nonlinear analysis of
framed structures, [online], available from URL: http//www.seismosoft.com.
Tada, M., Ohsaki, M., Yamada S., Motoyui, S., Kasai, K. [2007], E-Defense Test on Full-Scale Steel
Buildings: Part 3- Analytical Simulation of Collapse, NEES/E-Defense collaborative research
program on steel structures.
Appendix A: Structural Drawings
A1

APPENDIX A: Structural Drawings

Structure description 7 Steel structure
Complete Collapse Simulating Specimen 7.2 Material
Building 7.2.1
description Steel
Strucrure
description
7.2.2
7.2.4
Anchor bolt
7.10.3
Installation of
anchor bolts
Particular Specifications
5 Reinforcement bar
5.2 Material,5.3 Fabrication and Assembly
5.2.1
Re-bar
5.2.2
Welded wire mesh 7.2.7 Deck plate type
5.3.4 Deck plate
J oint and anchorage
Re-bar, welded wire mesh shall use J IS qualied product.
Mill certificate shall be confirmed by the purchaser.
6 Concrete 7.2.9
6.1 General ,6.2 Quality of concrete Column base
6.1.3 Concrete shall be as follows. mortar
Concrete type
6.1.4
Concrete strength
6.2.3
Workability
and slump
6.2.1 a.
Strength of
structural concrete
1) 28 days standard water curing or site water curing.
2) Over 28 days and less than 91 days site can sealed curing
6.2.4 Concrete shall be as follows.
Material and
blend condition
,
,
,
,
,
,
,
Position
Type
Nominal (Design)
Yield Stress
Nominal (Design)
Ultimate Stress
Girder Welding H.T.B SN400B 235 N/mm
2
400 N/mm
2
Beam,Small column SS400 235 N/mm
2
400 N/mm
2
= H.T.B
Diaphram Welding = SN400C 235 N/mm
2
400 N/mm
2
Base plate
SN490C
(TMCP)
325 N/mm
2
490 N/mm
2
Welding =
Steel base Welding = SM490A 325 N/mm
2
490 N/mm
2
High tension
bolt(H.T.B)
Type Material
H.S.B Torshear type S10T =0.45;
note
Material
Type Note Nominal (Design)
Ultimate Stress
Nominal (Design)
Yield Stress
Type
Anchor bolt
(Specimen column)
490 N/mm
2
Double nuts
SNR490B
(ABR490)
325 N/mm
2
BCR295 295 N/mm
2
400 N/mm
2
Anchor bolt
(Steel base)
Supplied by E-defense
S45CM48.etc.;
Steel
Shop connection Site connection
Double nuts
The anchor bolt of specimen column shall be J SSC certified Architectural rolled thread anchor
bolt: conformto ABR490 specification.Material certificate shall be confirmed by the purchaser.
Position Use Type
Amount of
galvanizing
Note
2F4F
slab
Composite slab
Composite deck
plate
120g/m
2
h=75,t=1.2
RF slab Form Flat deck plate 120g/m
2
t=1.2
The supporting method of specimen base plate is post-filling up method to use the level adjusting
plate in the center.
Base mortar shall use non-shrink mortar.The strength of non-shrink mortar shall greater than
45N/mm
2.
185 Air quantity 4.5%
Column
Note
Max.size of
coarse aggregate
,
20(Crushed stone)
25(Ballast)
Position
Concrete
type
Water
cement ratio
Min.cement
weight per n
Max.water
weight per m3
The strenght of structural concrete can be the compression strength of the specimen sampled at
the factory. The difference between the strength of structural concrete and the specimen can be
considered to be ZF(=3N/mm
2
).The curing method of specimen is as shown below and shall refer
to the administrated age.
Note
Welding =
Floor slab Plain 65 270
Slump
(
Floor slab Plain 21 24 15
Position
Concrete
type
Design
strength
N/,
2
Quality
strength
N/,
2
2F4F
slab
Main reinforcement Wire mesh 6-150x150 Lap joint
Additional bar SD295A Lap joint
Position Material,Type J oint
Roof
slab
Main reinforcement SD295A Lap joint
Additional bar SD295A Lap joint
Structure
Classification
of Main Elements
Steel columnbase
type
Structure of base Connected to the shaking table with steel base
Reinforcement concrete slab(formdeck)
External wall ALC panel(t=125,vertical allocation,locking method)
Interior wall LGS backing board
Beam Steel structure
Floor deck
Column
Exposed type
Steel structure
Outline of the Structure and Particular Specifications
S04.1
National Research Institute for Earth Science and disaster Prevention
Steel Building Specimen
2007.5.20
Bottomof base
Structural type
Composite deck plate floor
Roof
Longitudinal Moment frame structure(Y-direction=main loading direction)
Latitudinal Moment frame structure(X-direction)
Steel structure
Girder
Test building 1FL+1,550Bottomof base plate;
Steel Structure
Hyogo Earthquake Engineering Rearch Center, NIED
1501-21 Nishikameya, Mitsuda, Shijimicho, Miki-shi, Hyogo
Steel structural specimen (Complete collapse simulating specimen)
Location
Type of building
4F
Test building 1FL+16,475 (Top of parapet)
Story
Max of height
Frame type
Appendix B: Material Test Results
B1

APPENDIX B: Material Test Results

Column 1
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

Column 2
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

2G1- 2G11 - 3G11 - 3G12 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

2G1 - 2G11 - 3G11 - 3G12 Web
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

4G12 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

4G12 Web
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

B2
RG1 - RG11 - RG12 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

RG1 - RG11 - RG12 Web
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

4G1 - 4G11 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

4G1 - 4G11 Web
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

2G12 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

2G12 Web
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

3G1 Flanges
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

3G1 Web
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain (mm/mm)
S
t
r
e
s
s

(
M
P
a
)

B3
PL-32 Diaphragm
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain
S
t
r
e
s
s

(
M
P
a
)

PL-50 Base Plate
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain
S
t
r
e
s
s

(
M
P
a
)

PL-22 Diaphragm
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain
S
t
r
e
s
s

(
M
P
a
)

PL-19 Diaphragm
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain
S
t
r
e
s
s

(
M
P
a
)

Anchor Bolt
0
100
200
300
400
500
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain
S
t
r
e
s
s

(
M
P
a
)

Blind Prediction of a Full-Scale Steel Frame Test

Încărcat de

Informații document

Descriere originală:

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Blind Prediction of a Full-Scale Steel Frame Test

Încărcat de

Drepturi de autor:

Formate disponibile

BLIND PREDICTION OF A FULL-SCALE

3D STEEL FRAME TESTED UNDER

S-ar putea să vă placă și