Blast Resistance of Steel-Concrete Composite

BLAST RESISTANCE OF STEEL-CONCRETE COMPOSITE
STRUCTURES
KANG KOK WEI
B.Eng (Hons.), NUS
A THESIS SUBMITTED
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
DEPARTMENT OF CIVIL & ENVIRONMENTAL
ENGINEERING
NATIONAL UNIVERSITY OF SINGAPORE
2012
i
DECLARATION
I hereby declare that this thesis is my original work and it has been written by me in
its entirety. I have duly acknowledged all the sources of information which have been
used in the thesis.
This thesis has also not been submitted for any degree in any university previously.
_________________
Kang Kok Wei
29th October 2012
ii
ACKNOWLEDGEMENT
First and foremost, I would like to thank God to providing me with the opportunity go
through a candidature for a PhD as such opportunities do not befall most. I would like
also to thank my wife. Kareen, for all the physical and emotional support and love
that she has showered upon me throughout these years, especially now when we have
one additional member in our family.
At the university, I would like to express my heartfelt gratitude to my supervisor,
Professor Richard Liew. It is an honour to be under his supervision and I do
appreciate the support especially during periods when my candidature was converted
to part-time. He still finds the time to talk and remind me constantly of the objectives
of the PhD. I would also like to thank Dr. Lee Siew Chin for constantly pushing me to
improve the contents of my thesis and her guidance, discussions and encouragement
in numerical techniques. In addition, I would like to thank the staff at the structural
laboratory for their assistance and guidance in the conduct of the laboratory tests that
were carried out in this thesis. I would like to mention fellow researchers, Patria and
Andy, who went through a week of field tests during ETSC08 at Pulau Senang. The
conditions during the tests weren’t the best but the comradeship fostered in working
together to pursue the success of one another’s test on that island.
The completion of the PhD was not an easy and straight forward one. There are ups
and there are downs. Regardless of the outcome of this dissertation, the intangible
fruits that came with all the arduous process will have a longer lasting effect than the
few pages bounded in this book.
iii
TABLE OF CONTENTS
DECLARATION ...........................................................................................................ii
ACKNOWLEDGEMENT ........................................................................................... iii
TABLE OF CONTENTS .............................................................................................. iv
SUMMARY ............................................................................................................... viii
LIST OF TABLES ........................................................................................................ xi
LIST OF FIGURES .....................................................................................................xii
LIST OF SYMBOLS .................................................................................................. xix
INTRODUCTION ......................................................................................................... 1
LITERATURE REVIEW .............................................................................................. 6
2.1 General .......................................................................................................... 6
2.2 Explosive Attacks .......................................................................................... 8
2.2.1 Types of Explosives ........................................................................... 13
2.2.2 Nature of blast loading ....................................................................... 15
2.3 Background on Protective and the Protection of Structures........................ 18
2.3.1 Protection of Key Structural Elements .............................................. 19
2.4 Methodologies in Explosive Dynamic Analysis ......................................... 21
2.4.1 Analytical Methodologies .................................................................. 22
2.4.2 Experiment Methodologies ................................................................ 24
2.4.3 Numerical Methodologies .................................................................. 29
2.5 Structural Materials under Blast Loading ................................................... 33
2.5.1 Masonry ............................................................................................. 34
2.5.2 Reinforced Concrete .......................................................................... 35
2.5.3 Steel.................................................................................................... 38
2.5.4 Composite (Steel-Concrete) ............................................................... 42
ANALYSIS OF CONCRETE FILLED STEEL TUBULAR COLUMN SUBJECTED
TO BLAST LOADING ............................................................................................... 44
3.1 General ........................................................................................................ 44
3.2 Material Response Under Dynamic Loading .............................................. 47
3.2.1 Concept of High Strain Rate Effects .................................................. 47
3.2.1.1 Fundamentals ......................................................................... 48
3.2.1.2 Concepts behind the Phenomenon ......................................... 51
3.2.2 Experimental Programme to Examine the Basis of DIF of Concrete 52
3.2.2.1 Objective ................................................................................ 53
3.2.2.2 Specifications of Concrete Studied (Granite & Stalite) ......... 53
3.2.2.3 Experimental Setup and Instrumentation ............................... 54
iv
3.2.2.4 Test Results and Discussion ................................................... 59
3.3 Design of CFST Column ............................................................................. 66
3.3.1 Assumptions....................................................................................... 66
3.3.2 Methods of Analysis .......................................................................... 68
3.3.2.1 Equivalent System of Structural Element based on SDOF .... 68
3.3.2.2 Equivalent System of Structural Element based on Rigid-
Plastic Analysis .................................................................................. 80
3.4 Analytical Study of CFST Column ............................................................. 84
3.4.1 Specification of Composite Column .................................................. 84
3.4.2 Static and Blast Loading .................................................................... 86
3.4.3 Comparison and Discussion of Results.............................................. 88
3.5 Numerical Study of CFST Column ............................................................. 92
3.5.1 Finite Element Solution Scheme ........................................................ 92
3.5.1.1 Geometry and Design of Column .......................................... 93
3.5.1.2 Type of Elements Selected ..................................................... 93
3.5.1.3 Selection of Material Models ................................................. 94
3.5.1.4 Steel-Concrete Interface Modelling ....................................... 98
3.5.1.5 Blast Loading ......................................................................... 98
3.5.2 Comparison with Analytical Models ................................................. 99
3.5.3 Further Discussion on CFST Columns ............................................ 103
3.6 Summary ................................................................................................... 109
EXPERIMENTAL PROGRAMME OF STEEL-CONCRETE-STEEL (SCS)
SANDWICH PANELS UNDER STATIC AND BLAST LOADINGS ................... 112
4.1 General ...................................................................................................... 112
4.1.1 Concept of SCS Sandwich Panels.................................................... 116
4.1.2 Objectives ........................................................................................ 118
4.1.3 Design and Construction of Specimen ............................................. 118
4.2 Material Static Properties .......................................................................... 127
4.2.1 Steel.................................................................................................. 128
4.2.1.1 Instrumentation .................................................................... 128
4.2.1.2 Results and Discussions ....................................................... 129
4.2.2 Concrete ........................................................................................... 132
4.2.2.2 Results and Discussions ....................................................... 135
4.3 SCS Sandwich Panels Design Capacity under Static Loading.................. 138
4.3.1 Analytical Properties of SCS Sandwich Panels ............................... 142
4.4 Experimental Study on the Static Capacity of SCS Sandwich Panels ...... 144
v
4.4.1 Experimental Setup and Instrumentation ......................................... 144
4.4.2 Test Results ...................................................................................... 145
4.4.3 Discussion and Comparison with Analytical Solution .................... 154
4.5 Numerical Study on Static Capacity of SCS Sandwich Panels ................. 156
4.5.1 FE Solution Scheme ......................................................................... 156
4.5.1.1 Geometry and Design of SCS Panel .................................... 157
4.5.1.2 Type of Elements Selected ................................................... 159
4.5.1.3 Selection of Material Models ............................................... 160
4.5.1.4 Contact Interface Modelling ................................................ 162
4.5.2 Comparison of FE Model with Experimental and Analytical
Results ....................................................................................................... 163
4.5.2.1 Specimen SP ........................................................................ 163
4.5.2.2 Specimen SCSNE ................................................................ 165
4.5.2.3 Specimen SCSN ................................................................... 167
4.5.2.4 Specimen SCSN4 ................................................................. 169
4.5.2.5 Specimen SCSL ................................................................... 171
4.5.2.6 Specimen SCSH ................................................................... 173
4.6 SCS Sandwich Panels Design Capacity and Failure Modes under Blast
Loading ............................................................................................................... 175
4.6.1 Analytical Solution of the Blast Response of SCS Sandwich Panel
Specimen to Blast Loadings ...................................................................... 176
4.7 Experimental Programme on the Blast Resistance of SCS Sandwich
Panels .................................................................................................................. 177
4.7.1 Experimental Setup and Instrumentation ......................................... 178
4.7.1.1 Procedure and Design of Experiment .................................. 178
4.7.1.2 Blast Loading ....................................................................... 179
4.7.1.3 Reinforced Concrete Support Structure ............................... 180
4.7.2 Test Results ...................................................................................... 185
4.7.2.1 Blast Response and Failure Mode ........................................ 185
4.7.2.2 Pressure Signal Records ....................................................... 192
4.7.2.3 Acceleration Signal Records ................................................ 194
4.7.2.4 Strain Gauge Records ........................................................... 194
4.7.3 Discussion and Deductions from Experimental and Analytical Results
199
4.8 Numerical Study on the Blast Resistance of SCS Sandwich Panels ......... 205
4.8.1 FE Solution Scheme ......................................................................... 206
4.8.1.1 Geometry and Design of Test Setup and Specimens ........... 206
vi
4.8.1.2 Type of Elements Selected ................................................... 208
4.8.1.3 Selection of Material Models ............................................... 208
4.8.1.4 Steel-Concrete Interface Modelling ..................................... 208
4.8.1.5 Blast Loading ....................................................................... 209
4.8.2 Comparison of FE Model with Experimental and Analytical Results
211
4.8.2.1 Specimen SCSNE ................................................................ 211
4.8.2.2 Specimen SCSN ................................................................... 213
4.8.2.3 Specimen SCSN4 ................................................................. 216
4.8.2.4 Specimen SP ........................................................................ 217
4.8.2.5 Specimen SCSL ................................................................... 218
4.8.2.6 Specimen SCSH ................................................................... 220
4.8.3 Further Discussion on the Dynamic Design of SCS Sandwich Panels
220
4.9 Summary ................................................................................................... 226
CONCLUSION .......................................................................................................... 229
BIBLIOGRAPHY ...................................................................................................... 234
APPENDIX A ............................................................................................................A-1
APPENDIX B ............................................................................................................ B-1
APPENDIX C ............................................................................................................ C-1
vii
SUMMARY
Steel-concrete composite structural design is becoming common and more prominent
in the modern construction industry and this can be attributed to the facility of
construction and its capacity to harness the strength of both concrete and steel.
However, modern structures face an increasing threat due to the increasing presence
of terrorism with their access to destructive technologies through asymmetric warfare.
One of these concerns which arose is the use of explosives against commercial or
governmental buildings. Therefore it is now important for civil engineers to
understand dynamic designs and incorporate them into buildings to resist loads
generated from such an environment.
This study attempts to develop an analytical method to accurately capture the dynamic
inelastic behaviour of concrete filled steel tubular (CFST) columns subject to blast
loading. The proposed approach will possess a closed form solution approach and the
capability to analyse a structure, which respond both in flexural as well as in shear.
The thesis will also study the blast resistant performance of steel-concrete-steel (SCS)
sandwich panels through analytical, experimental and numerical study.
In the design of structural members against blast loading, the Single-Degree-of-
Freedom (SDOF) method is commonly used to approximate the dynamic response of
structures. One of the limitations of this method is the inability to capture the multi-
failure modes of the structural members. The Rigid-Plastic method is thus proposed in
this thesis to estimate the blast response of CFST columns. The Rigid-Plastic results
are compared with SDOF calculations as well as validated numerical models in order
to assess the competency of this proposed method. Due to the assumption of rigid-
viii
plastic material behaviour, the accuracy of this method is influenced by the extent of
plastic deformation of the structural member. For the case of impulsive blast loading,
the Rigid-Plastic estimations are found to be closer to the numerical results than those
obtained using the SDOF method. This study also encompasses a study into the
performance of a composite column as compared to that of a reinforced concrete one
and a significant improvement in the blast resistance of the composite column was
observed.
Another phase of this study includes an experimental study to investigate the response
of SCS sandwich panel of various configurations under quasi-static and dynamic
loadings. The quasi-static experiment series utilised a three-point laboratory load
setup and the dynamic study was carried out with actual explosives in an outdoor
firing range. The differences in response of six configurations of sandwich composite
panels, which differed in the thickness of steel plates, the concrete properties of the
sandwich core and the connectors, were investigated under both quasi-static and
dynamic loads. Both experimental series showed the enhancement effects by the
increased steel plate thickness and the presence of concrete core. In addition, the
comparison between quasi-static and the dynamic test series has emphasised the
differences between static and dynamic resistance. Specimens of high static resistance
may not necessarily perform well under dynamic load due to the brittle nature of the
concrete cores. Results from the experimental study are also used to validate the
numerical models and the analytical design approach, which has shown to be
conservative in static and in most dynamic cases. These numerical models are further
extended to demonstrate the effectiveness of incorporating steel plates between the
top and bottom steel face plates to enhance the blast resistance. In addition, the use of
ix
lightweight concrete could be used in blast resistance SCS panels provided sufficient
strength is designed in the concrete.
x
LIST OF TABLES
Table 3-1 Dynamic increase factor for yield strength of (a) steel and (b) concrete .... 51
Table 3-2 Specifications of granite and stalite ............................................................. 53
Table 3-3 Mix content of concrete specimens ............................................................. 54
Table 3-4 Specifications of SHS specimens ................................................................ 73
Table 3-5 Material properties of specimens................................................................. 73
Table 3-6 Maximum values from the tests .................................................................. 77
Table 3-7 Structural response regimes ......................................................................... 82
Table 3-8 Material properties of concrete-infilled steel column ................................. 86
Table 3-9 Applied blast loading .................................................................................. 87
Table 3-10 Blast response of column at stand-off distance of 10 m (impulsive regime)
...................................................................................................................................... 88
Table 3-11 Blast response of column at stand-off distance of 12.5m (dynamic regime)
...................................................................................................................................... 88
Table 3-12 Blast response of column at stand-off distance of 15m (dynamic regime)88
Table 3-13 Blast response of column at stand-off distance of 10m ............................. 99
Table 3-14 Blast response of column at stand-off distance of 12.5 m ......................... 99
Table 3-15 Blast response of column at stand-off distance of 15 m .......................... 100
Table 4-1 Specifications of the configurations of the specimens .............................. 119
Table 4-2 Tabulation of the key parameters from the tensile test.............................. 132
Table 4-3 Lightweight concrete mix .......................................................................... 133
Table 4-4 HSC concrete mix...................................................................................... 133
Table 4-5 Information derived from compressive cylinder tests ............................... 137
Table 4-6 Properties of steel used for analytical study .............................................. 143
Table 4-7 Properties of concrete used for analytical study ........................................ 143
Table 4-8 Properties of specimens used for analytical study ..................................... 143
Table 4-9 Important parameters from the comparison of the panel specimens which
are subjected to a three point load test ....................................................................... 148
Table 4-10 Tabulation of the parts in each model ..................................................... 159
Table 4-11 Tabulation of the analytical results of the peak and permanent deformation
.................................................................................................................................... 176
Table 4-12 Comparison of permanent deformation of specimens ............................. 201
Table 4-13 Comparison of permanent deformation of specimens ............................. 224
xi
LIST OF FIGURES
Figure 2-1 Effect of the gas explosion at Ronan Point, UK .......................................... 9

