Sharaf Tarek A 201008 PHD PDF

FLEXURAL BEHAVIOUR OF SANDWICH PANELS
COMPOSED OF POLYURETHANE CORE AND
GFRP SKINS AND RIBS
By
Tarek Abdel Moneim Sharaf
A thesis submitted to the Department of Civil Engineering

in conformity with the requirements for the degree of
Doctor of Philosophy
Queens University
Kingston, Ontario, Canada
August, 2010
Copyright Tarek A. Sharaf, 2010

Abstract
ABSTRACT
This study addresses the flexural performance of sandwich panels composed of a
polyurethane foam core and glass fibre-reinforced polymer (GFRP) skins. Panels with
and without GFRP ribs connecting the skins have been studied. While the motivation of
the study was to develop new insulated cladding panels for buildings, most of the work
and findings are also applicable to other potential applications such as flooring, roofing
and light-weight decking. The study comprises experimental, numerical, and analytical
investigations.
The experimental program included three phases. Phase I is a comprehensive
material testing program of the polyurethane core and GFRP skins and ribs. In Phase II,
six medium size (2500x660x78 mm) panels with different rib configurations were tested
in one-way bending. It was shown that flexural strength and stiffness have increased by
50 to 150%, depending on the rib configuration, compared to a panel without ribs. In
Phase III, two large-scale (9150x2440x78 mm) panels, representing a cladding system
envisioned to be used in the field, were tested under a realistic air pressure and discrete
loads, respectively. The deflection under service wind load did not exceed span/360,
while the ultimate pressure was about 2.6 times the maximum factored wind pressure in
Canada.
A numerical study using finite element analysis (FEA) was carried out. The FEA
model accounted for the significant material nonlinearities, especially for the
polyurethane soft core, and the geometric nonlinearity, which is mainly a reduction in
thickness due to core softness. Another independent analytical model was developed
based on equilibrium and strain compatibility, accounting for the core excessive shear
i
Abstract
deformation. The model also captures the localized deformations of the loaded skin,
using the principals of beam-on-elastic foundation. Both models were successfully
validated using experimental results. Possible failure modes, namely core shear failure,
and compression skin crushing or wrinkling were successfully predicted.
A parametric study was carried out to explore further the core density, skin
thickness, and rib spacing effects. As the core density increased, flexural strength and
stiffness increased and shear deformations reduced. Also, increasing skin thickness
became more effective as the core density increased. The optimal density was 95-130
kg/m3. Reducing the spacing of ribs enhanced the strength up to a certain level; It then
stabilized at a spacing of 2.9 times the panel thickness.
ii
Acknowledgements
Acknowledgements
First and foremost, I thank God through whom all things are possible. I would also like to
recognize and thank all the people who made my time at Queen's University during the
past four years unforgettable.
I would like to express my deepest gratitude to my supervisor, Dr. Amir Fam, for his
unwavering support and guidance throughout this research project. His patience,
leadership, and never ending encouragement gave me the confidence to focus and
proceed. I owe him an unbelievable amount of gratitude for his prominent role in helping
me to achieve one of the greatest accomplishments in my life. The support of the staff has
been a vital part of my success. Thanks go to Fiona Froats, Maxine Wilson, Cathy
Wagar, Diann King, Lloyd Rhymer, Neil Porter, Paul Thrasher, Jamie Escobar, and Bill
Boulton. Special thanks go to Dave Tryon, who provided great technical experience and
guidance. He will never be forgotten. I would also like to acknowledge my fellow
graduate students, who helped me along the way. Thanks go to Hart Honickman, Jeff
Mitchell, Wojciech Mierzejewski, Yu Ching Lai, John Gale, Reza Saeidi, Jose Aguilera,
Grag Woltman, Sarah Zakaib, Doug Tomlinson, Mike Brown, Hale Mathieson, and
Yazan Qasrawi. Special thanks got to Mark Nelson for his strong support during the
experimental phase of this study.
I wish to acknowledge the financial support provided by the Egyptian Government
through the Scholarship department. Thanks also go to Mr. Bruce Taylor, formerly with
iii
Acknowledgements
Res Precast Inc. and the Ontario Center of Excellence (OCE) for the financial and in-kind
support of this project.
I could not have survived the duration of this study without my family. I would like to
thank my parents who provided me with their love and prayers. Also, I would like to
thank my brothers, sisters, brothers-in-law and sisters-in-law for their love, support and
encouragement throughout my entire life.
From all my heart, I would like to thank my wife, Asmaa, who enlightened my life with
her love. Without her, I would not have been able to accomplish that work. I would like
to thank her for believing in me and for all her support throughout these years. For all
your love, patience and dedication, I am truly grateful.
iv
Table of Contents
TABLE OF CONTENTS