Figure 2-2 Illustration of the collapse at Ronan Point, UK ........................................... 9
Figure 2-3 Devastation from the bomb blast of Alfred Murrah Federal Building,
Oklahoma ..................................................................................................................... 10
Figure 2-4 Position of the columns and transfer girder which failed .......................... 11
Figure 2-5 3D illustration of the collapse area of the Alfred Murrah Federal Building
...................................................................................................................................... 11
Figure 2-6 Damage as a result of the terrorist attack on Khobar Tower...................... 12
Figure 2-7 Difference in pressure-time history of low and high explosives................ 14
Figure 2-8 Examples of blast walls .............................................................................. 20
Figure 2-9 Simplification of SDOF method ................................................................ 22
Figure 2-10 An example of a lab-based scale experiment setup ................................. 26
Figure 2-11 Laboratory test setup in the use of LPG as explosives............................. 27
Figure 2-12 Baker Risk large shock tube ..................................................................... 28
Figure 2-13 Pressure-time history obtained from the large shock tube ....................... 29
Figure 2-14 (a) Before and (b) after effects of protected and non-protected masonry
wall using Life Shield ® Panel Technology ................................................................ 35
Figure 2-15 Improvement in the use of FRP in columns ............................................. 38
Figure 2-16 Different modes of failure of beams under blast loading: (a) Mode I, (b)
Mode II and (c) Mode III ............................................................................................. 39
Figure 3-1 Range of strain rates under different loadings ........................................... 48
Figure 3-2 Difference between DIF from various sources .......................................... 50
Figure 3-3 Compilation of DIF values from experimental data .................................. 50
Figure 3-4 Proposed Crack pattern under quasi-static and dynamic loads .................. 52
Figure 3-5 Stalite coarse aggregates ............................................................................ 54
Figure 3-6 Stress -Strain plot of (a) granite and (b) stalite from unconfined
compressive cylinder load tests ................................................................................... 55
Figure 3-7 Schematic of SHPB used in study .............................................................. 57
Figure 3-8 Sample of the readings from the data logger ............................................. 57
Figure 3-9 Strain rate data from (a) one, (b) two and (c) three wave analysis on granite
concrete mix ................................................................................................................. 60
Figure 3-10 Stress data from (a) one, (b) two and (3) three wave analysis on granite
concrete mix ................................................................................................................. 61
Figure 3-11 Comparison of (a) one, (b) two and (3) three wave analysis on DIF of
granite concrete mix with current research .................................................................. 62
Figure 3-12 Strain rate data from a three wave analysis on stalite concrete mix ........ 63
Figure 3-13 Stress data from a three wave analysis on stalite concrete mix ............... 64
xii
Figure 3-14 Strain rate data from a three wave analysis on granite (solid line) and
stalite (dashed line) concrete mix ................................................................................ 64
stalite (dashed line) concrete mix ................................................................................ 65
Figure 3-16 Concept of the drop hammer-airbag setup (a) prior to loading and (b)
during the loading of the specimen .............................................................................. 70
Figure 3-17 Plan and elevation views of the setup of specimen and airbag ................ 71
Figure 3-18 Setup of drop hammer rig with specimen and airbag .............................. 72
Figure 3-19 Instrumentation setup: (a) strain gauge, (b) laser sensor (displacement),
(c) pressure sensor, (d) load cell .................................................................................. 73
Figure 3-20 Readings from various instrumentation from MS-50 .............................. 74
Figure 3-21 Readings from various instrumentation from MS-500 ............................ 75
Figure 3-22 Readings from various instrumentation SS-500 ...................................... 75
Figure 3-23 Deformation of mild steel SHS in-filled with sand from test MS-500 .... 76
Figure 3-24 Comparison between experiment (EXP) and analytical (SDOF) results for
test MS-50 .................................................................................................................... 78
test MS-500 .................................................................................................................. 78
test SS-500 ................................................................................................................... 79
Figure 3-27 Notation of beam to the Rigid-Plastic method ......................................... 81
Figure 3-28 Response modes in dynamic plastic analysis ........................................... 81
Figure 3-29 Plastic mechanisms under blast loading ................................................... 83
Figure 3-30 CFST column used to compare the analytical and numerical results
(Dimensions in mm) .................................................................................................... 85
Figure 3-31 P-M Interaction curves of concrete-filled steel composite ...................... 86
Figure 3-32 Loadings on the CFST column ................................................................ 87
Figure 3-33 Comparison of analytical and numerical predictions for stand-off distance
at 10m........................................................................................................................... 89
at 12.5m........................................................................................................................ 89
at 15m........................................................................................................................... 90
Figure 3-36 Meridian profiles for *MAT_072R3 in (a) 2D and (b) 3D stress space .. 96
at 10m......................................................................................................................... 100
at 12.5m...................................................................................................................... 101
at 15m......................................................................................................................... 101
Figure 3-40 Deformed shapes of (a) RC and (b) CFST columns at t = 0.004 ........... 104
xiii
Figure 3-41 Comparison of displacement-time histories of RC and CFST columns 104
Figure 3-42 Comparison of effective mean stress-time histories of the element at the
mid-span of RC and concrete-filled steel composite columns................................... 105
Figure 3-43 RC column with comparable properties as the concrete-filled steel
composite column in Figure 3-36 .............................................................................. 106
Figure 3-44 P-M Interaction curves of CFST and RC columns ................................ 107
Figure 3-45 Comparison of displacement time histories of RC and CFST columns. 108
Figure 3-46 Deformed shapes of (a) RC and (b) CFST column at t = 0.0035sec ..... 108
Figure 4-1 Examples of (a) Corrugated and ribbed steel decks and (b) connection
details on beam to ensure composite actions between steel beam and concrete slab 113
Figure 4-2(a) J-hooked connectors on steel panel, (b) Arrangement of aligned J-
hooked connectors assembled in SCS panels ............................................................ 117
Figure 4-3 Notation for (a) SCS composite sandwich panel and (b) Cellular steel
panel ........................................................................................................................... 120
Figure 4-4 Schematic of cellular steel panel (SP) assembly ...................................... 121
Figure 4-5 Schematic of SCS composite sandwich panel with 4mm top and bottom
plates (SCSN4) assembly........................................................................................... 122
Figure 4-6 Schematic of SCS composite sandwich panel (SCSN, SCSL, SCSH)
assembly..................................................................................................................... 123
Figure 4-7 Schematic of SCS composite sandwich panel without J-hook connectors
(SCSNE) assembly..................................................................................................... 124
Figure 4-8 Schematic of J-hook connector ................................................................ 125
Figure 4-9 Specimen preparation photos of (a) positioning of jig, (b) welding of J-
hook connectors, (c) placement of top and bottom plate prior to welding of the side
and end plates, (d) preparation of specimens with J-hook connectors prior to casting
of concrete core, (e) preparation of specimens without J-hook connectors prior to
casting of concrete and (f) concrete casting completion ............................................ 127
Figure 4-10 Specification for steel coupon ................................................................ 128
Figure 4-11 (a) 1.5mm, (b) 3mm and (c) 4mm thick coupons after tensile test ........ 129
Figure 4-12 Strain signal recorded from strain gauges on (a) 1.5mm, (b) 3mm and (c)
4mm coupons ............................................................................................................. 130
Figure 4-13 Processed strain data recorded from extensometer on (a) 1.5mm, (b) 3mm
and (c) 4mm coupons ................................................................................................. 131
Figure 4-14 Sample pictures of Liapor aggregates .................................................... 133
Figure 4-15 View of the instrumentation on the NSC and LWC cylinders ............... 134
Figure 4-16 View of the (a) instrumentation and (b) test machineries used for the HSC
cylinder static tests ..................................................................................................... 134
Figure 4-17 Longitudinal stress-strain curve of (a) NSC, (b) LWC and (c) HSC
cylinders ..................................................................................................................... 136
Figure 4-18 Transverse stress-strain curve of (a) NSC, (b) LWC and (c) HSC
cylinders ..................................................................................................................... 136
xiv
Figure 4-19 Schematic of the setup of a three point quasi-static load test ................ 138
Figure 4-20 Illustration of (a) Section, (b) Equivalent Steel Section and (c) Stress
Block of a sandwich panel ......................................................................................... 138
Figure 4-21 Photo of the three point load test for panel specimens ........................... 144
Figure 4-22 Instrumentation setup of the panel specimens ....................................... 145
Figure 4-23 Normalised quarterspan and midspan displacement histories from LVDT
of (a) SP, (b) SCSN4, (c) SCSN, (d) SCSNE, (e) SCSL and (f) SCSH sandwich
panels ......................................................................................................................... 146
Figure 4-24 Comparison of the midspan displacement of panel specimens during the
(a) elastic and (b) elasto-plastic response .................................................................. 147
Figure 4-25 Strain gauge readings from (a) SP, (b) SCSN4, (c) SCSN, (d) SCSNE,150
(d) SCSL and (e) SCS ................................................................................................ 150
Figure 4-26 Comparison of the midspan bottom strain of panel specimens during the
(a) elastic and (b) elasto-plastic response .................................................................. 151
Figure 4-27 Buckling of top steel plate that initiated softening for specimen SCSN 152
Figure 4-28 Buckling of top steel plate that initiated softening for specimen SCSNE
.................................................................................................................................... 153
Figure 4-29 Plots of the analytical properties from Table 4-8 of (a) SP, (b) SCSN4, (c)
SCSN, (d) SCSNE, (e) SCSL and (f) SCSH against experimental results ................ 155
Figure 4-30 Model of the J-hook connectors ............................................................. 157
Figure 4-31 Numerical model with support and load blocks illustrated .................... 158
Figure 4-32 Plane of symmetry of the SCS specimen ............................................... 159
Figure 4-33 Mesh used for the top steel plate of Specimen (a) SP and (b) SCSN .... 160
Figure 4-34 Constitution of the J-hook connector model used .................................. 160
Figure 4-35 Experimental and numerical stress-strain relationship .......................... 161
Figure 4-36 Force-Displacement used to specify the material model J-hook in NSC,
LWC and HSC ........................................................................................................... 162
Figure 4-37 Debonding failure model adopted in numerical model of SCS sandwich
panels ......................................................................................................................... 163
Figure 4-38 Force-displacement plot of Specimen SP .............................................. 164
Figure 4-39 Response of Specimen SP at (a) yield and (b) at 10mm midspan
deflection.................................................................................................................... 164
Figure 4-40 Force-displacement plot of Specimen SCSNE ...................................... 165
Figure 4-41 Response of Specimen SCSNE at 10mm midspan deflection ............... 166
Figure 4-42 Predicted crack or damage within the concrete core of Specimen SCSNE
.................................................................................................................................... 166
Figure 4-43 Force-displacement plot of Specimen SCSN ......................................... 167
Figure 4-44 Response of Specimen SCSN at 10mm midspan deflection .................. 168
xv
Figure 4-45 Predicted crack or damage within the concrete core of Specimen SCSN
viewed from (a) the elevation and (b) from a isometric angle with the fringe values
isolated ....................................................................................................................... 168
Figure 4-46 Force-displacement plot of Specimen SCSN4 ....................................... 170
Figure 4-47 Response of Specimen SCSN4 at 10mm midspan deflection ................ 170
Figure 4-48 Predicted crack or damage within the concrete core of Specimen SCSN4
isolated ....................................................................................................................... 170
Figure 4-49 Force-displacement plot of Specimen SCSL ......................................... 171
Figure 4-50 Response of Specimen SCSL at 10mm midspan deflection .................. 172
Figure 4-51 Predicted crack or damage within the concrete core of Specimen SCSL
isolated ....................................................................................................................... 172
Figure 4-52 Force-displacement plot of Specimen SCSH ......................................... 173
Figure 4-53 Response of Specimen SCSH at 10mm midspan deflection .................. 173
Figure 4-54 Predicted crack or damage within the concrete core of Specimen SCSH
isolated ....................................................................................................................... 174
Figure 4-55 Analytical displacement-time histories of the 6 specimens ................... 176
Figure 4-56 Alignment of the charges to the specimens............................................ 179
Figure 4-57 Reflected Overpressure and Impulse based on scaled distance of
1.077m/kg1/3 ............................................................................................................... 180
Figure 4-58 Isometric view of RC support structure ................................................. 181
Figure 4-59 Elevation view of RC support structure with the embedment details .... 181
Figure 4-60 Actual picture of RC support structure with two specimens.................. 182
Figure 4-61 Assembly to secure potentiometers (a) prior and (b) after the installation
.................................................................................................................................... 184
Figure 4-62 Adapters for potentiometers (left and right) and accelerometer (centre)
.................................................................................................................................... 184
Figure 4-63 (a) Schematic and (b) Actual positions of instrumentation .................... 185
Figure 4-64 Deformation of Specimen SP (a) onsite (left) and (b) in the laboratory 186
Figure 4-65 Deformation of Specimen A was obstructed by the steel I-beam .......... 186
Figure 4-66 Local buckling of the steel plates of Specimen SP ................................ 187
Figure 4-67 Steel fracture on the top steel plate at the midspan of the specimen...... 187
Figure 4-68 Deformation of Specimen SCSN (a) onsite (left) and (b) in the laboratory
.................................................................................................................................... 188
Figure 4-69 Actual and filtered displacement time histories of SCSN ...................... 189
Figure 4-70 Final deformed profile of SCSNE .......................................................... 189
Figure 4-71 Displacement time history of SCSNE .................................................... 190
xvi
Figure 4-72 Deformation of Specimen SCSL (a) onsite (left) and (b) in the laboratory
.................................................................................................................................... 190
Figure 4-73 Displacement time history of SCSNE .................................................... 191
Figure 4-74 (a) Rupture of steel side plate and (b) the failure of J-hook connectors due
to excessive shear response ........................................................................................ 191
Figure 4-75 Local buckling of the top plate (a) observed onsite and (b) measured in
laboratory ................................................................................................................... 192
Figure 4-76 Reflected pressure and impulse for Blast 2 ............................................ 193
Figure 4-77 Reflected pressure and impulse for Blast 3 ............................................ 194
Figure 4-78 Strain recordings for SCSN .................................................................... 195
Figure 4-79 Strain recordings for Specimen SCSNE ................................................ 196
Figure 4-80 Strain recordings for Specimen SCSL ................................................... 196
Figure 4-81 Strain recordings for Specimen SCSH ................................................... 197
Figure 4-82 Comparison of strain time histories at mid-span of Specimens SCSN,
SCSNE, SCSL and SCSH .......................................................................................... 197
Figure 4-83 Comparison of reflected (a) overpressure and (b) impulse .................... 200
Figure 4-84 Numerical model to study blast response of SCS specimens ................ 207
Figure 4-85 Illustration of (a) the nodes defined to form the rigid body and (b) the
centre of the rigid body in constraints applied to the steel bracket plate and steel roller
rods............................................................................................................................. 208
Figure 4-86 Comparison of the three blast pressure and impulse loads applied to the
models ........................................................................................................................ 210
Figure 4-87 Midspan displacement histories of numerical models of Specimen
SCSNE ....................................................................................................................... 211
Figure 4-88 Comparison of deformed shapes from (a) explosive test and (b) FE
analysis ....................................................................................................................... 213
Figure 4-89 Midspan displacement histories of numerical models of Specimen SCSN
.................................................................................................................................... 214
Figure 4-90 Comparison of various connection assumptions .................................... 215
Figure 4-91 Comparison of various connection assumptions .................................... 216
Figure 4-92 Midspan displacement histories of numerical models of Specimen SCSN4
.................................................................................................................................... 217
Figure 4-93 Midspan displacement histories of numerical models of Specimen SP . 218
Figure 4-94 Midspan displacement histories of numerical models of Specimen SCSL
.................................................................................................................................... 219
Figure 4-95 Deflection profile of Specimen SCSL with fringe levels of deflection . 219
Figure 4-96 Midspan displacement histories of numerical models of Specimen SCSH
.................................................................................................................................... 220
Figure 4-97 Comparison of the response of Specimen SCSN and SCENE .............. 221
Figure 4-98 Comparison of the response of Specimen SCSN and SCEN4 ............... 222
xvii
Figure 4-99 Comparison of the response of the SCS sandwich panels which are
infilled with NSC, LWC and HSC ............................................................................. 222
infilled with NSC, LWC and a concrete core with the strength of NSC but the density
of LWC ...................................................................................................................... 224
xviii
LIST OF SYMBOLS
H Depth of member
L Length of member
m Mass per unit length
M Moment
Mp Plastic moment capacity
p Force applied per unit length
P Blast overpressure
Peff Effective blast overpressure
Prmax Peak reflected blast overpressure
Q Shear force
Qd Dynamic shear
Qo Shear capacity
t Time
T Natural period
α Wave form parameter
η Ratio of peak reflected overpressure to static elastic collapse pressure
μ Ductility ratio
τ Blast pressure-time duration
τeff Effective blast pressure-time duration
ω Displacement
xix
CHAPTER 1:
INTRODUCTION
The presence of terrorism has slowly been increasing globally since the turn of the
21st century and their access to advanced technologies have concerned governments.
One of these concerns is the use of explosives against commercial or governmental
infrastructures. Without the considerations of extreme loadings in the design of
critical structural components such as columns, the structural members may fail and
lead to subsequent progressive collapse in the event of blast, which may cost the lives
of hundreds and thousands of occupants. Therefore it is now important for civil
engineers to understand dynamic design and incorporate passive protective measures
into the detailing of the structural members to resist loads generated from such
environment.
Steel-concrete composite structural members are commonly used in modern
construction due to the facility of construction, which can be derived from the options
to perform prefabrication and the reduction in the need of formworks. Composite
structures harness the strength of both concrete and steel to optimise on the usage of
materials in design. In view of the significant performance of steel-concrete
composite structural members over conventional steel or reinforced concrete
structures in static design, there is a need to research on such system to quantify the
performance of these columns against blast loading.
This study attempts to deepen the understanding of the design and response of steel-
concrete composite structural components which are subjected to blast loading
through reviews and to come up with a proposal of analytical design approach for
steel-concrete composite columns and slabs against blast loading. The analytical study
1
is coupled with numerical modelling and an experimental programme to ensure
validity of the result and ascertain the performance of certain assumptions that were
made in the design process. The specific objectives of this thesis are as follow:
• Develop an analytical method to accurately capture the dynamic inelastic
behaviour of concrete filled steel tubular (CFST) subject to blast loading
• Study the blast resistant performance of steel-concrete-steel (SCS) sandwich
panels through analytical, experiments and numerical simulations
In order to achieve these objectives, below documents a brief description of the scope
of work, of which the sequence of these work will be detailed in the subsequent
paragraphs of this chapter:
 Review the state-of-the-art in analytical, numerical and experimental works in
deriving the structural response to blast loading
 Conduct of quasi-static and dynamic tests of construction materials against
dynamic loading
 Validation of application of the Single Degree of Freedom (SDOF) method
 Comparison of SDOF method with a proposed analytical method for a CFST
columns which is subjected to various loadings
 Comparison of analytical results of the CFST columns with proposed
numerical models
 Conduct of quasi-static and dynamic tests on SCS sandwich panels and
compare the structural performance of these specimens from the two loading
regimes
 Perform numerical simulation of the quasi-static and dynamic response of the
SCS sandwich panels and compare the results with those that were derived
analytically and experimentally
2
In the design of structural members against blast loading, the SDOF method is
commonly used to predict the dynamic response. The applicability of this method to
CFST column will be reviewed through comparisons with experimental and
numerical approaches. One of the limitations of this method is that it cannot capture
the multi-failure modes of the structural members. The Rigid-Plastic method is thus,
proposed in this thesis to estimate the blast response of CFST columns. The Rigid-
Plastic results are compared with SDOF calculations as well as numerical simulations
in order to assess the competency of this proposed method.
An experimental study to investigate the response of SCS sandwich panel of various
configurations under quasi-static and explosive loadings was carried out. The quasi-
static experiment series utilised a three-point load test and the dynamic study involved
an explosive test setup. The differences in response of six configurations of sandwich
panels in the quasi-static and dynamic load series were investigated. In addition,
numerical simulations will be conducted to complement and provide limited
validation of the experimental results.
The thesis is organised into 6 chapters. The first chapter will provide the background
as well as the motivation of this study. The overarching objectives as well as the
scope of this thesis will also be documented in this chapter. Chapter Two will
describe the literature review of the current state-of-the-art of the analytical design
approaches for blast loading in the public domain. Concepts of structural components
that are currently used for blast-resistant as well as the numerical approach that will
be covered in this literature survey.
3
Chapter Three documents a brief investigation of the material response under
dynamic blast loading will also be documented through an experimental study using
the Split Hopkinson Pressure Bar (SHPB). A check on the validity of the SDOF
method in the analysis of steel composite structures is also included in this chapter.
This work is done through to dynamic impact tests that utilised an airbag to distribute
the forces across the span of square hollow steel sections, which are in-filled with
sand. As the SDOF approach has certain deficiencies, the Rigid Plastic approach that
emphasises on the use of closed form solutions and ability to capture multi failure
modes is proposed. The basis and derivation of this method together with the
assumptions made will be documented in this chapter. This approach is applied and
compared to the SDOF method for the design of CFST columns under blast loading
and their differences are examined and explained. The analytical results are then
compared with results from numerical models to ascertain the validity of the SDOF
and Rigid Plastic methods in the analysis of CFST columns. Further analysis with the
numerical model also demonstrated the superior performance of CFST columns as
compared to reinforced concrete ones.
Chapter Four focuses on the study of SCS sandwich panels. The study entails a series
of static three-point load tests that were carried out to determine the load-
displacement curves and failure mechanisms of specimens with different
configurations. This will be followed by a detailed insight into the experimental
programme conducted in collaboration with the Defence Science and Technology
Agency (DSTA) to investigate the blast performance of the SCS sandwich panels.
The results from the experimental study are compared with SDOF analytical and
numerical solutions and the findings are highlighted in this chapter.
4
Finally, the conclusions from the analytical, experimental and numerical studies that
are conducted in this thesis are detailed in Chapter Six. Subjects that require further
studies to understand and bridge some of the gaps that are highlighted in this work are
also listed in this chapter.
5
CHAPTER 2:
LITERATURE REVIEW
2.1 General
This chapter attempts to summarise some of the current basis of research in the design
of structures against blast. These state-of-the-art reviews will be based on studies of
actual accidental blast incidents as well as experimental programmes, which are
conducted by various research agencies and institutes, in hope of enhancing the
understanding of structural dynamic response to explosive loadings and build a solid
foundation to base the analysis that is conducted throughout the research work in this
thesis.
In the analysis of the blast effects, it is essential to understand the fundamentals. In an
explosion, the environment is composed of the following:
 Donor system (Blast pressure, primary and secondary fragments, ground
shock)
 Acceptor system
 Protective system (Protective structural and non structural elements or
barriers)
The literature review will commence with a survey of actual incidents involving
structures that were subjected to various dynamic and impulsive loadings. Most of
these incidents were terrorist related but amongst them were cases whereby such
explosive forces were ignited by accident. With respect to the donor system, the
different kind of explosions with their individual characteristics will also be
expounded upon in this section. Fragments and ground shock will not be covered to
limit and concentrate the effort in this study. A brief explanation behind the science
behind these explosions will be done. Subsequently, the protection systems of
6
structures against such loading will be reviewed. Different standards will be studied
and comparisons will be made between them to show the differences and mostly
similarities in their approach towards structural design against blast loading.
A review of the methodologies in the study of structural blast response will also be
included. This section will be split into three sub-sections, which are analytical,
experimental and numerical methods. Different analytical methods had been adopted
by various researchers and agencies for the past fifty years in hope of producing a
design guide in structural blast design. Therefore, a review of these approaches will
document the progress that has been attained thus far. It will also aid in providing an
overview of the purpose and objective to develop a design guide in blast analysis.
Numerous blast experiments together with their results have also been carried out
since World War Two to support the analytical study. However, many of these are not
available in the public domain. Hence, the author will attempt to the best of his
knowledge and resource to provide a comprehensive overview of the experimental
work that has been done. In the third sub-section, the basis for numerical method will
be evaluated to provide a correct foundation for the numerical simulation that will be
done throughout the thesis. Lastly, amongst the vast number of structural materials
that are used in modern structures, four kinds of structural materials, namely masonry,
reinforced concrete, steel and composite, will be highlighted. The strengths and
weaknesses of these materials and the scientific advances in the use of such materials
in blast design will be pointed out in this section.
7
2.2 Explosive Attacks
Research in blast design was initiated by the military due to the necessity to design
their structures against military ordnance such as bombs, missiles and other malicious
explosive devices. Extensive work started after World War Two where military
technology started to advance to optimise the limited material and human resources.
Important structures such as communication centres and unit headquarters, which
house both important personnel and assets for the success of the war campaigns, are
specially designed to resist explosive loads, in view of protecting these contents. Such
works were limited to the military sector as it was deemed unnecessary to design
commercial and residential buildings using such extreme design codes. Furthermore,
such data are classified to protect the individual military capabilities. These designs
would be deemed conservative.
However, it was not until an incident in Newham in east London, United Kingdom,
that sparked the need to design structures against explosive loads. This incident,
known infamously as the Ronan Point Disaster, happened in the morning of the 16th
of May 1968 whereby an explosion on the 18th floor of the new 23 storey residential
block caused the catastrophic collapse of an entire section of the building (Griffiths et
al, 1968). Figure 2-1 illustrates the damage which resulted from the explosion. The
investigation traced the source to a gas explosion and the load blew out a wall panel
which initiated the collapse of the floors above. The dynamic loading of such falling
structural elements from above caused the successive failure of the floors below. This
was termed as progressive collapse. Figure 2-2 shows graphically the domino effect
of incident. Inquiry to the design of Ronan Point found neither violation of the
building standards at that time nor defects in the workmanship (Crowder, 2005).
8
Interest in the design of commercial structures against such intense loading was then
started.
Figure 2-1 Effect of the gas explosion at Ronan Point, UK
Figure 2-2 Illustration of the collapse at Ronan Point, UK
9
Another occurrence of a commercial building that was subjected to blast loading is
the Alfred Murrah Federal Building incident which was being attacked by a Vehicle
Borne Improvised Explosive Device (VBIED) as seen in Figure 2-3. The bomb that
was detonated was fabricated with more than 6200lb (2,800kg) of ammonium nitrate
fertiliser, nitromethane and diesel fuel mixture and it was ignited in front of the north
side of the nine storey reinforced concrete building (Rogers and Koper). The intense
load, which was equivalent to 4000lb (1800kg) of TNT, resulted in the failure of three
columns which were used to support a transfer girder (Corley et al., 1998). The
positions of the columns and transfer girder are illustrated in Figure 2-4. The loss of
the transfer girder initiated the progressive collapse of the floors above and that
collapse resulted in the loss of almost half the occupancy space as shown in Figure 2-
5. This disaster resulted in the death of 168 people, both in and around the structure.
Figure 2-3 Devastation from the bomb blast of Alfred Murrah Federal Building,
Oklahoma
10
3 Columns which failed under
Transfer Girder blast load
Figure 2-4 Position of the columns and transfer girder which failed
Figure 2-5 3D illustration of the collapse area of the Alfred Murrah Federal Building
Opposed to the use of improvised explosive substances that was used in Alfred
Murray Federal Building incident, religious extremist used a truck loaded with fuel in
the attack of Khobar Towers in Saudi Arabia. The loading was believed to be
equivalent to a TNT weight of 20000lb (9000kg) (Crowder, 2005). On contrary to the
attacks that was mentioned in the preceding two paragraphs, the structure did not
11
collapse as precautionary measures, such as installing perimeter fences and waist-high
“Jersey” barriers, were put in place by the military prior to such attacks (Ziegler,
1998). Figure 2-6 shows crater created by the explosive and the damage on the front
walls of the building whilst the overall structure remained in-tact.
Original location
of barrier wall
Figure 2-6 Damage as a result of the terrorist attack on Khobar Tower
Other than terrorist attacks on commercial and military-related structures, another
concern for civil engineers lies in the offshore industry. A lot of offshore structures
are primarily used to extract, store or transport crude oil but, due to the volatility of
such a substance, combustion of such materials will cause substantial damage to these
structures. One example of such an incident is the Piper Alpha incident. Piper Alpha
was an oil production platform which was later converted to a gas production in the
1980s. In 1999, a hydrocarbon explosion and the subsequent fire that followed caused
the death of 167 personnel and left only 59 survivors (Cullen, 1988). It was deemed as
12
one of the most tragic offshore accident in terms of lives lost. Actions were then
undertaken by the authorities to device measures to prevent the recurrence of such a
disaster, of which one of them is the use of blast walls to separate personnel and
critical machineries from hazardous areas (Louca and Boh, 2004).
2.2.1 Types of Explosives
In general, there are two kinds of explosives that would concern engineers in the
design of structures against such abnormal loads, namely, low and high explosives.
They differ in the explosive charge used, which is one of the main factors in the
determination of nature of the blast load. It is to be understood that an explosion is the
result of a chemical decomposition of a chemical or energetically unstable substance
which leads usually to sudden production of heat and pressure. Low explosives
usually involve the deflagration or burning of the material and the detonation will
occur when the chemical reaction spreads rapidly like a wave from the point of
initiation (Bulson, 1997). Hydrocarbon fuel is one the most commonly known low
explosives that should concern engineers in the design of structures.
High explosives involve the detonation of the substances as opposed to the
deflagration that is found in low explosives. They decompose extremely rapidly upon
initiation which usually involves either a mechanical blow or a smaller detonation.
This will lead to an exothermic reaction of the substances which in turn will produce
intense pressure and heat. Common high explosives are trinitrotoluene (TNT),
cyclotrimethylenetrinitramine (RDX) and petaery-thritol-tetranitrate (PETN) which
are found in military ordnance. Improvised high explosives such as a mixture of
13
aluminium nitrate and fuel oil (ANFO), which was used in the Alfred Murrah Federal
Building incident, may also pose a serious threat to the design of structures.
Due to the nature of decomposition of the substance, the loadings on the structure
from low and high explosives are different. Figure 2-7 shows the differences between
the idealised pressure-time history of low and high explosive loadings. An actual
record of a low explosive from a hydrocarbon explosion is documented by Boh et al.
(Boh et al., 2005) and one of the characteristics of low explosive is the similarity
between the ramp and decay time. Another important observation is the load duration
which ranges between ten to a hundred milliseconds. This is comparatively long as
compared to load time duration from a high explosive blast, which is in the order of
10 milliseconds or less. This comparison is not evident from Figure 2-7 as the figure
is only used to show the pressure-time shape. Ramp time is almost negligible as
compared to the decay time in the case of high explosive.
Pressure
Low explosive
High explosive
Time
Figure 2-7 Difference in pressure-time history of low and high explosives
14
2.2.2 Nature of blast loading
In this thesis, high explosives will be considered as the primary donor system. As
observed in Figure 2-7, there are phases where the pressure is positive and negative
and they are commonly known as the positive phase and negative phase of the blast
loading. In most designs, only the positive phase will be regarded and, in order to
obtain the pressure-time history, two parameters, namely, the charge weight (W) and
the stand-off distance (Z), are necessary.
In most studies, TNT is used as the benchmark by which other explosives are
quantified. The way to quantify the charge is through its weight and, for blast resistant
design, the effective charge weight WE is calculated based on the heat of detonation of
d d
actual explosive H EXP and TNT H TNT , which are obtained through design codes
such as TM5-1300 and open literature (Baker et al, 1983), and the actual charge
weight (WE). The equation is as follow (TM5-1300, 1990):
d
H EXP
WE = d WEXP (2-1)
H TNT
The other parameter to determine the nature of the blast loading is the distance
between the charge, which is known as ground zero, and the target or acceptor
system. This distance is termed as stand-off distance Z. In the calculations, it is
commonly scaled according to the charge weight to calculate the pressure-time
history as follow:
Z
R 1 (2-2)
3
W
This method used is the Hopkinson or ‘cube root’ scale law. This law states that when
two charges of the same explosive and geometry, but of different size are detonated in
15
the same atmosphere, the shock wave produced are similar in nature of the same
scaled distances (Hopkinson, 1915). This scaling method is also applied to time t.
With the knowledge of the weight and stand-off distance, the important parameters
such as the peak reflected overpressure Prmax, reflected impulse ir and positive time
duration to can be obtained through charts in design codes (TM5-1300, 1990). In
general the positive phase of the pressure-time history can be simplified to an
exponentially curve which is described by the modified Friedlander model (Baker et
al, 1980; Baker et al, 1983):
(2-3)
This approximation is applicable in the case whereby the surface of the target is
normal to the direction of the wave propagation. Other expressions that describe the
pressure-time history shape on surface such as those by the side and the back of a
structure are reviewed by Beshara (Beshara, 1992). He reviewed and provided
procedures based on available unclassified literature to analytically model external
blasts caused by different sources of unconfined explosives on aboveground
structures. It is thus concluded that “precise loading information is hard to obtain and
may be not justified because of the many uncertainties involved in the interaction
process between the blast wave and the structure and the ideal gas assumption in the
derivation of relevant relations; linear idealizations of time history of airblast loads
are consequently adopted”.
Blast shape is also another factor to be considered in the analysis of blast. Beshara
(Beshara, 1992) has elaborated on the different loading shapes that can be obtained in
16
an unconfined blast loading in aboveground structure. As for confined blasts,
numerous literatures have based their pressure load model on a bilinear triangular load
(Baker et al, 1980; Krauthammer, 1999; TM5-1300, 1990). In addition to the shock
front, which explains the fast decaying part of the bilinear curve, the gradual decay
phase of the curve is attributed to a subsequent expansion of hot combustion gases
after the explosion (Ananth et al, 2008). A similar shaped pressure-time history is
observed for walls which are subjected to unconfined blast loading in which the
pressure phase with the slower decay is known as the stagnation pressure and this is
attributed to side-on over pressure and dynamic pressure (Beshara, 1992; TM5-1300).
In order to accurately model this bilinear blast curve, a new methodology (Rickman
and Murrell, 2007) was proposed as it claims to be capable of addressing a pressure-
relief phenomenon. Comparing the results from this methodology to results obtained
through a series of small scale experiments, it predicted the onset and magnitude of
the pressure relief on a directly loaded wall quite well.
Another simplification to the blast pulse shape is proposed by Youngdahl (Youngdahl,
1970; Youngdahl, 1971) to eliminate the influence of the pressure load shape on the
permanent plastic deformation of a structure. With the introduction of the effective
pressure Peff and duration τeff, which are defined as follow:
(2-4)
(2-5)
Youngdahl further simplified the triangular load to a rectangular load which has
eliminated the dependency of the dynamic plastic deformation on the pulse shape and
17
converted the residual displacement of a structure to be a function of only the impulse
and effective pressure for a rigid, perfectly plastic deformation.
2.3 Background on Protective and the Protection of Structures
In the design of structures against blast loading, there are, in general, two types of
protective measures to be considered: active and passive. Active protection involves
the implementation of non-structural means to mitigate the hazard by preventing the
detonation or ignition of the blast. Measures such as the provision of security
personnel around the perimeter of a structure, the surveillance of human traffic
through check points and regular checks on articles and goods are just three of the
numerous ways in active protection. Such steps are more economical and efficient in
mitigating blast effects as all the expected effects from a blast are eliminated.
However, it is not sufficient to just implement active measures as most of them are
prone to human error and this could be further aggravated by the fact that, as per all
accidental loading, the arrival time and place are not predictable. As a result, passive
measures must be applied. They can be defined as the precautions taken into the
design of the structure to minimise the structural, asset and personnel damage in the
case of an actual bomb blast. There are two approaches to minimise these damages,
direct and indirect (McCann and Smith, 2007). In view of the scope of the study, the
review will be based on the indirect approach. This section of the chapter reviews the
current state-of-the-art technology in mitigating blast effects by passive indirect
means.
18
2.3.1 Protection of Key Structural Elements
In the economical sense, it is impossible to ensure the protection of every single
element in the structure. Thus, it is important to identify the areas or structural
elements of a building that are more important which require more protection than the
rest of the structure. The concept of key elements is an idea to ensure all the principle
elements of the structure are identified and protected against blast loading. Some
simple guidelines in their identification are:
 Removal of element will lead to progressive collapse
 Damage of element will cause surrounding elements to fail
 Elements which are prone to attacks due to easy access
 Elements which are used to protect personnel and assets
Hence, there is a need to analyse structures to identify them. These elements may be
either hardened through the enhancement of their dynamic capacity or by the
protection with the use of another material or structure component which is applied
between ground zero and the elements. These key elements are usually columns but
other elements such as beams, floor slabs and even shear walls can also be identified
as key elements.
There are several means to protect these structural elements against blast. One of the
most common methods is the use of blast walls as illustrated in Figure 2-8. Figure 2-
8(a) shows the use of stainless-steel profiled steel panels on an offshore structure
(Schleyer, 2007). Figure 2-8(b) illustrates the use of geotextile and it has been widely
used in military operations in desert environment (http://www.defencell.com). Lastly,
Figure 2-8(c) shows the combined utilization of steel and concrete in a blast wall
(Crawford and Lan, 2006).
19
(a) (b)
(c)
Figure 2-8 Examples of blast walls
Regardless of the materials used, the main purpose in the use of these walls is to
reflect, absorb and diffract the blast waves and reduce the pressure that is transmitted
onto the key elements. Methodologies to calculate the reduction of pressure behind
the blast wall is present but it is restricted for official use only and thus unavailable to
commercial or academic usage. Furthermore, these formulations are not validated
with full-scale experiments (Remennikov and Rose, 2007). Numerical simulations
have shown also that effectiveness of blast walls in attenuating the pressure and
delaying the arrival time of the blast (Ngo et al, 2004; Zhou and Hao, 2008).
In view of the distribution limits of the guidelines to calculate the pressure behind a
barrier, several formulations have been derived in the literature to further the research
in improving the design guidelines of blast walls. Zhou and Hao used a numerical
20
simulation package called AUTODYN to derive approximate but reliable formulae to
estimate the reflected pressure-time history on a structure behind a barrier (Zhou and
Hao, 2008). Based on an approximately one-tenth scale target structure, Chapman et
al. developed a prediction technique to quickly assess the peak reflected and impulse
on a full scale structure behind a blast wall (Chapman et al, 1995). However it is
commented that a full scale experiment would be desirable to further validate these
claims.
In addition to the more traditional use of blast walls, other means to protect key
elements have also developed. One of the more recent innovations is the use of water
in mitigating blast effects on structures (Schwer and Kailasanath, 2007). Water was
either placed in proximity of the explosives or it is sprayed around as a mist. Schwer
and Kailasanath claimed that through their numerical study, water-mist can be used
to mitigate the shock-front pressure through the extraction of energy and momentum
in an unconfined environment. It is almost similar to having a denser ‘water-wall’
surrounding the explosive, provided that the total mass of the water is similar.
However, this type of protection, together with that of geotextile, will not be further
considered in the thesis as the study will be solely based on conventional structural
materials.
2.4 Methodologies in Explosive Dynamic Analysis
There are numerous approaches to analyse the response of structural elements to blast
loading. However the common goal of it all is to finally put all the work into design
guidelines to aid engineers in the construction of blast resistant structures. In recent
years, experimental programmes are conducted more frequently by non-military
21
research institutions but, due to the lack and limitation of past experiences on such
tests, many of these works have not been fully utilised. Therefore, the objective of
these studies is to provide simplified models of the transient blast load and dynamic
response of the structural elements so as to expand the knowledge in this field. This
section will expound on the progress made through the years in terms of analytical,
experimental and numerical methods.
2.4.1 Analytical Methodologies
The simplest and thus most popular analytical model that is used in research is the
single-degree-of-freedom (SDOF) method, which has been used extensively in the
design of structural elements (Biggs, 1964; TM5-1300, 1990). This method simplifies
the continuous structural elements under dynamic distributed loads into single point
masses which are point loaded. In addition, the blast resistance of the structure is
simplified to a spring. Figure 2-9 illustrates this model and Biggs provides an in-depth
explanation on the principles and derivation of the SDOF method (Biggs, 1964). This
model has been correlated with experimental results.
Pe(t)
P (t)
ymax
ymax
Ke
Figure 2-9 Simplification of SDOF method
22
There are some deficiencies in the SDOF method, which need to be addressed. As a
result, modified versions of the SDOF approach could be found in literature to obtain
more accurate results. Krauthammer et al. proposed an advanced SDOF approach
which took into consideration the tensile and compressive membrane effects, rate
effects and shear response of reinforced-concrete slabs (Krauthammer et al., 1986).
For steel structures, the rate sensitivity was also added to improve on a proposed
SDOF method, which took into consideration the general support conditions both in
flexibility and strength as well as the caternary action due to the axial restraint at the
support (Izzuddin, 2001; SCI, 2006). An overview on the use and history of SDOF
approach is reviewed and summarised by Morison (Morison, 2006) and he concluded
that the flaws in SDOF model for two-way spanning members and the failure to
provide any information with regards to the time history of the reaction forces from
tables in TM5-1300 and by Biggs must be noted in the use of such a simplified model
in design. However, the distribution of the documentation relating to the original
proposed model is limited.
With regards to one-way spanning structural elements, SDOF approach, as its name
suggests, is only capable of analysing one response type at a time and it is not able to
capture the multi-failure modes (Kang, 2005). It is very common that structures react
both in bending and shear under high intensive blast loading. With SDOF, methods
such as the use of loosely coupled and uncoupled SDOF systems have been designed
to obtain the shear and flexural response (Low and Hao, 2001; Krauthammer, 1986).
However, it is also argued that uncoupled analysis may yield inaccurate results due to
closely spaced natural frequencies between structural members and components (Jan
and Gurbuz, 2008). This could be improved through the use of an alternative
23
approach. A rigid-plastic assumption for ductile materials such as steel and various
composite structures can be made to derive a closed-form formulation to derive the
blast response (Jones, 1989; Kang, 2005; Kang and Liew, 2006).
In view of the constant progress in search of the appropriate formulations in the
design of blast-resistant structures, it is only natural to turn to design codes that are
currently available in the industry. Ngo et al. (Ngo et al, 2007) provides a
comprehensive coverage of various military design manuals, which can be applied to
commercial designs to predict blast loads and the response of structural systems,
although some of them may not be readily available to the public. Other available
design codes include the UFC-010-01, which provides several guidelines in designing
buildings to avoid mass casualties in times of terrorist attacks, PDC-TR 06-08, which
advices on the use of SDOF in anti-terrorist design of structures, and EM-1110-345-
417, which advices on the blast design of single-storey structures. Section 5 of
Eurocode 1 Part 7 provides several design guidelines to resist internal blast although
they are very general as opposed to the preceding ones that are mentioned.
In conclusion, analytical solutions allow designers to efficiently approximate values
in the design of structures. However, most of these methods and guidelines are very
general and they do not cover the full spectrum of structures. Therefore, it is
important to further the study through actual experiments to validate the design.
2.4.2 Experiment Methodologies
There are far less publications on experimental works to investigate the blast response
of structures. Unlike laboratory testing of structural materials and members using
24
quasi-static loads, there has yet to be a single institution to standardise blast test
procedures. Therefore, the only way to device actual blast tests of structural members
is through the review of the available literature which are based on actual blast tests.
In general, there are two kinds of blast test: small-scaled and full-scaled experiments.
Small-scaled tests are usually conducted in laboratories and thus they have the
advantage of having the experiment under controlled conditions, which minimises the
uncertainties during the test. Another advantage over a full or large scale experiment
is the economisation of logistics, equipment and the materials used.
Under laboratory conditions, some tests involved the use of actual explosive charges
while others utilised innovative methods to simulate the blast pressure. Silva and Lu
(Silva and Lu, 2006) tested one-way RC slabs with nominal dimensions of 1200 x
1200 x 90mm using RDX charges, which was varied between 0.45kg and 1.35kg with
standoff distances of 300 and 910mm. The experimental results correlated well with a
simplified analytical solution to gauge the damage level of the RC slabs. Another lab-
based blast test was demonstrated by Leppanen to investigate the combined effect of
blast waves and fragmentation (Leppanen, 2004). A method of maintaining the stand-
off distance in small-scaled laboratory-based tests is done by separating the specimens
and the explosive by polystyrene foam (Langdon et al., 2003). Yuen and Nurick
devised an improved approach to test steel plates in sandwiching the test specimen
between two support plates and laying the explosives in the form of an annuli as seen
in Figure 2-10 (Yuen and Nurick, 2003). This ensured that the uniform distribution of
explosive load over the specimen as opposed to a concentration of explosives at one
location of the specimen.
25
Figure 2-10 An example of a lab-based scale experiment setup
Other than the use of explosive charges, there are also other alternatives. LPG and
oxygen when ignited has been used to generate pressure to be applied on panels, as
seen in Figure 2-11 (Turkmen, 2002). A detonation tube was used to concentrate and
direct the pressure on the specimen. Zyskowski et al. conducted small scale test with
the use of explosions of a gaseous mixture cloud (hydrogen-air) to correlate with the
analytical approach for an internal explosion in a unvented area (Zyskoski et al.,
2004) and the result was that only the normal incidence overpressure was fairly
accurate.
26
Figure 2-11 Laboratory test setup in the use of LPG as explosives
Actual scaling of structures was also put into practice. Assumption in charge
geometry is required to obtain reasonable correlation between small and large scale
experiments (Smith et al., 1992) and it is recommended that scales of 1:30 to 1:40
would yield more satisfactory results. A one tenth scaled test was conducted to
investigate the pressure behind the cantilever wall and pressure transducers were
positioned on 4 points on a structure behind the wall (Chapman et al., 1994). The test
utilised TNT weighing between 0.03 to 0.08kg. The structure was constructed of
plywood and a 20mm steel plate simulated the wall. The charge was placed on top of
a layer of polystyrene foam, under which a steel plate was positioned. This technique
allowed “the assessment of peak reflected pressure and reflected impulse at a point on
a full-scale target structure”.
27
With respect to large scale blast tests, less data is found in literature. The testing of the
composite reinforced concrete and steel-concrete sandwich slabs are well documented
in various articles and the air blast parameters were in good agreement with design
values with TNT charge weights ranging from 8 to 100kg (Lok et al, 1996; Lan and
Heng, 2002; Lan et al, 2005). The use of crude plasticine displacement gauges was
used and produced reasonable data. Schenker provided another well documented full-
scale experiment on concrete walls (Schenker et al., 2006). He moulded 1000kg of
TNT charge to a hemi-sphere to best simulate the hemi-spherical shock front of the
explosion and he also used a crude device known as displacement “combs” to gauge
the overpressure from the blasts. Similar to small scaled tests, there are alternative
sources to obtain the blast pressure such as a shock tube. Figure 2-12 shows the size
and design of the shock tube and Figure 2-13 shows a sample of the experimental
pressure-time variation that can be produced by the shock tube (Schleyer et al., 2007).
This machine was fabricated to test steel panels. Another alternative method is the
combined use of the drop hammer impact machine and an airbag, which is used to
distribute the dynamic load (Jones, 2007).
Figure 2-12 Baker Risk large shock tube (Schleyer et al., 2007)
28
Figure 2-13 Pressure-time history obtained from the large shock tube
2.4.3 Numerical Methodologies
This section will present a brief introduction to dynamic finite element analysis and
the fundamental theories that are used the numerical analysis in this thesis. The latter
part of this chapter will attempt to summarise the important milestones achievements
in the use of finite element analysis to study the blast response of structures.
One of the most common methodologies in the numerical simulation is the use of
finite element method. This approach attempts to analyse the response of actual
processes such as heat transfer, fluid motion, electromagnetism and structural
response by discretising a continuous structure into numerous finite elements
connected by nodes, which would form a mesh. Based on this mesh, a series of
differential equations are solved approximately. This review will focus on the
approach and terms necessary to solve structural mechanical problems.
29
The basis of finite element method is the formulation of the equation of motion of the
elements within the structure. The most popular derivation is based on creating a so-
called weak formulation of the equation of motion with the use of the principles of
virtual work (Ottosen and Ristinmma, 2005). Based on the second law of motion and
the divergence theorem of Guass which relates the area integral of the divergence of a
vector field to the contour integral of a vector field, the following equation, which
holds for an arbitrary region V of a body, is formulated:
ij , j  bi  ui (2-6)
where ij is the stress tensor, bi is the body force vector per unit volume,  is the mass
per unit volume and ui is the displacement vector with the dot notations representing
the order of differentiation with respect to time. By multiplying the Equation (2-6) by
a weight vector i and re-arranging this equation, the weak formulation of the
equation of motion is achieved as follow:
  u dV     dV    t dS    b dV

V
i i
V
ij ij
S
i i
V
i i (2-7)
where ij is the strain tensor that is related to the weight vector, ti is the traction
vector and S is the outer surface of an arbitrary body.
As mentioned earlier, the finite element approach attempts to discretise continuous
structures and the scalar variable displacement solutions u which is obtained by
relating the nodal displacements ai in the following manner (Zienkiewicz and Taylor,
2000):
30
n
u   N i ai  Na (2-8)
i 1
where Ni is the shape function adopted by the user in presenting the element and n is
the total number of such functions used.
Assuming that element strain i can be related to the element displacement by a
differential operator S, it can be defined as follow:
i  Su  SNa  Ba (2-9)
Using Equations (2-8) and (2-9) into Equation (2-7) and by manipulating the
equation, we are able to derive the following fundamental equilibrium equation of
motion in finite element analysis:
Ma   BT dV  f (2-10)

V
with the mass matrix, M   ρN T NdV and external force vector,

V
f   NTtdS   NTbdV . Equation (2-10) is applicable in cases of dynamic analysis

S V
such as those that are used extensively in the study of structures subjected to transient
blast loading. Under linear static analysis, the nodal acceleration term is neglected and
the problem can be simplified with the elastic stiffness matrix D as follow:
Ka  f where K   BT DBdV (2-11)

V
However, in the blast analysis as well as various other studies, large displacement and
plastic response of the structure are normally taken into consideration. The numerical
study that was conducted throughout this thesis was done using the finite element
package LS-DYNA. With this tool, non-linearity can be applied to both geometry of
31
the elements and the material model. If non-linear material is being considered,
Equation (2-11) would no longer be valid. Instead, a different approach with the use
of a path-dependant stiffness matrix Dt, which relates the stress rate σ and strain rate
ε , is more applicable and that will yield the following expression:
K t a  f where K t   BT Dt BdV (2-12)

V
Prior to solving Equation (2-12), the matrix M, K and f must be derived and that will
involve the integration of the individual terms in the matrix. These operations can be
estimated numerically with the Gauss Quadrature technique which is one of the most
efficient techniques in solving polynomial or nearly polynomial functions (Fish and
Belytschko, 2007).
There are generally two approaches in solving Equation (2-12), which are namely
explicit and implicit methods. The explicit approach is the simpler of the two
methodologies. For dynamic analysis, the solution is based on the second central
difference approximation which estimates the acceleration of time step n on the
displacement of the previous two steps by the following expression:
an 
1
an1  2an  an1  (2-13)
t 2
By using Equation (2-13) in Equation (2-10), the following expression is derived to
solve the nodal displacements of the next time step:

Ma n1  M2an  an1   t 2 fn   BT n dV
V
 (2-14)
The other approach in solving the differential problem in Equation (2-10) is the use of
the implicit scheme. As opposed to the explicit scheme which utilises the information
from the preceding steps, this method requires information from the current step in
32
order to proceed. One of the most popular methods of the implicit scheme is that of
Newmark-β method (Weaver and Johnston, 1987) which approximates the velocity
and displacement as follow:
a n1  a n  1   an  an1 t (2-15a)
 1  
an1  an  a n t     an  an1  t 2 (2-15b)
 2  
The parameter γ produces numerical damping while β controls the variation of
acceleration within the time step. Another popular implicit approach is the Wilson-θ
method which assumes that the acceleration varies linearly over an extended time step
t  tn . The value of θ is normally rounded to 1.42.
With such fundamental understanding of numerical analysis, numerical simulation
with finite element analysis can be correctly conducted to validate experiment as well
as analytical results. In addition, these studies can also aid in performing parametric
study of structures which are subjected to blast since it is resourcefully more
economical as opposed to actual experiments.
2.5 Structural Materials under Blast Loading
This last section will present the research that has been carried out on various
structural materials and the contributions from these findings to the field of structural
analysis against blast loading. Four materials will be covered in this section: masonry,
reinforced concrete, steel and, lastly, composite.
33
2.5.1 Masonry
In the current context of modern structural design, masonry has been used mainly as a
non-structural component of a building. The low carrying capacity of such a material,
especially in the transverse axis, as compared to its contemporaries has left it to be
used primarily in compartmentalising floor areas. However, research into the effects
of blast on masonry structure is for the purpose of personnel protection due to the
possibility of fragmentation of the individual bricks.
The continuity of bricks to ensure the integrity of the wall was the main idea in the
protection of masonry walls. Different materials were used for this process. Fibre
reinforced polymer (FRP) which is wrapped along the surface of masonry walls has
proved to ensure the strengthening of the wall and this technology is well documented
by Buchan and Chen in the state-of-the-art review of such structures against blast
(Buchan and Chen, 2007). The overall consensus from this review is that FRP
improved the strength as well as the ductility of the material but the anchorage as well
as the ability to prevent the debonding of FRP from the walls are the challenges.
Polyurea is also another material that has been used to improve masonry walls. Sold
commercially, it can be procured either as films that are applied on the surface of the
masonry wall or in liquid form which is sprayed onto areas which are to be protected.
An example of the former is Life Shield ® Panel Technology and of the Defend-X ®.
Both companies have demonstrated the effects of protection through experiments and
results can clearly be observed in Figure 2-14. Experiment and numerical studies was
also conducted to observe the effects of masonry wall as a non-structural component
34
on the loading of the columns and the loading inside the structure (Baylot and Bevins,
2007).
(a) (b)
Figure 2-14 (a) Before and (b) after effects of protected and non-protected masonry
wall using Life Shield ® Panel Technology
2.5.2 Reinforced Concrete
One of the most researched materials in the field of passive structural protection is
reinforced concrete (RC). Research work on RC has been done extensively through
analytical, numerical and experimental studies. This can be attributed to the brittle
nature of concrete which is not desirable in providing blast resistance, which depends
mostly on the ductility of the material. This is clearly demonstrated through the study
of the collapse of the Alfred Murrah Federal Building which showed the breaching
effect of the three columns that initiated the progressive collapse of the structure
(Mlakar et al, 1998).
The design of RC structural members can be widely found in manuals and literature
(TM5-1300, 1990; Baker et al, 1983) has also been published to compile various
standards to follow in the design of such structures (McCann and Smith, 2007).
Despite the fact that they are brittle, the mass of concrete will prove important in the
35
resistance of blast load due to the dynamic nature of the response under blast loading.
Another important characteristic of concrete is in its increase of compressive capacity
under confinement. A review of past experiments on RC roofs by Ishikawa and Beppu
evaluated the design of roofs to withstand 250kg bomb drops from a height of 800m
(Ishikawa and Beppu, 2007). Critical curves were proposed to estimate the safe stand-
off distances to resist terrorist bombings scenarios are studied numerically and
presented (Zhou and Hao, 2008).
Through the years of research, concrete mix has also advanced and these new mixes
are put to the test through a series of experiment and numerical analyses. Scaled
experiments showed that steel fibre reinforced concrete exhibited lower damage with
a higher degree of reinforcement and longer fibres are preferred in resisting cracking
and spalling damage (Lok et al.,1996). High strength concrete (HSC) columns are
studied numerically by varying the reinforcement detailing, geometry and loading
conditions (Ngo et al., 2003). HSC performed better under impulsive loading as they
exhibited higher energy absorption.
In order to improve the blast resistance of current RC structures, retrofits are needed
and there are several materials which can be used. One of the most common retrofit in
the market today is the use of fibre reinforced polymer (FRP). They are being applied
on all the exposed surfaces of the structure and their application improves the ductility
and capacity of the structure (Silva and Lu, 2007). A very detailed review of the use
of FRP is currently published by Buchan and Chen and their review showed the
effectiveness of FRP in improving the blast resistance of various structural RC
components: beams, columns, slabs and walls (Buchan and Chen, 2007). The
36
economy in the use of various fibres is also presented in this journal paper. Carbon
fibre seems to be one of the better choices due to its stiffness, strength and cost. Other
more common fibres include glass and steel fibres. With respect to the improvement
in shear capacity, the above mentioned two citations seem to contradict each other.
More studies have to be made in this aspect. An example of the improvement through
the use of FRP is shown in Figure 2-15. The use of aluminium foam has also been
introduced and a full scale test was conducted. However, its effectiveness in
improving concrete slabs is debatable and needs to be further studied (Schenker et al.,
2008).
(a) RC Column (b) FRP Wrapped Column
37
(c) Shear Failure on RC Column (d) Concrete intact on FRP Column
Figure 2-15 Improvement in the use of FRP in columns (Crawford et al., 2001)
2.5.3 Steel
As opposed to concrete, steel possesses the ductility as well as strength to resist blast
loading. However, due to its cost, steel structural elements tend to be slender and less
massive than concrete structures. As a result, failure modes such as global and local
buckling are common failure modes of steel structures which are subjected to blast
loading. In general, steel structure members that are subjected to blast loading have
three failure modes: (a) Mode I, large ductile flexural deformation failure, (b) Mode
II, tensile tearing at support failure and (c) Mode III, transverse shear failure at
supports, as illustrated in Figure 2-16 (Jones, 1989).
38
(a)
(b)
(c)
Figure 2-16 Different modes of failure of beams under blast loading: (a) Mode I, (b)
Mode II and (c) Mode III
Currently, the most common approach in designing steel structure members is through
the SDOF approach that has been introduced through design manuals (TM5-1300,
1990) and further encouraged through practicing engineers and researchers (Biggs,
1964; Krauthammer, 1999; Longinow and Alfawakhiri, 2003; Ngo el al., 2007).
Similar to the research on RC structures, the focus on research has been based on the
analysis of the response of key elements and one of them is the research on columns.
Unlike beams, columns are subjected to both lateral and axial loads thus the ductility
limits assigned by TM5-1300 for steel may not be always applicable (TM5-1300,
1990; Barker, 1993). Liew and Chen modeled and analysed the effects on a single
steel column which is subjected to blast loading (Liew and Chen, 2005) and that
model is further improved with the addition of moment-thrust load to define a failure
surface that characterises the effects of shock on the component (Leigh and Earls,
39
2008). Other methods and standards can also be found in literature for the design of
columns (Kang, 2005; Godinho, 2007) but the standardisation in the design is far
from being concrete. Experiments on columns which are subjected to blast loading are
scarcer as compared to other components due to the fact that it is very hard to impose
an axial load on the member whilst loading it under an explosion. However, Morrill et
al. conducted such a complex experiment to test the beam-column connections under
blast loading (Morrill et al, 2006). Exact details of the setup are not fully evaluated.
Other researchers have also focus on the local effects of blast through the evaluation
of the local buckling of steel columns (Lee et al, 2009).
Another steel component that received massive attention is the wall panel. Although
concrete may be more suitable as barriers, they are simply too massive to be used in
applications such as marine and offshore structures. Therefore, steel is used in such
situations. Early research in blast resistant steel wall is geared towards the ship
construction design (Houlston et al, 1987; Houlston and DesRochers, 1989; Slater,
1994). Numerical and experimental studies have studied the use of stiffeners to
improve the blast-resistance of these panels. Such design was then extended to
offshore platforms which are prone to hydrocarbon explosions. Analytical solutions
were proposed in the design of such panels (Louca et al, 1996) and experiments and
numerical simulation were carried out to show the adverse effects of over-stiffening
such panels (Yuen and Nurick, 2005). Different configurations of stiffeners were used
and results showed that by increasing the stiffness either by increasing the thickness
or number of stiffeners will results in a lower displacement. However, it should also
be noted that the panel response also changed from a flexural response to tensile
tearing response with the increase. Similar tests were conducted based on localised
40
blast effects and they showed that these loadings may lead to an increase in the
threshold impulse up to a factor of 4 as compared to uniform distributed load
(Langdon et al, 2005). It is also observed that stiffeners play a much larger role in
reducing the deformation in localised blast.
Corrugated and profiled panels were also studied and in view of the difference in
boundary conditions of such panels, new methodologies such as introducing new
transformation factors and the incorporation of rigid-plastic assumption had to be
proposed to obtain more accurate results (Liang et al., 2007). Experiment and
numerical simulation were carried out on stainless steel panels in which the study
focused on the effect of the boundary condition design on the overall dynamic
response of the structure (Langdon and Schleyer, 2005; Langdon and Schelyer, 2006;
Schleyer et al, 2007). It was observed that the end connections which form the
boundary condition may affect the onset of the membrane effect. Subjected to large
impulsive loading and producing large displacements, such panels, which yield with
large in-plane forces, require optimisation to ensure that these forces do not cause
extensive damage to the primary structures. Buckling resistance is also appraised to
perform well for corrugated panels. To further improve these panels, an innovative
passive barrier system is introduced to reduce the membrane effects and minimise
tearing of the horizontal welds at the ends (Boh et al, 2005). In these numerical
simulations, strain rate effects and failure criteria were considered in the model to
produce agreeable results.
Other novel ideas include a sandwich system which positions a layer of corrugated
plate between two flat steel plates (Liang et al., 2001) and other systems that use
41
metallic square honeycomb and I-core as core materials to improve the blast resistant
of these steel panels (Liang et al., 2007)
2.5.4 Composite (Steel-Concrete)
There are numerous composite materials in the market and even more that are in the
process of research. Therefore, in keeping to the objective of the thesis, the focus of
this review will solely be based on the blast resistance of steel-concrete composite
structures. Unlike reinforced concrete and steel structural members, there is currently
little work to ascertain the failure limits for steel-concrete composite structures. In
fact, research into the blast resistance of steel-concrete composite structures has been
less extensive as compared to the work done on the individual components
themselves.
The most common composite structural element that has been researched into is the
wall panel. Bi-steel utilises both steel and concrete to fabricate a panel (Pryer and
Bowerman, 1998; Corus Group Ltd, 1999; Bowerman and Chapman, 2000). Concrete
is grouted between two steel plates which are held 200mm apart by shear connectors.
These shear connectors are friction welded onto the steel plates to ensure composite
action between steel and concrete. Bi-steel has published information on the design
and construction of these panels against static and various other kind of loadings
(Corus Group Ltd, 1999; Yu et al., 2006; Xie and Chapman, 2006; Foundoukos et al.,
2007, Xie et al., 2007) but information with regards to the study of the blast resistance
of these panels has been very limited (http://www.corussecurity.com/en/
design_and_innovation/blast_protection/; Bowerman et al, 2002).
42
However, there are other data on the blast capability of such construction. An early
documentation of such works showed that the confinement of the concrete core yields
responses which were lower than those predicted through analytical methods such as
the SDOF method (Heng et al., 1995). A series of tests on such sandwich structures
were tested by the Defence Science and Technology Agency (DSTA) of Singapore
and showed that the use of concrete core led to a significant deflection reduction
under blast loading and it performed relatively well under repeated loadings (Lan and
Heng, 2002; Lan et al, 2005). Preliminary works on producing an analytical solution
for the blast design of such sandwich panels has also started (Kang et al., 2007).
Other steel-concrete composite structures that have been tested against blast include
the use of profiled steel plates and they performed relatively well as compared to steel
reinforcement bars (Lan and Heng, 2002; Lan et al., 2005). Such tests could be
applied in the construction of floor panels in composite structures.
A key element that has received little attention is columns. Research on composite
columns has not been found in existing literature and thus special attention has to be
made to study the response of such structural elements to blast loading.
43
CHAPTER 3:
ANALYSIS OF CONCRETE FILLED STEEL TUBULAR COLUMN
SUBJECTED TO BLAST LOADING
3.1 General
One of the main key elements in the design of a structure is the column. The column
is one of the most important structural components because if redundancy is not
properly incorporated into the design of the building structure, disproportionate
collapse may result. This can be clearly seen in various examples that have been
discussed in the previous chapter. This chapter will attempt analyse concrete filled
steel tubular (CFST) column which is subjected to blast loading. These proposed
analytical solutions may provide a deeper insight into the design of CFST columns as,
at present, there is yet to be a design code that has been specifically devised to design
composite structures under blast loading. However, as it will be observed in this
chapter, the design of the steel-concrete composite columns are based on Eurocode 4
(Eurocode 4, 2004), which has similarities to the design of steel structures such as the
assumption of elastic-plastic material response and deduction of shear and moment
capacities. It will be seen how such assumptions will aid in the formulation of an
effective analytical and physics-based methodology in the design of such composite
columns against blast loading.
The use of CFST columns has gained popularity due to its advantages as compared to
those of conventional steel and reinforced concrete (RC) (Liew, 2004). These
columns have higher axial capacities due to the interaction between the steel section
and concrete core. In addition, the buckling tendency of steel sections is reduced in
due to the high flexural rigidity of the composite structure. The confining effect of
concrete by the steel section also enables relatively higher transverse ductility and
44
capacity to be achieved as compared to their RC counterparts of the same dimensions
(Eurocode 4, 2004). In view of the various advantages when subjected to static loads,
these columns should be studied under blast loads to observe whether such traits are
also applicable in dynamic situations.
An investigation is first carried out to study the material response under dynamic
loading. Blast loadings, especially close-in explosions, subject structures to be
subjected to much higher strain rates than in cases where structures are under static
loads. The strain rate effect will influence the material response by altering the
material strength and the resistance of the structures involved. In this study, a brief
experimental study using the Split Hopkinson Pressure Bar (SHPB) will be carried out
to investigate the high strain rate effects on concrete. This is done to provide further
proof of a hypothetical scientific theory behind the effects of high strain rate as well
as compare the experimental results with those that are found in literature such as
design codes, experiment and numerical research.
In view of the difficulty in obtaining blast experimental data on CFST columns to
compare and validate the analytical results, the validity of the Single-Degree-of-
Freedom (SDOF) method will be compared to a dynamic impact test that utilises an
airbag to distribute the forces across the span of square hollow section. This test was
being conducted in collaboration with Dr. Alex Remmenikov and Mr. Kong Sih Ying
from the University of Wollongong. The test setup and the specimens that were used
will be documented in the chapter. The results obtained from the experiments will
then be compared to the analytical method. This was done so as to verify the accuracy
of this analytical method.
45
As shown in the literature review, SDOF approach does have certain deficiencies and,
as a result, another method is proposed and described in this chapter. This approach
assumes that the composite moment and shear capacity of the steel and concrete
material are rigidly plastic in nature. The elastic response of the material is ignored
since the final deformed response exceeds the elastic deformation limit by a few
orders. This chapter will attempt to explain the basis of this method and use the
assumptions and basic mechanics to derive the essential equations to build an easy-to-
use table for the design of steel-concrete composite columns under blast loading. This
approach will be compared to the more standardised SDOF method and their
differences will be examined and explained.
There was not any information in open literature that documents the response of an
axially loaded CFST column. Such tests are difficult to conduct due to the reliability
of the mechanism to apply the axial load on the member under intense blast loading.
Therefore, in order to affirm the analytical results, numerical modelling is harnessed
and conducted. The results can be used to further the understanding of axially loaded
CFST column.
Therefore, in view of the need to conduct a numerical study on CFST columns, the
non-linear finite element solver LS-DYNA is adopted as the tool to carry out this
study. LS-DYNA is predominantly used in crash and dynamic analysis and it will be a
useful and reliable tool to assess the response of these structural components against
high dynamic loads.
46
3.2 Material Response Under Dynamic Loading
The two materials that are under the scope of study on composite structures are steel
and concrete. It is suggested that under high strain rates, the properties of the
materials will deviate from those which have been derived from static loadings. Strain
rate is evaluated as the change in strain over a period of time, as indicated in Equation
(3-1).
 
 (Discrete)
   t
d (3-1)
 (Continuous)
 dt
Blast, earthquake and impact loadings are some of the common scenarios that will
subject specimens to high strain rates. Of the two materials, the study on the effects of
high strain rate will be limited to concrete. The objective of this study is to determine
the dynamic properties of concrete experimentally through the use of the Split
Hopkinson Pressure Bar (SHPB) and, from the results, demonstrate that one of the
hypothesis on physical mechanism behind the dynamic properties of concrete. The
former has been studied extensively by numerous researchers but the latter has yet to
be conducted experimentally. Therefore, although this study may be academic and it
may not be directly applicable in the design or modelling of composite structures, it
proves to be a step ahead in understanding the natural phenomenon that is behind the
effects of high strain rate on concrete.
3.2.1 Concept of High Strain Rate Effects
Under dynamic loads such as blast, earthquake and impact, the rate of loading differs
from static loads to which conventional structural elements are designed. This can be
attributed to the relatively shorter duration of these dynamic loads and the magnitude
of the force and acceleration to which structures are being subjected. As a result of
47
these two conditions, the materials will be subjected to high strain rates. Figure 3-1
illustrates the general range of strain rates from different loadings (Bishcoff and Perry,
1991).
Figure 3-1 Range of strain rates under different loadings
3.2.1.1 Fundamentals
The strength of the concrete is one of the main characters that is altered when
subjected to high strain rates. It is related to the change in stress flow under different
loading rates. The change in the concrete strength is one of the main deterministic
factors in the design of structures which involves the use of concrete against blast
loading. One of the most common terms that describe this change is called the
Dynamic Increase Factor (DIF) which can be defined as the ratio of the dynamic to
the quasi-static strength. Under dynamic loading, concrete will exhibit an
enhancement in strength as compared to the response from quasi-static loading.
Therefore, DIF is higher than unity. Due to this dynamic strength enhancement of
concrete, designs of structures against dynamic loads such as blast will be less
conservative as compared to designs that ignore the DIF. In view of the benefits in
considering DIF, much interest has been raised in order to quantify these values
accurately to optimise the designs of structures subjected to blast loading. One of the
premier design bodies to publish the use of DIF values of concrete in design is
European Comité Euro-International du Béton who recommended DIF formulas for
concrete in both tension and compression (Comité Euro-International du Béton, 1993),
which are defined as follow respectively:
48
  1.026
for   30s
1
 
 S 
DIFcomp  1 (3-2a)
    3
for   30s
1
  S  
  S 
  1.016
for   30s
1
 
 S 
DIFten   1 (3-2b)
    3
for   30s
1
    
 S 
,where α = 1/(5 + 9fc/fco), log γs = 6.156α – 2, δ = 1/(10+6 fc/fco) log β = 7.11δ - 2.33
and fco = 10MPa
Below are three other DIF expressions that describe the enhancement in compressive
strength of concrete which differ from those in Equation (3-2) (Ross et al, 1995;
Grote et al, 2001; Li and Meng, 2003) and their differences is highlighted in the
Figure 3-2. It is observed that the CEB values are less conservative as compared to the
other three expressions, whose values are close to one another. Figure 3-3 illustrates a
compilation of the variation of DIF with strain rate under compression (Bishcoff and
Perry, 1991) and, by comparing the two figures, it is observed that CEB values are
based on these literature values.
0.00965 log   1.058 for   63.1s

1
DIFcomp   (3-3)
 0.758 log   0.289 for   63.1s
1
 1  (log   3)0.03438 for   266s

1
DIFcomp   (3-4)
1.729 log   7.1372 log   8.5303 for   266s
2 1
for   100s
1
 0.00235 log   1.07
DIFcomp  2 (3-5)
0.882 log   4.4 log  7.22 log   2.64 for   100s
1
3
49
6
4
CEB, 1993
DIF
3 Ross et al, 1995

Grote et al, 2001
2 Li and Meng, 2003
0
0 200 400 600 800 1000 1200
Strain Rate (s-1)
Figure 3-2 Difference between DIF from various sources
Figure 3-3 Compilation of DIF values from experimental data (Bishcoff and Perry,
1991)
The measurements of DIF under dynamic loadings can be experimentally conducted
using several techniques such as drop-hammer tests, split Hopkinson pressure bar
(SHPB) and explosive test setups. The current work will utilise the SHPB to ascertain
the parameters and mechanisms involved in the high strain rate effects that are
observed in concrete.
50
In the design process of assessing the material strength of the structure, fixed DIF
values are factored to the design strengths. These values are summarised in Table 3-1
(TM5-1300, 1990).
Table 3-1 Dynamic increase factor for yield strength of (a) steel and (b) concrete
Dynamic Increase Factor (DIF)
Yield Strength
Material Yield Strength
(Axial Compression Ultimate Strength
(Pure Bending)
and Bending)
Steel S275 1.36 1.24 1.10
Steel S355 1.24 1.15 1.05
Reinforcement Steel 1.23 1.13 1.05
(a)
Dynamic Increase Factor (DIF)
Ultimate Strength
Material Ultimate Strength
(Axial Compression and Bending)
(Pure Bending)
Concrete 1.25 1.16
(b)
3.2.1.2 Concepts behind the Phenomenon
In view of that high strain effects are exhibited when concrete is subjected to intense
dynamic loadings, there are several hypotheses on physical mechanism that explains
the material response under high strain rates, amongst which there are two major
groups of thoughts which prevail. One of them states that the strain rate effects
observed is due to confinement of concrete (Li and Meng, 2003; Forrestal et al,
2007). Through numerical analysis Li and Meng concluded that the lateral inertia
force of the specimen increases the lateral confinement in a SHPB test, which causes
an apparent increase of the DIF for concrete and concrete-like materials. This effect
becomes significant when the nominal strain-rate is around 100s-1, which coincides
with experimental data. Thus, the observed strain-rate sensitivity from 100s-1 in SHPB
test is a pseudo-strain-rate effect, which is caused actually by the material strength
sensitivity to the hydrostatic stress (Li and Meng, 2003).
51
Another perspective on the origin of the strain rate effect approaches the problem
from the micro and meso scale. Rossi et al published a paper on the presence of water
that may influence the response of the concrete under high strain rates (Rossi et al,
1994). This work attributed that Stefan effect on free water within the concrete may
change the cracking patterns of the concrete under dynamic loading. Under quasi-
static loads, cracking occurs through the cement matrix and when it encounters a
coarse aggregate, the cracking will continue around the boundaries of the aggregate.
However, due to the Stefan effect, the path of the crack penetrates through the
aggregate as illustrated in Figure 3-4. Since cracks are occurring through the coarse
aggregates, more energy is required which will then translate to the concrete
exhibiting a higher resistance to external forces to which it is subjected. The increase
in resistance explains the increase in stress-flow in concrete. This hypothesis has thus
far not been verified experimentally and that will be one of the objectives of the
experiment that is proposed in this project.
Quasi-static loading crack
Dynamic loading crack
Figure 3-4 Proposed Crack pattern under quasi-static and dynamic loads
3.2.2 Experimental Programme to Examine the Basis of DIF of Concrete
In order to verify the hypothesis that has been proposed in Section 3.2.1.2, a
comparison of two concrete mixes will be tested using the SHPB. The difference in
the mix will be the coarse aggregates used. The results from the SHPB test on the two
mixes which includes the variation of the DIF with respect to different strain rates will
52
be compared to determine the influence of the coarse aggregates on influencing the
high strain rate effects.
3.2.2.1 Objective
The objectives of the SPHB test on concrete under high strain rates are:
 To verify that cracking does penetrate through the aggregates
 To show coarse aggregates do influence the DIF of concrete
 To provide experimental data to support the influence of the Stefan effect
3.2.2.2 Specifications of Concrete Studied (Granite & Stalite)
Two types of coarse aggregates were selected for this test, namely granite and stalite.
Granite is generally used in most if not all the construction of reinforced concrete
structures in Singapore and, being a very much established building material, it will
be a good comparison with other uncommon aggregates. Stalite is an expanded slate
aggregate and it is used in producing lightweight concrete. The difference in the
properties of granite and stalite is tabulated in Table 3-2 and Figure 3-5 shows the
10mm graded stalite used in the experiment. All these tests were conducted according
to experimental standards (ASTM C29/C29M-97, 2003; ASTM C127-04, 2004;
BS812-112, 1990; BS812-110, 1990).
Table 3-2 Specifications of granite and stalite

Water Bulk Apparent Aggregate Impact Value Aggregate
Void
Aggregate Absorption Density Specific (AIV) Crush Value
(%)
(%) (kg/m3) Gravity Range Average (ACV)
Granite 41.5 0.548 1509 2.624 17.85-19.3 18.7 26.38
Stalite 45.9 6.576 782 1.6 29.14-30.29 29.6 35.45
53
Figure 3-5 Stalite coarse aggregates
From the above data, stalite is almost 50% less dense than granite which can be
attributed to higher porosity in the aggregates, as seen in Figure 3-5. This is also
evident in the water absorption capacity of the aggregate. As a result of its density,
stalite do not possess much strength as compared to granite as demonstrated in the
Aggregate Impact Value (AIV) and Aggregate Crush Value (ACV) test results
(BS812-112, 1990; BS812-110, 1990). For this reason, stalite is chosen as a
comparison with granite due to the vast property differences.
3.2.2.3 Experimental Setup and Instrumentation
In order to generate useful data from the comparison between the two aggregates, the
mix should be the same. Therefore, in view of that, a mix was formulated to compare
concrete that constitutes the two types of coarse aggregates. The mix details are as
follow:
Table 3-3 Mix content of concrete specimens

Cement Water Fine Aggregate Coarse Aggregate W/C
VA:VC VFA:VCA
(kg/m3) (kg/m3) (kg/m3) (kg/m3) Ratio
845 (for granite),
420 230 800 7.7 0.95 0.52
435 (for stalite)
54
The densities of the granite and stalite mixes are 2327.3kg/m3 and 2052.6kg/m3. Tests
on the static strength of the two mixes are also carried out after curing the specimens
for a period of 28 days. The cube strength of granite and stalite mixes are 37.38MPa
and 38.83MPa. The results from the 28-day cylinder test are plotted out in Figure 3-6.
35 35
30 30
25 25
Stress (MPa)
Stress (MPa)
20 20
15 15
10 10
5 5
0 0
0 0.005 0.01 0 0.005 0.01
Strain Strain
(a) (b)
Figure 3-6 Stress -Strain plot of (a) granite and (b) stalite from unconfined
compressive cylinder load tests
It is observed that although the two concrete mixes differ in density, the stress-strain
response to quasi-static loading is similar as their yield strength moduli are similar to
each other. However, the softening curve of the two concrete mixes with the granite
mix possessing a higher residual capacity, which is partly due to the higher ductility,
as compared to stalite. This can therefore be attributed to the difference in the strength
and the high crushability of the coarse aggregates used.
In view of the similarities in strength of in the two mixes under static loading, the
SHPB will be used in testing the dynamic strength of the concrete specimens to
observe the influence of the coarse aggregates. The SHPB is the most commonly used
method for determining material properties at high rates of strain. The setup of the
SHPB that was used in this study consists of the short cylindrical concrete specimen
being sandwiched between two long high strength steel elastic bars of 80mm diameter.
55
Figure 3-7 is a schematic of the SHPB setup. A 100mm diameter cylindrical steel
striker bar is projected by the sudden release of pressurised gas onto end of the input
or incident bar. For our test, the pressures used prior to the projection of the striker bar
were 15, 20, 25, 30 35MPa. The velocity of the striker bar will increase with the
amount of pressure used to propel it forward and high velocities will yield higher
strain rates from the specimen.
Upon the impact of the striker onto the incident bar, one-dimensional compressive
stress wave, which will travel along the length of the bar, will be generated. Upon
arriving at the interface between the bar and the specimen, the wave is partially
reflected back into the incident bar and partially transmitted through the specimen and
subsequently into the output bar. The reflected pulse is transmitted back to the
incident bar as tension wave, while the transmitted pulse remains in compression.
These data is recorded by the strain gauges at the midspan of the two elastic bars and
a sample of the recorded data is plotted in Figure 3-8. As observed, the first
compression (-ve) strain (εi) wave recorded has a duration of 0.5msec. After the
compression wave reaches the interface, the reflected tension strain (εr) wave is
recorded. Considering that the distance of the strain gauge of the incident and output
is equidistance from the specimen and the velocity of the wave is conserved, the
compressive transmitted strain (εr) wave from the output bar will be recorded in the
same time frame as of the reflected wave in the incident bar.
56
Propelling Pressure Incident Bar Specimen Output Bar
Stiker Bar Strain Guage Strain Guage
Digital oscilloscope
Figure 3-7 Schematic of SHPB used in study

εr
1000
800
Tension
600
400
200
Strain
0
-200 0 0.001 0.002 0.003 0.004 0.005 0.006
Compression
-400
εt
-600
-800
-1000
εi Time (sec)
Figure 3-8 Sample of the readings from the data logger
In order to analyse, the data from readings such as Figure 3-8, the following
assumptions were made in the experiment to ensure the validity of the subsequent
analyses:
 Stress within the incident and output bars remain within the elastic limits
 All waves are one-dimensional and uniaxial
 The stresses in specimen is assumed to be uniform and homogeneous
 Friction between all interfaces in the SHPB is negligible
With the two strain readings identified, a series of analyses can be performed to
determine the stress flow that is within the specimen. Kolsky (Kolsky, 1949)
developed the following relation for calculating the specimen stress σ and strain ε:
57
 t   Eo t t 
Ao
(3-6)
A
 t   o   r  d
2C t
(3-7)
L 0
,where Ao = Cross-sectional area of the steel bars, A = Cross-sectional area of the
specimen, Eo = Young modulus of the steel bars, Co = Elastic wave sped in steel bars,
L = Length of the specimen.
This is treated as a one wave analysis as it involves taking the strain reading directly
from the incident and output bars. There are two other methods in deriving the stress
and strain of the specimen, namely the two and three wave analyses. As their names
suggest, the two and three wave analyses involve the use of two and three waves
respectively to obtain the variation of stress and strain with time. The expressions for
the two wave analysis is stated in Equations (3-8) and (3-9) while those of the three
wave analysis are expressed in Equations (3-10) and (3-11). The influence of the use
of various approaches will be evaluated in next two sections.
 t   Eo  i t    r t 
Ao
(3-8)
A
 t   o   i     t  d
2C t
(3-9)
L 0
 t   o Eo  i t    r t    t t 
A
(3-10)
2A
 t   o   i     t     r  d
C t
L 0 (3-11)
Two of the critical parameters that will influence the recorded data are the length and
diameter of the specimen between the incident and output bars. Therefore, it is
imperative that appropriate dimensions are used in the SHPB test, especially in cases
where the specimens are large, which will lead to stress uniformity not being achieved
along the specimen (Malvern et al, 1985). The diameter of the specimen used is
78mm (Davies and Hunter, 1963; Gary, 2000). This is to ensure that the propagating
58
waves in both specimen and bars remain one-dimensional and in-plane throughout the
whole experimental process. In order to attain stress equilibrium within the specimen
when it is loaded by the propagating wave, it is important to minimise its length.
However, the specimen of concrete will not have been correctly tested if the length is
comparable to the nominal size of the aggregate (Dioh et al, 1993; Gary and Bailly,
1998). A general rule-of-thumb in the sizing of the specimen is to ensure that the
characteristic length is three times that of the nominal aggregate size. The nominal
aggregate size used in the granite and stalite is 10mm. In addition, it is recommended
that the ratio of the length to the diameter vary from 0.3 to 1.
Therefore, the cylindrical specimens are moulded with a length of 45mm. These
specimens were initially cast in PVC tubes with a length of 100mm and prior to the
testing of the specimens. These concrete-infilled PVC pipes are cut into half, grinded
at both ends to attain the desired length of 45mm and the concrete is extruded from
the pipes. For each pressure setting (15, 20, 25, 30, 35MPa), three specimens will be
tested. In total, 15 specimens are fabricated for each of the two concrete mixes.
3.2.2.4 Test Results and Discussion
The stress and strain histories as well as the strain rates will be used for comparison.
However, the preceding section has also shown that there are three methods: one
wave, two waves and three waves analysis (Equations (3-6 to 3-11)), to attain the
strain rates and stresses. Therefore, a comparison of the one, two and three wave
methods is made to ascertain the most appropriate method derive the DIF in
comparison of these values of the granite and stalite concrete mixes. The comparison
59
between the strain rates and stress of the granite concrete mix is plotted in Figures 3-9
and 3-10.
300 300
250 250
Strain Rate (s-1)

Strain Rate (s-1)
200 200
150 150
100 100
50 50
0 0
0 0.0002 0.0004 0.0006 0 0.0002 0.0004 0.0006
Time (sec) Time (sec)
(a) (b)
300
250
Strain Rate (s-1)
200 15bar
20bar
150
25bar
100 30bar
35bar
50
0
0 0.0002 0.0004 0.0006
Time (sec)
(c)
Figure 3-9 Strain rate data from (a) one, (b) two and (c) three wave analysis on granite
concrete mix
80 80
70 70
60 60
Stress (MPa)
Stress (MPa)
50 50
40 40
30 30
20 20
10 10
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Strain Strain
(a) (b)
60
80
70
60
Stress (MPa) 50
Static
40
30
15bar
20 20bar
10 25bar
0 30bar
0 0.01 0.02 0.03 0.04 0.05
35bar
Strain
(c)
Figure 3-10 Stress data from (a) one, (b) two and (3) three wave analysis on granite
concrete mix
From the comparison of the stress-strain and strain rate histories, it is difficult to
decipher which method to use for our study since all of them yield different results.
Therefore, the results will be compared to past research that has been done on the
dynamic compressive strength of concrete (Ross et al, 1995; Grote et al, 2001; Li and
Meng, 2003). By taking the peak strain rate and the peak stress of the specimen from
each test, the DIF of the granite concrete mix from the three analyses is shown in
Figure 3-11.
61
3.5 3.5
3 3
2.5 2.5
DIF 2 2
DIF
1.5 1.5
1 1
0.5 0.5
0 0
0 100 200 300 400 0 100 200 300 400
Strain Rate (s-1) Strain Rate (s-1)
(a) (b)
3.5
3
2.5 1 wave analysis
2
2 wave analysis
DIF
3 wave analysis
1.5
CEB
1 Ross et al
0.5 Grote et al.
0 Li and Meng
0 100 200 300 400
Strain Rate (s-1)
(c)
Figure 3-11 Comparison of (a) one, (b) two and (3) three wave analysis on DIF of
granite concrete mix with current research
It is observed that all three analysis yielded results which are closer to the research
works as compared to the CEB recommendations. Another observation is that in
general, the two wave analysis is higher than the one wave analysis with the three
wave analysis being the average of the two. Therefore, in view that all three results
are satisfactory close to recommendations of past research works, our study will be
based on the three wave analysis which utilise data from the incident, reflected and
transmitted strain readings.
Similar tests were conducted on the stalite mix concrete to compare the DIF of these
two concrete mixes. Considering the only difference in the mix content is the coarse
aggregates used while keeping fabrication, curing and testing processes the same, the
62
differences in the dynamic stress flow between the two concrete mixes can be isolated
to the coarse aggregates used. The results from SHPB test on the stalite concrete mix
is plotted in Figures 3-12 and 3-13. The results of the strain rates of the both concrete
mixes are also plotted in Figure 3-14. It can be observed that the strain rates to which
the two concrete mixes are subjected are similar.
300
250
200
Stress (MPa)
15bar
150 20bar
25bar
100 30bar
35bar
50
0
0 0.0002 0.0004 0.0006
Strain
Figure 3-12 Strain rate data from a three wave analysis on stalite concrete mix
63
60
50
40
Stress (MPa)
15bar
30 20bar
25bar
20 30bar
35bar
10
0
0 0.01 0.02 0.03 0.04 0.05
Strain
Figure 3-13 Stress data from a three wave analysis on stalite concrete mix
300
250
200
Strain Rate (s-1)
15bar
150 20bar
25bar
100 30bar
35bar
50
0
0 0.0002 0.0004 0.0006 0.0008 0.001
Time (sec)
stalite (dashed line) concrete mix
From the data collected from the SHPB tests on stalite concrete mix, the DIF obtained
from this series of test is plotted against those from granite concrete mix and the result
64
is drawn out in Figure 3-15. The following are the observations derived from the DIF
plot of the SHPB in the strain rates in the range of 100s-1 to 300s-1:
 Stalite produces results that are closer to predictions based on past research
works.
 Granite concrete mix specimens exhibit more erratic response than the stalite
specimens.
 Granite produces slightly higher enhanced strength than stalite.
 The values obtained from the tests tend to be closer to the trend lines proposed
by Ross et al, Grote et al and Li and Meng than the one proposed by CEB.
 Compared to the trend lines proposed by Ross et al, Grote et al and Li and
Meng, the collected data tend to be slightly higher.
2.5
Granite
1.5
DIF
Stalite
1 CEB
Ross
0.5 Grote
Li
0
100 150 200 250 300
Strain Rate (s-1)
stalite (dashed line) concrete mix
Based on the experimental data, the expressions which can describe the DIF as a
function of strain rate for granite and stalite can be expressed respectively as follow:
DIFgranite  0.32 ln   0.162 (3-12)
65
DIFstalite  0.482 ln   1.16 (3-13)
With reference to the assumptions made in the experiment, the differences between
the results can be attributed to the types of the coarse aggregates in the concrete mix.
Considering that lower DIF values are derived from the stalite concrete mix under
similar strain rates, the hypothesis on the crack patterns penetrating through the coarse
aggregates can be supported since stalite has a lower AIV and ACV, which will
translate to a lower resistance to fracture as compared to granite. Similarly, this also
supports the hypothesis on the influence of Stefan effect in concrete under high strain
rate.
3.3 Design of CFST Column
With a gauge on the design parameters which are used for the loading and material
response of a structural member which is subjected to blast loading, the process in the
design of the steel-concrete composite column can commence. This section will cover
the assumptions made in the design of the member, the fundamentals as well as
validation of the SDOF method, the introduction of the Rigid-Plastic approach and
comparison between these two analytical methods.
3.3.1 Assumptions
The following assumptions are taken into consideration in the dynamic analysis of
steel-concrete composite member subjected to blast loading, which are largely based
on the Eurocode (Eurocode 4, 2004) and TM5-1300 blast design manual (TM5-1300,
1990):
66
 Dynamic increase factors (DIF) given in the blast design manual (TM5-1300,
1990) will be factored into the design strength of steel and concrete to
simulate the effects of strain-rate.
 Moment and transverse shear capacities of the composite member are based
on the Eurocode 4 (Eurocode 4, 2004).
 Transverse shear resistance is equal to that of steel only (Eurocode 4, 2004).
 The reduction of moment capacity in a column due to axial forces is to be
accounted for (Eurocode 4, 2004).
 Scabbing, spalling or disengagement of concrete due to the blast load is
ignored
 Concrete under tension is omitted from the calculation of component capacity
(Eurocode 4, 2004).
 The longitudinal shear resistance of the member is not exceeded in order to
ensure full composite actions which can be achieved through proper design of
shear connectors.
 It is assumed that the column end connections do not fail and are able to
sustain the blast loads.
 Damping is not considered in the analysis.
 The natural period of the structural member is derived from equivalent SDOF
system (Biggs, 1964; TM5-1300, 1990).
 The peak deflection of the member does not exceed the ductility and end
rotation limits both steel and concrete respectively (TM5-1300, 1990).
67
3.3.2 Methods of Analysis
One of the most common methods in analysing dynamic problems is the use of the
SDOF approach. In view that there are deficiencies in the SDOF method, an
alternative method, which utilises the rigid-plastic material response of the structural
element to yield the peak displacement through the use of close-form solutions, will
be introduced.
3.3.2.1 Equivalent System of Structural Element based on SDOF
In SDOF analysis, the peak displacement of a continuous member is treated to be
equal to that of an idealized mass-spring system that has a single displacement
variable and the structural response is presumed prior to the analysis so as to obtain
the equivalent dynamic system in the SDOF analysis (Biggs, 1964; TM5-1300, 1990).
The equivalent SDOF system can be analyzed through calculus, the use of non-
dimensional charts such as those in TM5-1300 or numerical methods.
Several deficiencies are found in the SDOF method with the main setback being the
incapacity to analyse simultaneously the different responses of a structure. Thus, the
SDOF method is limited as it is only capable of analyzing the flexural and shear
responses independently (Yandzio and Gough, 1999) and is not able to capture multi-
failure modes. Since a structural element may be damaged due to shear followed by
flexural yielding, the two responses should not be decoupled.
3.3.2.2.1 Experiment Setup for Verification
Although multiple research projects have studied the SDOF methodology and
compared the theoretical results with those which were derived experimentally, there
68
are few which are able to simulate uniform distributed load on a member in a
controlled environment. For such cases, researchers will either conduct actual blast
test using explosives or, at the current state-of-the-art, build shock tubes to simulate a
distributed load on a structural member. Both these methods are expensive and a lot
of safety measures have to be taken into consideration prior to testing. Therefore, in
view of such constraints, a novel experiment setup is devised to provide a distributed
load across the span of a steel composite structural member.
The objective of the test is to test the feasibility of using a drop hammer setup
together with an airbag which is able to distribute the load across the face of the
structural member. Airbags have been used in several experimental setups which
require a uniformly distributed load to be applied on a specimen (Flanagan and
Benett, 1993; Mackes et al, 1995). The common objective of these experiments was
to obtain the loading capacity of the components and these resistances are restricted to
statics. Therefore, in order to validate the SDOF approach on a column with a lateral
distributed load across one face of the structural member, the load applied should be
dynamic. Figure 3-16 illustrates the concept of the experimental setup. The airbag will
be secured into position on top of the specimen and the assembly will be put below a
drop weight which will be raised above the specimen-airbag assembly. The airbag
will then be inflated to a certain initial pressure Pi. The drop weight will then be
released and, upon impact with the airbag, the pressure within the airbag will increase
by ΔP(t). With the assumption that the airbag is in full contact with the specimen, a
pressure confined within the airbag can to be evenly transferred to the specimen. This
action will simulate the distributed lateral load on a column. This pressure will
69
fluctuate with time as, upon the rebound of the drop weight, the pressure in the airbag
will start to decrease.
Drop weight
Airbag
Pi Pi + ΔP(t)
Specimen
(a) (b)
Figure 3-16 Concept of the drop hammer-airbag setup (a) prior to loading and (b)
during the loading of the specimen
The experiment was conducted in the laboratory in the civil engineering department
of the Faculty of Engineering in the University of Wollongong. The final setup of the
experiment is shown in the schematics in Figure 3-17 and the actual setup can be seen
in Figure 3-18. The supports are first erected below the drop weight on the strong
floor of the laboratory and subsequently three specimens are arranged across the
supports (see Figure 3-18(c)). The boundary conditions of the experiment can be
assumed to be simply supported as seen in Figure 3-17 due to the use fixed steel rods
at each end of the specimens. The main specimen to be tested is to be placed at the
centre of the other two specimens. The purpose of having these two other specimen is
to allow the load to be fully transferred to the test specimen by supporting the
download motion of the airbag. A 4mm thin plywood plank is placed above the
specimens to distribute the load across the specimen followed by a box which will be
used to house and limit the displacements of the airbag. The airbag is then placed in
the box and covered by a thin polymer fibre cover (see Figure 3-18(b) and (d)).
70
Polymer fibre cover
Plywood
Airbag
Steel rod (Simply supported)
Steel support secured to strong floor

c/c 375mm c/c 375mm
c/c 995mm
Support specimen Test specimen
Figure 3-17 Plan and elevation views of the setup of specimen and airbag
71
(b)
(a) (c)
(d) (e)
Figure 3-18 Setup of drop hammer rig with specimen and airbag
As seen in Figure 3-17, the specimens are positioned with a spacing of 375mm and
the effective length of the specimen is 995mm. The two specimens that were studied
consist of a mild steel square hollow section (SHS) infilled with sand and a stainless
72
steel SHS with no in-filled material. The specifications of the specimens and the
material properties are as follow:
Table 3-4 Specifications of SHS specimens

Specimen b (mm) d (mm) t (mm)
b
Mild Steel SHS in- d
t
50 50 1.2
filled with sand
t d
Stainless Steel SHS 50 50 1.2 b
d
Table 3-5 Material properties of specimens

Specimen Yield Strength (MPa) Modulus (GPa) Density (kg/m3)
Mild Steel 350 200 7800
Stainless Steel 515 193 8000
Sand - - 1600
Instrumentation for the setup includes a strain gauge being applied along the bottom
fibre at the midspan of the specimen, a laser sensor which traces the displacement of
the midspan, a pressure sensor which measures the pressure in the airbag and a load
cell at the drop hammer to measure the impact force (see Figure 3-19).
(a) (b)
(c) (d)
Figure 3-19 Instrumentation setup: (a) strain gauge, (b) laser sensor (displacement),
(c) pressure sensor, (d) load cell
73
The drop height of the drop hammer will be varied to obtain various responses from
the two specimens. The mild steel specimen will be loaded twice: once from a height
of 50mm and, subsequently, another from a height of 500mm. These two tests will be
denoted by MS-50 and MS-500 from henceforth. The stainless steel specimen will be
loaded from the hammer being dropped from a height of 500mm and this test will be
denoted by SS-500.
3.3.2.2.2 Results and Discussion
The recorded data from the instrumentation of the test MS-50, MS-500 and SS-500
are shown in Figure 3-20, Figure 3-21 and Figure 3-22 respectively.
30 15
Displacement (mm) / Strain (millistrain)