ABSTRACT. i
ACKNOWLEDGMENTS iii
TABLE OF CONTENTS v
LIST OF TABLES xi
LIST OF FIGURES... xii
NOTATIONS xxv
CHAPTER 1: INTRODUCTION. 1
1.1 Introduction..... 1
1.2 Research Objectives..... 2
1.3 Scope and Contents.. 4
CHAPTER 2: LITERATURE REVIEW... 9
2.1 Introduction.... 9
2.2 Polyurethane Foam Material Composition and Structure. 10
2.3 Sandwich Panels with Solid Cores.. 11
2.4 Sandwich Panels with Cellular Cores. 17
2.5 Mechanical Properties of Sandwich Panels . 22
2.6 Buckling and Skin Wrinkling of Sandwich Panels . 25
2.7 Delamination and Skin Debonding of Sandwich Panels 30
2.8 Numerical Analysis of Sandwich Panels .... 32
v
Table of Contents
CHAPTER 3: MATERIAL TEST PROGRAM...... 47
3.1 Introduction ... 47
3.2 Experimental Program. 48
3.2.1 Tests on Glass Fibre-Reinforced Polymer (GFRP).. 48
3.2.1.1 Tension Coupons 48
3.2.1.2 Compression Coupons............. 49
3.2.1.3 Shear Coupons.............. 50
3.2.2 Polyurethane Foam Tests.... 51
3.2.2.1 Tension Coupons 51
3.2.2.2 Compression Coupons............. 51
3.2.2.3 Shear Coupons.............. 52
3.3 Test Results 53
3.3.1 GFRP Coupons. 53
3.3.2 Polyurethane Foam Coupons......... 54
CHAPTER 4: EXPERIMENTAL STUDY ON SANDWICH PANELS WITH
DIFFERENT RIB CONFIGURATIONS. 73
4.1 Introduction 73
4.2 Description of Test Specimens and Fabrication 74
4.3 Fabrication... 75
4.4 Test Setup.... 76
4.5 Instrumentation.. 76
4.6 Test Results, Failure Modes ... 77
vi
Table of Contents
4.7 Discussion 80
4.8 Analysis of Stiffness and Deflections at Service Load Levels.. 81
4.9 Comparison with Reinforced Concrete Panels . 83
CHAPTER 5: TESTING OF LARGE SCALE CLADDING SANDWICH
PANELS.. 98
5.2 Description of the Wall Panels and Fabrication 99
5.3 Test Specimens.. 100
5.4 Materials 101
5.5 Test Setups and Instrumentation.. 102
5.5.1 Test I - Full Scale Panel Under Simulated Uniform Air
Pressure 102
5.5.2 Test II Half Full Scale Panel Under Discrete Line Loads........ 104
5.6 Results of the Experimental Program.. 105
5.6.1 Test I - Full Scale Panel Under Uniform Air Pressure. 105
5.6.1.1 Load-deflection responses ...... 105
5.6.1.2 Load-strain responses ............. 106
5.6.1.3 Failure mode.............. 106
5.6.2 Test II Half Full Scale Panel Under Discrete Line Loads........ 106
5.6.2.1 Load-deflection responses 106
5.6.2.2 Deflection distribution across the width . 107
5.6.2.3 Load-strain responses ............. 108
5.6.2.4 Strain distribution across the width.. 108
vii
Table of Contents
5.6.2.5 Failure modes . 108
5.6.3 Comparison Between the Responses in Tests I and II 109
5.6.4 Performance Relative to Design Wind Loads... 110
CHAPTER 6: FINITE ELEMENT MODELING OF SANDWICH PANELS.. 136
6.2 Summary of Another Experimental Program (Shawkat, 2008).. 138
6.3 Finite Element Model. 139
6.3.1 The 2-D Model.......... 140
6.3.2 The 3-D Model.......... 142
6.3.2.1 The 3-D Model for Panels Tested by Shawkat (2008).. 142
6.3.2.2 The 3-D Model for Panels with different Rib Configurations
(Chapter 4)...... 144
6.3.2.3 The 3-D Model for Large Scale Cladding Panels
(Chapter 5).. 146
6.3.3 Material Models..... 146
6.3.4 Failure Criteria 148
6.4 Evaluation of the Finite Element Model....... 148
6.4.1 Panels tested by Shawkat (2008)... 148
6.4.2 Sandwich Panels with different Rib Configurations
(Chapter 4)............ 149
6.4.3 Large Scale Cladding Panels (Chapter 5). 151
CHAPTER 7: SEMI-ANALYTICAL MODELING OF SANDWICH PANELS 192
viii
Table of Contents
7.2. The Semi-Analytical Model.... 193
7.2.1 Strain Profile.......... 193
7.2.2 Nonlinear Material Properties......... 194
7.2.3 Meshing.......... 197
7.2.4 Force Equilibrium and Moments. 198
7.2.5 Moment-Curvature Response......... 200
7.2.6 Generation of Full Load-Deflection Response.. 202
7.2.6.1 Flexural Effect.. 202
7.2.6.2 Shear Effect. 203
7.2.6.3 Winkler Model................ 207
7.2.6.4 Nonlinear Geometric Effects. 212
7.2.7 Failure Criteria........... 213
7.3 Illustration of Key Features of the Model. 214
7.4 Model Validation........ 216
7.4.1 Sandwich Panels with different Core Densities and different
Loadings. 216
7.4.2 Sandwich Panels with different Rib Configurations. 218
CHAPTER 8: PARAMETRIC STUDY ON SANDWICH PANELS. 241
8.1 Introduction .... 241
8.2 Material Modeling of Polyurethane Core..... 242
8.2.1 Behaviour under Compression ... 242
8.2.1.1 Modelling of the Linear Domain.............. 243
8.2.1.2 Modelling of the Nonlinear Domain.. 244
ix
Table of Contents
8.2.2 Behaviour under Tension 246
8.2.2.1 Modelling of the Linear Domain.............. 246
8.2.3 Behaviour under Shear 249
8.2.3.1 Modelling of the Linear Domain....... 250
8.3 Parametric Study.... 252
8.3.1 Skin Thickness and Core Density Effects.. 252
8.3.1.1 Effect of Skin Thickness. 253
8.3.1.2 Effect of Core Density............... 254
8.3.2 Rib Configuration Effect........... 254
CHAPTER 9: SUMMARY AND CONCLUSIONS 284
9.1 Summary ..... 284
9.2 Conclusions 285
9.2.1 Simply-Supported Panel Tests with different Rib
Configurations 285
9.2.2 Two-Span Full Scale Cladding Panel Tests.. 287
9.2.3 Numerical and Analytical Modelling........... 288
9.2.4 Parametric Study....................................... 288
9.3 Recommendations for Future Work.. 289
REFERENCES.. 291
APPENDIX A: FORTRAN CODE FOR FLEXURAL ANALYSIS OF