25
Load (kN) / Pressure (kPa)
20 10
Load
Strain
15 Displacement
Pressure
10 5
0 0
0 200 400 600 800
Time (msec)
Figure 3-20 Readings from various instrumentation from MS-50
74
70 35

60 30
50 25
40 20 Load
Strain
Displacement
30 15 Pressure
20 10
10 5
0 0
0 100 200 300 400 500 600
Time (msec)
Figure 3-21 Readings from various instrumentation from MS-500
50 25

45
40 20
35
Load
30 15 Strain
Displacement
25 Pressure
20 10
15
10 5
0 0
0 100 200 300 400 500 600
Time (msec)
Figure 3-22 Readings from various instrumentation SS-500
It is observed in all three recordings that there is a spike in the load cell reading prior
to more gradual increase, which is in phase with the pressure-histories recorded in the
airbag. This spike can be attributed to the first contact between the drop weight and
75
the surface of the polymer fibre cover, which is on top of the airbag. As observed, this
initial impulsive load was not transferred to the airbag and specimen and thus will be
ignored the in analysis.
Pressure recordings from MS-50 and SS-500 shows two peaks due to the rebound of
the drop hammer and the reason for the single peak reading from MS-500 is due to the
damage of the airbag which ruptured shortly after the first impact. There was also a
signal cutoff from the strain gauge during the loading phase of that test due the limits
setup in the data logger.
(a) (b)
Figure 3-23 Deformation of mild steel SHS in-filled with sand from test MS-500
Compared to the other two tests, MSS-500 shows significant plastic deformation with
a peak displacement of 33mm and a permanent deformation of 21.5mm at the
midspan. Figure 3-23 shows the deformation of the specimen. In addition, the top
fibre at the midspan has signs of buckling under the significantly large dynamic load.
The other two tests did not yield any plastic response from the specimens but the data
will be useful to verify the accuracy of the SDOF method in predicting the dynamic
response of steel hollow sections. Table 3-6 tabulates the maximum values from the
data collected from the three test runs.
76
Table 3-6 Maximum values from the tests
Midspan Load Cell
Test Pressure (kPa) Millistrain
Displacement (mm) Reading (kN)
MS-50 5.41 23.3 28.9 1.4
MS-500 33.2 58.3 46.1 4.11
SS-500 11.7 50.5 67.1 2.95
The test result will be used to benchmark the SDOF method in the analysis of steel
SHS under dynamic loading. From the average strain readings from the test the strain
rates for test MS-50, MS-500 and SS-500 is 0.014/s, 0.09/s and 0.03/s respectively.
Under such strain rates, influence of DIF will not be significant and thus this effect
will be ignored in the analytical solution. By applying simply supported assumptions
at both ends of the member, assuming that span is 995mm, the response of the
specimens to be purely flexural and the loadings follow the experimental recordings,
the SDOF method is being applied to validate itself against experimental results. In
addition, the strength contribution of sand core of the mild steel specimen is assumed
to be negligible and its only contribution to the analytical solution is the mass it adds
to the specimen. Figures 3-24, 3-25 and 3-26 compares the displacement and pressure
loadings from the experimental results and analytical solution of test MS-50, MS-500
and SS-500 respectively.
77
35
30
Pressure (kPa)/Displacement (mm)
25
20 Exp Load
SDOF Load
Exp Displacement
SDOF Displacement
15
10
0
0 50 100 150 200 250 300
Time (msec)
test MS-50
50
40
Pressure (kPa)/Displacement (mm)
30
Exp Load
SDOF Load
20 Exp Displacement
SDOF Displacement
10
0
0 100 200 300 400
-10
Time (msec)
test MS-500
78
80
Pressure (kPa)Displacement (mm) 70
60
50
Exp Load
40 SDOF Load
Exp Displacement
SDOF Displacement
30
20
10
0
0 50 100 150 200 250 300
-10
Time (msec)
test SS-500
From the figures above, the results from the SDOF method relates well to the
experiment. Under elastic response in MS-50 and MS-500, the analytical results
match the experimental variation of the midspan displacement. Under plastic response
in MS-500, although the peak displacement from the SDOF approach may be very
close to the experimental readings, the same cannot be said of the permanent
deformation whereby the SDOF predicts a displacement of 25mm as compared to
21.5mm from the experiment. This may be attributed to the local buckling effect as
seen previously in Figure 3-25 which is not considered in the analytical solution.
Overall, it can be seen through this simple dynamic experiment that SDOF approach
is an efficient tool to estimate the dynamic response of steel hollow sections and it is
capable of predicting the peak displacements well. The accuracy of permanent
79
displacements may vary due to the local deformation of the section which may affect
the final deformation of the member.
3.3.2.2 Equivalent System of Structural Element based on Rigid-Plastic Analysis
The rigid-plastic method is an alternative analytical method which takes into account
flexural and transverse shear displacements simultaneously in estimating the peak
displacement response of steel-concrete composite structural element. No charts or
numerical procedures are required as opposed to the SDOF approach (Biggs, 1964;
TM5-1300, 1990) and the Rigid-Plastic solutions are determined from closed-form
equations. By taking into considerations the assumptions stated in Section 3.3.1, the
Rigid-Plastic method is described in this section for the analysis of steel-concrete
composite members subjected to blast loading.
3.3.2.2.1 Concept
The Rigid-Plastic method was first proposed by Jones (Jones, 1989) for the analysis
of metal beams. In the present study, the method is extended for steel-concrete
composite members. As stated in the name, the Rigid-Plastic method assumes that the
material of the structural element behaves in a perfectly plastic manner which ignores
the elastic resistance. The fundamentals of Rigid-Plastic method are based on
Equations (3-14) and (3-15) that is derived from Figure 3-27.
M
Q (3-14)
x
Q w2
 p m 2 (3-15)
x t
where Q = shear force, M = bending moment, p = force per unit length, m = mass per
unit length and w = deflection.
80
p
M
H x
w Q
Figure 3-27 Notation of beam to the Rigid-Plastic method
There are three response modes to be considered in the analysis, namely ductile
deformation (Mode I), tensile tearing at support (Mode II) and transverse shear failure
at support (Mode III) as illustrated in Figure 3-28. Mode II is ignored in this study by
assuming that the connections are able to sustain the applied blast loading as stated in
Section 3.3.1. Mode I is associated with flexural failure of the structure and Mode III
with transverse shear failure.
Mode I: Large ductile deformations
Mode II: Tensile failure at supports
Mode III: Transverse shear failure

at supports
Figure 3-28 Response modes in dynamic plastic analysis
3.3.2.2.2 Methodology
Based on the works by Jones (Jones, 1989), the following steps can be followed to
determine the response of a one-way structural element which is subjected to blast
loading
i. Determine the ultimate shear capacity, Qo, ultimate moment capacity, Mp,
and natural period of vibration, T, of the member.
ii. Determine the blast loading parameters. For Rigid-Plastic analysis, the
blast loading time history is simplified to a rectangular shape with a
81
constant effective pressure, Peff, and effective loading time, τeff (see
Equations 2-4 and 2-5).
iii. Determine the blast loading regime as described in Table 3-7.

Table 3-7 Structural response regimes
Loading regime τ/T
Impulsive τ/T < 0.4
Dynamic 0.4 < τ/T < 2
Quasi-static τ/T > 2
iv. Determine the blast regime (dynamic or impulsive).
v. For dynamic loading:
o Check the following equations for transverse shear yielding
Qd 2 3 M p
  1.0 (3-16)
Qo SS
L Qo
Qd 4 3 M p
or   1.0 (3-17)
Qo FE
L Qo
Subscript SS denotes simply-supported condition and FE denotes fixed
end, η is the ratio of peak overpressure to static elastic yield pressure
and Qd is the dynamic transverse shear at the end of the member.
o If Qd / Qo > 1,
 For η < 3, the mechanism consists of two transverse shear
slides at the two ends and three rotating hinges (Figure 3-
29(a)).
 For η > 3, the mechanism consists of two transverse shear
slides at the two ends and four rotating hinges (Figure 3-29(b)).
o If Qd / Qo < 1,
 For η < 3, the mechanism consists of three rotating hinges
(Figure 3-29(c)).
82
 For η > 3, the mechanism consists of four rotating hinges
(Figure 3-29(d)).
vi. For impulsive loading:
o Calculate the coefficient ν = QoL/4Mp
o If ν < 1 (SS) or 2 (FE), the mechanism consists of two transverse shear
slides at the two ends (Figure 3-29(e)).
o If 1 (SS) or 2 (FE) < ν < 1.5 (SS) or 3 (FE), the mechanism consists of
two transverse shear slides at the two ends and three rotating hinges
(Figure 3-29(a)).
o If ν > 1.5 (SS) or 3 (FE), the mechanism consists of two transverse
shear slides at the two ends and four rotating hinges (Figure 3-29(b)).
(a) (b) (c)
(d) (e)
Figure 3-29 Plastic mechanisms under blast loading
The governing equations, boundary conditions and response solutions for various
plastic mechanisms that are described in steps i. to v. are provided in the appendices.
The charts in Appendix A shows the closed form equations that describe the blast
responses of one-way structural member in the dynamic loading regime while
Appendix B provides solutions for the impulsive response of such structural members.
The dynamic responses of these structural members, where transverse shear response
is taken into consideration are given in Appendix C.
83
Therefore, from the analytical study, shear response does not exclusively occur under
impulsive loading. The correct evaluation of such responses is necessary to ensure the
proper evaluation of the blast resistance of structural members, especially those which
are under consideration of reinforcements. Enhancement of the bending moment
capacity does not necessarily translate to an improvement in the shear resistance and
vice-versa.
In addition, the time variation of the reaction forces at the ends of load transfer
members of a structure such as beams and slabs will vary according to the response of
the structure. Therefore, it is important to estimate the dynamic nature of the forces
that is being imposed onto the primary support elements in order to determine the
design.
3.4 Analytical Study of CFST Column
In order to better illustrate the two methods that has been discussed in the preceding
section, it will be appropriate to size up the differences between the SDOF method
and the Rigid-Plastic approach. A concrete-filled rectangular composite column is
analysed in the study to determine its vulnerability to blast threats. A series of blast
loading will be introduced to the specimen to investigate the response of the column
under various response regimes.
3.4.1 Specification of Composite Column
The CFST column that is studied consist of a rectangular hollow section which
measures 200mm in width, 400mm in depth and 4000mm in length. The thickness of
84
the hollow section is assumed to be 10mm. Within the core of the column are steel
reinforcement bars. Four (4) longitudinal reinforcement bars of 20mm diameter run
along the length of the column and they are positioned 40mm away from each of the
four corners of the column. Reinforcements were provided in the composite column
as a form of fire protection since the outer steel tube is exposed to fire, which usually
occur in the event of blast. In addition, stirrups are also introduced into the element
and each stirrup is tied around the outer perimeter of the four main reinforcement
bars. The diameter of these stirrups is 10mm and they spaced 200mm along the
length of the column. Finally, the core of the hollow section is in-filled with concrete.
The boundary ends of the column specimen are fixed. The column is illustrated in
Figure 3-30.
200
Rectangular
hollow section
Concrete Core
4000
400
Stirrup
Main reinforcement
bar
Figure 3-30 CFST column used to compare the analytical and numerical results
(Dimensions in mm)
The materials used for the specimen are tabulated in Table 3-8 and Figure 3-31 uses
this information to plot the P-M interaction curve according to the specifications of
the Eurocode:
85
Table 3-8 Material properties of concrete-infilled steel column
Strength Modulus Density Poisson
Component
(MPa) (GPa) (kg/m3) Ratio
Rectangular hollow section 355 200 7850 0.3
Main reinforcement steel bars 460 200 7850 0.3
Stirrups 460 200 7850 0.3
Concrete Core 30 28 2320 0.2
7000
6000
Axial Force P (kN)
5000
4000
3000
2000
1000
0
0 100 200 300 400 500 600 700
Moment M (kNm)
Figure 3-31 P-M Interaction curves of concrete-filled steel composite
3.4.2 Static and Blast Loading
Both static and dynamic loadings are considered in the study to fully illustrate the
effectiveness of the analytical approaches in the design of steel-concrete composite
columns. A static axial force of 2500kN is being applied at the top end of the column.
In the design of the column according to Eurocode (Eurocode 4, 2004) and a
moderate amount of conservativeness with a safety factor of 2 in the compressive
capacity of the column was considered in the design to cater for the subsequent lateral
blast loading.
The lateral blast loadings of the problem are varied according to the equivalent TNT
charge weight and the stand-off distance. These variations is documented in Table 3-9
and Figure 3-32 shows the loading directions of both the static and dynamic forces.
86
Despite the fact that blast loading may vary across the length of the column, the loads
will be assumed to be uniformly distributed to ensure the simplicity of the analytical
study. The table also includes a column specifying the blast loading regime, which
derived from the ratio of the blast time duration and the natural period of the CFST
column.
Table 3-9 Applied blast loading

Stand-off distance, R (m) TNT charge weight, W (kg) Blast loading regime
750
800
10 Impulsive
900
1000
1000
1100
1200
12.5 Dynamic
1300
1400
1500
2000
2100
2200
15 Dynamic
2300
2400
2500
2500kN
Uniformly
distributed lateral
dynamic blast load
Explosive Charge
Stand-off distance
Figure 3-32 Loadings on the CFST column
87
3.4.3 Comparison and Discussion of Results
The peak displacements and ductility ratios (μ), which is defined as the ratio of the
maximum displacement of a structure to the yield displacement, of the CFST column
calculated using the SDOF and Rigid-Plastic methods are tabulated in Tables 3-10, 3-
11 and 3-12 respectively. Graphical comparison of the results are illustrated by
plotting the ductility ratio at the peak displacement against equivalent TNT charge
weight in Figures 3-33, 3-34 and 3-35.
Table 3-10 Blast response of column at stand-off distance of 10 m (impulsive regime)

TNT weight (kg) 750 800 900 1000
SDOF (mm) 19.98 21.76 24.90 27.59
SDOF (μ) 5.65 6.17 7.06 7.82
Peak
RP (mm) 16.75 18.44 21.92 25.50
displacement
RP (μ) 4.74 5.22 6.21 7.22
SDOF-RP/SDOF (%) 16.1062 15.3971 12.0397 7.67263
Table 3-11 Blast response of column at stand-off distance of 12.5m (dynamic regime)
TNT weight (kg) 1000 1100 1200 1300 1400 1500
SDOF (mm) 13.04 15.21 17.22 19.64 22.12 24.17
Peak SDOF (μ) 3.70 4.31 4.88 5.57 6.27 6.85
Displace- RP (mm) 10.49 12.74 14.96 17.29 19.7 22.19
ment RP (μ) 2.97 3.61 4.21 4.90 5.58 6.29
SDOF-RP/SDOF (%) 19.7297 16.2413 13.7295 12.0287 11.0048 8.17518
Table 3-12 Blast response of column at stand-off distance of 15m (dynamic regime)
TNT weight (kg) 2000 2100 2200 2300 2400 2500
SDOF (mm) 21.07 23.14 24.47 25.57 26.99 28.34
SDOF (μ) 5.97 6.55 6.94 7.25 7.65 8.03
Peak 20.16 22.04 23.97 25.94 27.96 30.02
RP (mm)
Displace-
ment RP (μ) 5.71 6.24 6.79 7.35 7.92 8.50
LS-DYNA (mm) 21.17 22.59 24.08 25.72 26.69 28.32
SDOF-RP/SDOF (%) -5.0099 -2.4955 -0.4589 0.84811 4.5422 5.66289
88
10
9
8
Ductility Ratio 7
6
SDOF
5
Rigid-Plastic
4
3
2
1 10m
0
600 700 800 900 1000 1100
TNT Weight (kg)
at 10m
10
9
8
7
Ductility Ratio
6
SDOF
5
Rigid-Plastic
4
3
2
1 12.5m
0
800 1000 1200 1400 1600
TNT Weight (kg)
at 12.5m
89
10
9
8
Ductility Ratio 7
6
SDOF
5
Rigid-Plastic
4
3
2
1 15m
0
1800 2000 2200 2400 2600
TNT Weight (kg)
at 15m
The following observations are noted from the comparison and discussion on the
factors that may contribute to the difference in the results from the SDOF and Rigid
Plastic method:
 Overall, the SDOF and Rigid-Plastic analytical results are found to be
reasonably close to each other. The differences are within 20 %. This shows
that the use of conventional design codes, such as the Eurocode, to determine
the transverse moment and shear carrying capacities of CFST columns is
satisfactory.
 The similarity of the results also indicate that the Rigid Plastic method may be
an alternative to estimate the peak displacement of a CFST column, based on
the validity of the SDOF method as explained in Section 3.3.2.2.1.
 Comparing the percentage differences in the peak displacement results
between the SDOF and Rigid Plastic method, it is observed that their
difference diminish as the ductility ratio increases. This is a general
90
observation for each of the three sets of data. This can be attributed to the fact
that the Rigid Plastic method does not take into consideration the elastic
resistance of the material as explained in Section 3.3.2.2. Therefore, under
large displacements, the results from the SDOF and Rigid Plastic analysis will
converge due to the reduction in the influence of the elastic response in the
overall deformation of the member.
 From the tabulation of the variation between the results from the SDOF and
Rigid Plastic methods, the percentage difference is within 10% if the peak
ductility ratio is more than 7. In view of that, from the dynamic and impulsive
case studies, it is recommended that Rigid Plastic method be limited to cases
whereby the maximum ductility ratio of the concrete-infilled steel member is
more than 7.
It is observed that the Rigid-Plastic method does have its deficiencies when the CFST
column does not exhibit large lateral displacements. However, the advantage of being
capable in predicting multi-failure modes of structures in the Rigid-Plastic method are
outweighs the limitations of the proposed method in the analysis of CFST. This is
because critical columns of the structure are normally located on the lower levels of
buildings and these members are normally designed to be stocky members. As a
result, in addition to flexural response, these members will be subjected to shear
failure. Hence, it is important to be able to capture both flexural and shear response of
these CFST columns under blast loading to accurately design them against designated
threat loadings. Thus, the Rigid-Plastic method is recommended to be used under such
circumstances.
91
3.5 Numerical Study of CFST Column
This section focuses on the modelling of the CFST column which is subjected to blast
loading. The design of the column will be directly adopted from the column that was
studied analytically in the preceding section. The column will be loaded axially before
being subjected to lateral loading. The results from this section will be correlated to
the analytical methods and deductions will be drawn from the comparison.
3.5.1 Finite Element Solution Scheme
The finite element (FE) scheme that was adopted in this model will be explicit
schemes as the responses are largely and highly transient. In addition, it is noted by
the author that the column must be preloaded axially to create some form of stress
initialisation within the steel-concrete composite column. If stress initialisation is not
considered, the stresses in the materials, when laterally loaded with dynamic blast
load, may not be in a realistic state and thus produce results which may not reflect the
response accurately.
In view of this consideration, the model includes the use of dynamic relaxation to
evaluate the quasi-static state prior to the highly transient response. Dynamic
relaxation is a feature found in LS-DYNA to find the solution to quasi-static problems
using transient integration (Hallquist, 2006). In this stage, the axial load is applied to
the column and damping is automatically applied to remove the kinetic energy in the
system to obtain the quasi-static response of the structural component. The explicit
scheme is adopted for this stage. This method provides a fast and effective manner to
estimate the elastic response of a column subjected to quasi-static loading.
92
3.5.1.1 Geometry and Design of Column
The CFST column considered in this study is identical to the member that is described
in Section 3.4.1. In numerical model, there are a total of four parts: the steel hollow
section, the concrete core, the vertical reinforcement bars and the stirrups. Each of
these parts will be assigned a particular element model, material model and equation
of state model.
3.5.1.2 Type of Elements Selected
The steel hollow section comprises of 20 x 20 mm2 8-nodes thick-shell plain stress
elements with a thickness of 10 mm. Thick-shell element was adopted in this study
because of its advantages from a contact standpoint since all potential contact surfaces
of the thick-shell are discretely represented by 4-node segments which is in contrast to
normal shell elements which only meshes the mid-plane. This is, in fact, one of the
motivations amongst others behind the development of thick shell elements in LS-
DYNA (Stuhmeyer, 2005). It is also inherently easier to transition from thick-shells to
solids than from thin-shells, whereby the nodes of the thick-shells can be merged to
that of solids to simulate a perfect-bond condition.
Hughes-Liu beam elements were applied for the reinforcement bars and stirrups
(Hallquist, 2006). These beam elements are conventionally integrated element and are
modelled to have a circular cross section area (20mm diamater for the vertical
reinforcement bars and 10mm diamater for the stirrups). Although the formulation of
this element is simple, as it is only capable of generating a constant moment across its
length instead of varying the moment across the length of the beam, this element is
93
chosen due to its compatibility to the material models to which the reinforcement bar
parts are going to be assigned.
The concrete core of the column was modeled using 20 x 20 x 20 mm3 8-nodes
hexahedral elements. The concrete elements are merged at the nodes with the
reinforcement bar beam elements. This is a simplified approach, which assumes a
fully bonded interface between the steel bars and concrete with interaction beween the
two parts, to model the effects of reinforced concrete and this technique has been
widely used and recognised in the prediction of the dynamic response of RC
structures (Wang et al., 2008; Sanji and May, 2009; Riedel et al., 2010).
3.5.1.3 Selection of Material Models
Two material models will be chosen for the four parts. The plastic kinematic model
(*MAT_003) will be used to model the steel components, which consists of the steel
hollow section, the vertical reinforcement bars and the stirrups, while the concrete
damage model release 3 (*MAT_072R3) was used to model concrete. The rationale
behind the selection of these materials as well as the actual parameters will be
discussed.
The Plastic-Kinematic material model (*MAT_003) features a bi-linear stress-strain
curve (Hallquist, 2006). The DIF-strain-rate relationships for the steel material is
taken into account by the Cowper-Symond equation defined in Equation 3-18 (Jones,
1989), where  is the dynamic strain-rates and C (unit of 1/s) and P are the Cowper-
Symonds strain-rate parameters. This approach is different from the analytical
94
solutions whereby single value DIFs were multiplied to the respective yield strength
of the materials (TM5-1300, 1990).
(3-18)
For the concrete core, the concrete damage material release 3 (*MAT_072R3) is used
to model the concrete core (Malvar et al., 1997). This material model is developed to
mainly analyse structures under blast and impact loadings. The features that made this
material suitable for the model includes the ability to incorporate strain rate effects for
both tension and compression into the model, the capacity to model and modify
softening of concrete through three independent deviatoric failure surfaces and
damage mechanics and the inclusion of shear dilation. The deviatoric and hydrostatic
(pressure against volumetric strain) parts of the response are uncoupled. The
deviatoric portion is represented by three failure surfaces while the hydrostatic
behaviour is governed by a compaction curve or an equation of state.
The stress-strain behaviors of the concrete material under various response phases and
confining pressures are defined in this plasticity material model by three independent
failure surfaces (see Figure 3-36), which are given as follow:
Initial yield surface, (3-19a)
Residual failure surface, (3-19b)
Maximum failure surface, (3-19c)
95
Maximum failure
surface
Initial yield surface
Residual failure
surface
(a) (b)
Figure 3-36 Meridian profiles for *MAT_072R3 in (a) 2D and (b) 3D stress space
The parameters with various subscripts of Equations (3-19) modify the shape of the
failure surfaces. The parameters of Equation (3-19c) can be derived through an
unconfined cylinder test and triaxial tests with various confinement pressures. The
parameters in Equation (3-19a) can be derived as it is predicted that the initial yield
surface is directly proportional to the maximum failure surface. The material model
adopts an octahedral cross section for the surfaces based on the William-Warnke’s
Lode angle function.
The strain rate effects is introduced into the model through an enhancement factor or
DIF which scales the strength surface when the material is subjected to high loading
rates. For this work, the DIF-strain-rate relationships described by Equation (3-20)
were adopted to simulate the strain-rate effect for the concrete core (Bischoff and
Perry, 1991, Malvar and Ross, 1999).
( / s ) for   1 s 1 with   1 /(1  8 f ' c / 10MPa)

DIF (Tens) =  (3-20a)
  ( / s ) for   1 s log   6  2
1/ 3 1
with
( / s )1.026 for   30s 1 with
s
 s  1 /(5  9 f 'c / 10MPa)
DIF (Comp) =  (3-20b)
 s ( / s ) for   30s log  s  6.156 s  2
1/ 3 1
with
In evaluating the softening of concrete, there are three parameters, which are b1, b2
and b3, that control this response in the model. b1 governs the softening part of the
96
unconfined uniaxial compression stress-strain curve from the maximum to the
residual failure surface. As documented by Malvar et al, softening is governed by a
localisation limiter or characteristic length and thus it is mesh-dependent (Malvar et
al, 1997). In order to eliminate this mesh dependency, b1 has to be determined by
iterative calculations such that the area under the stress-strain curve for an uniaxial
unconfined compressive test coincides with the quotient of the fracture energy of the
concrete and the localisation width, which is either one element width or the cube root
of the element volume in 3D. The fracture energy of concrete can be obtained through
the CEB Model Code (CEB, 1990). Similar to b1, b2 and b3 govern the softening part
for the unconfined uniaxial tensile stress-strain curve. b2 can be obtained in the same
manner as b1 except that the uniaxial force is in tension instead of compression. b3 is
obtained through a triaxial tensile test. As observed, it is clear that b1 and b2 can be
easily obtained through conventional laboratory equipment while b3 may require more
specialised equipment and services to derive.
Due to the complexity of the concrete material model, it would require quite a
substantial amount of effort to derive all the parameters in the model. Therefore, the
model has the capability to generate all the parameters including the equation of state
through the unconfined compressive strength of the concrete in use. This is achieved
factoring certain parameters by the ratio of the unconfined cylinder strength of the
concrete in the study to the default unconfined cylinder strength of the concrete that
was used to develop this model. The unconfined compressive strength of the default
concrete is 6582psi (45.4MPa) with 19mm nominal aggregate size. Therefore,
modifications have to be implemented into the model parameters after the auto-
generation in order to obtain an accurate structural response from the model. In this
97
model, the concrete material is assumed to have an unconfined compressive strength
of 30MPa.
3.5.1.4 Steel-Concrete Interface Modelling
In modelling the contact interface between the concrete core and steel hollow section,
a contact algorithm is selected. Merging the exterior nodes of the concrete core
elements with the inner nodes of the steel hollow section, which would assume full
composite interaction between these components. It is presumed that the concrete core
will undergo certain levels of shrinkage and thus the assumption of full composite
may not be appropriate.
The interfacial algorithm chosen between steel and concrete is based on the penalty-
based approach. In every time step, the code will check for penetration of the nodes of
the designated slave part or segment through the surface of the master part or
segment. Upon detection, a force proportional to the penetration depth will be applied
to resist and eliminate the penetration. This force is automatically generated and can
be based on the stiffness of the material, the area of the segment that is penetrated and
the mass of the parts in involved. In addition, the contact model between steel and
concrete includes friction between the two components and the coefficient of friction
used in the model is 0.35.
3.5.1.5 Blast Loading
In an actual blast scenario, the blast pressure acts on all four faces of the column with
varying intensity along the height of the column. The forces on the opposite faces will
counteract each other, resulting in the reduction of the peak displacement. In addition
98
to preserve the level of conservativeness, simplified single face loading was assumed
in the FE analysis so as to maintain consistency with the SDOF method in TM5-1300
and to obtain a fair comparison between the numerical and analytical solutions. For
similar reasons, idealized triangular blast loading profiles, like those used in the
SDOF method (see Table 3-9), were applied in the FE analysis.
3.5.2 Comparison with Analytical Models
The peak displacements and ductility ratios (μ) of the CFST column calculated using
the SDOF, Rigid-Plastic and FE methods are tabulated in Tables 3-13, 3-14 and 3-15.
Graphical comparison of the results are illustrated by plotting the ductility ratio at the
peak displacement against charge weight in Figures 3-37, 3-38 and 3-39.
Table 3-13 Blast response of column at stand-off distance of 10m

TNT weight (kg) 750 800 900 1000
SDOF (mm) 19.98 21.76 24.90 27.59
SDOF (μ) 5.65 6.17 7.06 7.82
RP (mm) 16.75 18.44 21.92 25.50
Peak RP (μ) 4.74 5.22 6.21 7.22
displacement LS-DYNA (mm) 18.05 18.92 21.58 24.30
LS-DYNA (μ) 5.12 5.36 6.12 6.89
SDOF/LS-DYNA (%) 110.37 115.06 115.33 113.51
RP/LS-DYNA (%) 92.58 97.39 101.47 104.79
Table 3-14 Blast response of column at stand-off distance of 12.5 m

TNT weight (kg) 1000 1100 1200 1300 1400 1500
SDOF (mm) 13.04 15.21 17.22 19.64 22.12 24.17
SDOF (μ) 3.70 4.31 4.88 5.57 6.27 6.85
RP (mm) 10.49 12.74 14.96 17.29 19.7 22.19
Peak RP (μ) 2.97 3.61 4.21 4.90 5.58 6.29
Displacement LS-DYNA (mm) 14.93 16.67 18.43 19.92 21.9 23.86
LS-DYNA (μ) 4.23 4.72 5.22 5.64 6.20 6.76
SDOF/LS-DYNA (%) 87.40 91.33 93.49 98.71 101.15 101.35
RP/LS-DYNA (%) 70.21 76.48 80.65 86.88 90.00 93.05
99
Table 3-15 Blast response of column at stand-off distance of 15 m
TNT weight (kg) 2000 2100 2200 2300 2400 2500
SDOF (mm) 21.07 23.14 24.47 25.57 26.99 28.34
SDOF (μ) 5.97 6.55 6.94 7.25 7.65 8.03
RP (mm) 20.16 22.04 23.97 25.94 27.96 30.02
Peak
Displacement RP (μ) 5.71 6.24 6.79 7.35 7.92 8.50
LS-DYNA (mm) 21.17 22.59 24.08 25.72 26.69 28.32
SDOF/LS-DYNA (%) 141.18 138.77 132.87 128.53 123.37 118.83
RP/LS-DYNA (%) 134.99 132.20 130.08 130.32 127.74 125.74
10
7
Ductility Ratio
5 SDOF
Rigid-Plastic
4
LSDYNA
3
2 10m
1
0
600 700 800 900 1000 1100
TNT Weight (kg)
at 10m
100
10
7
Ductility Ratio
5 SDOF
4 Rigid-Plastic
LSDYNA
3
2 12.5m
0
800 900 1000 1100 1200 1300 1400 1500 1600
TNT Weight (kg)
at 12.5m
10
7
Ductility Ratio
5 SDOF
4 Rigid-Plastic
LSDYNA
3
2 15m
0
1800 1900 2000 2100 2200 2300 2400 2500 2600
TNT Weight (kg)
at 15m
The following observations are noted from the comparison:
 Overall, the SDOF and Rigid-Plastic analytical results are reasonably close to
the FE predictions. The differences are within 15 % except for a few cases
which are explained in the subsequent paragraphs. This shows that the use of
101
conventional design codes, such as the Eurocode 4, to determine the transverse
moment and shear carrying capacities of CFST columns is satisfactory.
 The results for the case of impulsive blast loading at stand-off distance of 10
m are given in Table 3-13 and Figure 3-14. The peak displacements calculated
using the Rigid-Plastic method are closer to those obtained using LS-DYNA
as compared to that of the SDOF calculation. This could be attributed to the
occurrence of transverse shear yielding in the column (Kang et al., 2007)
which was not accounted for in the SDOF method. Thus, it is important that
the effect of the shear response is taken into consideration in the analysis of
composite columns subjected to impulsive blast loading.
 Table 3-14 provides the results for the case of dynamic blast loading at a
stand-off distance of 12.5 m. The average percentage of difference between
the Rigid-Plastic method and the FE simulation is 17.0 %, which is higher
than the 5.3 % difference for the SDOF method. As discussed earlier, the
elastic response is ignored in the Rigid-Plastic analysis since the structural
member usually exhibit significant post-yield displacement under blast
loading. In this study, the calculated transverse displacement of the column is
3.53 mm at first yield. Hence, for peak displacements in the relatively low
range of 10 to 15 mm, the Rigid-Plastic method may not reflect accurately the
total deflection of the member. As the significance of the elastic phase
decreases with the increase of the peak displacement, the Rigid-Plastic
estimations become closer to the FE results as seen from Figure 3-38.
 The results for the case of dynamic blast loading at stand-off distance of 15 m
are given in Table 3-15 and Figure 3-39. It was observed from the results that
the predictions of the Rigid-Plastic method are comparable to those obtained
102
using the SDOF approach. Both analytical methods produced results that are
satisfactorily close to the FE predictions as shown in Figure 3-39.
3.5.3 Further Discussion on CFST Columns
To illustrate the effectiveness of composite over RC members, the blast response of
the CFST column in the preceding study is compared to that of the RC columns using
the FE analysis. Two different configurations of the RC column compared.
In the first comparison, the CFST column and the RC column have the same external
dimensions, concrete strength and reinforcements. The aim of this comparison is to
identify the usefulness of the steel hollow section in strengthening the column against
blast loading. The explosive used in the analysis has a charge-weight of 2200 kg TNT
and is placed at a stand-off distance of 15 m. The blast pressure-time history was
defined as a bi-linear pressure-time curve, which takes into consideration the
clearence distance due to the slender nature of the column (TM5-1300, 1990). From
the comparison of the deformed shapes of the columns at time, t = 0.004 sec after the
initiation of the blast loading in Figure 3-40, it can be seen that the RC column
underwent excessive transverse shear deformation near the supports while the CFST
column responded in a flexural manner. This shows that the steel hollow section is
effective in significantly improving the lateral ductility of the column and
considerably reducing its peak displacement as shown in Figure 3-41.
103
Figure 3-40 Deformed shapes of (a) RC and (b) CFST columns at t = 0.004
100
80
RC column
60
Displacement (mm)
40
CFST column
20
0
0 0.002 0.004 0.006 0.008 0.01
-20
Time (sec)
Figure 3-41 Comparison of displacement-time histories of RC and CFST columns
The improvement is attributed to the enhancement of flexural bending moment and
more significantly, the shear resistance of the composite column as compared to the
RC column. In addition, the confinement of the concrete core by the steel section may
also be one of the factors that contributed to the improved blast resistance of the
104
composite column. This is evident from the comparison of the effective mean stress-
time variation of the concrete element at the mid-height of the composite and RC
columns as plotted in Figure 3-42. The selected elements are on the loaded face and is
primarly subjected to compressive loading. In the case of the RC column, the element
is directly exposed to the blast loading. It was observed that the element in the
composite column attained a higher effective mean stress level as compared to the one
in the RC column, eventhough the CFST column has a lower deflection. This could be
the result of the confinement effect on the concrete core, which enhance the moment
resistance and contributed to the improved blast performance of the CFST column.
25
Composite column
20
Effective Mean Stress (MPa)
15
10
RC column
0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014
-5
Time (sec)
Figure 3-42 Comparison of effective mean stress-time histories of the element at the
mid-span of RC and concrete-filled steel composite columns
In the second comparison, the blast response of the CFST column obtained in the first
comparison is compared to that of the RC column shown in Figure 3-43, which has a
square cross section and twelve T20 reinforcement bars.The RC column was designed
105
such that it has a comparable P-M interaction curve for the bending axis perpendicular
to the blast loading direction to that of the CFST column. Although the two columns
have different cross-sections, the load carrying capacities are similar as depicted by
their interaction curves in Figure 3-44. This ensure that both of the columns have
similar strength under combined static axial load and moment. The arrangement of
reinforcement bars in the RC column reflects that of a simplified design for
comparison purpose. In an actual scenario, reinforcements should be provided in both
directions since the location of the blast load is not known beforehand.
Bar diameter = 10 mm
Yield strength = 460 MPa 600 mm
Passion ratio = 0.3
E modulus = 200 GPa
Density = 7850 kg/m3
4000 mm
600 mm
Spacing = 200 mm
Loaded face 100mm

Compressive strength = 30 MPa Bar diameter = 20 mm
Tensile strength = 2.9 MPa Yield strength = 460 MPa
Passion ratio = 0.2 Poisson ratio = 0.3
E modulus = 28 GPa E modulus = 200 GPa
Density = 2320 kg/m3 Density = 7850 kg/m3
Figure 3-43 RC column with comparable properties as the concrete-filled steel

composite column in Figure 3-36
106
7000
6000
CFST column
5000
Axial Force P (kN)
4000
3000 RC column
2000
1000
0
0 100 200 300 400 500 600 700 800
Moment M (kNm)
Figure 3-44 P-M Interaction curves of CFST and RC columns
The same blast loading as the one applied in the first comparison is maintained in this
analysis. In addition, the columns also support a static vertical axial load of 2500kN.
As expected, both columns exhibit similar flexural response at the initial loading stage
(t < 0.003 sec) due to their comparable loading capacities as seen from the
comparison of the mid-span displacement time histories in Figure 3-45 as well as the
deformed shapes at t = 0.0035 sec in Figure 3-46.
107
80
70
60
50 RC column
Displacement (mm)
40
30
CFST column
20
10
0
0 0.002 0.004 0.006 0.008 0.01
-10
Time (sec)
Figure 3-45 Comparison of displacement time histories of RC and CFST columns
(a) (b)
Figure 3-46 Deformed shapes of (a) RC and (b) CFST column at t = 0.0035sec
However, subsequently the RC column buckled and failed under the applied axial
load. This shows that the RC column is not able to sustain the design axial load after
108
blast and the load has to be redistributed to the surrounding columns. Consequently,
there is a potential risk of progressive collapse if the other columns were not designed
to take additional load. As demonstrated from the FE analysis, the risk is averted in
the case of the CFST column although it has a smaller dimensions and lower lateral
stiffness than that of the RC column. The concrete material softens drastically upon
attaining its compressive and tensile strengths while the steel material is able to
sustain further load after yield. Thus, the latter is capable of absorbing more energy
which also translate to higher residual strength as evident from the FE results. The
results from the two comparisons are clear indications of the advantages of CFST
over RC columns as well as the critical importance of residual capacity in the design
of columns against blast loading.
3.6 Summary
This chapter of the thesis covered the basics parameters to be considered in the design
of CFST column which is subjected to blast loading. The study dwelled into the
dynamic material response, the analytical approach to predict the response and the
generation of numerical models.
The strain rate effects on concrete is validated using SHPB experimental setup and the
results from the test shows significant correlation to the experimental results from
various research and is more conservative than the commonly used CEB strain rate
plots. By comparing two concrete mixes which differ in the coarse aggregate used, it
was shown that the DIF for the concrete with aggregates which have a lower density
and load capacity yield lower DIF values as compared to the concrete with aggregates
109
of higher density and load capacity. This experiment may support the significance of
Stefan effects of concrete under high strain rate.
The shortcomings of the commonly used SDOF method for the analysis of structural
members subjected to blast loading are briefly highlighted. In view of this, the Rigid-
Plastic method is proposed and its advantage over the SDOF method is discussed. A
step-by-step methodology in determining the response shape and peak displacement
of a blast-loaded structure by using the Rigid-Plastic method is shown. Closed-form
solutions for the Rigid-Plastic method are also devised for steel-concrete columns to
assist engineers in the preliminary blast design of such structural members (see
Appendices A, B and C).
The Rigid-Plastic method was applied for the analysis of CFST column subjected to
various blast loadings. It was shown that the Rigid-Plastic method is applicable for
such members and is capable of predicting satisfactory results, which are similar those
obtained using SDOF method and numerical models. However, the limitation of the
Rigid-Plastic analysis was identified for cases that involve small deflections.
Nevertheless, the proposed Rigid-Plastic method offers a more efficient and easier
approach through the use of closed form equations to gauge the response shape as
well as peak displacement of CFST columns. In addition, it is also highlighted that
due to the design of critical columns, they are under high risk of failing in shear.
Therefore, the Rigid Plastic method is a superior approach in the analysis of CFST
columns.
110
FE analysis was also carried out to compare the blast performance of the CFST and
RC columns. It was demonstrated that the composite column is much more resilient
against blast loading as compared to the RC columns in terms of improved ductility,
enhanced shear resistance and improved residual load-carrying capacity.
111
CHAPTER 4:
EXPERIMENTAL PROGRAMME OF STEEL-CONCRETE-STEEL (SCS)
SANDWICH PANELS UNDER STATIC AND BLAST LOADINGS
4.1 General
Massive concrete structures are one of the primary approaches in blast mitigation. The
principle of this method relies on the mass of concrete to reduce the structural
response. This approach has been tested since the two world wars where bunkers and
critical facilities were designed with either thick layers of overburden soil or
extremely thick reinforced concrete slabs. These are in the effort to attenuate the
propagation of blast wave and protect the personnel and assets within the facility. A
classic example of such engineering in a civil structure can be located in the Cabinet
War Rooms in London where slabs ranging from one to three metres in thickness
were used to protect Winston Churchill and his staff against the bomb raids from the
Germans. In order to further enhance the protection within such bunkers, a layer of
steel is constructed under the slab. The purpose of the steel serves mainly two
purposes: one, to distribute the load from the massive concrete slab to the rest of the
supporting members and, two, to protect the personnel and assets within the facilities
from spalling high velocity concrete debris.
As technology advances, the use of steel as a structural enhancer to concrete is being
identified. It is an appropriate material to reinforce the use of concrete without
increasing the dimensions in numerous civil structures. The high strength and ductility
of steel compliments the mass and rigidity of concrete to yield a composite material
which will meet some of the spatial requirements of certain projects. The
conventional slab design is usually based on having a layer of concrete which is cast
over a layer of steel. This layer of steel will be ribbed or possess connectors to ensure
112
the composite action between the steel and concrete. In the design of connections
between supporting beams and the slab, connectors or studs will be welded along the
span of the beam and the sufficient studs must be installed to ensure the required
composite action of the design. The design of such a composite slab is clearly
documented in design codes such as the Eurocode 4. Figure 4-1 illustrates some
examples of designs of composite slabs.
(a) (b)
Figure 4-1 Examples of (a) Corrugated and ribbed steel decks and (b) connection
details on beam to ensure composite actions between steel beam and concrete slab
In such a design, there is no requirement to have layers of steel on both the tension
and compression surfaces of a section as the tension forces across a section can
normally be balanced by compressive forces from the concrete. However, there are
situations where steel is also required on compression side of the section. Such
sections can be described as steel-concrete-steel (SCS) sandwich panels and they may
be applied when:
 the direction of the load cannot be determined
 spalling of concrete may pose a threat to personnel or asset in proximity to the
113
slab as well as reduce the loading capacity of the element due to the loss of
material
 confinement of concrete ,which will increase the load capacity of the element,
is required to meet the design specification of the structure
In cases where blast loading is considered on a slab, the above three scenarios are
critical in the design considerations. The origins of the bomb threats are often difficult
to determine and thus such slabs have to be designed to take bi-directional loads.
When high explosives are positioned in close proximity to concrete, due to the
propagation of stress waves in the structural elements, the tension waves will cause
concrete spalling. The loss of concrete from this phenomenon will reduce the axial
load capacity column and, at the same time, the reduced flexural resistance from
reinforcement bars as they lose their restraints and buckle under axial loads.
In view of the need to study SCS sandwich structures, a programme was planned and
executed to study the differences in response mechanisms in both the static and
dynamic loading. This chapter will attempt to document the work that was carried in
the study. In addition to studying the difference in static and dynamic responses of
these sandwich panels, the study also covered the levels of influence of the concrete
core, steel panel thickness and plate connectors to the overall responses.
This chapter will first highlight the fundamentals to the materials used in the
specimens that are studied. Various design procedures are referred to in the design of
such sandwich panels. Subsequently, a series of static three-point load tests were
carried out to determine the load-displacement curves of various specimens. The
failure mechanisms of these specimens can also be clearly observed and compared in
114
the test. The results from the experiment are then compared to the analytical and
numerical predictions and the differences between the two results will be deduced
based on the observations from the experiment and assumptions from the analytical
approach.
The following section provides a detailed insight into the experimental programme
which leverages on Explosive Testing of Structural Components (ETSC) 2008 which
was conducted by the Defence Science and Technology Agency (DSTA) to
investigate the blast performance of SCS sandwich panels. In total, 3 load cases of
100kg TNT at stand-off distance of 5m were conducted to test 6 specimen panels. The
details behind the motivation of the experimental setup are documented in this
chapter. Instrumentation such as strain gauges, potentiometers, pressure sensors and
accelerometers were installed to capture as much of the dynamic data as possible. Pre-
and post-test photos are also included in this chapter to observe the differences
between the responses of the various specimens. The data plots from the
instrumentations are also included. Similar to the study on the quasi-static response of
SCS sandwich panels, the experimental result will be compared to analytical
predictions and deductions on the analytical approach will be listed. In addition,
numerical models, which are based on the experimental study, will be generated. In
addition to the comparison with the experimental result, these models will be used to
further identify critical parameters to design an effective blast resistance SCS
sandwich panel.
Overall, this chapter will cover both quasi-static and dynamic response of SCS
sandwich panels and provide a holistic view on:
115
 the differences between static and dynamic response
 the use of different concrete cores in such composite panels
 the differences between composite and steel panels
4.1.1 Concept of SCS Sandwich Panels
The steel-concrete-steel (SCS) composite sandwich panel is formally designed for
offshore applications in the construction of ship hulls. However, it could also be used
for the blast protection in the form of perimeter walls as well as protective barriers
between personnel and areas which contain explosively volatile materials. The
advantages in the use of such composite sandwich panels as protective walls are as
follow:
 Taking advantage of the characteristics of both steel and concrete, an optimum
design can minimise the thickness of these walls as compared to the RC
members
 Being sandwiched between steel plates, the wall will possess excellent residual
strength as compared to RC structures
 Having the top and bottom steel face plates to confine the concrete within, the
possibility of spalling and scabbing of concrete fragments, which may injure
personnel who are in proximity of the wall, is eliminated
 Being a relatively thin wall, these panels can be replaced upon damage easily
 The materials of such walls can be found locally and the fabrication process of
such SCS walls, which will be documented in latter part of this chapter, can be
easily made and assembled by the local suppliers
Blast tests had been conducted on such sandwich panels and they have been deployed
in the protection of critical structures. They performed relatively well under actual
116
explosions and research has shown that the massiveness and ability to absorb energy
has proven that the combination of steel and concrete makes an appropriate blast wall
(Heng et al, 1995; Lan et al, 2005).
Therefore, this study aspires to further understand the mechanism of such sandwich
panels with the introduction of two important factors. One of the factors is the
influence of the concrete core to the overall performance of sandwich panels.
Therefore, lightweight concrete (LWC) and ultra-high strength concrete (HSC) will
be compared with normal strength concrete (NSC). Another factor to be considered is
the connectors used. It is understood that connectors are very important in
maintaining the integrity of composite structures to increase the composite action
between steel and concrete. Therefore, in understanding the importance of the
connectors, the J-hook connectors will be used in place of standard headed stud
connectors that are commonly used in composite slabs. These connectors are welded
onto steel face plates with the alignment of the hooks being different for the top and
bottom face plates as seen in the Figure 4-2.
(a) (b)
Figure 4-2(a) J-hooked connectors on steel panel, (b) Arrangement of aligned J-
hooked connectors assembled in SCS panels
117
4.1.2 Objectives
Therefore, with the considerations from the preceding section, the specific objectives
of this study will be as follow:
a. Study the effectiveness of SCS sandwich panel in resisting quasi-static and
blast loadings
b. Study the influence of the thickness of steel face plates in the performance of
the SCS sandwich panels to quasi-static and blast loadings.
c. Study the effectiveness of J-hook connectors in the quasi-static and blast
performance of SCS sandwich panels
d. Study the influence of different concrete cores on the performance of SCS
sandwich panels to quasi-static and blast loadings.
e. Use the data collected from the quasi-static and blast load tests as benchmarks
for numerical simulation
f. Study the differences in the influence of the parameters discussed in the above
points a, b, c and d on the response of SCS sandwich panels under quasi-static
and dynamic loadings situations.
4.1.3 Design and Construction of Specimen
Six specimens were fabricated for the tests with each having different specifications.
All the specimens have the same length and width of 1200mm and 495mm
respectively. Each specimen will share the same spacing of 70mm between the inner
surfaces of the top and bottom steel face plates. The spacing of the J-hook installed in
some of the SCS panels is 100mm. Steel plates are fillet welded along the span, which
will be termed as side plates, and at both ends, which will be termed as end plates, of
the specimens to fully confine the concrete core within the specimen. The side and
118
end plates are incorporated into the structure to provide formwork for the casting of
concrete as well as a means to confine the concrete core to enhance the structural
capacity. Figure 4-3 clearly illustrates the terminologies used in the description of the
specimens. The specifications of the steel plates and concrete core and the references
to computer aided drawing of each of the six specimens are documented in Table 4-1.
Table 4-1 Specifications of the configurations of the specimens

Thickness of Thickness of
Thickness of
top and side plate/ Type of Figure
Label end plate
bottom plate stiffener concrete core Reference
(mm)
(mm) (mm)
SP 4 3 3 - 4-4
SCSN4 4 1.5 3 NSC 4-5
SCSN 3 1.5 3 NSC 4-6
SCSNE 3 1.5 3 NSC 4-7
SCSL 3 1.5 3 LWC 4-6
SCSH 3 1.5 3 HSC 4-6
Note:
SP = Cellular steel panel with stiffeners in place of concrete core in-fill
SCSN4 = SCS composite sandwich panel 4mm thick top and bottom plates
SCSN = SCS composite sandwich panel
SCSNE = SCS composite sandwich panel without J-hook connectors
SCSL = SCS composite sandwich panel with lightweight aggregate concrete core
SCSH = SCS composite sandwich panel with ultra-high strength concrete core
NSC = Normal strength concrete

LWC = Lightweight aggregate concrete
HSC = Ultra-high strength concrete
119
J-hook Connectors
Top plate
Side Plate
Bottom Plate
End Plate
(a)
Stiffeners
Top plate
Side Plate
Bottom Plate
End Plate
(b)
Figure 4-3 Notation for (a) SCS composite sandwich panel and (b) Cellular steel
panel
120
Figure 4-4 Schematic of cellular steel panel (SP) assembly
121
Figure 4-5 Schematic of SCS composite sandwich panel with 4mm top and bottom
plates (SCSN4) assembly
122
Figure 4-6 Schematic of SCS composite sandwich panel (SCSN, SCSL, SCSH)
assembly
123
Figure 4-7 Schematic of SCS composite sandwich panel without J-hook connectors
(SCSNE) assembly
The steel components (plates and hooks) of the SCS composite sandwich panels were
fabricated in the Structural Engineering Laboratory at the National University of
Singapore (NUS) and procured from external suppliers. The steel plates exhibited
124
similar stress-strain responses as S275 steel from tensile coupon tests conducted. Due
to the different thickness of steel required to construct the specimen, coupons of
various thicknesses were tested and the test setup and results are documented in
Section 4.2.1. The steel J-hook connectors were then fabricated according to the
specifications in Figure 4-8.
Figure 4-8 Schematic of J-hook connector
Other than NSC, the other two concrete types are mixed in the laboratory. NSC was
supplied from a local concrete suppler and was specified as Grade 45 concrete. The
LWC is based on a mixture of expanded clay aggregates, which were supplied by
Liapor Group, sand, ordinary Portland cement, silica fume and water. This concrete
had a specific density of 1250kg/m3. The concrete product that was used for the HSC
was D4, which was supplied by Densit ApS with claims that its cylinder strength can
be as high as 200MPa. However, the mix that was used for the test could only attain
125
190MPa. Concrete cylinder test procedures and comparisons between the unconfined
stress-strain curves of the three concrete mixes are documented in Section 4.2.2.
All the panel specimens were fabricated in the laboratory. The J-hook connectors
were welded on the steel plate with the use of a welding gun and a jig, which aided in
the alignment of all the connectors. In general, the fabrication procedure of the panels
is as follow:
 Position the top steel plate
 Position the jig for the first row of connectors as seen in Figure 4-9(a)
 Install the ceramic ferrules on one end of the connectors
 Lock the connector in place the welding gun
 Weld the connector as seen in Figure 4-9(b)
 After welding all the connectors, repeat the same process on the bottom steel
plate
 After the completion of the two face plates, hook the panels and place spacers
between them as seen in Figure 4-9(c)
 Fillet weld the side plates followed by one of the end plates, which will leave
one side open to allow the casting of concrete as seen in Figure 4-9(d) and (e)
 Casting of concrete into the steel box will then follow
 The last end plate will be positioned welded to close the box to obtain the final
product.
126
(a) (b)
(c) (d)
(e) (f)
Figure 4-9 Specimen preparation photos of (a) positioning of jig, (b) welding of J-
hook connectors, (c) placement of top and bottom plate prior to welding of the side
and end plates, (d) preparation of specimens with J-hook connectors prior to casting
of concrete core, (e) preparation of specimens without J-hook connectors prior to
casting of concrete and (f) concrete casting completion
4.2 Material Static Properties
In order to determine the design capacities of the composite structure, it is important
to assess the properties of each material that constitutes the panels. The tests that were
carried out are solely based on static test methods. Coupon tests in accordance to the
ASTM standards (ASTM E8M, 2004) were carried out on the steel that was used to
127
fabricate the top, bottom, side and end plates. In addition, utilising the same ASTM
laboratory test standards (ASTM C39/C39M, 2005) cylinder and cube tests were
carried on the three types of concretes that were used as the core materials in the
sandwich panel specimens. This section documents the test procedures as well as the
results from static tests.
4.2.1 Steel
Nine specimens were fabricated for this test. Each specimen was shaped (as seen in
Figure 4-10) according to the ASTM standards (ASTM E8M, 2004). The thicknesses
of these specimens were varied to reflect the thickness of the steel plates used in the
panel specimens. Three thicknesses were chosen: 4mm, 3mm and 1.5m. Thus, three
specimens will be assigned to each thickness.
Figure 4-10 Specification for steel coupon
4.2.1.1 Instrumentation
A strain gauge and an extensometer is being utilised in capturing data of the test. The
former being applied longitudinally on the centre of the coupon. A gauge length of
50mm was used and an appropriate extensometer, which was centred about the
midspan of the coupon, was used for the test. Therefore, there would be two raw
signal records during each of the coupon tensile test. The one end of the coupon is
attached to the loading arm test machine while the other end is secured to the
128
immobile base. The test is displacement controlled and the specimens are all tested till
failure.
4.2.1.2 Results and Discussions
Figure 4-11(a), 4-11(b) and 4-11(c) shows the 1.5mm, 3mm and 4mm steel coupons
respectively after the end of the tests. It can be observed that the failure zone is not
limited to the centre of the gauge length. As a result, it is important that the
extensometer records the post yield displacement of the coupon because the strain
gauge is only able to record the changes at the centre. Furthermore, the strain gauges
were detached from the surface of the specimen upon attaining yield stress due to
weak bond with the steel surface.
(a) (b)
(c)
Figure 4-11 (a) 1.5mm, (b) 3mm and (c) 4mm thick coupons after tensile test
Figure 4-12 plots the strain gauge signal from the 1.5mm, 3mm and 4mm steel
coupons respectively while Figure 4-13 records the signal from the extensometer.
129
Although the extensometer records the changes in displacement across the gauge
length of 50mm, this signal is converted to strain with the pre-determined gauge
length of 50mm to be plotted against the stress. In the plots, the strain that is recorded
can be considered as Cauchy or engineering strain. This would explain the drop in
stress level as the coupon is near its failure strain.
400 500
Coupon 1.5-1 Coupon 3-1
360 Coupon 1.5-2 450 Coupon 3-2
320 400
280 350
Stress (MPa)
Stress (MPa)
240 300
200 250
160 200
120 150
80 100
40 50
0 0
0 5000 10000 15000 20000 25000 30000 35000 40000 45000 0 20000 40000 60000 80000 100000 120000
Strain () Strain ()
(a) (b)
450
Coupon 4-1
Coupon 4-2
400 Coupon 4-3
350
300
Stress (MPa)
250
200
150
100
50
0
0 6000 12000 18000 24000 30000 36000 42000 48000 54000 60000
Strain ()
(c)
Figure 4-12 Strain signal recorded from strain gauges on (a) 1.5mm, (b) 3mm and (c)
4mm coupons
130
400 500
360 450
320 400
280 350
Stress (MPa)
Stress (MPa)
240 300
200 250
160 Coupon 1.5-2 200 Coupon 3-2
120 150
80 100
40 50
0 0
0 50000 100000 150000 200000 250000 300000 350000 0 25000 50000 75000 100000 125000 150000 175000 200000 225000
(a) (b)
450
400
350
300
Stress (MPa)
250
200
Coupon 4-1
150 Coupon 4-2
Coupon 4-3
100
50
0
0 50000 100000 150000 200000 250000 300000 350000
Strain ()
(c)
Figure 4-13 Processed strain data recorded from extensometer on (a) 1.5mm, (b) 3mm
and (c) 4mm coupons
By comparing the results, the following observations and deductions can be made
with a summary of the details tabulated in Table 4-2:
1. The yield strength of steel is in the range of 300MPa to 350MPa. This is
similar to the S275 structural steel of the British Standard. From the recorded
data, the mean yield strength is 315MPa.
2. With regards to the ultimate strength, the comparison of the steel coupon of
different thickness produce more varied results. 1.5mm thick coupons tend to
have a lower ultimate strength as compared to the 3mm and 4mm steel
coupons. The highest ultimate stress attained by the 1.5mm coupon was
395MPa. On the other hand, 3mm and 4mm coupons produced ultimate stress
of 475MPa and 440MPa respectively. By taking the mean of all the records,
131
the ultimate strength of the material is 427MPa.
3. The ultimate strain are more scattered for the 3mm and 4mm coupons as
compared to the 1.5mm coupons. Taking the average of the values tabulated in
Table 4-2, it can be assumed the ultimate strain of the material can be taken as
0.23.
Table 4-2 Tabulation of the key parameters from the tensile test
Specimen Yield Strength Ultimate Strength
Ultimate Strain
thickness (mm) (MPa) (MPa)
300 385 0.32
1.5 310 395 0.275
310 385 0.325
330 445 0.125
3 350 475 0.2
330 450 0.225
310 440 0.2
4 300 425 0.195
300 440 0.225
4.2.2 Concrete
As mentioned in preceding sections, three types of concrete were used. The NSC used
for the test was a ready-mixed concrete labelled as Grade 45 in the product catalogue.
It was used in specimens SCSN4, SCSNE and SCSN.
LWC which composes primarily of low density aggregates called Liapor was mixed
for specimen SCSL in the test. This aggregate is a type of expanded clay and it was
procured from Liapor GmbH & Co. KG. These aggregates can be seen in Figure 4-14.
The volumetric mix that had been used for the test is tabulated in Table 4-3.
132
Figure 4-14 Sample pictures of Liapor aggregates
Table 4-3 Lightweight concrete mix

Ordinary Lightweight Lightweight
Portland Water Silica Fume Fine Coarse
W/C
Cement (kg/m3) (kg/m3) Aggregate Aggregate
(kg/m3) (kg/m3) (kg/m3)
750 200 80 125 205 0.27
The HSC used in Specimen SCSH was D4 concrete which was obtained from Densit
ApS. The aggregates and cement are premix in barrels and they are mixed with water
to obtain the concrete. The mix is tabulated in Table 4-4.
Table 4-4 HSC concrete mix

D4 ready-mix (kg/m3) Water (kg/m3)
2630 200
Nine concrete cylinders were moulded for this test. Each specimen was prepared
using moulds which were according to the ASTM standards (ASTM C39/C39M,
2005). The end products were cylinders of 50mm diameter and 100mm height. For
each of the concrete type, three specimens were made.
Both strain gauges and linear variable differential transformers (LVDT) were used in
measuring the response of concrete under the quasi-static loading. For each specimen,
4 strain gauges were used, amongst which two were used to measure the longitudinal
strain response while the other two were used to record the changes in the transverse
133
axis. To evaluate the post yield or softening response of the concrete cylinders, four
LVDT were used to measure the movement of the loading piston.
For the NSC and LWC cylinders, the MTS machine was used. It has a loading
capacity of 1000MN. Figure 4-15 shows the setup of the cylinder with the necessary
instrumentation. However, due to the high ultimate stress that can be achieved by the
HSC, another machine with a larger load capacity had to be used. Figure 4-16(a)
shows the close-up view of the specimen together with the strain gauges and LVDT
that were used for the test and, as seen in the figure, the configuration of all the
instrumentations is similar to ones used for the NSC and LWC cylinders.
Figure 4-15 View of the instrumentation on the NSC and LWC cylinders
(a) (b)
Figure 4-16 View of the (a) instrumentation and (b) test machineries used for the HSC
cylinder static tests
134
4.2.2.2 Results and Discussions
Figure 4-17 shows the stress-strain history obtained from the instrumentation.
Records prior to the peak stress were obtained from the strain gauge readings whereas
the post-peak response was derived from the LVDT that measured the total
deformation of the cylinder. Figure 4-18 recorded from the strain gauges plots the
transverse stress-strain history. The critical information which was derived from these
figures are tabulated in Table 4-5 based on the ASTM C469-02 Static Modulus of
Elasticity and Poisson’s Ratio of Concrete in Compression (ASTM C469, 2002).
It should be noted that HSC Cylinder 1 was loaded twice as featured in Table 4-5. The
reason being that no softening curve was recorded due to the fact that the piston of the
machine automatically unloaded upon reaching the ultimate load. Upon failure, the
HSC cylinder produces a loud bang sound and that was missing in the first specimen.
Therefore, the same specimen was reloaded and it failed on the second try.
40 25
Cylinder 1 Cylinder 1
35 Cylinder 3 Cylinder 3
20
30
25
Stress (MPa)
Stress (MPa)
15
20
10
15
10
5
0 0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 0 500 1000 1500 2000 2500 3000 3500 4000 4500
(a) (b)
135
200
Cylinder 1
180 Cylinder 1 (Reload)
Cylinder 2
Cylinder 3
160
140
Stress (MPa)
120
100
80
60
40
20
0
0 500 1000 1500 2000 2500 3000 3500 4000 4500
Strain ()
(c)
Figure 4-17 Longitudinal stress-strain curve of (a) NSC, (b) LWC and (c) HSC
cylinders
40 22.5
Cylinder 2 21 Cylinder 2
36
Cylinder 3 19.5 Cylinder 3
32 18
16.5
28
15
Stress (MPa)
Stress (MPa)
24 13.5
12
20
10.5
16 9
7.5
12
6
8 4.5
3
4
1.5
0 0
0 2000 4000 6000 8000 10000 12000 0 60 120 180 240 300 360 420 480 540 600
Strain (() Strain ()
(a) (b)
210
Cylinder 1
195 Cylinder 1 (Reload)
180 Cylinder 2
Cylinder 3
165
150
135
Stress (MPa)
120
105
90
75
60
45
30
15
0
0 100 200 300 400 500 600 700 800 900
Strain ()
(c)
Figure 4-18 Transverse stress-strain curve of (a) NSC, (b) LWC and (c) HSC
cylinders
136
Table 4-5 Information derived from compressive cylinder tests
Ultimate Strength Modulus of
Concrete Cylinder No Poisson Ratio
(MPa) Elasticity (GPa)
1 35.26 24.11091 0.231576
NSC 2 39.15 25.68654 0.206692
3 35.24 24.7823 0.198602
1 20.41 11.21809 0.25901
LWC 2 20.10 10.98703 0.205134
3 18.99 11.25838 0.236801
1 196.68 58.83788 0.257031
1 (Reload) 172.05 42.58773 0.189577
HSC
2 184.16 62.46412 0.263459
3 172.62 63.6143 0.27019
By comparing the table and figures above, the following conclusions can be drawn:
1. HSC exhibited the highest strength, which went as high as 196.68MPa, among
the three concrete tested and it was followed by NSC and LWC respectively.
2. The same could be said of the static modulus of elasticity whereby the
stiffness of the HSC cylinders was more than two times of NSC. On the other
hand, the modulus of elasticity of NSC cylinders was two times that of LWC.
3. The Poisson ratios of all three concretes were similar as they do not differ as
significantly as the ultimate stress or the modulus of elasticity. In fact, the
Poisson ratios for all the specimens are more scattered.
4. The strain at ultimate stress of NSC and HSC are in the same range of between
0.0025 and 0.0035 while LWC has a lower value of 0.002.
5. Prior to the ultimate stress for LWC and HSC, the gradient of the stress-strain
curve are very constant. This is a stark contrast to that of NSC whereby the
gradient changes as the load reaches the peak
By observing the softening curve of the different concrete types, NSC clearly
possesses a larger residual capacity than that of LWC and HSC. HSC almost does not
have a softening response as the stress dropped very drastically upon reaching the
ultimate stress
137
4.3 SCS Sandwich Panels Design Capacity under Static Loading
Under the three point test setup (see Figure 4-19), the specimen panels will primarily
exhibit flexural response. The design confines the concrete within the panel which
reduces the effects of bond slip between concrete and steel and thus this will be
ignored in the analysis. Therefore, the capacity of the specimens will be largely based
on the flexural stiffness and capacity.

Load
Figure 4-19 Schematic of the setup of a three point quasi-static load test
The equivalent steel beam approach is used in the derivation of the analytical
properties of the specimen (McKinley and Boswell, 2002). Figure 4-20 illustrates the
derivation of the rectangular stress block which assumes that the neutral axis z is
measured from the underside of the compression plate.
b
tt Top Plate Ft
z Fc+Fsc
z 2z/3
h NA NA
h Side Plate h
h
b/n 2(h-z)/3
Concrete Fst
Core Fb
tb
Bottom
ts ts Plate
(a) (b) (c)

Figure 4-20 Illustration of (a) Section, (b) Equivalent Steel Section and (c) Stress
Block of a sandwich panel
From Figure 4-20(b), based on assumptions considered by McKinley and Boswell and
the additional properties of the side plates, the elastic neutral axis may be predicted as
follow:
138
h
h h
(4-1)
where n is the ratio between the modulus of elasticity of steel and concrete. The stress
of the side plate is assumed to vary linearly from the top and bottom plates to the
elastic neutral axis. With these preceding assumptions, the elastic moment capacity of
the section can be derived by taking the moment about the line of action of the
compressive forces in the concrete and the side plates (Fc + Fsc) and the expression is
as follow:
h (4-2)
As sandwich panels are designed to be similar to under-reinforced concrete beams
(McKinley and Boswell, 2002), which implies the bottom plate will yield first (fb =
fy). With Equation (4-1) and the assumption that the plane sections remain plane after
straining, ft at yield can be evaluated as a function of fy. Assuming that the maximum
tensile stress of the side plate under the elastic regime is similar to the stress in the
bottom plate (fst = fb = fy), the elastic moment capacity of the beam can be expressed
as:
(4-3)
h
Under further loading, the section will respond plastically which is characterised by
the rising of the neutral axis. The rise of the neutral axis will reduce the contribution
of the concrete and the side plate to the enhancement of the moment capacity and this
139
will subject the top plate to excessive compressive load. This is characterised in
experiments by the generation of tensile cracks in concrete towards the compressive
fibre of the concrete (Sohel and Liew, 2011). When the neutral axis has reached the
underside of the top plate (z = 0), the contribution of the concrete will be negligible
and without the consideration of the concrete core, the plastic moment capacity of the
sandwich panel can be assessed as follow:
(4-4)
Further loading will result in the local buckling of the top plate. This response will be
observed and described in the subsequent sections.
As mentioned in the first paragraph of this section, assuming that the specimens
undergo pure bending with no slip between the steel plate and the concrete core, there
will not be any shear deformation in the core. Therefore, the bending deflection of the
specimen for the three-point load test can expressed as follow:
(4-5)
where EIeff is the effective stiffness of the composite panel which is based on
uncracked section.
The above assumptions and equations are based on the full composite action of the
section. However, it is important to verify whether a composite structural component
possesses full or partial composite properties in order to assess accurately the
structural response under various loadings.
140
Based on the concepts introduced in Eurocode 4, the design capacities of the
connectors must be able to resist the longitudinal shear. The maximum longitudinal
shear force can be calculated based on the tensile capacity of the bottom steel plate (Rs
= fybtb) and the capacity of the shear connectors Rq according to BS5950 Part 3
Section 3.1 can be calculated based on the following expression:
(4-6)
where ns is the number of shear connectors between points of zero and maximum
moment, k is the reduction factor for decking (0.8 for sagging moment regions and 0.6
for hogging moment regions) and Qk is the characteristic resistance of the studs. The
values of Qk can be derived from Table 5 of BS5950 Part 3 Section 3-1.
Equation (4-6) does not take into consideration of the crushing of the concrete. This is
considered in Eurocode 4, in which the design resistance of the shear connectors can
also be calculated in the following manner (Eurocode 4, 2004; Ranković et al, 2002):
(4-7a)
(4-7a)
where,  = 0.2(hs/d +1) for 3 ≤ hs/d ≤ 4 or  = 1.0 for hs/d  4, hs = overall height of
the stud, d = diameter of the stud shank, f u = specified ultimate tensile strength of the
stud but  500 MPa, f ck = characteristic cylinder strength of concrete, Ecm= secant
modulus of concrete and the partial material factor is normally taken as 1.25. Taking
the lesser of the Equations (4-7), the capacity of the shear connectors can be
expressed as:
(4-8)
141
For the case, whereby Rq is less than the maximum longitudinal shear stress Rs, the
specimen should be designed as a partial composite beam. Assuming maximum
capacity of the bottom steel plate is Rq (<Rs) which would result in the maximum
yield stress attainable by the bottom steel plate to be accessed as follow:
(4-9)
By using Equation (4-6) into Equation (4-3), the elastic moment capacity under partial
composite action can be computed.
To evaluate the vertical transverse resistance of the sandwich panels, it will be
assumed that the concrete does not contribute the shear resistance of the specimen and
the resistance will be evaluated as follow:
(4-10)
Where Av = shear area of the steel section = 2(h + (tb + tt)/2)ts (for specimens SCSN4,
SCSN, SCSNE, SCSL, SCSH), 4(h + (tb + tt)/2)ts and fd = design strength of steel.
4.3.1 Analytical Properties of SCS Sandwich Panels
This section will document the assessment of the static properties of the six specimens
that are described in Table 4-1. The results derived from this section will be validated
in the Section 4.4. The properties that are studied in the analysis are as follow:
 Full or partial composite interaction
 Elastic moment capacity
 Elastic displacement at midspan
 Plastic moment capacity
 Vertical shear capacity
142
Parameters described in Table 4-6 and Table 4-7, which is derived from the data
collected from various material tests (see Section 4.2), will be used to calculate the
five properties above. Table 4-8 compares these properties with additional data of the
6 specimens.
Table 4-6 Properties of steel used for analytical study

Steel Plate J-hook connectors
Mass Density (kg/m3) 7860 7860
Characteristic Strength, fyk (MPa) 300 300
Ultimate Strength, fu (MPa) 400 400
Modulus of Elasticity, E (GPa) 200 200
Diameter, d (mm) - 10
Height, hs (mm) - 45
Table 4-7 Properties of concrete used for analytical study

NSC LWC HSC
Mass Density (kg/m3) 2310 1450 2600
Characteristic Strength, fck (MPa) 35 20 185
Material Partial Factor of Safety 1.5 1.5 1.5
Modulus of Elaticity, E (GPa) 24 10 60
Table 4-8 Properties of specimens used for analytical study

SP SCSN4 SCSN SCSNE SCSL SCSH
Full or partial
NA Partial Full NA Partial Full
composite
Elastic moment
44.80 23.24 22.66 25.96 11.88 21.64
capacity (kNm)
Elastic displacement at
3.21 1.56 2.02 2.40 1.13 1.75
midspan (mm)
Elastic Point Load at
195.24 179.81 133.61 133.61 133.61 133.61
midspan (kN)
Plastic moment
48.81 44.95 33.40 33.40 33.40 33.40
capacity (kNm)
Vertical shear capacity
153.81 38.45 37.93 37.93 37.93 37.93
(kN)
It is to be noted that SCSNE is classified as NA as there are no shear connectors in the
specimen and, in the calculations, it is assumed that the composite panel has full
composite interaction between concrete and steel. The result of this assumption will
be evaluated with the experimental results from the following section.
143
4.4 Experimental Study on the Static Capacity of SCS Sandwich Panels
With the design capacities of the specimens ascertained through the design theories
introduced in the preceding section, an experimental programme is devised to validate
these predictions. The programme will validate the assumptions that are made in the
analytical approach and observations from the test will aid in the understanding of
certain mechanisms analytically possible to calculate.
4.4.1 Experimental Setup and Instrumentation
The experiment entails the use of a three point load test to record the response of
sandwich panels which are subjected to quasi-static loading (see Figure 4-21). The
boundary conditions at the two ends of the specimen are assumed to be pinned of
which one end is free to translate in the axial direction of the specimen. The load is
being applied by a 50T actuator. In order to perform a three point test on the panels,
the actuator arm is attached to a bar to create line load across the width of the
specimen. All six specimens are loaded till they commence to soften with the
exception of SCSH.
Figure 4-21 Photo of the three point load test for panel specimens
In total, five strain gauges and five LVDT are used in the instrumentation of each
specimen. The positions of five strain gauges which were arranged parallel to the span
of the specimen and three of the LVDT are illustrated in Figure 4-22. These gauges
measured the strain and displacement histories at the quarter and midspan of the
144
panels. The other two LVDT measured the vertical displacement at the two boundary
conditions which are essential to ensure proper data is acquired for benchmarking the
numerical simulation. The experiment is displacement controlled with the loading rate
maintained at 0.1mm/min initially and accelerated as the specimen shows plastic
response. The results from these instrumentation will be documented in the following
section.
Simple Support on
Roller
: Strain guage
: LVDT
Load Line
Simple Support
Fixed
Top Side Bottom Side