SANDWICH PANELS..... 303
x
List of Tables
LIST OF TABLES
Table 3.1 E-Glass fabric manufacturer data (3 Tex, Inc)... 57
Table 3.2 Epoxy resin manufacturer data (Epoxical, Inc).. 57
Table 3.3 GFRP lamina manufacturer data (Comtek, Inc) 57
Table 3.4 CORAFOAM U020 (low density polyurethane foam)

manufacturer data (DUNA-USA)..
57
Table 3.5 CORAFOAM U040 (high density polyurethane foam)

manufacturer data (DUNA-USA)..
57
Table 3.6 GFRP material properties based on tension coupons tests 58
Table 3.7 GFRP material properties based on compression coupons

tests....
59
Table 3.8 GFRP material properties based on shear coupons tests. 60
Table 3.9 Polyurethane foam material properties based on tension

coupons tests...
61
Table 3.10 Polyurethane foam material properties based on

compression coupons tests
62
Table 3.11 Polyurethane foam material properties based on shear

coupons tests
63
Table 4.1 Specification of test specimens. 85
Table 4.2 Summary of test results.. 85
Table 6.1 Summary of the experimental program by Shawkat (2008).. 154
Table 8.1 Summary of the parametric study and results. 257
xi
List of Figures
LIST OF FIGURES
Figure 1.1 Typical reinforced concrete sandwich panels and proposed
7
composite sandwich panels.............................................................
Figure 1.2 Typical failure modes of composite sandwich panels without ribs
8
(Shawkat, 2008) .
Figure 1.3 Typical flexural responses of composite sandwich panels. 8
Figure 2.1 Micrographs showing the structure of man-made cellular materials
[Gibson and Ashby, 1989] . 38
Figure 2.2 (a) Geometry, (b) internal resultants, (c) sign convention, and (d)
deformations of sandwich panels [Frostig et al, 1992] .. 39
Figure 2.3 Vertical displacement fields in the core
[Petras and Sutcliffe, 1999] ... 40
Figure 2.4 Buckling of truss core sandwich panel [Kocher et al, 2002] . 41
Figure 2.5 Corrugated-core cellular panels [Valdevit et al, 2004] .. 41
Figure 2.6 Fibrous sandwich panel [Zhou and Stronge, 2005] ... 41
Figure 2.7 Typical lattice truss topologies [Liu et al, 2006] ... 42
Figure 2.8 Bending behaviour of the sandwich panel [Fan et al, 2007] . 42
Figure 2.9 Comparisons of normalized buckling load results from different
methods for various debond ratios [Sleight and Wang, 1995] .. 43
Figure 2.10 Finite element prediction and deformation of axially loaded panels
[Falzon and Hitchings, 2003] . 43
Figure 2.11 Comparison of load-deflection predictions with the experimental
results for four-point bend test of sandwich beam with a soft foam
core at mid-span [Sokolinsky et al, 2003] . 44
Figure 2.12 Sandwich element, geometry and decomposition.
[Demiray et al, 2004] . 44
Figure 2.13 Local buckling loads determined using FEA and experimentally for
sandwich panels with circular debonds
[Avils and Carlsson, 2006] ... 45
Figure 2.14 Equivalent orthotropic thick plate for truss-core sandwich panel
xii
List of Figures
[Lok and Cheng, 2000] .. 45

Figure 2.15 Corrugated-core sandwich panel [Chang, 2005] ... 45
Figure 2.16 Sandwich panel test setup and load-deflection response at mid-span
[Boni et al, 2008] ........... 46
Figure 3.1 GFRP skin structure (3TEX, Inc) .. 64
Figure 3.2 GFRP Coupons dimensions 64
Figure 3.3 Polyurethane foam tension coupons............... 65
Figure 3.4 Polyurethane foam compression coupons... 65
Figure 3.5 Polyurethane foam shear coupons.. 65
Figure 3.6 Tensile stress-strain responses in the longitudinal direction for
GFRP coupons 66
Figure 3.7 Tensile stress-strain responses in the transverse direction for GFRP
coupons... 66
Figure 3.8 Typical failure modes for GFRP coupons in tension.. 67
Figure 3.9 Compressive stress-strain responses in the longitudinal direction for
GFRP coupons 67
Figure 3.10 Compressive stress-strain responses in the transverse direction for
GFRP coupons.... 68
Figure 3.11 GFRP Compression coupons test setup and failure modes in both
fiber. 68
Figure 3.12 Load-normal strain response for GFRP coupons in shear.. 69
Figure 3.13 Shear stress-strain responses for GFRP coupons 69
Figure 3.14 GFRP Shear coupons test setup and failure mode.. 70
Figure 3.15 Tensile stress-strain responses for polyurethane
foam 70
Figure 3.16 Tension failures for polyurethane foam coupons... 71
Figure 3.17 Compressive stress-strain responses for polyurethane foam.. 71
Figure 3.18 Compression failures for polyurethane foam coupons... 72
Figure 3.19 Shear stress-strain responses for polyurethane foam coupons 72
Figure 3.20 Shear failure for Polyurethane foam coupons. 72
Figure 4.1 Sandwich panels (S1 to S6) with different rib configurations 86
xiii
List of Figures
Figure 4.2 VARTM fabrication process for the sandwich panels 87

Figure 4.3 Test setup.... 88
Figure 4.4 Sandwich panels instrumentation 89
Figure 4.5 Load-deflection responses for all specimens.. 90
Figure 4.6 Load-strain behaviour of sandwich panels. 91
Figure 4.7 Strain distributions along the two axes of panel S1 92
Figure 4.8 Failure mode of specimen S1.. 93
Figure 4.14 Effect of ribs and core density on strength and stiffness of sandwich
panels.. 96
Figure 4.15 Experimental versus analytical load-deflection responses. 96
Figure 4.16 Comparison between sandwich panels and conventional RC panels. 97
Figure 5.1 General description and composition of the cladding panels. 112
Figure 5.2 Panel during shipping and handling 112
Figure 5.3 Full scale sandwich panel dimensions 113
Figure 5.4 Overview of test apparatus. 114
Figure 5.5 Back and side forms of the pressure box.... 114
Figure 5.6 Cross-section through the sandwich panel and the test apparatus.. 115
Figure 5.7 Pressure box with the Polycrepe seal layer. 115
Figure 5.8 Wooden supporting blocks (pads) . 116
Figure 5.9 Pressure Load Actuators (PLAs) ... 116
Figure 5.10 Air pressure inlets and pressure reading sensors 117
Figure 5.11 Deflection measurements using Linear Potentiometers (LPs) 117
Figure 5.12 Longitudinal strain measurements.. 118
Figure 5.13 Transverse strain measurements. 118
xiv
List of Figures
Figure 5.14 Half-width sandwich panel in Test II. 119

Figure 5.15 Deflection measurements using Linear Potentiometers (LPs) ... 120
Figure 5.16 Strain gauges distribution for top and bottom skins... 120
Figure 5.17 Load-deflection response for the first test on full scale panel 121
Figure 5.18 Load-deflection response for the second test on full scale panel... 121
Figure 5.19 Load-deflection responses for the second and third tests for full
scale panel in Test I.... 122
Figure 5.20 Load-longitudinal strain responses for the loaded skin of the full
scale panel in Test I 122
Figure 5.21 Load-longitudinal strain responses for the opposite skin of the full
scale panel in Test I 123
Figure 5.22 Load-transverse strain responses for the loaded skin of the full scale
panel in Test I. 123
Figure 5.23 Load-transverse strain responses for the unloaded skin of the full
124
scale panel in Test I
Figure 5.24 Skin wrinkling and crushing failure of the full scale panel in Test I.. 124
Figure 5.25 Load-deflection responses for half-width panel in Test II.. 125
Figure 5.26 Load-deflection responses of the half-width panel in Test II over
the edge support.. 125
Figure 5.27 Load-deflection responses over the middle support for the half-
width panel in Test II.. 126
Figure 5.28 Sandwich panel deformations in Test II 126
Figure5.29 Deflection distribution over an axis passing through points 1-2-3 in
Test II...................................... 127
Figure 5.30 Deflection distribution over an axis passing through points 4-5-6 in
Test II.. 127
Test II.. 128
Figure 5.32 Deflection distribution over an axis passing through points 10-11-
12 in Test II. 128
xv
List of Figures
Figure 5.33 Load-strain responses for loaded skin at the mid-span regions for
the half-width panel in Test II. 129
Figure 5.34 Load-strain responses for loaded skin at the middle support regions
for the half-width panel in Test II... 129
Figure 5.35 Load-strain responses for opposite skin at the middle support
region for the half-width panel in Test II 130
Figure 5.36 Load-strain responses for opposite skin at the mid-span region for
Figure 5.37 Loaded skin strain distribution over cross section passing through
axis 1-2-3-4-5 in Test II.. 131
axis 14-13-11-10-7 in Test II.. 131
axis 16-12-9 in Test II. 132
Figure 5.40 Opposite skin strain distribution over cross section passing through
axis 15-1-2 in Test II... 132
Figure 5.41 Opposite skin strain distribution over cross section passing through
axis 11-10-7-8-4 in Test II.. 133
Figure 5.42 Compression skin wrinkling and crushing over the middle support
for the half-width panel in Test II... 133
Figure 5.43 Core shear failure mode near the middle support for the half-width
panel in Test II 134
Figure 5.44 Top skin crushing and wrinkling failure mode at the mid-span for
Figure 5.45 Moment-curvature responses for both full and half-width sandwich
panels in Test I and II. 135
Figure 6.1 Test setup of the panels tested by Shawkat (2008) .... 155
Figure 6.2 Sandwich panel geometry and the corresponding 2D FEM (for
specimens tested by Shawkat (2008)) 156
Figure 6.3 Elements used in the 2D FEM (ANSYS) 156
Figure 6.4 Finite element mesh used in the convergence study for the 2-D
xvi
List of Figures
FEA. 157
Figure 6.5 Convergence study results for the 2-D FE model (case of panels
tested under simulated uniform load Shawkat (2008)) 158
Figure 6.6 Final 2-D FEM mesh and loading configuration for panels tested by
Shawkat (2008) .. 158
Figure 6.7 Sandwich panel geometry and the corresponding 3-D FEM for
specimens tested by Shawkat (2008) . 159

Figure 6.8 Elements used in the 3-D FEM (ANSYS )... 159
Figure 6.9 Finite element mesh used in the convergence study for the 3-D FE
model for the panels tested by Shawkat (2008) . 160
tested under simulated uniform load Shawkat (2008)) 161
Figure 6.11 Final 3-D FEM mesh for the panels tested by Shawkat (2008) . 161
Figure 6.12 Finite element mesh used in the convergence study for the 3-D FE
model for panels with different rib configuration (Chapter 4) .. 162
Figure 6.13 Convergence study results for the 3-D FE model for panels tested
in Chapter 4. 163
Figure 6.14 Final 3-D FEM mesh panels with different rib configurations
(Chapter 4) . 163
Figure 6.15 FEA geometric models and assumptions of large scale cladding
panels (Chapter 5) .. 164
Figure 6.16 Final 3-D FEM for the large scale sandwich panels (Chapter5) 164
Figure 6.17 Bi-linear tension and compression stress-strain curve for GFRP
skins.... 165
Figure 6.18 Bi-linear Shear stress-strain curve for GFRP ribs.......... 165
Figure 6.19 Polynomial curve fitting for soft and hard polyurethane foam
stress-strain curve... 166
Figure 6.20 Polynomial curve fitting for Shear stress-strain curve for both soft
and hard polyurethane foam... 166
Figure 6.21 Load-deflection response for panel (P1) with soft polyurethane
xvii
List of Figures
foam core and loaded in three-point bending 167

foam core and loaded in four-point bending... 167
Figure 6.23 Load-deflection responses for panels (P3, P4 and P5) with soft
polyurethane foam core and tested under a uniform load... 168
Figure 6.24 Load-deflection response for panels (P7, P8 and P9) with hard
polyurethane foam core and tested under uniform load 168
Figure 6.25 Load-strain response for panel (P1) with soft polyurethane foam
core and loaded in three-point bending... 169
core and loaded in four-point bending 169
Figure 6.27 Load-strain response for panels (P3, P4 and P5) with soft
polyurethane foam core and tested under a uniform loading.. 170
Figure 6.28 Load-strain response for panels (P7, P8 and P9) with hard
polyurethane foam core and tested under a uniform loading.. 170
Figure 6.29 Deformed shape under three-point bending for panel P1... 171
Figure 6.30 Deformed shape under four-point bending for panel P2 172
Figure 6.31 Deformed shape under uniform load for panels P3, P4 and P5 with
soft polyurethane foam core... 173
Figure 6.32 Deformed shape under uniform load for panels P7, P8 and P9 with
hard polyurethane foam core.. 174
Figure 6.33 Shear failure modes within the soft and hard polyurethane foam
core under uniform loads 175
Figure 6.34 Load-deflection responses for sandwich panels with different rib
configurations reported in Chapter 4.. 176
Figure 6.35 Load-strain responses for panel S1. 176
Figure 6.38 Load-strain responses for panel S4..... 178
xviii
List of Figures

Figure 6.41 Failure mode of panels S1.. 179
Figure 6.42 Failure mode of panel S2.... 180
Figure 6.43 Failure mode of panel S3.... 180
Figure 6.44 Failure mode of panel S4 181
Figure 6.47 Pressure-deflection responses for the full size cladding tested under
air pressure.. 182
Figure 6.48 Load-deflection responses near mid-span for the half cladding
panel tested under discrete loading..... 183
Figure 6.49 Load-deflection responses at supports for the half-width panel
tested under discrete loading... 184
Figure 6.50 Pressure-longitudinal strain responses for the loaded skin of the
large scale panel tested under air pressure...... 185
Figure 6.51 Pressure-transverse strain responses for the loaded skin of the large
scale panel tested under uniform air pressure. 185
Figure 6.52 Pressure-longitudinal strain responses for the opposite skin of the
large scale cladding panel tested under uniform air pressure. 186
Figure 6.53 Pressure-transverse strain responses for the opposite skin of the
large scale cladding panel tested under uniform air pressure. 186
Figure 6.54 Load-longitudinal strain responses on the loaded skin for the half-
width panel near middle support. 187
Figure 6.55 Load-longitudinal strain responses on the loaded skin of the half-
width panel near mid-span ......... 187
Figure 6.56 Load-longitudinal strain responses on the opposite skin for half-
width panel near the middle support... 188
Figure 6.57 Load-longitudinal strain responses on the opposite skin of the half-
width panel near the mid-span 188
Figure 6.58 Compression skin wrinkling and crushing failure modes and the
xix
List of Figures
corresponding Tsai-Wu failure criteria index value... 189