Figure 4-22 Instrumentation setup of the panel specimens
4.4.2 Test Results
This section documents the readings from the LVDT and strain gauges. In addition to
the plots, photos showed local failure which is prevalent in most of the specimens.
The displacement histories of the mid and quarter span are plotted in Figure 4-23 and
145
Figure 4-24 which compared the deformation of the specimens. The displacement
plots have take into consideration the vertical displacement of the pivots at the ends of
the specimens. In addition, Table 4-9 extracts some of the more critical parameters
from the comparison.
300 300
250 250
200 200
Load (kN)
Load (kN)
150 150
100 100
50 50
0 0
0 10 20 30 40 0 20 40 60 80
Displacement (mm) Displacement (mm)
(a) (b)
300 300
250 250
200 200
Load (kN)
Load (kN)
150 150
100 100
50 50
0 0
0 20 40 60 80 0 20 40 60 80
(c) (d)
300 300
250 250
200 200
Load (kN)
Load (kN)
150 150
100 100
50 50
0 0
0 10 20 30 40 0 50 100 150
(e) (f)
Quarterspan 1 Quarterspan 2 Midspan
Figure 4-23 Normalised quarterspan and midspan displacement histories from LVDT
of (a) SP, (b) SCSN4, (c) SCSN, (d) SCSNE, (e) SCSL and (f) SCSH sandwich
panels
146
225
SP
200
SCSN4
175
SCSH SCSNE
150
SCSN SCSL
Load (kN)
125
100
75
50
25
0
0 2 4 6 8 10
Displacement (mm)
(a)
300
SCSN4
250
SCSH
200
SCSNE
Load (kN)
SP SCSN
150
SCSL
100
50
0
0 20 40 60 80 100 120 140
Displacement (mm)
(b)
Figure 4-24 Comparison of the midspan displacement of panel specimens during the
(a) elastic and (b) elasto-plastic response
147
Table 4-9 Important parameters from the comparison of the panel specimens which
are subjected to a three point load test
Midspan Quarterspan
Specimen Load at yield (kN) displacement at displacement at Ultimate load (kN)
yield (mm) yield (mm)
SP 205 8.5 5.2 205
SCSN4 210 9.8 6.0 266
SCSN 145 7.0 4.4 190
SCSNE 160 9.4 5.6 190
SCSL 150 8.8 5.5 161
SCSH 156 7.3 3.9 217
Five strain gauges were installed in each specimen but it can be observed that some of
these readings are incomplete due to the debonding of the strain gauges from the
surface of the steel plates. The two readings that are plotted in the negative strain axis
were recorded from top steel plate which was under compression. There are two sets
of plots for each specimen: one that plots the load against the strain and the other that
plots strain against the midspan displacement. The latter is plotted to illustrate the
history of the strain readings during the response.
250 15000
200
10000
Load (kN)
150
Strain (µ)
100 5000
50
0
0 0 10 20 30 40
-5000 0 5000 10000 15000
-5000
Strain (µ) Midspan Displacement (mm)
(a)
300 25000
250 20000
200
Load (kN)
15000
Strain (µ)
150
10000
100
5000
50
0 0
-5000 0 5000 10000 15000 20000 0 10 20 30 40
-5000
(b)
148
250 25000
200 20000
Load (kN)
150 15000
Strain (µ)
100 10000
50 5000
0 0
-5000 0 5000 10000 15000 20000 25000 0 20 40 60 80
-5000
(c)
250 15000
200 10000
Load (kN)
150
Strain (µ)
5000
100
0
50 0 20 40 60 80 100
-5000
0
-10000 -5000 0 5000 10000 15000
-10000
(d)
200 15000
150 10000
Load (kN)
Strain (µ)
100
5000
50
0
0 0 10 20 30 40
-5000 0 5000 10000 15000
-5000
(e)
149
250 40000
35000
200
30000
Load (kN)
150 25000
Strain (µ)
20000
100 15000
50 10000
5000
0 0
-10000 0 10000 20000 30000 40000
-5000 0 50 100 150
(f)
Quarterspan 1 Quarterspan 2 Midspan
(Bottom) (Bottom) (Bottom)
Quarterspan 1 Quarterspan 2
(Top) (Top)
Figure 4-25 Strain gauge readings from (a) SP, (b) SCSN4, (c) SCSN, (d) SCSNE,
(d) SCSL and (e) SCS
250
SCSN4
200
SP
SCSNE
150
Load (kN)
SCSN SCSH SCSL

100
50
0
0 1000 2000 3000 4000 5000
Strain (µ)
(a)
150
300
SCSN4
250
200 SP SCSN
Load (kN)
SCSNE SCSH
150
SCSL
100
50
0
0 5000 10000 15000 20000 25000 30000 35000 40000
Strain (µ)
(b)
Figure 4-26 Comparison of the midspan bottom strain of panel specimens during the
(a) elastic and (b) elasto-plastic response
Due to the manner in which the specimen panels were loaded, all the specimens failed
in a flexural manner. SCSN4, SCS, SCSL and SCSH yielded similar response in
terms of overall response shape and failure mode. Upon being loaded at the midspan,
the panel will respond in a flexural manner and, beyond the yield point, and with
increasing load, the top steel plate buckles under compression near the loading bar.
The local buckling normally occurs between two rows of connectors which are closest
to the loading bar (see Figure 4-27). This is the point where the actuator will decrease
the load being applied to the specimen and also the point when the experiment is
stopped.
151
Figure 4-27 Buckling of top steel plate that initiated softening for specimen SCSN
The elastic response of SP is similar to the rest of the specimens but, upon yielding,
the specimen is not capable of taking any further loading and thus softens. The top
steel plate buckles between the stiffeners the transferred load from the top plate will
cause the stiffeners to buckle along its weak axis.
The sandwich panel without connectors SCSNE, again exhibits similar elastic
response as the rest of the specimens but the observations from post-yield records
show otherwise. Due to the absence of the connectors across the span of the specimen,
instead of local buckling which is observed in the other concrete-filled specimens, the
top steel plate exhibits global buckling where the steel plates detaches itself from the
concrete core and deforms in a convex manner between the loading bar and the ends
of the panel (see Figure 4-28).
152
Figure 4-28 Buckling of top steel plate that initiated softening for specimen SCSNE
The points below summarises the observations from the test records:
1. Comparing SP and SCSN4, it is clear that although they do not share the same
structural configuration, the elastic stiffness of these specimens are close,
which was as specified in the design.
2. It is observed that SCSN, SCSNE, SCSL and SCSH the elastic stiffness and
the yield point are very close to each other. This can be attributed to the fact
that elastic stiffness and yield points are largely controlled by the steel plates.
3. SP and SCSN4 have a higher yield load capacity as compared to the rest of the
sandwich panel specimens.
4. Beyond the yield point, SCSN4 exhibits are hardening while SP softens as the
panels are further loaded with SCSN4 attaining an ultimate load capacity of
266kN.
5. Both SCSN and SCSNE share the same ultimate load capacity of 190kN but
the point at which the specimen softens are different. SCSN exhibit a steeper
hardening curve and thus attains the ultimate load capacity earlier than
SCSNE.
6. Of all the specimens SCSL has the lowest ultimate load capacity of 161kN and
the gradient of the hardening curve is the lowest while SCSH has the highest
ultimate load capacity of 217kN and the corresponding the hardening curve is
153
the highest.
7. The readings from the midspan strain gauge on the bottom steel plate
correlates well with the displacement response where higher strain is attained
for SP and SCSN4 as compared to the rest.
8. With exception to SCSNE, the strain will increase (positively for the strain
gauges on the bottom steel plate and negatively for those on the top steel plate)
as the specimen is loaded till yield. Due to the formation of the plastic hinge at
midspan after the yield is attained, the strain gauges at the quarterspan cease to
increase and remain constant due to the hardening process. Upon reaching the
ultimate strain, these strain gauges showed that the top steel plate starts to go
slightly into tension and the bottom plate into compression.
9. The difference of the response of SCSNE was clearly read from the
quarterspan strain histories at the top steel plate. Instead of exhibiting
compression, the strain increased positively, which means that the material at
the quarterspan is actually under tension during the elastic stage. This shows
that bulging of the top has occurred not just during the post-yield stage, as
mentioned in one of the preceding paragraphs, but also during the elastic
response.
10. The post yield quarterspan strain is larger for SCSNE and SCSH (see Figure 4-
25(d) and (e) respectively) than the rest of the specimens.
4.4.3 Discussion and Comparison with Analytical Solution
As described in one of the preceding paragraphs, the experimental load-displacement
variation of the specimens will be used to assess the analytical approach. From the
154
analytical results obtained in Table 4-8, the elastic and plastic yield points will be
compared with the experimental results, which are shown in Figure 4-29.
250 250
Midspan Point Load (kN)

200 200
150 150
100 100
50 50
0 0
0 2 4 0 2 4
Midspan Displacement (mm) Midspan Displacement (mm)
(a) (b)
250 250
200 Midspan Point Load (kN) 200
150 150
100 100
50 50
0 0
0 2 4 0 2 4
(c) (d)
250 250
200 200
150 150
100 100
50 50
0 0
0 2 4 0 2 4
(e) (f)
Analytical Elastic Analytical Plastic Experimental
Yield Point Yield Point Result
Figure 4-29 Plots of the analytical properties from Table 4-8 of (a) SP, (b) SCSN4, (c)
SCSN, (d) SCSNE, (e) SCSL and (f) SCSH against experimental results
155
The initial observation is that the modulus of elasticity from both analytical and
experimental results is close as the analytical plots are on the load-displacement
histories of all the six specimens. Therefore, it can be deduced that the method
introduced in Section 4.3 is appropriate for predicting the modulus of elasticity of
such sandwich panels.
In addition, it is also observed that the predictions are conservative as the plastic
limits predicted by the analytical methods are lower than the point in which the
specimen response starts to exhibit strain hardening. Therefore, it can be deduced that
the method introduced in Section 4.3 is conservative. It is most conservative when
used with LWC and least conservative with steel stiffened plate panels. This
observation can be attributed to confinement of the concrete within the specimen. The
confinement will increase the hydrostatic pressure of concrete which will in turn
increase the shear or deviatoric strength of the concrete. This is difficult to quantify
analytically and thus it will not investigated within the scope of this study.
4.5 Numerical Study on Static Capacity of SCS Sandwich Panels
This section documents the numerical simulation work that was carried out using LS-
DYNA to extend the work that was carried out analytically and experimentally. One
of the main objectives of this study is to verify the static response of the specimens to
create a calibrated model.
4.5.1 FE Solution Scheme
As opposed to the transient analysis of the CFST column in Chapter 3, the FE analysis
that is carried out on the quasi-static response of the SCS sandwich panels is based on
156
the implicit scheme. The nonlinear solution with Davidson-Fletcher-Powell (DFP)
updates is selected for the model and the time step is deliberately kept constant to
ensure the capture of some of the plastic response of the specimen (Hallquist, 2006).
4.5.1.1 Geometry and Design of SCS Panel
The geometries of the six specimens are the same to the ones described in Table 4-1.
The locations of the components of the specimens are similar to the description except
for the J-hook connectors, which are used in Specimen SCSN4, SCSN, SCSL and
SCSH.
Due to the complexity of including the actual geometry of the J-hook and interfacing
the J-hook with the concrete core, the J-hook connectors are simplified to two rows of
solid elements connected by a discrete beam element, which reacts like a spring.
Figure 4-30 illustrates the J-hook connector model in a cross section of a specimen
that has J-hook connectors. Details on the properties and formulations of the J-hook
connectors will be furnished in the following sections.
(a) (b)
Figure 4-30 Model of the J-hook connectors
The load actuator and the boundary conditions were also modelled. Figure 4-31
illustrates the actuator and the roller supports at the two ends. As the quasi-static load
157
test is displacement controlled, the actuator is modelled as a load block and a row of
nodes is assigned at the top which displaced transversely (negative z-direction) to
model the displacement-controlled mechanism that was used in the static test setup.
The roller supports at the end are modelled as support blocks on both ends of the
specimen: one being fixed while the other being able to displace along the span of the
specimen. Figure 4-31 illustrates the arrangements of the rotating support blocks with
the appropriate boundary conditions.
Nodes under displacement

Load Block (Actuator) control in z direction
x Boundary Condition: x rotation

y
Boundary Condition: x rotation and y translation free
Figure 4-31 Numerical model with support and load blocks illustrated
Benefitting from the symmetrical setup of the experiment, only half the setup is
modelled (see Figure 4-32). The two support block is fully bonded with the nodes on
the underside of the bottom steel plate. Each of the specimens has a different number
of parts due to the various configurations of the specimens. Table 4-10 attempts to
clarify the parts which are incorporated in the setup of the models of the six
specimens.
158
Plane of
Symmetry
Figure 4-32 Plane of symmetry of the SCS specimen
Table 4-10 Tabulation of the parts in each model

SP SCSN4 SCSN SCSNE SCSL SCSH
Top Steel Plate      
Bottom Steel Plate      
Side Steel Plate      
End Steel Plate      
Concrete Core     
Load and Support Block      
J-hook Connectors (Solid)    
J-hook Connectors (Beam)    
A single layer of 10x10mm2 thick shell elements were used primarily to build up the
top, bottom, side and end steel plates. The thickness of the elements depends on the
thickness of the steel plates used. The concrete core constitutes of 10x10x10mm3
constant stress solid elements. It is to be noted that not all elements share the same
geometry. In order to ensure a continuous element mesh for the parts which are
merged at their interfaces, certain elements have different aspect ratios. Figure 4-33
shows the mesh pattern that was adopted for Specimen SP and SCSN, which also
represents the model adopted for SCSN4, SCSL and SCSH. The mesh used to model
Specimen SCSNE is similar to Figure 4-33(b) except that the grid is uniform along
the face of the steel plate part due to the absence of the J-hook connectors.
159
(a) (b)
Figure 4-33 Mesh used for the top steel plate of Specimen (a) SP and (b) SCSN
As described in preceding chapter, the J-hook is composed of three parts: one part
which is composed of solid elements, bonded to the top steel plate, one part which is
also composed of solid elements, bonded to the bottom steel plate, one discrete beam
element that connects the two preceding parts, which has zero initial length. Figure 4-
34 shows a close-up view of the connectors in view of the other components and an
exploded view of the three parts that constitutes the model.
Solid part attached Discrete beam

to top steel plate element (Spring)
Solid part attached

to bottom steel plate
Figure 4-34 Constitution of the J-hook connector model used
The same elasto-plastic material model with a user defined stress versus strain curve
(*MAT_PIECEWISE_LINEAR PLASTICITY).with a user material that was
assigned to the top, bottom, side and end face plates was also applied to the solid parts
of the J-hook connectors and the load and support blocks. The properties of this
material were derived from one the coupon test discussed in Section 4.2.1.2. The true
stress versus true strain relationship that was used in the model is plotted against the
engineering stress-strain plots from the experiment in Figure 4-35. It is to be noted
160
that although this material model has the capacity to consider strain rate effects, this
factor is not considered for steel and the rest of the other materials in this study
600
500
400
Stress (MPa)
300
Experiment (Engineering)
200 Numerical (True)
100
0
0 0.05 0.1 0.15 0.2 0.25
Strain
Figure 4-35 Experimental and numerical stress-strain relationship
The concrete model (*MAT_CONCRETE_DAMAGE_REL3) that is used in the
models is similar to the one discussed in Section 3.5.1.3. Three types of concretes are
being considered in this study: NSC, LWC and HSC. The constitutive material model
of the three concrete are based on the plots documented in Section 4.2.2.2. In
addition to the characteristic cylinder strength, the secant modulus, the poisson ratio
and the softening rate derived, the mass density and the aggregate size are also taken
into consideration in calibrating the constitutive material models.
The properties that were used in modelling the discrete beam in the J-hook model are
described in terms of the force with respect to the elongation. The properties of the J-
hooks for various concrete types differ which is illustrated in Figure 4-36.
161
60
50
40
Force (kN)
30 NSC
LWC
20
HSC
10
0
-100 -50 0 50 100
-10
Displacement (mm)
Figure 4-36 Force-Displacement used to specify the material model J-hook in NSC,
LWC and HSC
4.5.1.4 Contact Interface Modelling
Three approaches were taken into account for defining the contact conditions between
the various parts that constitutes the model. The first approach is similar to the contact
properties that were used for the concrete-steel interface in Section 3.4.1.4. In
addition, the model includes steel-steel contact models. The approach is similar to the
concrete-steel model with a difference in the coefficient of friction. The value used in
the steel-steel contact model is 0.25.
The second approach assumes full composite action between the steel and concrete
components by having all the components to share their neighbouring nodes.
162
The third approach adapts the two approaches described in the previous two
paragraphs. The initial assumption is based on the full composite action between steel
and concrete. As the specimen is loaded, the steel and concrete nodes which are
attached to one another will detach themselves when a certain stress value is attained,
after which the assumption is similar to the first approach. This approach attempts to
simulate the natural adhesion between the steel plates and the concrete core and the
debonding that would occur under certain stress conditions. Figure 4-37 illustrates the
debonding failure model that is adopted in the model.
Figure 4-37 Debonding failure model adopted in numerical model of SCS sandwich
panels
The results of these three approaches will be compared in Section 4.5.2.
These six specimens are simulated and the results from the models will be compared
with those that are derived through the experiment and analytical approaches.
4.5.2.1 Specimen SP
Figure 4-38 compares the numerical, experimental and analytical results. The
numerical force values are derived based the reaction forces calculated at the two
boundary conditions and the displacement is taken at the midspan of the specimen.
163
Figure 4-39 shows deformed shape of Specimen SP from LSDYNA. The deflections
in Figure 4-39 is scaled up by 5 times to observe the response of the steel plates.
250
200
Reaction Force (kN)
150
Experiment
100 Numerical
Analytical
50
0
0 2 4 6 8 10
Midspan Displacement (mm)
Figure 4-38 Force-displacement plot of Specimen SP
(a)
(b)
Figure 4-39 Response of Specimen SP at (a) yield and (b) at 10mm midspan
deflection
It is observed that the elastic modulus and the yield points of the numerical model
match that of the experimental and the analytical approaches. However, whereas the
numerical model shows hardening of the stiffened steel panel, the experiment exhibits
a softening response. The softening response is attributed to the buckling of the top
164
steel plates in the experiment. The point of buckling failure is highly dependent on the
construction of the panel. Weld failure as well as imperfections in the steel plates may
have led the experimental specimen to buckle earlier. In addition, the modelling of
local buckling is highly mesh dependant and, thus, would require further study to
evaluate a more physics based model to tackle this issue.
4.5.2.2 Specimen SCSNE
Figure 4-40 compares the numerical, analytical and experimental results as it plots the
reaction forces against the displacement at the midspan of the specimen. In Figure 4-
40, the numerical results from the three contact interface modelling approaches are
compared. Figure 4-41 shows deformed shape of Specimen SCSNE from LSDYNA
and again the deflections are scaled by a factor of 5. Figure 4-42 shows the cracking
and damage within the concrete core.
200
180
160
Reaction Force (kN)
140
Experiment
120
Numerical (Penalty)
100
Numerical (Full Composite)
80
Numerical (Nat Adhesion)
60
Analytical
40
20
0
0 2 4 6 8 10
Figure 4-40 Force-displacement plot of Specimen SCSNE
165
Figure 4-41 Response of Specimen SCSNE at 10mm midspan deflection
Figure 4-42 Predicted crack or damage within the concrete core of Specimen SCSNE
From the comparison of the contact approaches, it is clearly observed in Figure 4-40
that the modelling of the contact interface between concrete and steel is critical in
achieving a model that represents the elastic response of the SCS sandwich panel. The
assumption of full composite would yield an elastic modulus that is similar to the
experiment but as the bottom steel plate yields, the full composite action continues to
exhibit the same stiffness as the initial stage, which results in a higher stiffness than
the experimental results. For the model which utilises a penalty contact algorithm
from the start of the response yielded a response which is less stiff than the test result.
Therefore, it could be concluded from this study that a model which can incorporate
the two contact methodologies is necessary. Through the use of the
*CONTACT_TIEBREAK option in LSDYNA to model the natural adhesion between
steel and concrete, a sufficiently accurate prediction of the stiffness evolution as well
as yield stress can be achieved. Ignoring the natural adhesion between steel and
concrete will yield conservative results. The normal and shear failure stress used in
the numerical model is 1Pa.
166
In Figure 4-42, it can be clearly seen through the crack in the concrete that a plastic
hinge is formed at the midspan but the top and bottom steel plates allows the panel to
exhibit further strain hardening prior to failure, which is not modelled in this study.
Comparison of the experimental and numerical response with those from the
analytical method show that the level of conservativeness of this approach.
4.5.2.3 Specimen SCSN
reaction forces against the displacement at the midspan of the specimen. From the
preceding section, contact approach, which models the natural adhesion between the
steel and concrete, was applied, considering the capacity of this technique to achieve
an accurate prediction of the elastic response of steel-concrete composite structures.
Figure 4-44 shows deformed shape of Specimen SCSN from LSDYNA and again the
deflections are scaled by a factor of 5. Figure 4-45 shows the cracking and damage
within the concrete core.
250
200
Reaction Force (kN)
150
Experimental
100 Numerical
Analytical
50
0
0 2 4 6 8 10
Midspan Deflection (mm)
Figure 4-43 Force-displacement plot of Specimen SCSN
167
Figure 4-44 Response of Specimen SCSN at 10mm midspan deflection
(a)
(b)
Figure 4-45 Predicted crack or damage within the concrete core of Specimen SCSN
isolated
As observed in Figure 4-43, the setup of the model produced results which yield an
elastic response which is largely similar to the experiment. The flexural responses as
well as the crack locations are similar to that of the Specimen SCSNE and this is also
evident in the experiment. The same level of conservativeness is observed in the
analytical result.
168
In addition to validation, the numerical model shows the stress concentration and
possibly the internal cracking within the core, which may not be observed in the test.
Figure 4-45(b) shows areas of damaged or cracked concrete within the core.
Compared with SCSNE where the cracks are isolated to the midspan, the model of
SCSN actually shows cracked concrete near the interface between the bottom steel
plate and the connectors. The experimental results show that although Specimens
SCSN and SCSNE share the same ultimate load at 190kN, SCSN is subjected to
softening earlier in the deformation process. This can be attributed to cracking within
the SCSN is more extensive than in SCSNE.
4.5.2.4 Specimen SCSN4
reaction forces against the displacement at the midspan of the specimen. The same
approach that was described in Section 4.5.2.3 was carried out for Specimen SCSN4.
As mentioned in Chapter 4, the only difference is the thickness of the top and bottom
steel plates in the SCSN4 model. Figure 4-47 shows deformed shape of Specimen
SCSN4 from LSDYNA and again the deflections are scaled by a factor of 5. Figure 4-
48 shows the cracking and damage within the concrete core.
169
250
200
Reaction Force (kN)
150
Experiment
100 Numerical
Analytical
50
0
0 2 4 6 8 10
Figure 4-46 Force-displacement plot of Specimen SCSN4
Figure 4-47 Response of Specimen SCSN4 at 10mm midspan deflection
(a)
(b)
Figure 4-48 Predicted crack or damage within the concrete core of Specimen SCSN4
isolated
170
Similar to the SCSN model, Figure 4-46 shows that the SCSN4 model yields a force-
displacement history of the elastic response which is close to the experimental setup
and the analytical result which is conservative. The rest of the observations in Figures
4-47 and 4-48 are also similar to that of SCSN that is described in Section 4.5.2.3.
4.5.2.5 Specimen SCSL
approach that was described in Section 4.5.2.4 was carried out for Specimen SCSL.
As mentioned in Chapter 4, the only difference is the concrete core material in the
SCSL model. Figure 4-50 shows deformed shape of Specimen SCSL from LSDYNA
180
160
140
Reaction Force (kN)
120
100
Experiment
80
Numerical
60
Analytical
40
20
0
0 2 4 6 8 10
Figure 4-49 Force-displacement plot of Specimen SCSL
171
Figure 4-50 Response of Specimen SCSL at 10mm midspan deflection
(a)
(b)
Figure 4-51 Predicted crack or damage within the concrete core of Specimen SCSL
isolated
Similar to the preceding models, the stiffness and yield locations of the model of
Specimen SCSL are close to the experimental results and the analytical results are
conservative. However, comparing the damage of the concrete core of Specimen
SCSL and the sandwich panels with NSC cores, Specimen SCSL exhibits more
extensive damage within the core. This can be observed in Figure 4-51. This probably
explains the experimental phenomenon whereby Specimen SCSL possesses the
lowest hardening capacity among the five steel-concrete composite sandwich panels.
172
4.5.2.6 Specimen SCSH
approach that was described in Section 4.5.2.3 was carried out for Specimen SCSH.
As mentioned in Chapter 4, the only difference is the concrete core material in the
SCSH model. Figure 4-53 shows deformed shape of Specimen SCSH from LSDYNA
250
200
Reaction Forice (kN)
150
Experiment
100 Numerical
Analytical
50
0
0 2 4 6 8 10
Figure 4-52 Force-displacement plot of Specimen SCSH
Figure 4-53 Response of Specimen SCSH at 10mm midspan deflection
173
(a)
(b)
Figure 4-54 Predicted crack or damage within the concrete core of Specimen SCSH
isolated
The stiffness of the Specimen SCSH predicted by the model is close to that of the
experimental results but the plastic yield point predicted by the model exceeds the
experimental output by 5% (see Figure 4-51). The difference could be attributed to the
compressive strength of the HSC used in the specimen. The HSC material model is
based on cylinder tests that were described in Section 4.2.2. The cylinder was steam
cured and high strength cementitous materials such as Densit will perform differently
according to the curing conditions, of which steam curing will achieve higher
compressive strength. Noting that the Specimen SCSH is cured under ambient
conditions, the strength of the HSC in the core of the specimen may be lower than that
of the cylinder. Therefore, using the cylinder properties will yield SCSH models with
higher capacities and this is observed in Figure 4-51. Despite the differences between
174
the numerical and experimental result, the analytical method yields results that is
conservative.
Another observation in the numerical model of the SCSH model is the increase
number of cracks, as observed in Figure 4-54(b). Although such a configuration may
possess superior compressive capacities, under tensile load, HSC tend to be exhibit
higher brittleness with lower energy absorption capacity. This is observed in the
numerical model as there are more areas where crack is predicted as compared to the
models of sandwich panels infilled with NSC.
4.6 SCS Sandwich Panels Design Capacity and Failure Modes under Blast
Loading
Prior to the subjecting the six specimens from Table 4-1 to actual blast loading,
analytical solutions were carried out to predict the response of the specimens. The
single-degree-of-freedom (SDOF) method was used in the analysis and the specimens
are predicted to response in a flexural manner.
The SDOF methodology requires the ultimate moment capacity and flexural stiffness
of the specimens. The ultimate moment capacity is based on the plastic moment
capacity and this parameter together with the flexural stiffness will be used to derive
the dynamic response analytically. In addition to these assumptions, dynamic increase
factors (DIF) will be implemented in for the strength of steel (DIFsteel = 1.36) and
concrete (DIFconcrete = 1.25).
175
The approach taken for the analysis is based on the average acceleration approach
which yields the displacement-time history of the specimen (TM5-1300, 1990) and
the response regime of all the specimens are assumed to be dynamic. Damping is
ignored in the analysis.
4.6.1 Analytical Solution of the Blast Response of SCS Sandwich Panel Specimen
to Blast Loadings
Figure 4-55 illustrates predicted response of the six specimens. Table 4-11 tabulates
the prediction of the peak and permanent deformation of the specimens at their
midspan. The latter is obtained by averaging the maxima and minima of the
oscillations for each specimen.
120
100
Displacement (mm)
80 SP
SCSN4
60
SCSN
40 SCSNE
SCSL
20 SCSH
0
0 0.005 0.01 0.015
Time (msec)
Figure 4-55 Analytical displacement-time histories of the 6 specimens
Table 4-11 Tabulation of the analytical results of the peak and permanent deformation
Specimen Peak Deformation (mm) Permanent Deformation (mm)
SP 100 95
SCSN4 37 33
SCSN 53 50
SCSNE 53 50
SCSL 75 70
SCSH 48 45
Observations from the comparison of the tabulated values are as follow:
176
 Despite possessing the highest yield load capacity, which corresponds to the
highest resistance in the SDOF method, and one of the highest stiffness, the
analysis predicts that SP will tend to deform more than the rest of the
specimens. This can be attributed to the low mass density of the specimen and
indicates the importance of mass in dynamic resistance of structures.
 Specimens SCSN4 has predictably deformed less than SCSN and this showed
the effectiveness of utilising thicker steel face plates.
 Specimen SCSL yielded the highest deformation among the specimens with
concrete cores and this can be attributed to both the low strength as well as the
low mass density.
 Specimen SCSH displayed the least deformation for the sandwich panels with
3mm steel face plates. Compared to SCSN and SCSNE, Specimen SCSH
deformed 8mm less than SCSN and SCSNE although the static compressive
strength of HSC is more than 4 times that of NSC. This is due to the resistance
properties of sandwich panels being much controlled by steel components.
4.7 Experimental Programme on the Blast Resistance of SCS Sandwich
Panels
As described in the preceding sections, the author leveraged on the Explosive Testing
of Structural Components (ETSC) test programme, which is organised by the Defence
Science and Technology Agency (DSTA), to carry out the blast tests of the six
specimens that are described in Table 4-1. The test was carried out on an offshore
island, Pulau Senang, which is an hour from the mainland by ferry. The onsite pre-test
preparation took three days and the test was carried out over a period of two days.
177
4.7.1 Experimental Setup and Instrumentation
This section will attempt to describe the design of the experimental setup. The six
specimens as tabulated in Table 4-1 with the reinforced concrete (RC) support
structure were fabricated in Structural Laboratory at NUS and shipped to Pulau
Senang for the tests. Some of the works that was contracted out includes:
 Onsite logistics and manpower by Semborp Design and Construction Pte Ltd
 Instrumentation by ST Kinetics
 Fabrication of specimens and RC support structure by local contractor
4.7.1.1 Procedure and Design of Experiment
The test programme entailed three repeated blasts with a pair of specimens loaded in
for each test. The pairing of the specimens is as follow:
 Blast 1: SP and SCSN4
 Blast 2: SCSN and SCSNE
 Blast 3: SCSL and SCSH
In Blast 1, the performance of the SCS sandwich panel with NSC core (SCSN4) was
compared to that of the cellular steel panel (SP). These two specimens were designed
such that the stiffness and bending moment capacity were relatively similar. The
influence of J-hook connectors on the blast performance of the SCS sandwich panel
was evaluated in Blast 2 while the last blast was intended to demonstrate the effects of
using different concrete cores.
All three specimens were subjected to the same loading and thus the charge-weight
and standoff distance were maintained though the test series. The details of the
loading are documented in the following section. In each of round, the charge will be
178
positioned such that two specimens were subjected to similar loads and this can be
achieved by aligning the centre of the charge perpendicularly to the front surface of
the support structure (see Figure 4-56).
Reinforced concrete
supporting structure
Specimen
Ground
zero of
blast load
90°
Figure 4-56 Alignment of the charges to the specimens
With the above assumption, the effect of using thicker steel face plates can be
identified by comparing SCSN4 and SCSN. SCSN can also be compared to those in
Blast 3 in order to identify the effects of core materials on the blast response of the
SCS sandwich panels.
The specimens were subjected to loading which is based on a scaled distance of
1.077m/kg1/3 and explosive used for the test are bare TNT charges. Based on
CONWEP, the expected positive phase of the blast overpressure is plotted in Figure
4-57.
179
6000
Overpressure (kPa) / Impulse (kPa-msec)

5000
4000
3000
Overpressure
Impulse
2000
1000
0
0 2 4 6 8 10 12 14
Time (msec)
Figure 4-57 Reflected Overpressure and Impulse based on scaled distance of

1.077m/kg1/3
4.7.1.3 Reinforced Concrete Support Structure
The specimens were subjected to immense forces in the blast test. Thus, the
supporting structure had to fulfil the following requirements:
 Design to eliminate any possibility of sliding and toppling of the structure by
having sufficient mass.
 Possess the capacity to hold up two specimens, which will be positioned
adjacent to each other in the test.
 Capacity to be re-used for three explosions.
Based on these three requirements and the reviews of past blast tests that had been
conducted on similar specimens (Lok et al, 1996; Lan and Heng, 2002; Lan et al,
2005), a RC support structure was proposed and the design was conceived as shown
in Figures 4-58 and 4-59.
180
Roof to minimise pressure
equalisation on specimen
Stiffeners to strengthen
steel bracket
Pressure sensor adapter
Large base to ensure

stability
Steel I-beam for

instrumentation
Removable steel
bracket
Figure 4-58 Isometric view of RC support structure
Cavity to access to
Steel rods to create instrumentation
simply suppoted
boundary condition
Aluminum tube leading

from the pressure
sensors to the cavity
behind specimens
Soil
Figure 4-59 Elevation view of RC support structure with the embedment details
181
The RC support structure was designed with a roof in order to reduce the amount of
pressure acting on the back of the specimen, which will cause an equalising effect
against the pressure from the front and affect the overall response of the specimens.
The structure has a large base that was partially embedded in the soil to ensure
stability as depicted in Figure 4-59. Two specimens can be fitted into the cavity of the
RC support structure in a manner seen in Figure 4-60, with the span of the specimens
in the vertical axis.
Figure 4-60 Actual picture of RC support structure with two specimens
To ensure that the support structure can be re-used for three repeated blasts, 10mm
thick steel brackets were designed to support the specimens. The brackets have three
stiffeners welded onto them to improve the bending capacity. They were secured onto
the top and at the base of the RC support structure by using four threaded bolts which
182
were welded to the reinforcement steel cage within the RC structure. The use of these
bolts enables easy installation and dismantling of the specimens before and after each
blast, respectively.
A steel I-beam for instrumentation was laid horizontally across the cavity of the RC
support structure and was also welded onto the reinforcement steel cage. Details on
the use of this I-beam will be discussed in the subsequent section.
As noted in Figure 4-60, two pressure sensors were installed on the front surface of
the two shear walls of the RC support structure. They were positioned at the mid-span
level of the specimens and are 250 mm away from the interior edge of the shear walls.
In order to ensure that the diaphragm of the pressure sensors were flushed with the
wall surface, adapters fabricated by ST Kinetics Ltd. were installed before the casting
of the RC support structure. Figure 4-59 illustrates that the adapters are connected to
aluminium tubes which leads to the cavity behind the specimens.
Potentiometers, which were used to record the mid-span displacement time histories
of the specimens, were mounted onto the steel I-beam in the RC support structure.
Each potentiometer was secured to the I-beam by using 4 stainless steel threaded bolts
(see Figure 4-61), which were welded to the I-beam, and a 10 mm thick aluminium
plate, which sandwiched the potentiometer between itself and the I-beam by means of
tightening the four nuts against it. The bolts restrained the potentiometer from moving
horizontally while the aluminium plate constrained its vertical movement. As seen in
Figure 4-62, aluminium adapters, which secure the specimen to the shaft of the
183
potentiometers, were adhered on the specimen. The same figure also shows the
accelerometer and its adapter which was welded onto the back face of the specimen.
The strain gauges were mounted on the mid-span and quarter-spans of the specimen
as indicated in the schematic of the instrumentation layout in Figure 4-63.
(a) (b)
Figure 4-61 Assembly to secure potentiometers (a) prior and (b) after the installation
Figure 4-62 Adapters for potentiometers (left and right) and accelerometer (centre)
184
Accelerometer
Potentiometer
Strain gauge
(a) (b)
Figure 4-63 (a) Schematic and (b) Actual positions of instrumentation
4.7.2 Test Results
Blast 1 was performed on the 22nd of August 2008 whereas Blasts 2 and 3 were
conducted on the 26th day of the same month. Preparation prior to the test, which
includes transportation and installation of the specimens and RC support structure as
well as the setting up of instrumentation, took five days.
4.7.2.1 Blast Response and Failure Mode
The test observations on all the six specimens are documented in this section
according to the order of the test. Description of the failure mode, extent of
deformation and the damage to which the specimen are documented.
 Specimen SP: Due to the absence of a core layer, the specimen underwent
severe local buckling and global bending when being subjected to the blast
loading. The deformed shape of the specimen is shown in Figure 4-64 and it
can be seen that the global displacement of the specimen has exceeded the
value predicted prior to the test. This can be attributed to the excessive local
185
buckling on the front and side steel plates which wasn’t taken into account in
the analytical approach. The displacement of the specimen was so excessive
such that it collided with the steel I-beam, which was positioned 175mm from
the back of the panel, as seen in Figure 4-65. The permanent displacement at
the midspan of the specimen was 160 mm. However, this measurement would
be an underestimate since the response was obstructed by the I-beam. The
potentiometers did not yield readings for this test due to a failure to trigger the
data acquisition system and, in addition, the 100 mm potentiometer was
crushed due to the excessive displacement of the specimen. The supporting
bolts for the potentiometers were also sheared off due to the deformation of
the specimen. Figure 4-64(b) shows the overall deformed shape of the
specimen which can be considered as a flexural response.
(a) (b)
Figure 4-64 Deformation of Specimen SP (a) onsite (left) and (b) in the laboratory
Figure 4-65 Deformation of Specimen A was obstructed by the steel I-beam
186
In addition to the extensive global deformation of the specimen, local buckling was
also observed for the top steel face plate and side plates. It can be seen in Figures 4-
66(a) that the top face plate was deformed in a concave manner between the steel
stiffeners. The side plates were being pulled toward the centre of the specimen near
the mid-span. Figure 4-66(b) indicates local buckling of the side plate near the bottom
support of the specimen, which resulted in a reduced bending moment capacity of the
specimen, which have attributed to the underestimation of the maximum deformation
prediction.
(a) (b)
Figure 4-66 Local buckling of the steel plates of Specimen SP
Lastly, there was a fracture crack observed at the side edge of the top steel face place
around the mid-span as seen in Figure 4-67. This could have been the result of
excessive compressive response of the top plate due to the large deformation of the
specimen.
Figure 4-67 Steel fracture on the top steel plate at the midspan of the specimen
187
 Specimen SCSN4: Presuming that this specimen is subjected to the same
loading as Specimen SP, it sustained relatively less damage and the permanent
deformation at the mid-span of the specimen was measured onsite was 27 mm
(see Figure 4-64(a)). Specimen B responded to the blast loading in a flexural
manner. Neither local buckling nor fracture is observed in the response.
 Specimen SCSN: This specimen is positioned on the left in the Figure 4-68(a).
The specimen deformed in a flexural manner similar to that of SCSN4 (see
Figure 4-68(b)). The mid-span permanent deformation of Specimen C was
53mm. The data acquisition system was triggered successfully in this blast and
the recorded data from the potentiometer is plotted in Figure 4-69, which plots
the original and filtered signals. According to the readings from the
potentiometer, the mid-span permanent deformation of Specimen C was
43mm, which is 10 mm less than the onsite measurement. Verification of the
specimen in the laboratory shows that the onsite record is more accurate. The
underestimation by the potentiometer could be attributed to the debonding of
shaft of potentiometer from the adapter. In addition, the large force from the
blast might have misaligned the potentiometer and caused the reading to be
slightly erroneous.
(a) (b)
Figure 4-68 Deformation of Specimen SCSN (a) onsite (left) and (b) in the laboratory
188
100
Actual Recording
Filtered Recording
80
Displacement (mm)
60
40
20
-20
0 20 40 60 80 100 120
Time (msec)
Figure 4-69 Actual and filtered displacement time histories of SCSN
 Specimen SCSNE: As compared to SCSN, this specimen sustained less
damage as in Figure 4-68(a). The specimen deformed in a flexural manner (see
Figure 4-70). The permanent deformation at the midspan of SCSNE measured
31mm onsite which is similar to the data captured by the potentiometer (see
Figure 4-71). However, the displacement time history by the potentiometer
does not seem reflect the response correctly as peak displacement was
sustained over a period of 90msec before rebounding to the final displacement.
The author can only attribute this problem to the potentiometer.
Figure 4-70 Final deformed profile of SCSNE
189
60
Actual Recording
40
Displacement
20(mm)
-20
0 20 40 60 80 100 120 140 160 170
Time (msec)
Figure 4-71 Displacement time history of SCSNE
 Specimen SCSL: Figure 4-72 indicates that this specimen was sheared at the
ends of the specimen. The permanent displacement at the midspan of the
specimen measures 86mm onsite, which is higher than the 75 mm
displacement plotted by the potentiometer as shown in Figure 4-73.
(a) (b)
.
Figure 4-72 Deformation of Specimen SCSL (a) onsite (left) and (b) in the laboratory
190
125
Actual Recording
100
75
Displacement (mm)
50
25
0
0 20 40 60 80 100 120
Time (msec)
Figure 4-73 Displacement time history of SCSNE
The excessive shear response of the specimen resulted in the buckling of the
side steel plates as evident from Figure 4-72(b). This buckling effect
contributed to the failure of the weld between the top and side steel plates as
well as the fracture of the side plate as shown in Figure 4-74(a). In addition,
steel holes were also visible on the rear side of the specimen at the positions of
the J-hook connectors as shown in Figure 4-74(b). This indicates the failure of
the J-hook connectors due to excessive shear forces.
(a) (b)
Figure 4-74 (a) Rupture of steel side plate and (b) the failure of J-hook connectors due
to excessive shear response
191
 Specimen SCSH: This specimen exhibited flexural response under the blast
loading as seen in Figure 4-72(a). The permanent displacement at the midspan
of the specimen measured 79mm onsite. The potentiometer for this specimen
was found to be damaged after test and thus, the data acquired was discarded.
Due to the large deformation of the specimen, the top steel plate sustained
local buckling under large compressive loads which is similar to the response
observed in the quasi-static tests (see Figure 4-75).
(a) (b)
Figure 4-75 Local buckling of the top plate (a) observed onsite and (b) measured in
laboratory
4.7.2.2 Pressure Signal Records
As mentioned in Section 4.7.1.3, pressure histories were recorded by two pressure
sensors that were mounted onto the RC support structure. Pressure records were not
complete. Blast 1 did not yield results as it failed to trigger. In addition, the sensors
were damaged and, due to budgetary constraint which limited the procurement of only
one spare pressure sensor, there was only one pressure history plot from the
subsequent two blasts.
The blast pressure recordings from Blast 2 and 3 are plotted in Figures 4-76 and 4-77
respectively and, incorporated in the plots, are the impulse histories. A small rise in
192
the blast pressure was observed prior to the peak pressure in both graphs, which could
be attributed to noise, and thus, it is ignored. The impulse history is transposed to
reflect this assumption.
15000 4500
13000 3900
11000 3300
Impulse (kPa-msec)
Overpressure (kPa)
9000 2700
7000 Reflected Pressure 2100