Figure 6.59 Load-deflection response of panel Test II, as is, and assuming it
was of a full width.. 189
Figure 6.60 Shear stress distribution for panel Test II in the foam core at the
cut-off edge. 190
Figure 6.61 Shear stress distribution for panel Test II in the GFRP interior rib 191
Figure 7.1 Normal stress and strain distributions. 220
Figure 7.2 Cross-section meshing 220
Figure 7.3 Mesh configuration and convergence study results 221
Figure 7.4 Normal-strain profile ..... 221
Figure 7.5 Flow chart of the procedure to obtain the moment-curvature
response of a cross-section 222
Figure 7.6 Schematic for the process for determining deflection due to flexure. 223
Figure 7.7 Shear deflections in sandwich panel... 224
Figure 7.8 Section transformation accounting for variable modulus in core and
skin in tension and compression. 225
Figure 7.9 Winkler effect of polyurethane foam softness at loading points 226
Figure 7.10 Flow chart of procedure used to obtain the load-deflection response 227
Figure 7.11 Illustration of significance of various features of the model for
panels with soft cores, P3 to P5 (Shawkat, 2008) .. 228
Figure 7.12 Illustration of significance of various features of the model for
panels with hard cores, P7 to P9 (Shawkat, 2008) . 228
Figure 7.13 Load-deflection response of panel P1 (Shawkat, 2008) 229
Figure 7.14 Deflection responses along the span of specimen P1 (Shawkat,
2008) .. 229
Figure 7.15 Load-longitudinal strain responses for specimen P1 (Shawkat,
2008) .. 230
Figure 7.16 Load-deflection response of panel P2 (Shawkat, 2008) 230
Figure 7.17 Deflection responses along the span of specimen P2 (Shawkat,
2008) .. 231
xx
List of Figures
Figure 7.18 Load-longitudinal strain responses for specimen P2 (Shawkat,

2008) .. 231
Figure 7.19 Load-deflection responses of panels P3 to P5 (Shawkat, 2008) 232
Figure 7.20 Load-deflection responses of panels P7 to P9 (Shawkat, 2008) 232
Figure 7.21 Deflection responses along the span of panels P3 to P5 (Shawkat,
2008) .. 233
Figure 7.22 Deflection responses along the span of panels P7 to P9 (Shawkat,
2008) .. 233
Figure 7.23 Load-longitudinal strain responses for panel P3 to P5 (Shawkat,
2008) ...... 234
Figure 7.24 Load-longitudinal strain responses for panel P7 to P9 (Shawkat,
2008) .. 234
Figure 7.25 Load-deflection response for panel S1 (Chapter 4) ... 235
Figure 7.31 Load-strain responses for panel S1 (Chapter 4) . 238
Figure 8.1 Schematic for the compressive stress-strain curve for polyurethane
foam.... 258
Figure 8.2 Schematic for an idealized cubic cell model for closed-cell foam,
showing the edge thickness, t e , and the face thickness, t f . [Gibson
258
and Ashby (1988)] .
Figure 8.3 Experimental and predicted compressive stress-strain relationship
xxi
List of Figures
for 32 kg/m3 and 64 kg/m3 polyurethane foam densities 259

Figure 8.4 Predicted compressive stress-strain curves for polyurethane foam
densities ranging from 32 kg/m3 to 192 kg/m3... 259
Figure 8.5 Schematic for the tensile stress-strain curve for polyurethane foam . 260
Figure 8.6 Normalized tensile stress-strain curve for polyurethane foam
material with 32 kg/m3 and 64 kg/m3 densities.. 260
Figure 8.7 Experimental and predicted tensile stress-strain curve for
polyurethane foam with 32 kg/m3 and 64 kg/m3 densities. 261
Figure 8.8 Predicted tensile stress-strain curve for polyurethane foam with
Figure 8.9 Schematic for the shear stress-strain curve for polyurethane foam ... 262
Figure 8.10 Normalized shear stress-strain curve for polyurethane foam
material with 32 kg/m3 and 64 kg/m3 densities.. 262
Figure 8.11 Experimental and predicted shear stress-strain curve for
polyurethane foam with 32 kg/m3 and 64 kg/m3 densities. 263
Figure 8.12 Predicted shear stress-strain curve for polyurethane foam with
Figure 8.13 Load-deflection response of sandwich panel M1t1 264
xxii
List of Figures

Figure 8.31 Effect of skin thickness on load-deflection response of sandwich
panels with core density of 32 kg/m3.. 273
panels with core density of 128 kg/m3 274
Figure 8.37 Effect of skin thickness on behaviour of panels. 276
Figure 8.38 Effect of core density on load-deflection response of sandwich
panels with skin thickness = 1.6 mm.. 277
Figure 8.41 Effect of skin thickness on behaviour of panels 279
Figure 8.42 Rib configurations used in the parametric study 280
Figure 8.43 Predicted failure mode of panel PS1... 280
xxiii
List of Figures

Figure 8.49 Effect of rib configuration on strength and stiffness of sandwich
panel 283
Figure A.1 Sandwich panel load scheme.. 304
Figure A.2 Rib configurations used in the FORTRAN program.. 305
xxiv
Notations
NOTATIONS
a Distance from beam left end to the point of calculation
A Ci Area of polyurethane foam core at layer i
Ai Area at layer i
a j ,b j ,c j ,d j Spline function coefficients
A Si Area of GFRP skin at layer i
A a, A b Constant used to calculate slopes due to elastic foundation effect for a
uniform load q at the uniform load ends a, b.
A x Constant used to calculate deflections due to elastic foundation effect for a
certain concentrated load P
b Distance from beam right end to the point of calculation
B1 Constant used to calculate critical wrinkling stress for the top skin
b l,i Original cross-section width at layer i and segment l
b t,l,i Width of transformed section at layer i at certain span segment l.
B a, B s Constant used to calculate moments due to elastic foundation effect for a
B x Constant used to calculate slopes due to elastic foundation effect for a
C1 C4 Integration constants for Winkler model
C a, C b Constant used to calculate shear forces due to elastic foundation effect for a
C x Constant used to calculate moments due to elastic foundation effect for a
xxv
Notations
D Constant use to determine foam properties at any load level
dx Span segment length
D a, D b Constant used to calculate deflections due to elastic foundation effect for a
D x Constant used to calculate shear forces due to elastic foundation effect for a
EA Modulus of elasticity after yielding
EC Core modulus of elasticity
E fc Compressive modulus of elasticity for polyurethane foam
E ft Tensile modulus of elasticity for polyurethane foam
EI Flexural rigidity of the GFRP skin
E l,i Secant modulus of elasticity for material at layer i and segment l
ES Skin modulus of elasticity
Es Compressive modulus of elasticity for solid polyurethane foam
Gf Polyurethane foam shear modulus
H Section overall thickness at any segment
H new,l New section thickness at segment l
H old,l Old section thickness at segment l
i Layer number
I t,l Moment of inertia of the transformed section at certain span segment l
k Elastic modulus of the subgrade (polyurethane foam)
l Segment number
xxvi
Notations
M Section total moment
m Constant use to determine foam properties at any load level
M` 1 Induced moments in the top skin at end 1 based on the fact that the top skin
has infinite length for the submodel I
M o End condition moment for infinite beam submodel II
M o End condition moment for infinite beam submodel I
M 1 Induced moments in the top skin at end 1 based on the fact that the top skin
has infinite length for the submodel II
M1 Moment at end 1 of infinite beam for model I
M1, M2 Induced moments in the top skin at ends 1, 2 based on the fact that the top
skin has infinite length
M2 Moment at end 2 of infinite beam for model II
Mi Moment at certain span segment
M o1 Moment at left end 1 of the top skin
M o2 Moment at left end 2 of the top skin
n Total number of layers
n bot Number of layers below section neutral axis
n top Number of layers above section neutral axis
P Concentrated load
P o End condition force for infinite beam submode II
P o End condition force for infinite beam submodel I
pa Atmospheric pressure
po Foam cell internal pressure
xxvii
Notations
P o1 Force at left end 1 of the top skin
P o2 Force at left end 2 of the top skin
Q Shear forces at any section according to Winkler
q Uniform load or pressure
Q` 1 Induced shear force in the top skin at end 1 based on the fact that the top
skin has infinite length for the submodel I
Q 1 Induced moments in the top skin at end 1 based on the fact that the top skin
has infinite length for the submodel II
Q1, Q2 Induced shear force in the top skin at ends 1, 2 based on the fact that the top
skin has infinite length
Q t,l,i First moment of area for the transformed section at certain span segment l
for layer i
S j (x) Spline function
Vl Total applied shear force at specific span segment l
w Vertical deflection according to elastic foundation assumption
y bar Distance for the extreme tension fibres to the neutral axis
yi Layer location
1 Constant used to calculate end straining actions
2 Constant used to calculate end straining actions
Characteristic of the system, (1/) is the characteristic length of beam on
elastic foundation
Shear strain
D Shear limiting strain
xxviii
Notations
l,i Average shear strain at layer i at segment l
pl Plastic shear strain
m,l Deflection due to flexure at segment l
tot,l Total deflection due to flexure, shear and Winkler at segment l
v,bot Deflection due to shear of the bottom skin at mid-span
v,i Shear deflection for layers i at mid-span
v,l Deflection due to shear at segment l
v,l,bot Average shear deflection for layers below section neutral axis at segment l
v,l,i Shear deflection for layers i at segment l
v,l,top Average shear deflection for layers above section neutral axis at segment l
v,top Deflection due to shear of the top skin at mid-span
w,l Deflection due to Winkler at segment l
y i Change in elevation between two successive segments
i Change in slope between two successive sections
Strain level
c Extreme compressive strain
cu Ultimate compressive strain
Dc Compressive limiting strain
Dt Tensile limiting strain
i Strain at any layer i
t Extreme tensile strain
tu Ultimate tensile strain
i Slope at midpoint of any segment
xxix
Notations
c Core Poissons ratio
f Poissons ratio for polyurethane foam
f Density of polyurethane foam(foam with voids)
s Density of solid polyurethane foam(foam without voids)
A Stress increase after the linear domain
c Compressive stress for polyurethane foam
Ci Stress at polyurethane foam core at any layer i
cr Minimum critical wrinkling stress
i Stress at any layer i
plc Plastic limit compressive stress for polyurethane foam
plt Plastic limit tensile stress for polyurethane foam
Si Stress at GFRP skin at any layer i
t Tensile stress for polyurethane foam
ys Yielding stress for solid polyurethane foam
Shear stress
f Polyurethane foam shear stress
l,i Shear stress at layer i at segment l
pl Plastic shear stress for polyurethane foam
Section curvature
i Curvature of certain span segment
s Polymer fraction in the strut
xxx
Chapter 1 Introduction
CHAPTER 1: INTRODUCTION
1.1 Introduction
Sandwich panels have traditionally been used in aerospace and marine applications. They
are typically composed of two thin skins made from fibre-reinforced polymers (FRPs) or
metallic materials such as steel or aluminium and a core material sandwiched in between.
This core material takes several forms such as honeycomb cores, corrugated cores, truss
cores, Z-cores, C-cores, I-cores or solid foam cores. The skins are bonded to the core by
means of an adhesive polymer such as epoxy resin. Composite sandwich panels refer to
those with FRP skins bonded to any of the previously mentioned core types. The FRP
skin carries the in-plane compressive and tensile stresses resulting from bending, while
the main function of the core is to keep the two FRP skins apart at the desired distance
and to resist and transmit the induced shear forces to the supporting points. The core may
also provide thermal insulation.
In the building construction industry, cladding walls generally provide an
envelope to the building for protection against the elements, thermal insulation, fire
protection and may also provide a certain architectural design to enhance the aesthetics of
the building. Cladding panels are typically non-load-bearing in the plane of the panel and
are generally designed for out-of-plane loading due to wind loads. Conventional cladding
panels are usually made from precast reinforced concrete thick skins with a foam core
material used to provide the required insulation. Although these types of cladding panels
have been widely used for years, they have several drawbacks. Being exposed to the
elements, they are susceptible to corrosion of the steel reinforcement, especially in humid
1
environments. Also, they add much dead weight to the building, which in turns adds
significantly to the size of members, including beams, columns and foundation. Other
disadvantages of the heavy-weight reinforced concrete panels are the difficulties
associated with shipping and handling, which impact the speed of installation (Fig.
1.1(a)). Cladding composite sandwich panels with FRP skins are durable and relatively
very light in weight (Figure 1.1(b))
1.2 Research Objectives
The research program carried out in this thesis aims to study in depth the flexural
performance of composite sandwich panels. Although the study was motivated by the
building cladding application, the methods and findings of this study can indeed be
generalized and serve several other potential structural applications. This may include
flooring, decking, platforms and roofs. A preliminary and rather small experimental study
carried out at Queens University several years ago (Shawkat, 2008) on small scale
composite sandwich panels without any internal or exterior ribs demonstrated the
potential of this new system. However, it pointed out the low stiffness as a weakness and