Reflected Impulse
5000 1500
3000 900
1000 300
-1000 -300
-3000 -900
0 0.8 1.6 2.4 3.2 4 4.8 5.6 6.4 7.2 8
Time (msec)
Figure 4-76 Reflected pressure and impulse for Blast 2
193
15000 4500
13000 3900
11000 3300
Impulse (kPa-msec)
Overpressure (kPa)
9000 2700
7000 Reflected Pressure 2100

Reflected Impulse
5000 1500
3000 900
1000 300
-1000 -300
-3000 -900
0 0.6 1.2 1.8 2.4 3 3.6 4.2 4.8 5.4 6
Time (sec)
Figure 4-77 Reflected pressure and impulse for Blast 3
4.7.2.3 Acceleration Signal Records
As mentioned earlier in Section 4.7.1.4, each specimen had an accelerometer attached
to its mid-span. However, none of the data collected were useable for the following
reasons:
 Failure to trigger data acquisition system (e.g. Blast 1)
 Data collected exceeded the threshold limit setup on the data logger
 Data collected exceeded the threshold limit of the accelerometer
Therefore, improvements in the data acquisition techniques are very much needed for
future tests of such close-in explosions.
4.7.2.4 Strain Gauge Records
No data was recorded for Specimens SP and SCSN4. The strain gauge readings for
Specimens SCSN, SCSNE, SCSL and SCSH are plotted in Figures 4-78, 4-79, 4-80
194
and 4-81, respectively. It is observed that the strain time histories were not as smooth
as those recorded from quasi-static loading (see Figure 4-25). The blue line in the
figures indicates the strain data at the mid-span of the specimen while the readings
from the strain gauges that were positioned at the quarterspan are plotted in red and
green. The strain-time histories of the four specimens are compared in Figure 4-82. In
view that some of the displacement time histories obtained using the potentiometers
were erroneous or incomplete, the records from the strain gauges can be used to
derive the displacement-time histories through certain assumptions.
10000
Quarterspan Top
9000 Midspan
Quarterspan Bottom
8000
7000
6000
5000
Strain ()
4000
3000
2000
1000
-1000
-2000
0 30 60 90 120 150 180 210 240 270 300
Time (msec)
Figure 4-78 Strain recordings for SCSN
195
10000
Quarterspan Top
9000 Midspan
Quarterspan Bottom
8000
7000
6000
5000
Strain ()
4000
3000
2000
1000
-1000
-2000
0 30 60 90 120 150 180 210 240 270 300
Time (msec)
Figure 4-79 Strain recordings for Specimen SCSNE

10000
Quarterspan Top
9000 Midspan
Quarterspan Bottom
8000
7000
6000
5000
Strain ()
4000
3000
2000
1000
-1000
-2000
0 30 60 90 120 150 180 210 240 270 300
Time (msec)
Figure 4-80 Strain recordings for Specimen SCSL
196
10000
9000
8000
7000
6000
5000
Strain ()
4000
3000
Quarterspan Top
2000 Midspan
Quarterspan Bottom
1000
-1000
-2000
0 30 60 90 120 150 180 210 240 270 300
Time (msec)
Figure 4-81 Strain recordings for Specimen SCSH

10000
9000 Specimen C (SCSN)

Specimen D (SCSNE)
8000
Specimen E (SCSL)
Specimen F (SCSH)
7000
6000
5000
Strain ()
4000
3000
2000
1000
-1000
-2000
0 30 60 90 120 150 180 210 240 270 300
Time (msec)
Figure 4-82 Comparison of strain time histories at mid-span of Specimens SCSN,

SCSNE, SCSL and SCSH
197
The following are the observations and conclusions that are drawn from the above-
mentioned figures:
a) The strain at the midspan of Specimen SCSNE and SCSH exceeded the limit
of the strain gauges due to the underestimation of the panels’ displacements.
b) With the exception of Specimen SCSNE, it was observed that the recorded
strain time histories at the two quarter-span positions for the other concrete
infilled sandwich specimens are not identical. It is interesting to note that
Specimen SCSNE was the only specimen without the J-hook connectors and
the absence of connectors may have allowed the steel to yield at the midspan
instead of distributing it across the span of the steel face plates. It is observed
that the residual strain on the quarter spans of specimen SCSNE is almost at
zero while the residual strains at the similar locations of the other SCS panels
have yielded and are responding plastically.
c) The strain-time histories at the quarterspans of Specimen SCSNE decayed to
zero after some oscillations, which indicated that the material did not yield at
those locations.
d) Specimen SCSL exhibited the lowest peak strain among the four specimens
despite having the largest permanent deformation at the mid-span. This is a
clear indication of the shear response of the specimen whereby the mid-section
of the specimen between the two areas of shear yielding did not undergo
significant bending.
e) From Figure 4-80, it can be seen that Specimen SCSL exhibited the largest
drop in the midspan as strain readings decay from 8500μ to 1900 μ, which is a
78 % drop as compared to 27 % and approximately 50 % drops for Specimens
SCSN and SCSNE respectively.
198
4.7.3 Discussion and Deductions from Experimental and Analytical Results
The recorded blast overpressures and impulses are compared to CONWEP predictions
in Figures 4-83 at a scaled distance of 1.077m/kg1/3. It can be seen in Figure 4-83(a)
that the recorded peak overpressure and the decay of both Blast 2 and Blast 3 are
almost identical. The peak impulses for both blasts are also similar with Blast 3
having a slightly steeper rise time than that of Blast 2. This supports the assumption of
subjecting all specimens to the same loading.
In comparison with the blast pressure curves predicted using CONWEP, the impulse
variations from the actual test seem to fit the CONWEP plots well, particularly for
Blast 2. This signifies the reliability of CONWEP in predicting the impulse for close-
in explosions and that the CONWEP impulse curves are not conservative. On the
contrary, there seem to be a considerable underestimation of the actual peak
overpressure by CONWEP. While the CONWEP plot peaked at 5600kPa, the actual
readings from the pressure sensors documented 12700kPa and 13200kPa peak
overpressures for Blast 2 and Blast 3, respectively. It may be argued that these peaks
may be the result of noise but still a higher peak overpressure will be obtained after
filtering the data.
199
14000
12000
10000
Overpressure (kPa) 8000

Blast 2
Blast 3
6000 CONWEP
4000
2000
-2000
0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4
Time (msec)
(a)
4000
3600
3200
2800
Impulse (kPa-msec)
2400
Blast 2
2000 Blast 3
CONWEP
1600
1200
800
400
0
0 0.6 1.2 1.8 2.4 3 3.6 4.2 4.8 5.4 6
Time (msec)
(b)
Figure 4-83 Comparison of reflected (a) overpressure and (b) impulse
Table 4-12 compiles the mid-span permanent displacements of the six specimens
tested.
200
Table 4-12 Comparison of permanent deformation of specimens
Experimental Permanent Analytical Permanent
Blast No Specification
Deformation (mm) Deformation (mm)
SP >200 95
1
SCSN4 28 33
SCSN 55 50
2
SCSNE 34 50
SCSL 85 70
3
SCSH 81 45
By comparing the permanent deformations, extent of damage and the modes of failure
of the specimens, the following conclusions on the blast resistance of the specimens
could be drawn. Below are the observations and deductions which are based on the
experimental result:
a) The concrete core provided the SCS sandwich composite specimens with
added mass and rigidity which minimises the local buckling steel face plates
upon yielding. This is evident from the comparison of the deformed shapes of
Specimens SP and SCSN4. Although they have relatively similar static
stiffness and bending moment capacity, Specimen SP exhibited much more
damage as compared to Specimen SCSN4. The ability to maintain the integrity
and shape of the specimen is thus, attributed to the concrete core layer. This
shows that the concrete core is more effective in enhancing the residual
capacity of the specimen as well as reducing the post yield buckling effects of
the steel plates as compared to the steel plate stiffeners.
b) The performance of J-hook connectors in maintaining the integrity of SCS
sandwich specimens, which responded in a flexural manner, was found to be
satisfactory as little or no separation was observed between the steel face
plates and the core. Specimen SCSL underwent shear response and was an
exception whereby several J-hook connectors debonded from the steel face
plates at the areas of shear failure. However, it must be noted that the side
201
plates of all the sandwich specimens appeared to play a significant role in
holding the two steel face plates together throughout the blast response. This
was evident from the deformed shape of Specimen SCSNE which had no
connectors and yet little or no separation was observed between the steel face
plates of this specimen. Since the specimens were relatively small in size, the
influence of the connectors may not be as significant as in the case where
larger specimens are used. The J-hook connectors are expected to be effective
in maintaining the composite action between the steel plates and concrete core
in areas that are further away from the side plates.
c) From the comparison of Specimens SCSN4 and SCSN, it was found that the
thickness of the steel face plates influenced the permanent deformation
significantly. By increasing the thickness of the steel face plates by 1 mm, the
permanent deformation of the SCS sandwich specimen was halved.
d) The damage incurred by the SCS sandwich composite specimen with J-hook
connectors (Specimen SCSN) was observed to be more severe than the
specimen without connectors (Specimen SCSNE). More tests is recommended
to be carried out to investigate this issue as it is contradictory to both
analytical and numerical predictions, which will be discussed in the following
section and chapter respectively. One possible explanation is that there were
more points of discontinuity within the concrete with J-hook connectors, and
hence, that would have created more locations of stress concentration during
fracture.
e) SCSL produced the highest midspan displacement and deformed pre-
dominantly in shear. This response could be attributed to two factors: the
weight of the concrete core and the relatively lower shear strength of the
202
specimen. However, it was found that SP also underwent flexural response
despite possessing lowest mass density compared to the rest of the specimens.
Hence, it may be deduced that the shear response of Specimen SCSL was due
to its lower shear capacity of the concrete core. Since such shear response will
result in the buckling of the steel plates which will result in the reducing the
residual strength the panel, the use of LWC is thus, not recommended for blast
resistant structures.
f) The use of HSC did not decrease the permanent deformation as compared with
concrete of lower strength that was used in the test. Thus, the HSC material
may have an adverse effect on the SCS sandwich specimen and, hence, it may
not be suitable in the design of blast resistant structures. This could be
attributed to the minimal softening response of the HSC material which is
documented in Section 4.2.2. Consequently, the HSC may be more easily
shattered as the shock wave propagates through the core. This process prior to
the global bending response of the specimen might have rendered the concrete
from developing its full ultimate strength. Similar to the influence of the J-
hook connectors, further studies is required to understand the dynamic
response of the HSC.
g) Comparing the quarterspan strain readings of SCSNE in Figures 4-79 with the
records from the SCSN specimen, it may be deduced that the stress in the SCS
sandwich specimen without connectors is more evenly distributed as compared
to the specimens with connectors. This may be due to J-hook connectors
which introduced more stress concentration points on the steel panel and that
lead to local tensile yielding of the steel, which would explain the non-
uniformity of the strain at the quarterspans.
203
Comparing the results and deductions from the experiments with those from the
analytical approach, there are certain similarities and some discrepancies. The
following points out the author’s observation and deductions from this comparison:
 It is shown in both analytical and experimental studies through the response of
Specimen SP, which exhibited the largest extent of deformation among all six
specimens, that the use of concrete cores to resist blast loading as an effective
method to reduce the level of damage to the structural element.
 Both analytical and experimental results support that the implementation of
thicker steel face plates will results in less damage and deformation.
 The analytical approach is not able to capture the response of the Specimen SP
due to the excessive local buckling and failure of the welds that were observed
in the specimen.
 The analytical solutions are conservative for Specimens SCSN4 and SCSNE.
This is probably expected due to the conservative properties that were adopted
from the quasi-static analysis.
 The analytical solutions are not conservative for SCSN was not conservative
but the difference between the experimental and numerical result is only 10%.
 Although the analytical solutions of Specimens SCSN and SCSNE are
identical but the experimental results differ with SCSN yielding a larger
deformation than SCSNE. This can be attributed to a scenario whereby the
load on the two specimens are not similar due to the irregular shape of the
explosives used. Another possibility could be the impulse that was transferred
to the specimen by various fragments that impacted the loaded surface of the
SCS sandwich panels. It was observed that fragments penetrated through the
loaded steel face plate and Specimen SCSN showed significantly more points
204
of penetration than Specimen SCSNE. Therefore, being loaded more, the
specimen will deflect more.
 SCSL responded differently than predicted analytically. The failure mode was
different where the experiment exhibited shear failure that opposed to the
flexural response assumed in the analytical solution. One possibility could be
attributed to the low shear resistance of the concrete core. In addition, it is
observed that weld failure was observed in the specimen which caused more
deformation than expected.
 Although SCSH responded in a flexural manner as assumed in the analytical
method, the deformation observed in the experiment is much more. This again
can be attributed to the non-uniform loading on the panel.
4.8 Numerical Study on the Blast Resistance of SCS Sandwich Panels
The last section of this chapter documents the attempts in using the FE solver, LS-
DYNA, to perform a numerical study on the specimens (see Table 4-1) that were
subjected to blast pressure loadings. The comparison between the specimens in the
experimental has shown that specimens with shear connectors deform more than those
without. Another observation is that HSC infill in SCS sandwich panel did not
improve the dynamic performance as opposed to the improvement observed in the
static tests. Such structural behaviour will be further studied through numerical
modelling. In addition, the results from the numerical model will further reinforce the
reasons, which were proposed in Section 4.7.3, to explain the discrepancies between
the experimental and analytical result. In addition, the numerical study will be further
extended to indicate possible approaches to develop an effective blast resistant SCS
sandwich panel.
205
4.8.1 FE Solution Scheme
The time integration scheme adopted in the simulation is based on the explicit
scheme. As mentioned in Section 5.2.1, the explicit scheme is an excellent choice in
studying the transient structural response and, due to the short response time, this
scheme provides an excellent platform to perform explicit time integration with small
time steps to ensure stability and convergence to the actual structural response.
4.8.1.1 Geometry and Design of Test Setup and Specimens
The geometries of the specimens are similar to the ones described in 4.5.1.1. The
difference is in the boundary conditions to which the models of the specimens were
subjected. In the quasi-static load setup model, load and support blocks are modelled
to impose incremental loading and create a simply support boundary conditions
respectively. In order to recreate the boundary conditions of the test setup, the model
of the specimen included three additional parts: the steel bracket plates, the steel roller
rods and the RC support structure.
The full geometry of the specimen is modelled together with the part of the RC
support structure with appropriate boundary conditions. Figure 4-84 shows the
configuration of the model that was used in the numerical study. Each specimen was
modelled individually as opposed to the experimental setup where specimens were
tested in pairs.
206
Steel Bracket Plate RC Support Structure
with Stiffeners
SCS Specimen
Steel Roller
Rods to ensure
Simple Support
at boundaries
Direction of Blast
Loading Pressure
Figure 4-84 Numerical model to study blast response of SCS specimens
All the nodes of the RC support structure part were restricted in all axes of translation
to provide a rigid body assumption in the analysis. The same boundary conditions
were applied to the steel roller rods that were sandwiched between the specimen and
the RC support structure. The nodes were shared between the steel bracket plates and
the RC support structure at their contact interfaces. Nodes were also shared between
the steel bracket plates and the steel roller rods, that were sandwiched between the
steel bracket plates and specimen, at the contact interfaces (see Figure 4-85). An
additional constraint that defined a rigid body based on the specified nodes
(*CONSTRAINED_NODAL_RIGID_BODY) was applied to the nodes of the steel
bracket plates and steel roller rods in the vicinity of their contact interfaces. This
technique was applied to limit the rotation of the steel roller rods. The nodes specified
in the set to form the rigid body are shown in Figure 4-85(a) with the centre of the
rigid body as seen in Figure 4-85(b).
207
(a) (b)
Figure 4-85 Illustration of (a) the nodes defined to form the rigid body and (b) the
centre of the rigid body in constraints applied to the steel bracket plate and steel roller
rods
The elements that were used for the specimens are similar to the specifications that
are described in Section 4.5.1.2. The 10mm solid constant stress elements were used
to model the RC support structure and the steel roller rods and thick shell elements
were applied to the steel bracket plates.
The material models used for the specimens are similar to the specifications that are
described in Section 4.5.1.3. The material model that was used for the RC structure
was similar to the ones used in the NSC model since the material used in the casting
of concrete core of the SCS panels and the RC structure originate from the same batch
of concrete. The material model used for the steel bracket plates and the steel roller
rods are similar to the ones in the steel plates of the specimens.
4.8.1.4 Steel-Concrete Interface Modelling
There are two additional contact interfaces that were introduced in this model as
opposed to the quasi-static load test model were the interaction between the steel face
plates of the specimen and the steel roller rod and the interaction between the steel
208
face plates of the specimens with RC support structure, as the specimens were at rest
on the RC support structure in the initial condition.
Although, in Section 4.5, it is deduced that the importance of modelling the natural
adhesion between steel and concrete is critical in producing a model that is validated
in the elastic range, there is a high probability that the steel and concrete will debond
prior to the global response of the specimen. Therefore, the contact interface
modelling between steel and concrete would be restricted to friction only, which is
similar to the penalty contact algorithm that was described in Section 3.5.1.4.
As documented in Section 4.7.3, the pressure histories recorded in the experiment
matched the impulse data from the empirical prediction from CONWEP but the peak
pressure from the experiment differs. Therefore, in the numerical models, three
pressure loads will be applied on the specimens to access the response. One will be
based on the empirical pressure history at a scaled distance of 1.077m/kg1/3, which is
the loading that was used in the experimental setup and the pressure and impulse are
plotted in green in Figure 4-86. The second load history, plotted in orange in Figure 4-
86, will be based on 0.8402 m/kg1/3, which yields the same peak pressure as the
experimental pressure readings. The third pressure will be the application of the actual
experimental load history on the specimen. The pressure and impulse of the last case
is plotted in violet. Figure 4-86 illustrates the three pressure and impulse histories
with the blue lines representing the actual blast pressure readings.
209
16000 9000
14000 8000
7000
12000
6000
10000
5000
Impulse (kPa-msec)
Pressure (kPa)
8000
4000
6000
3000
4000
2000
2000
1000
0 0
0 0.5 1 1.5 2
-2000 -1000
Time (msec)
Figure 4-86 Comparison of the three blast pressure and impulse loads applied to the
models
It is to be noted the first two pressures will be applied through the use of the
*LOAD_BLAST option in LS-DYNA as it applies the blast pressure on elements,
which is based on based on TNT explosive quantity and the standoff distance of these
surfaces of these elements to ground zero (Randers-Pehrson and Bannister, 1997).
The data, from which the pressure-time histories are derived, are based broadly on
CONWEP. Thus, this will result in slightly different pressure histories on various
locations on the specimen. The third pressure differs from the former two as the
forces are applied through a user-defined pressure-time history and thus is uniform on
the surface of the specimen.
210
These six specimens are simulated and the results from the models will be compared
with those that are derived through the experiment and analytical approaches.
4.8.2.1 Specimen SCSNE
Based on the three loadings that were described in the preceding section, numerical
models of Specimens SCSNE was analysed using LSDYNA. The residual midspan
displacement of the specimen was compared with that of the experimental results.
Figure 4-87 compares the models with SCSNE with the experimental results. No
comparison was done for the displacement-time history of the panel because the
potentiometers failed to capture the data during the test.
60
50
40
Numerical (1.077m/kg-3)
30
Numerical (Recorded)
Experiment
20 Analytical
10
0
0 0.005 0.01 0.015 0.02
Time (sec)
Figure 4-87 Midspan displacement histories of numerical models of Specimen

SCSNE
211
As observed in Figure 4-87, the residual displacement of the panel measured after test
correlates well with FE prediction using a uniform distributed pressure based on the
recorded pressure-time history. The FE model which assumed a scaled distance of
1.077kg/m3 and applied the pressure using the *LOAD_BLAST option, yielded
results which are slightly lower than the test results but the difference is acceptable.
Comparison with the analytical result show that the conservativeness of the method
and this can be attributed to the lower plastic capacity predicted using the SDOF
method (see Figure 4-29(b)).
From Figure 4-88, it can be observed that the recorded and simulated permanent
deformed shape of the specimen are similar with a global flexural response with a
slight localised shear yielding at one end of the specimen. To further validate the
model, a mesh convergence study was conducted and the mesh size was refined to a
nominal size of 5 mm. The FE results produced by the 5 mm mesh model was found
to be similar to that of the 10 mm mesh model. Therefore, it has been shown that the
FE model used in this study for SCSNE is able to produce reliable predictions of the
response of steel-concrete composite sandwich panels when subjected to blast
loading. In addition, this work also validates the numerical models described in
Section 3.5 as similar material and contact models amongst many others were used.
212
(a) (b)
Figure 4-88 Comparison of deformed shapes from (a) explosive test and (b) FE
analysis
4.8.2.2 Specimen SCSN
Using the similar approach to SCSNE, Figure 4-89 illustrates the comparison of the
numerical and experimental results of Specimen SCSN. Unlike the result from
Specimen SCSNE, it is shown that the model that best fits the test data in SCSNE was
not applicable in the case of Specimen SCSN. This is different from the best fit model
of Specimen SCSNE. In fact, the numerical results are correlates well with the
analytical results which predicted that the midspan deflection would be similar for
Specimen SCSN and SCSNE. However, this does not validate the analytical
approach. As described in the preceding section, the analytical method would yield
results which are conservative. Therefore, factor that could have caused this
difference would be explored.
213
60
Midspan Displacement (mm) 50
40
30
Experiment
20 Analytical
10
0
0 0.005 0.01 0.015 0.02
Time (sec)
Figure 4-89 Midspan displacement histories of numerical models of Specimen SCSN
One of the possibilities that could have attributed to the difference is the model of the
connectors as that is the sole difference between the SCSN and SCSNE models. In
view of this, another two models of SCSN was created: one with the removal of the
spring element between the top and bottom connectors and one with the top and
bottom connectors merged. The difference in the midspan deflection is described in
Figure 4-90 and it can be seen that the modelling of the spring does not affect the
peak displacement. However, by maintaining the integrity of the connectors, the
residual deflection could be decreased although the peak deflection remains
unchanged. Therefore, it can be deduced that the modelling of the connector is still
valid.
214
45
40
35
30
25
Spring
20 No Spring
Merged
15
10
0
0 0.005 0.01 0.015 0.02 0.025
Time (sec)
Figure 4-90 Comparison of various connection assumptions
As described in Section 4.7.3, there may be a chance whereby the load on SCSN may
be different from the load on the SCSNE specimen. Therefore, with the assumption
that the scaled distance is 0.8402m/kg3, the LOAD_BLAST option is selected and
applied to the SCSN model. The comparison between the analytical, experimental and
numerical results is shown in Figure 4-91. It is observed that the results from the
numerical model are close to that from the experiment. Therefore, this could have
explained the larger deflection observed in SCSN as compared to the SCSNE
specimen.
215
70
60
50
40
30
Experiment
20
10
0
0 0.005 0.01 0.015 0.02
Time (sec)
Figure 4-91 Comparison of various connection assumptions
4.8.2.3 Specimen SCSN4
Figure 4-92 illustrates the differences between numerical models of SCSN4 with
results from the experiment and analytical SDOF method. The similar phenomenon is
observed for SCSN4 as of SCSN where the model being loaded under scaled distance
at 0.8402m/kg1/3 is closer to the experiment. This discrepancy between the numerical
and experimental results could potentially be attributed to the difference in loading.
Notwithstanding the result from the experiment, it is again observed that the
analytical solution is more conservative than the numerical result.
216
50
45
40
35
30
25 Numerical (Recorded)
20
Experiment
15 Analytical
10
0
0 0.005 0.01 0.015 0.02
Time (sec)
Figure 4-92 Midspan displacement histories of numerical models of Specimen SCSN4
4.8.2.4 Specimen SP
Figure 4-93 illustrates the differences between numerical models of SCSN4 with
results from the analytical solution. It is observed from this comparison that the
analytical solution is more conservative than the numerical model as the analytical
plastic capacity is shown to be lower than that the experimental prediction. However,
specimen which was subjected to blast deformed more than 200mm, which is more
than two times of the analytical and numerical prediction. The difference is attributed
to the fracture of the welding between the face and side steel plates. The numerical
model does not include the fracture of the welding and, thus, it has a higher capacity
than that of the experimental specimen.
217
120
80
60
Numerical (1.077mkg-3)
Analytical
40
20
0
0 0.005 0.01 0.015 0.02
Time (sec)
Figure 4-93 Midspan displacement histories of numerical models of Specimen SP
4.8.2.5 Specimen SCSL
Figure 4-94 illustrates the differences between numerical models of SCSL with results
from the experimental and analytical solution. Different from the rest of the
comparisons between the analytical and numerical results, the peak deflection
predicted from the numerical model is larger than that from the analytical method
although the residual deflection from the analytical model is more conservative. This
can be attributed to the fact that the structure responded predominantly in shear rather
than flexural, in which the SDOF method assumed. Figure 4-95 illustrates the profile
of the model after loading, which indicates shear failure. This same phenomenon is
observed in the experimental specimen. The difference in deflection predicted is
possibly caused by the weld failure as seen in Figure 4-74(a). This fracture would
have lowered the shear capacity at the ends of the specimen and caused the specimen
218
to deflect more. Similar to the model of Specimen SP, the failure of the weld is not
modelled in and thus the numerical model would possess a higher capacity than the
failed experimental specimen.
90
80
70
60
50
40 Analytical
Experimental
30
20
10
0
0 0.005 0.01 0.015 0.02
Time (sec)
Figure 4-94 Midspan displacement histories of numerical models of Specimen SCSL
Figure 4-95 Deflection profile of Specimen SCSL with fringe levels of deflection
219
4.8.2.6 Specimen SCSH
Figure 4-96 illustrates the differences between numerical models of SCSH with
results from the experimental and analytical solution. Comparison of the analytical
and numerical results indicates that the analytical method is conservative but the
difference in the peak deflection between the two approach are less than what was
observed in the SCS panels which were infilled with NSC. Comparison between the
numerical and experimental results has shown a great discrepancy. This can be
attributed to the loading which could have been higher on the specimen than the
specified load in the numerical model.
90
80
70
60
50
40 Experimental
Analytical
30
20
10
0
0 0.005 0.01 0.015 0.02
Time (sec)
Figure 4-96 Midspan displacement histories of numerical models of Specimen SCSH
4.8.3 Further Discussion on the Dynamic Design of SCS Sandwich Panels
This final section of the numerical study will attempt to adopt the validated numerical
models from the previous section to identify factors in optimising the SCS sandwich
220
panel to resist blast loading. Design parameters such as the shear connectors, the
thickness of the steel face plates as well as the properties of the concrete core will be
studied to provide a platform to improve and optimise a blast resistant SCS sandwich
panel. The loading that is assumed in this section is similar to the recorded pressure-
time history from the experimental setup described in Section 4.7 (see Figure 4-86).
50
45
40
35
30
25
SCSNE
20 SCSN
15
10
0
0 0.005 0.01 0.015 0.02 0.025
Time (sec)
Figure 4-97 Comparison of the response of Specimen SCSN and SCENE
Figure 4-97 illustrates the difference in response of the numerical models of Specimen
SCSN and SCSNE. As expected, the shear connectors do improve the resistance of
the sandwich panel as a lower peak deflection is observed. However, it can be
observed that the residual deflection is almost similar. This can be attributed to the
side steel plates, which act as additional stiffeners to these SCS sandwich panels. It
can be deduced that the contribution of these steel stiffeners overrides the significance
and effectiveness of the shear connectors. Hence, it may be concluded that, in the
design of a blast resistant SCS sandwich panels, it may be sufficient to provide steel
stiffeners in place of shear connectors, which are more difficult to install.
221
45
40
30
25
20 SCSN
SCSN4
15
10
0
0 0.005 0.01 0.015 0.02 0.025
Time (sec)
Figure 4-98 Comparison of the response of Specimen SCSN and SCEN4
Figure 4-98 demonstrated the effectiveness of increasing the thickness of SCS
sandwich panels as the peak and residual deflection that were observed from the two
specimens, which differ only by the thickness of their steel face plates by 1mm.
90
80
70
60
50
NSC
40
LWC
30 HSC
20
10
0
0 0.005 0.01 0.015 0.02
Time (sec)
infilled with NSC, LWC and HSC
222
Figure 4-99 compares the response of three specimens which were infilled with
different concrete content. It is clear from the comparison that the use of LWC is not
an effective solution as compared to the specimens infilled with NSC. As explained in
the previous section, it can be attributed to the difference in the failure mode of the
specimen. Comparing the SCS sandwich panels which were infilled with NSC and
HSC has shown that, despite the superior static load carrying capacity of using HSC
in the panel, the dynamic resistance of the panel is comparable with that of the
sandwich panel infilled with NSC. Therefore, the use of HSC in SCS sandwich panel
may not yield a significant improvement in the blast resistance.
In view that the numerical model of the Specimen SCSL exhibited the highest
deflection, it would be interesting to use the same model to investigate the reason
behind this response. In dynamic design, the response of a structure is a function of
both the mass as well as the stiffness. In this study, in addition to the numerical model
of the Specimen SCSL, a model similar to the Specimen SCSL was created. The
difference in this new model is that the concrete core would possess the constitutive
model of NSC but have the density of LWC. The comparison of their response is
plotted in Figure 4-100.
223
90
80
60
50 NSC
40 LWC
30 NSC resistance, LWC

density
20
10
0
0 0.005 0.01 0.015 0.02
Time (se)
infilled with NSC, LWC and a concrete core with the strength of NSC but the density
of LWC
It is observed that the deflection the SCS sandwich panel with the new concrete core
is significantly reduced as compared to the response of Specimen SCSL. The peak
and residual deflection is almost reduced by half. Comparing this response with that
of Specimen SCSN, the difference between the two is less than 20%. This
demonstrates that the dynamic response of the SCS sandwich panel is controlled more
by the strength of the concrete core than by the density.
Table 4-13 Comparison of permanent deformation of specimens