the vulnerability of the panels to shear failure due to the lack of ribs (Figure 1.2).
Therefore, in that study, it was hypothesized that introducing FRP ribs of different
configurations, such as internal longitudinal ribs, internal transverse ribs, and exterior ribs
(diaphragm) would significantly enhance flexural strength and stiffness (Figure 1.3).
Furthermore, part of the study is to produce test results and model full scale composite
sandwich wall panels intended for cladding applications, specifically.
2
The specific objectives of the study are:
Establish the constitutive relationships of the rather unconventional constituent
materials of the sandwich panel, namely the polyurethane foam core and glass fibre
reinforced polymer (GFRP), through a comprehensive material testing program.
Experimental evaluation of the flexural performance of composite sandwich panels with
different rib configurations, in terms of flexural strength and stiffness and failure
modes.
Compare the flexural behaviour of the sandwich panels to that of comparable reinforced
concrete conventional panels.
Examine the flexural behaviour of a full scale production sandwich panel specifically
designed for cladding application, under a realistic air pressure loading. Another full
scale test is carried out using conventional mechanical loading to complement and
confirm results of the first test.
Assess the performance of the full scale panel relative to the building code requirements
for maximum design wind pressure in Ontario and Canada.
Develop a rigorous numerical model using the finite element method (the ANSYS
program) to predict and calculate the structural behaviour of the composite sandwich
panels, in terms of the full load-deflection response, and the failure mode. The model
accounts for geometric and material non linearities.
Develop an independent comprehensive analytical model using the FORTRAN
programming language to predict the flexural response of the panels and compare with
the finite element model. Key features of this model include accounting for the
3
considerable shear deformation of the core and the change in thickness of the panel due
to the soft core.
Perform a comprehensive parametric study using the developed models to examine a
wider range of geometric and mechanical property parameters, beyond the limitations of
the experimental program.
1.3 Scope and Contents
The scope of this study consists of experimental investigations and numerical and
analytical modelling. The experimental programs are designed to address the general
flexural behaviour of composite sandwich panels as well as to model a potential specific
application, namely cladding of buildings. The experimental results obtained from these
studies and from a previous study conducted at Queens University are used to verify the
developed analytical and numerical models. The models are then used in the parametric
study.
The experimental work was divided into three phases. Phase I was mainly to
evaluate the mechanical characteristics of the constituent materials of the composite
sandwich panel, namely the GFRP skins and the Polyurethane foam core. This was
achieved by testing the materials under different loading schemes, namely in tension,
compression and shear, according to ASTM standards. Phase II was an experimental
investigation of composite sandwich panels with various configurations of internal and
exterior longitudinal and transverse GFRP ribs integrated with the two skins. Six,
2500x660x78 mm, specimens were manufactured using GFRP skins and ribs, and low-
density polyurethane foam core, and were tested under a simulated uniform load by using
a group of discrete loads. Flexural strength, stiffness and various failure modes were
4
assessed and compared to those of conventional reinforced concrete panels. Also, a
simple design-oriented analytical model was developed for deflection calculations at
service load level, and validated using test results. In Phase III, two full scale panels,
9150x2440x78 mm, were fabricated specifically for cladding application. The first panel
was tested under a uniform load simulated by actual air pressure actuators, using a
specific setup, while the second one was tested under conventional mechanical loading
using a group of discrete loads, to confirm the strength of the first panel. The panels were
tested with the same boundary conditions as intended in the real life application.
For the theoretical work, two models have been developed to predict the complete
flexural behaviour of the composite sandwich panels. The first one is a numerical
nonlinear Finite Element Model (FEM) using the commercial code (ANSYS). It was
first verified using the results of Shawkat (2008) as well as those of Phase II tests, and
showed very good agreement. Given the complexity of the panels in Phase III, the model
was then used to predict the behaviour of the panels before carrying out the tests, to
design the instrumentation scheme to capture the maximum deflections and strains. The
second independent model is a nonlinear analytical model based on the concepts of strain
compatibility and force equilibrium. The model is capable of predicting the full flexural
response of the panels, accounting for the effects of the very soft core on the excessive
shear deformation and change in panel thickness. The model was successfully validated
against the experimental and numerical results. Based on the proposed models, a
parametric study has been performed to study the effect of skin thickness, core density,
and rib spacing, on the behaviour of the panels.
5
The following is a brief summary of the contents of each chapter:
Chapter 2 presents the literature review, including the background and fundamental
characteristics of both metallic-skin and composite-skin sandwich panels with different
core types. A summary of related experimental and analytical work is also presented.
Chapter 3 describes Phase I of the experimental program, which examines the
mechanical characteristics of the constituent materials used to fabricate the composite
sandwich panels.
Chapter 4 describes Phase II of the experimental program, which examines the flexural
behaviour of composite sandwich panels with different rib configurations.
Chapter 5 describes Phase III of the experimental program. This includes the first full
scale cladding panel tested under air pressure, and the second panel tested under
discrete loads as a simulation of a uniform load.
Chapter 6 presents the numerical (FEA) models developed to predict the flexural
responses of the panels with and without ribs of various configurations. The models are
verified using the experimental results from a previous study and the present study.
Chapter 7 presents the analytical model developed to predict the flexural behaviour of
the panels with and without ribs. The model is verified using the experimental results
from the previous and present study, and the FEA model.
Chapter 8 presents the parametric study performed to examine the effect of the different
parameters on the behaviour of the sandwich panels.
Chapter 9 provides a summary of the thesis and conclusions based on the experimental
and theoretical programs. The chapter also presents recommendations for future work.
Appendix A provides the FORTRAN code for the analytical model in Chapter 7.
6
(a) Crane lifting of a typical precast reinforced concrete panel
(b) Light weight composite sandwich panel
Figure 1.1 Typical precast reinforced concrete sandwich panels and proposed
composite sandwich panels
7
(a) Inward skin wrinkling (b) Core shear failure
Figure 1.2 Typical failure modes of composite sandwich panels without ribs
(Shawkat, 2008)
Panels enhanced by
ribs Full potential
flexural strength
Load
Panels with no
ribs
Deflection
Figure 1.3 Typical flexural responses of composite sandwich panels
8
Chapter 2 Literature Review
CHAPTER 2: LITERATURE REVIEW
2.1 Introduction
Sandwich panels are being used in many applications ranging from aerospace to
automotive and transportation, but are not as common in structural applications.
Typically, sandwich panels are composed of two thin skins made from metallic materials
such as aluminum or steel, or from fibre reinforced polymers (FRPs), with a core material
in between. This core material takes several forms such as honeycomb cores, corrugated
cores, truss cores, Z-shaped cores, C-shaped cores, I-shaped cores, or a solid core. The
core materials may be metallic or polymeric. The concept behind sandwich structure is
that the skins carry the in-plane compressive and tensile stresses resulting from the
induced bending moment, while the main function of the light-weight core is to keep the
two skins apart, at a desired distance, and also to resist and transmit the induced shear
forces to the supporting points.
Sandwich panels have a great potential for several structural engineering
applications such as light-weight decking, flooring, or roofing and as cladding panels for
buildings. The primary purpose of the exterior non-load-bearing cladding is to provide
shelter from the rain, wind and snow and to provide the necessary thermal insulation. The