Numerical Permanent Analytical Permanent
Blast No Specification
Deformation (mm) Deformation (mm)
SP 69 95
1
SCSN4 12 33
SCSN 33 50
2
SCSNE 30 50
SCSL 60 70
3
SCSH 28 45
Table 4-13 tabulates the permanent deformation predicted by the analytical and
numerical models. In general, it can be observed that:
224
 The analytical approach is conservative as all the results from the numerical
model are lower. This, again, can be attributed to the lower plastic capacity
predicted through the analytical design assumptions.
 The effectiveness of the increasing the thickness of the steel face plates are
demonstrated in both the numerical and analytical solutions although the
significance is more evident in the numerical model. This can be attributed to
the contribution of the strain hardening of the steel face plates, which was
taken into consideration in the numerical model but not in the analytical
method.
 Both the analytical and numerical method also showed that the effectiveness
of the shear connectors was overshadowed by the steel side plates as both
analytical and numerical predictions of SCSN and SCSNE were similar.
 Although the analytical results of Specimen SCSL were more conservative,
the difference is less significant as compared to the rest of the SCS sandwich
panels. This can be attributed to inability of the SDOF method to predict the
shear response, which was evident in the numerical model.
 The contribution of the strength of HSC to the response of SCS sandwich
panel is demonstrated in both analytical and numerical models whereby the
difference. Both analytical and numerical models show slight improvement
over the SCS sandwich panel infilled with NSC. Thus, showing the
effectiveness of the SCS sandwich panel is very much controlled by the top
and bottom steel face plates.
225
4.9 Summary
This chapter described the analytical, experimental and numerical study that was
carried out in the attempt to study the effects of the type of concrete, the steel plate
thickness and the J-hook connectors on the sandwich panels which are subjected to
both quasi-static and dynamic blast loading. Comparing the differences in the static
and dynamic response of these panels has also brought to light the differences in
design approach in both static and dynamic design. In order to achieve this, the study
was phased as follow:
 Analytical study of quasi-static properties of SCS sandwich panels
 Experimental study of quasi-static properties of SCS sandwich panels through
the use of a three point quasi-static load test
 Numerical modelling of the quasi-static response of SCS sandwich panels
 Analytical study of dynamic properties and response of SCS sandwich panels
using the SDOF method
 Experimental study of dynamic properties of steel-concrete composite
sandwich panels through a explosive test at a scaled distance of 1.077m/kg1/3
 Numerical modelling of the dynamic response of SCS sandwich panels to
understand the approach to obtain an efficient blast resistant SCS sandwich
panel
The comparison between the quasi-static and dynamic result gave insight to the
different response of the same material under different loading regimes. Firstly, the
use of concrete core and increments in the thickness of the steel face plates improved
the both static and dynamic performance of the sandwich panel specimens. It was also
demonstrated in both the quasi-static and dynamic response of the SCS sandwich
panel that the J-hook connectors were not significant in improving the resistance of
226
the structure and it is due to the steel side plates which were incorporated into the
specimens. The use of LWC in SCS sandwich panel is not recommended in both
static and dynamic loading conditions. The former will result in relatively low post
yield ductility and the latter has shown vulnerability to shear failure due to the low
shear capacity of the core material. Although yielding excellent static capacity, the
use of HSC in SCS sandwich panels will not yield the same benefits in dynamic blast
design as the specimen did not exhibit higher blast resistant capacity as compared to
the panel which were infilled with NSC.
Quasi-static properties, which were derived analytical, are conservative regardless of
the properties of the concrete, thickness of the steel plate and the presence of the J-
hook connectors. Therefore, it is recommended that if the concrete core of the SCS
panels is confined within the vicinity of the steel face plates, it can be assumed that
the panel posses full composite action in static design. In the use of the properties
derived from quasi-static analytical and experimental methods, the dynamic properties
of SCS sandwich panels were conservative when compared with numerical
predictions.
In the aspect of numerical modelling effort, it was shown that one of the critical
components of modelling the SCS sandwich panel accurately is the natural adhesion
between steel and concrete. As the concrete is cured within the confines of the steel
box, it is inevitable that there would be adhesion between steel and concrete and it
was demonstrated in the modelling of the specimens under quasi-static loading.
227
In extending the models which were calibrated from the tests, it demonstrated that the
modelling of the J-hook connectors has little effect on the peak displacement but its
influence can be seen in the residual displacement. Therefore, if the peak
displacement is the required data to determine the design, the J-hook can be simplified
to accelerate the computational time. The SCS sandwich panel which was infilled
with LWC yield results which do not show favour in its implementation. However, it
was illustrated that the performance of the concrete core is dependent on the strength
more than the mass. Therefore, the use of LWC core in SCS sandwich panels is
recommended, provided the concrete has sufficient strength properties.
228
CHAPTER5:
CONCLUSION
The study of this thesis was initiated with an in depth investigation and validation of
the SDOF approach to analyse the response of steel-concrete composite structures.
The strain rate effect on concrete was validated using SHPB experimental setup. The
results from the test has shown significant correlation to the experimental results from
various researches and are more conservative than the commonly used CEB strain rate
plots. In order to validate the SDOF method to predict the response of steel composite
columns, the setup of a drop hammer and airbag was implemented. The assumptions
in SDOF, as its name describes, is limited to structures with one predominant failure
mode. In view of this, the Rigid-Plastic method is proposed for one-way structural
members. This method has the capacity to predict theoretically the mode of failure.
In the study of the dynamic response concrete-filled steel tubular (CFST) column
subjected to various blast loadings, the Rigid-Plastic method was applied and the
results were compared with those from the SDOF method. The comparison was
further extended with a series of numerical models of the CFST column and the result
reinforced the validity of the Rigid-Plastic method. It was shown that the Rigid-
Plastic method is applicable for the member and is capable of predicting the response
satisfactorily. The Rigid-Plastic method offers a more efficient and easier approach
than the SDOF approach through the use of closed form equations to gauge the
response shape as well as peak displacement of CFST columns. There is no
requirement to assume a certain failure mode, which is necessary in the case of SDOF
application. In addition, it was highlighted that due to the design of critical columns,
229
they are under high risk of failing in shear. Therefore, the Rigid Plastic method is a
superior approach in the analysis of CFST columns.
Further numerical studies of the CFST column demonstrated that the composite
column is much more resilient against blast loading as compared to the RC columns
in terms of improved ductility, enhanced shear resistance and improved residual load-
carrying capacity.
In addition to the CFST column, the study also included an investigation to assess the
design of steel-concrete-steel (SCS) sandwich panels against blast loading. The study
was carried out in an attempt to study the effects of the type of concrete, the steel
plate thickness and the use of J-hook connectors on SCS sandwich panels which are
subjected to blast loading. Included in this study were analytical, experimental and
numerical studies of the both the dynamic and quasi-static properties of these
sandwich panels. The comparison between quasi-static and dynamic studies gave
insight to the different responses of the same material under different loading regimes.
In the quasi-static study, the experiments demonstrated that the use of concrete core
improved the ductility of the sandwich panel specimens as compared to a stiffened
panel of a similar load carrying capacity and stiffness. The same test series also
showed the effectiveness of increasing the thickness of the steel face plates in
improving the loading carrying capacity and ductility of the SCS sandwich panels.
Compared with SCS sandwich panels, which were infilled with normal strength
concrete (NSC), the use of high strength concrete (HSC) core improved the load
carrying capacity as well as the ductility of the panel. Comparing the experimental
230
results with proposed analytical solutions, the analytical method was shown to be
conservative in terms of the load carrying capacity of the SCS sandwich panels.
However, the stiffness matches the experiment results very well. Finite element (FE)
analysis of SCS specimens under quasi-static loading showed that one of the critical
components of modelling the elastic phase of a steel-concrete composite panel
accurately is the natural adhesion between steel and concrete. With consideration of
this modelling component, the numerical results match those of the experiments.
Similar to the static design of the SCS sandwich panels, the experimental
investigation demonstrated the effectiveness of using concrete core as well as thicker
steel face plates to improve the dynamic performance of the panel. It was also
observed that the use of light weight concrete (LWC) in SCS sandwich panels may
lead to shear failure due to the low shear capacity of the core material. In addition, the
use of HSC in SCS sandwich panels does not necessarily yield a higher blast
resistance. This observation is a stark difference from the improvement observed in
the static response. The FE analysis, in which the models are calibrated on the
experimental programme, deduced that the use of J-hook connectors do not reduce the
peak deflection not can improve the residual deflection slightly. Similar to the
experimental study, improvement is observed in the response of the numerical models
when the thickness of the steel face plates is increased. Comparing the numerical
models of the SCS sandwich panels with and without connectors, the response were
similar, which indicated that the dynamic response of SCS sandwich panels were
predominantly controlled with the side steel plates, which confines the concrete core.
The numerical study also indicated that the low blast resistance of the SCS sandwich
panel, which was filled with LWC, can be attributed mainly to the low strength of the
231
concrete material and not the density. Lastly, the FE study indicates that the use of
HSC in SCS sandwich panels produced similar responses as specimens which were
infilled with NSC. With the experimental and numerical result, comparisons with the
analytical SDOF predictions indicate that the designs are generally conservative and
thus appropriate to the used.
Summarising the scientific and engineering contribution from this study of SCS
sandwich panels, it was shown that the use of concrete core is essential to yield an
efficient SCS sandwich in both static and dynamic loading scenarios. Under static
loading, the load carrying capacity and the ductility is a function of the concrete core
strength and, if the concrete core is confined with steel side plates, full composite
action can be assumed in the static and dynamic design. Under blast loading, the
contribution of the shear connectors to the overall dynamic performance to reduce the
peak displacement is not significant if side steel plates are installed to confine the
concrete core of the SCS sandwich panels. In terms of the design of the concrete core,
the strength of the concrete core is significant to the resistance of the SCS panels but
there is probably a limit to the improvement obtained through increasing the concrete
strength. It is observed that this limit could be in the range of the strength of NSC.
Overall, the present study has achieved the objectives of this thesis. The first specific
objective was to develop an analytical method to accurately capture the dynamic
inelastic behaviour of the CFST columns when subjected to blast loading. This was
achieved through the proposal of the Rigid-Plastic method, of which the accuracy and
efficiency were validated through analytical, numerical and experimental studies. The
second objective to study the blast resistant of SCS sandwich panels was also
232
achieved through the conduct of a series of quasi-static and blast tests. Numerical
models which were based on the experiments were also setup to understand the
response and failure mechanisms of these structures in order to optimise the design
approach and finally it was shown that the analytical method was conservative and
thus appropriate to be used in the design of these structures.
The numerical, analytical, experimental study also highlighted some questions. It was
shown in the blast test that the use HSC in SCS sandwich panels caused the specimen
to deform more than specimens constructed with NSC. In view of the complexity of
explosive tests and limited resources of the present study, this topic is recommended
for future studies.
In conclusion, this study has explored a wide range of engineering methodology to
assess the performance of steel-concrete composite structural members against blast
loading. The study has enhanced certain assumptions and methods and also
ascertained the accuracy and the level of confidence of both current and novel
methods to predict structural response. In addition, the journey also uncovered some
irregularities which were not expected and it is recommended to conduct further study
in these areas.
233
BIBLIOGRAPHY
Ananth, R.; Ladouceur, H.D.; Willauer, H.D.; Farley, J.P.; Williams, F.W., Effect of
water mist on a confined blast, in: Presented before the Suppression and Detection,
Research and Applications- A Technical Working Conference (SUPDET 2008)
(March 11–13, 2008), Orlando, Florida, 2008
ASTM C29/29M-97 Standard Test Method for Bulk Density and Voids in Aggregate,
2003
ASTM C39/C39M-05 Standard Test Method for Compressive Strength of Cylindrical

Concrete Specimens, 2005
ASTM C127-04 Standard Test Method for Density, Relative Density (Specific
Gravity), and Absorption of Coarse Aggregate, 2004
ASTM E8M-04 Standard Test Methods for Tension Testing of Metallic Materials,
2004
Baker, W.E.; Kulesz, J.J.; Westine, P.S.; Cox, P.A.; Wilbeck, J.S., A manual for the
prediction of blast and fragment loadings on structures, San Antonio, Texas, South
West Research Institute, 1980
Baker, W.E.; Cox, P.A.; Westine, P.S.; Kulesz, J.J.; Strehlow, R.A., Explosions
Hazards and Evaluation, Elsevier Science Publishers B.V., Amsterdam, The
Netherlands, 1983
Barker, D.D.; Whitney, M.G.; Waclawczyk Jr, J.H.; Failure and Design Limit
Criteria for Blast Loaded Structures, Process Safety Progress, Volume 12, No. 4,
Pages 216-221, 1993
Baylot, J.T.; Bevins, T.L., Effect of responding and failing structural components on
the airblast pressures and loads on and inside of the structure, Computers and
Structures, Volume 85, Issue 11-14, Pages 891-910, 2007
Beshara, F.B.A., Modelling of blast loading on aboveground structures – I. General

Phenomenology and external blast, Computers and Structures, Volume 51, No. 5,
Pages 585-596, 1994
Biggs, J.M., Introduction to Structural Dynamics, McGraw-Hill Book Company,

1964
Bischoff, P.H., Perry, S.H., Compression behavior of concrete at high strain-rates,

Mater. Struct. 24, Pages 425–450, 1991
BS EN 1991-1-7:2006, Eurocode 1. Actions on structures. General actions.

Accidental actions, 2006
234
BS 812-110: 1990, Testing Aggregates – Part 110: Methods for determination of
aggregate crushing value (ACV), 1990
BS 812-112: 1990, Testing Aggregates – Part 112: Methods for determination of

aggregate impact value (AIV), 1990
Eurocode 4, Eurocode 4: Design of composite steel and concrete structures- Part 1-1
General - Common rules and rules for buildings, CEN, Brussels, 1994
Boh, J.W.; Louca L.A;Choo Y.S., Energy absorbing passive impact barrier for
profiled blastwalls,
International Journal of Impact Engineering 31, Pages 976–995, 2005
Bowerman, H.; Chapman, J.C., Bi-SteelTM Steel-Concrete-Steel Sandwich

Construction, Proceedings of the Fourth International Conference on Composite
Construction in Steel and Concrete Banff, Alberta, Canada, 2000
Bowerman, H.; Coyle, N.; Chapman, J.C., An innovative steel/concrete construction

system, The Structural Engineer 80(20), Pages 33–38, 2002
Bulson, P., Explosive loading of engineering structures, E&FN Spon, 1997
Century Dynamics. AUTODYN 2-D & 3-D User’s Manual, 2003
Comite Euro-International du Beton - Federation Internationale de la Precontrainte,

CEB-FIP model Code 90, Redwood Books, Trowbridge, Wiltshire, Great Britain,
1990
Chapman T.C., Rose T.A.; Smith P.D., Reflected Blast Wave Resultants Behind
Cantilever Walls - A New Prediction Technique, Int J Impact engineering, Volume
16, No. 3, Pages 397-403, 2005
Chen, H.; Liew, J.Y.R., Explosion and fire analysis of steel frames using mixed
element approach, Journal of Engineering Mechanics, Volume 131, No.6, Pages 606-
616, 2005
Corley, W.G.; Mlakar, P.F.; Sozen, M.A.; Thornton, C.H, The Oklahoma City
Bombing: Summary and Recommendations fo Multihazard Mitigation, J. Perform.
Constr. Fac. ASCE 12(3), Pages 100-112, 1998
CONWEP, Conventional Weapons Effects, US Army TM-855, 1992
Corus Group Ltd, Bi-Steel Design & Construction Guide, Volume 1-2, 1999.
Crawford, J.E;, Lan, S, Blast Barrier Design and Testing, Proceedings of the ASCE
Structures Congress, St Louis, MO, 2006
Carwford, J.E.; Malvar, L.J.; Morrill, K.B.; Ferritto, J.M., Composite retrofits to
increase the blast resistance of RC buildings, 10th international symposium on
interaction of the effects of munitions with structures, 2001
235
Crowder, B, Progressive Collapse – Historical Perspective, Naval Facilities
Engineering Command, 2005
Cullen, L, The public enquiry into the Piper Alpha Disaster, UK: HMSO, 1990
Davies E.D.H.; Hunter, S.C., The dynamic compression testing of solids by the
method of the split Hopkinson pressure bar, J Mech Phys solids, Vol 11, Pages 155-
179, 1963
Dioh, N.N.; Leevers, P.S.; Williams, J.G., Thickness effects in split Hopkinson
pressure bar tests, Polymer, Volume 34, No. 20, Pages 4230-4234, 1993
EM 1110-345-417, Design of Structures to Resist the Effects of Atomic Weapons
Flanagan R.D.; Benett, R.M., Uniform lateral load capacity of infilled frames, ASCE
Structures Congress, Atlanta, GA, 1993
Fish, J.; Belytschko, T, A., First Course in Finite Element, John Wiley & Sons, Ltd,
2007
Foundoukos, N.; Xie, M.; Chapman, J.C., Fatigue tests on steel-concrete-steel

sandwich components and beams, Journal of Constructional Steel Research, Volume
63, Issue 7, Pages 922-940, 2007
Forrestal, M.J.; Wright, T.W.; Chen, W., The effect of radial inertia on brittle
sameples during the split Hopkinson pressure bar test, Int J of Impacy Engineering,
Volume 34, Pages 405-411, 2007
Gary, G.; Bailley, P., Behaviour of quasi brittle material at high strain rate.
Experiment and modelling, Eur J Mech A/Solids, Volume 17, No. 3, Pages 403-420,
1998
Godinho, J., Analysis of steel columns for air-blast loads, Structure Magazine, Pages
13-14, 2007
Graves, M.A., A New Generation of Expeditionary Earth-filled Protective Barrier

(EEPB)
Griffiths, Hugh; Pugsley, A. G.; Saunders, Owen, Report of the Inquiry into the
Collapse of Flats at Ronan Point, Canning Town, Her Majesty’s Stationery Office,
London, 1968
Grote, D.L., Park, S.W., Zhou, M., Dynamic behavior of concrete at high strain-rates
and pressures: I. Experimental characterization, Int. J. Impact Engng, Volume 25,
Pages 869–886, 2001
Hallquist, J.O., LS-DYNA Theory Manual, Livermore Software Technology

Company, Livermore CA, 2006
236
Heng, L.; Kennedy, L.J.; Mays, G.C., Blast resistance of fully enclosed steel–
concrete–steel sandwich panels, Proceedings of 7th international symposium on
interaction of the effect of munitions with structures, Mannheim, Germany, Pages
189–97, 1995
Hopkinson, B., British ordnance board minutes 13565(1915).
Houlston, R., Nonlinear structural response of ship panels subjected to air blast
loading, Computers and structures, Volume 26, No ½, Pages 1-5, 1987
Houlston, R., Slater, J.E., Pegg.; DesRochers, G.C., On analysis of structural

response of ship panels subjected to air blast loading, Computers and Structures,
Volume 21, No. ½, Pages 273-289, 1985
Ishikawa, N.; Beppu, M., Lessons from past explosive tests on protective structures in
Japan, International Journal of Impact Engineering, Volume 34, Pages 1535–1545,
2007
Jan, S.F.; Gurbuz, O.; Dynamic Nonlinear Finite Element Analysis of Blast Resistant
Concrete Buildings in Petrochemical Facilities, Proceedings of the 2008 Structures
Congress, Vancouver, BC, Canada, Pages 1-8, 2008
Jones, J., Investigation of Dynamic Resistance of Fixed-End, One-way Reinforced

Concrete Panels,Faculty of Engineering, University of Wollongong, 2007
Jones, N., Structural Impact, Cambridge University Press. 1989
Kang, K.W., Design of Blast Resistant Steel Structures, B.Eng thesis, National
University of Singapore, 2005
Kang. K.W.; Liew, J.Y.R., Analytical Study of Blast Response of Steel-Concrete

Composite Structures, Proceedings of the 2006 International Colloquium on Stability
and Ductility of Steel Structures, Lisbon, Portugal, 2006
Kang, K.W.; Lee S.C.; Liew, J.Y.R., Analysis of Steel-Concrete-Steel Composite

Sandwich Panel Subjected to Impulsive Blast Loading, 5th International Conference
on Advances in Steel Structures, Singapore, 2007
Karagiozova, D., Nurick, G.N., Langdon, G.S., Chung Kim Yuen, S., Chi, Y., Bartle,
S., Response of flexible sandwich-type panels to blast loading, Composites Science
and Technology, 2008
Kolsky, H., An Investigation of the Mechanical Properties of Materials at VeryHigh

Rates of Strain, Proc. Roy. Phys. Soc., Volume 62, Pages 676-700, 1949
Krauthammer, T., Blast Effects and Related Threats. Technical Paper, Protective
Technology Center, Penn State University, 1999
Krauthammer, T., Blast-resistant structural concrete and steel connections, Int J of

impact eng, Volume 22, Pages 887-910, 1999
237
Krauthammer, T., Bazeos, N., Holmquist, T. J., Modified SDOF Analysis of R. C.
Box-
Type Structures, Journal of Structural Engineering, Volume 112, No. 4, Pages 726-
744, 1986
Lan S.; Heng, L., Composite structural components under blast loading, Structures
under shock and impact, Pages 13-21, 2002
Lan S.; Lok, T.S.; Heng, L., Composite structural panels subjected to explosive
loading, Construction and Building Materials, Volume 19, Issue 5, Pages 387-395,
2005
Langdon, G.S.; Yuen, C.K.; Nurick, G.N., Experimental and numerical studies on the
response of quadrangular stiffened plates, Part II – Localised blast loading, Int. J.
Impact Eng, Volume 31, No. 1, Pages 85-111, 2005
Langdon G.S.; Schleyer, G.K., Inelastic deformation and failure of profiled stainless
steel blast wall panels: Part I, experimental investigations, Int. J. Impact Eng,
Volume 31, No. 4, Pages 341-369, 2005
Langdon G.S.; Schleyer, G.K., Deformation and failure of profiled stainless steel
blast wall panels. Part III: FE simulation and overall summary, Int J of impact
engineering, Volume 32, Pages 988-1012, 2006
Lee, K.; Kim, T.; Kim, J., Local response of W-shaped steel columns under blast
loading, Structural engineering and mechanics, Volume 31, No 1, Pages 25-38, 2009
Leigh, E.; Earls, D., Simulating blast effects in steel lattice structures, Proceedings of
the 6th international conference on computational shell and spatial structures, 2008
Leppänen, J., Experiments and numerical analyses of blast and fragment impacts on
concrete, Int J Impact Eng, Volume 31, Pages 43–86, 2005
Li, Q.M.; Meng, H., About the dynamic strength enhancement of concrete-like
materials in a split Hopkinson pressure bar test, International Journal of Solids and
Structures. Volume 40, Pages 343–360, 2005
Liang, Q.Q.; Uy, B.; Bradford, M. A.; Ronagh, H. R., Closure to “Strength Analysis
of Steel-Concrete Composite Beams in Combined Bending and Shear, Journal of
Structural Engineering, Volume 131, Pages 1593-1600, 2005
Liang, Y.H.; Louca, L.A.; Hobbs, R.E., Corrugated panels under dynamic loads, Int
Journal of Impact Engineering, Volume 34, Pages 1185-1201, 2007
Liang, Y.; Spuskanyuk, A.V.; Flores, S.E.; Hayhurst, D.R.; Hutchinson, J.W.;
McMeeking, R.M.; Evans, A.G., The Response of Metallic Sandwich Panels to Water
Blast, Journal of Applied Mechanics, Volume 74, Pages 81-99, 2007
238
Liang, C.; Yang, M.; Wu, P., Optimum design of metallic correugated core sandwich
panels subjected to blast loads, Ocean engineering, Volume 28, Pages 825-861, 2001
Liew J.Y.R., Buildable design of multi-storey and large span steel structures. Journal
of Steel Structures, Korean Society of Steel Structures; 4:2, Pages 53-70, 2004
Lok, T.S.; Xiao J.R.; Lan S.R.; Heng, L., Response of SFR concrete panels to air-
blast loading, Proceedings of the Asia-Pacific Conference on Shock & Impact Loads
on Structures, Singapore, 1996
Longinow, A.; Alfawakhiri, F., Blast Resistant design with structural steel, Modern
Steel Construction Magazine, 2003
Louca,L.A.; Boh,J.W., Analysis and design of profiled blast walls, Health and Safety
Executive Report, RR146, Pages 1 – 63, 2004
Louca, L.A., Punjani, M.; Harding, J.E., Non-linear analysis of blast walls and
stiffeened panels subjected to hydrocarbon explosions, J Construct steel res, Volume
37, No 2, Pages 113, 1996
Low, H.Y., Hao, H., Reliability Analysis of Reinforced Concrete Slabs under
Explosive Loading, Structural Safety, Volume 23, Pages 157-178, 2001
Mackes, K.H.; Lofersk J.R.; White, M.S., A Pneumatic Pressure Bag Testing
Machine for Applying a Uniform Load to Panels and Pallets, Journal of testing and
evaluation, Volume 23, No. 4, Pages 295-299, 1995
Malachowski, J., Effect of blast wave on chosen structure - numerical and

experimental study , International journal of mathematics and computers in
simulation, Issue 3, Volume 2, Pages 228-235, 2008
Malvar, L.J.; Crawford, J.E., Dynamic increase factor for concrete, Twenty-Eighth
DDESB Seminar Orlando, FL, 98
Malvar, L.J., Crawford, J.E., Wesevich, J.W. and Simons, D., A plasticity concrete
material model for DYNA3D. Int J Impact Eng, Volume 19. Pages 847-873, 1997
McCann, D.M.; Smith, S.J., Blast resistant design of reinforced concrete structures,
Structure Magazine, 2007
McKinley, B., Boswell, L.F., Behaviour of double skin composite construction,

Journal of constructional Steel Research, Volume 58, No. 10, Pages 1347-1359, 2002
Morision, C. M., Dynamic response of walls and slabs by single-degree-of-freedom

analysis-a critical review and revision, International journal of Impact Engineering,
Volume 32, Pages 1214-1247, 2006
239
Morrill, K.B.; Crawford, J.E.; Magallanes, J.M.; Choi, H.J., Development of
simplified tools to predict blast response of steel beam-column connections,
Karagozian and Case, Burbank CA, P-06-10, 2006
PDC-TR 06-08, SDOF Structural Response Limits for Antiterrorism Design
Ngo, T.; Mendis, P.A.; Teo, D.; Kusuma, G., Behaviour of high strength columns
subjected to blast loading, Advances in Structures, Proceedings of the International
Conference on Advances in Structures (ASSCCA '03), Sydney, Australia, Pages
1057-1064. 2003
Ngo, T.; Mendis, P.A.; Gupta, A.; Ramsay, J., Blast Loading and Blast Effects on
Structures – An Overview, EJSE Special Issue: Loading on Structures, 2007
Ngo, T.; Nguyen, N.; Mendis, P., An investigation on the effectiveness of blast walls
and blast-structure interaction, Proceedings of ACMSM-18 conference, 2004
Ottosen, N.S.; Ristinmaa, M., The Mechanics of Constitutive Modeling, Elsevier

Science, Amsterdam, The Netherlands; 2005
Pan, F.X.; Zhu, J.S.; Helminen, A.O.; Vatanparast, R., Three Point Bending Analysis
of a Mobile Phone Using LS-DYNA Explicit Integration Method, 9th International LS-
DYNA Users Conference, 2006
Pryer, J.W.; Bowerman, H.G., The development and use of british steel bi-steel, J
Construct. Steel Res, Volume 46, No. 1-3, Pages 15, 1998
Randers-Pehrson, G.; Bannister, K.A., Airbast Loading Model for DYNA2D and
DYNA3D, Army Research Laboratory, Rept ARL TR 1310, 1997
Ranković, S.; Drenić, D., Static strength of the shear connectors in steel-concrete
composite beams: regulations and research analysis, Facta Universitatis Series:
Architecture and Civil Engineering Volume 2, No. 4, Pages 251 – 259, 2002
Remennikov, A; Rose, T.A., Predicting the effectiveness of blast wall barriers using
neural networks, International Journal of Impact Engineering, Volume 34, Pages
1907–1923, 2007
Riedel, W.; Mayrhofer, C.; Thoma, K.; Stolz, A., Engineering and Numerical Tools
for Explosion Protection of Reinforced Concrete, International Journal of Protective
Structures, Volume 1, No. 1, 2010
Rickman, D.D.; Murrell, D.W., Development of an improved methodology for

predicting airblast pressure relief on a directly loaded wall, Trans. ASME J. Pressure
Vessel Technol., Volume 129, Pages 195–204, 2007
Rogers, J.; Koper, D.; Keith, D., Some Practical Applications of Forensic Seismology,
Coursework reading
240
Rose, T.A., A Computational Tool for Airblast Calculations: Air3D version 9 users’
guide, Engineering Systems Department, Cranfield University, 2006
Rose, T.A., An approach to the evaluation of blast loads on finite and semi-finite
structure, PhD thesis, Engineering System Department, Cranfield University, 2001
Rose, T.A.; Smith, P.D.; McLennan, T.C.P., Blast loading and clearing on tall
buildings, Journal of battlefield technology, Volume 7, No. 3, Pages 1440-5113, 2004
Ross, A., Tedesco, J.W., Kuennen, S.T., Effects of strain rate on concrete strength.
ACI Mater. J., Volume 92, No. 1, Pages 37–47, 1995
Sanji, A.J.; May, I.M.; High-Mass, low-velocity impacts on reinforced concrete slabs,
7th European LSDYNA conference, 2009
Schenker, A; Anteby, I; Gal E;, Kivity, Y; Nizri, E; Sadot, O; Michaelis, R; Levintant

O; Ben-Dor, G, Full-scale field tests of concrete slabs subjected to blast loads,
International Journal of Impact Engineering, Volume 35, Issue 3, Pages 184-198,
2008
Schleyer, G.K.; Lowakb, M.J.; Polcynb, M.A.; Langdonc, G.S., Experimental

investigation of blast wall panels under shock pressure loading, International Journal
of Impact Engineering, Volume 34, Pages 1095–1118, 2007
Schwer, D.A.; Kailasanath, K., Numerical simulations of the mitigation of unconfined

explosions using water-mist, Proceedings of the Combustion Institute, Volume 31,
Pages 2361–2369, 2007
Silva, P.F.; Lu, B., Improving the blast resistance capacity of RC slabs with
innovative composite materials, Composites Part B: Engineering, Volume 38(5-6),
Pages 523-534, 2007
Slater, J.E., Selection of a blast-resistant GRP Composite panel design for naval ship
structures, Marine structures, Volume 7, Pages 417-440, 1994
Smith, P. D.; Mays, G. C.; Rose, T. A.; Teo K. G.; Roberts, B.J., Small scale models
of complex geometry for blast overpressure assessment, Int. J. Impact Engng. Volume
12, No. 3, Pages 345-360, 1992
Sohel, K.M.A.; Liew, J.Y,R.; Steel-Concrete-Steel sandwich slabs with lightweight

core -- Static performance, Engineering Structures, Volume 33, Issue 3, Pages 981-
992, 2011
Steel Construction Institue, Improved simplified response methods to blast loading.

Prepared for the Health and Safety Executive, 2006
Stuhmeyer, A., The Thick Shell Element for Metalforming and other Applications, 5th
European LSDYNA Users Conference, 2005
241
TM5-1300, Structures to Resist the Effects of Accidental Explosions, U.S.
Departments of the Army, Navy and Air Force, Technical Manual, Army TM5-1300,
Navy NAV P-397, Air Force AFR 88-54,
UFC 4-010-01, DoD Minimum Antiterrorism Standaeds for Buildings, Department of
Defenece, 2012
U.S. Department of Commerce, National Technical Information Service, Washington

D.C. 1990
Toro E.F.; Riemann solvers and numerical methods for fluid dynamics, Berlin:
Springer; 1997
Turkmen H.S., Structural response of laminated composite shells subjected to blast

loading: Comparison of experimental and theoretical methods, Journal of Sound and
Vibration Volume 249, Pages 663-678
Wada, Y.; Liou, M.S., An accurate and robust flux splitting scheme for shock and
contact discontinuities, J Sci Comput, Volume 18, No. 3, Pages 633–57, 1997
Wang, F.; Wan, Y.K.M.; Chong O.Y.K.; Lim C.H.; Lim, E.T.Z., Reinforced Slab
Subjected to Close-in Explosion, LSDYNA 7th German Forum, 2008
Weaver, W; Johnston, P.R., Structural dynamics by finite elements, Prentice Hall, NJ,
1987
Xie, M.; Foundoukos, N.; Chapman, J.C., Static tests on steel–concrete–steel

sandwich beams, Journal of Constructional Steel Research, Volume 63, Issue 6,
Pages 735-750, 2007
Xie, M.; Chapman, J.C., Developments in sandwich construction, Journal of

Constructional Steel Research, Volume 62, Issue 11, Pages 1123-1133, 2006
Xu, K., Lu, Y., Numerical simulation study of spallation in RC plates subjected to
blast loading, Computers and Structures, Volume 84, Pages 431-438, 2006
Yuen, S. C. K.; Nurick, G. N., Experimental and numerical studies on the response of
quadrangular stiffened plates. Part I: Subjected to uniform blast load, Intl. J. Impact
Engineering Volume 31, No. 1, Pages 55-83, 2005
Yuen, S. C. K.; Nurick, G. N.; Verster, W.; Jacob, N.; Vara, A.R.; Balden, V.H.;
Bwalya, D.; Govender, R.A.; Pittermann, M., Deformation of mild steel plates
subjected to large-scale explosions, International Journal of Impact, 2008
Youngdahl C.K., Correlation Parameters for Eliminating the Effect of Pulse Shape on
Dynamic Plastic Deformation, ASME Journal of Applied Mechanics, Volume 37,
Pages 744-752. 1970
Youngdahl C.K. Influence of Pulse Shape on the Final Plastic Deformation of a

Circular Plate, International Journal of Solids and Structures, Vol. 7, Pages 1127-
1142, 1971
242
Yu, C.; Huang, Z.; Burgess, I.W.; Plank, R., 3D modelling of bi-steel structures
subject to fire, Fourth International Workshop on Structures in Fire, Pages 393-404,
2006
Zhou, X.Q.; Hao, H., Predicting of airblast loads on structures behind a protective
barrier, International Journal of Impact Engineering, Volume 35, Pages 363–375,
2008
Zhou, X.Q.; Hao H, Numerical prediction of RC concrete exterior wall response to

blast loading, Advances in Structural Engineering, Volume 11, No. 4, Pages 355-367,
2008
Ziegler, J.J., From Beirut to Khobar Towers: Improving the Combating Terrorsim
Program, Air Command and Staff College, Air University, Maxwell Air Force Base,
Alabama, 1998
Zienkiewicz, O.C.; Taylor, R.L., The finite element method, Volume 1: The Basics,
Butterworth-Heinemann, 2000
Zyskowski, A.; Sochet, I.; Mavrot, G.; Bailly, P.; Renard, J., Study of the explosion
process in a small scale experiment--structural loading, Journal of Loss Prevention in
the Process Industries, Volume 17, Issue 4, Pages 291-299, 2004
243
APPENDIX A
Time Phase Deflection Shape Governing Equations Boundary Conditions Time (s) Velocity (m/s), Deflection (m)
Figure A.1 Response of a simply-supported member with η ≤ 3
A-1
where,
where, where,
=0
A-2
Figure A.2 Response of a simply-supported member with η > 3
A-3
Figure A.3 Response of a fixed-end member with η ≤ 3
A-4
where,
where, =0
where,
A-5
Figure A.4 Response of a fixed end member with η > 3
A-6
APPENDIX B
Figure B.1 Response of a simply-supported member with υ ≤ 1
B-1
Time Phase Deflection Shape Governing Equations Boundary Time (s) Velocity (m/s), Deflection (m)
Conditions
where,
Figure B.2 Response of a simply-supported member with 1 < υ ≤ 3/2
B-2
Conditions
where,
where,
where, where,
B-3
Figure B.3 Response of a simply-supported member with υ > 3/2
B-4
Figure B.4 Response of a fixed-end member with υ ≤ 2
B-5
Conditions
where,
Figure B.5 Response of a fixed-end member with 2 < υ ≤ 3
B-6
Conditions
where,
where,
where, where,
B-7
Figure B.6 Response of a fixed-end member with υ > 3
B-8
APPENDIX C
where,
C-1
Figure C.1 Response of a simply-supported member with η ≤ 3
C-2
Conditions
where,
where,
C-3
where,
where, where,
Figure C.2 Response of a simply-supported member with η > 3
C-4
where,
C-5
Figure C.3 Response of a fixed-end member with η ≤ 3
C-6
Conditions
where,
where,
C-7
where,
where, where,
Figure C.4 Response of a fixed-end member with η > 3
C-8

Blast Resistance of Steel-Concrete Composite

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Blast Resistance of Steel-Concrete Composite

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Blast Resistance of Steel-Concrete Composite

Uploaded by

Copyright:

Available Formats

BLAST RESISTANCE OF STEEL-CONCRETE COMPOSITE

KANG KOK WEI

B.Eng (Hons.), NUS

FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

DEPARTMENT OF CIVIL & ENVIRONMENTAL

NATIONAL UNIVERSITY OF SINGAPORE

used in the thesis.

Kang Kok Wei

29th October 2012

one additional member in our family.

At the university, I would like to express my heartfelt gratitude to my supervisor,

Professor Richard Liew. It is an honour to be under his supervision and I do

together to pursue the success of one another’s test on that island.

few pages bounded in this book.

Steel-concrete composite structural design is becoming common and more prominent

of terrorism with their access to destructive technologies through asymmetric warfare.

governmental buildings. Therefore it is now important for civil engineers to

generated from such an environment.

capability to analyse a structure, which respond both in flexural as well as in shear.

sandwich panels through analytical, experimental and numerical study.

In the design of structural members against blast loading, the Single-Degree-of-

Freedom (SDOF) method is commonly used to approximate the dynamic response of

performance of a composite column as compared to that of a reinforced concrete one

of SCS sandwich panel of various configurations under quasi-static and dynamic

loadings. The quasi-static experiment series utilised a three-point laboratory load

firing range. The differences in response of six configurations of sandwich composite

extended to demonstrate the effectiveness of incorporating steel plates between the

strength is designed in the concrete.

Figure 2-1 Effect of the gas explosion at Ronan Point, UK .......................................... 9

One of these concerns is the use of explosives against commercial or governmental

infrastructures. Without the considerations of extreme loadings in the design of

of hundreds and thousands of occupants. Therefore it is now important for civil

engineers to understand dynamic design and incorporate passive protective measures

Steel-concrete composite structural members are commonly used in modern

to perform prefabrication and the reduction in the need of formworks. Composite

materials in design. In view of the significant performance of steel-concrete

composite structural members over conventional steel or reinforced concrete

performance of these columns against blast loading.

concrete composite structural components which are subjected to blast loading

• Develop an analytical method to accurately capture the dynamic inelastic

behaviour of concrete filled steel tubular (CFST) subject to blast loading

• Study the blast resistant performance of steel-concrete-steel (SCS) sandwich

panels through analytical, experiments and numerical simulations

paragraphs of this chapter:

 Review the state-of-the-art in analytical, numerical and experimental works in

deriving the structural response to blast loading

 Conduct of quasi-static and dynamic tests of construction materials against

 Validation of application of the Single Degree of Freedom (SDOF) method

 Comparison of SDOF method with a proposed analytical method for a CFST

columns which is subjected to various loadings

 Comparison of analytical results of the CFST columns with proposed

 Conduct of quasi-static and dynamic tests on SCS sandwich panels and

 Perform numerical simulation of the quasi-static and dynamic response of the

analytically and experimentally

CFST column will be reviewed through comparisons with experimental and

in order to assess the competency of this proposed method.

An experimental study to investigate the response of SCS sandwich panel of various