primary design criterion in this case is to resist and transfer wind loads to the internal
skeleton of the building. The conventional exterior cladding is usually composed of a
rather different type of sandwich panels. These are heavy precast reinforced concrete
sandwich panels. Corrosion of steel reinforcement in these panels is a concern since the
panels are exposed to the elements. Composite FRP-sandwich panels are quite durable as
9
they do not corrode. They are also significantly lighter in weight than concrete panels.
The use of composite GFRP-sandwich panels in building applications started in the 1970s
(Pamela, 2007). The composite sandwich panels are constructed using either a rigid or
soft polyurethane foam core sandwiched between two layers of Glass-FRP (GFRP) and
adhered by a layer of resin such as epoxy. From the cost stand point, it was shown that
the [0/90o] GFRP configuration is the most appropriate for one-way loaded panels (Awad
et al, 2010). As indicated earlier, the GFRP layers resist the tensile and compressive
stresses resulting from the flexure induced by wind loading, which could be a pressure or
suction. The polyurethane foam core resists shear stresses and also contributes to the
moment of inertia of the panel by acting as a spacer that maintains the GFRP skins
spaced apart at a desired distance. In cladding application, it also serves another
important function, namely thermal insulation of the panel.
This chapter presents an overview of the sandwich panels fabricated from
different skins and cores of different materials and types. Also, it introduces a brief
summary of the analytical and numerical methods developed to study the behaviour of
sandwich panels.
2.2 Polyurethane Foam Material Composition and Structure
One of the commonly used materials as a core in sandwich panels is polyurethane foam.
This refers to a number of different types of foam consisting of polymers made of
molecular chains bound together by urethane links. It can be flexible or rigid, but
generally has a low density. Flexible polyurethane foam is most often used in bedding
and upholstery, while the more rigid variety is used for thermal insulation and in
automobile dashboards. Foams are made using different techniques. Polymers are foamed
10
by introducing a gas into the liquid monomer or hot polymer with a blowing agent,
allowing the bubbles to grow and stabilize, and then solidifying the foam by cross-linking
or cooling. Metallic foams are made by mixing organic beads such as carbon into a metal
melt in an inert atmosphere when the metal has cooled and solidified, the carbon is burnt
off, leaving a cellular matrix. Ceramic foams are made by infiltrating an open-cell
polymer foam with a fine slurry of the ceramic, in water or some other fluid. When the
aggregate is fired, the slurry bonds to give an image of the original foam, which, of
course, burns off. Additional, less common, techniques of making foams have been
described by Gibson and Ashby (1988).
The unique properties of foams, arising from their cellular structure, can be
exploited in engineering design. The small cell size and low volume fraction of solids in
closed cell foams make them excellent thermal insulators for applications ranging from
coffee cups to building cladding panels. Because of their low compressive strength and
high deformation capacity, they are outstanding energy absorbers. This property is
exploited in different packaging and protective padding. Their low density makes them
ideal core materials for light-weight structural sandwich panels used in modern aerospace
components and sporting equipment (Gibson and Ashby, 1988). Figure 2.1 shows
micrographs showing the structure of man-made cellular materials for flexible open-cell
and flexible closed-cell polyethylene foams (Gibson and Ashby, 1988).
2.3 Sandwich Panels with Solid Cores
Solid flexible cores refer to cores typically made of foams and are compressible such that
through-thickness deformations are possible. Some of the earliest applications of
sandwich panels in the 20th century have been in aircraft industry (Allen, 1969). This
11
was followed by expansion of applications into the aerospace, automotive and marine
industries, especially after introducing FRP composite materials. Early on, sandwich
panels fabricated from metallic cores were assumed to be Antiplane by many researchers.
Filon (1937) described the Antiplane core as one that is incompressible (i.e. with
negligible through-thickness deformation) and also negligible contribution to the flexural
stiffness, an assumption that Allen (1969) also made. The assumption of Antiplane core
was not an accurate representation of sandwich panels with flexible cores. Other
researchers attempted to enhance the analytical procedures. Reissner (1947, 1948) and
Hockman (1973) took the effect of shear in addition to bending on deflection, but ignored
the through-thickness deformation of the sandwich beam. Holt and Webber (1982) and
Pearce (1973) studied sandwich structures with anisotropic composite skins with
Antiplane cores. Ojalvo (1977) accounted for the through-thickness deformation but
ignored the peeling stresses between the skins and the core. Others made the assumption
that sandwich panels with a foam core act like an ordinary beam with equivalent sectional
properties (Ogorkiewicz and Sayigh, 1973). In the case of sandwich panels having
incompressible cores, in which the thickness of the core is unchanged, the core can be
analyzed using the shear deformable approach (Kant and Mallikarjuna, 1989, Kant and
Patil, 1991, Senthilnathan et al, 1988, and Chandrashekhara and Krishnamurthy, 1990).
Gordaninejad and Bert (1989) investigated a non-prismatic sandwich beam with
thick skins and the core was assumed to be of the Antiplane type. The approach was
sufficient to describe the general behaviour of the beam, but it did not take into account
the high-order effects and the normal stresses involved at the skin-core interfaces,
especially at the concentrated load locations, which may cause skin and core indentation.
12
Prismatic sandwich panels, which are symmetric about their mid-thickness surface, were
studied by Payder and Libove (1988). Their approach, using small deflection theory,
predicted the overall deflection of the panel within acceptable accuracy but was not able
to predict the deflection at localized regions such as concentrated loads and reaction
points.
When foam or low-strength honeycomb cores (described later in section 2.4) are
used, in which the core thickness changes due to its transverse flexibility, this effect must
be considered. Frostig and Baruch (1990) took into consideration this flexibility of the
core in the thickness direction, particularly the localized compressibility in the vicinity of
the applied loads. Closed-form equations for predicting deflection, normal stresses in
skins, and core shear stresses were developed earlier by Allen (1969), without
considering the core flexibility, while Frostig and Baruch (1990) developed the governing
differential equations for these engineering quantities based on superposition approach,
but without giving closed-form equations. A high-order bending theory based on virtual
work was later developed by Frostig et al. (1992), as an alternative to superposition, to
generate the governing deferential equation to predict the bending behaviour of a
sandwich beam with flexible core, as shown in Figure 2.2.
Frostig (1993a) presented analytically the bending behaviour of a sandwich beam
in the presence of discrete diaphragms running perpendicular to the longitudinal
direction. The study stated that when the diaphragm is located anywhere along the beam,
its effects on the beam's behaviour are implemented through boundary and continuity
conditions at the junction between skin and diaphragm. The behaviour of the diaphragm
is expressed using compatibility conditions of the upper and lower skin-diaphragm joints.
13
The presence of diaphragms causes localized effects in both the skins and the core,
especially at the applied load locations.
Frostig (1993b) took into account analytically the effect of stress concentration
under different types of loads in various regions of the panel. This resulted basically in an
improved general high-order theory. Several cases were studied. In one case, the effect of
modulus of elasticity of the core on the level of skin stresses at the locations of point
loads and supports was investigated. When the core modulus was increased, the moments
and deformations of the skin decreased and the peeling stresses increased. However,
according to the study, this does not mean that stiffer cores are more sensitive to
premature peeling failure than flexible ones, since the improvement of the stiffness also
increases the allowable stresses of such a core. In another case, localized delaminations
were introduced between top skin and the core at various locations. It was shown that the
length of the delaminated region greatly affected the stress level. Also, the stress level
was higher in the case of edge delamination than inner one with the same length. In the
last case, the effects of bonded versus unbonded diaphragm, at different locations,
namely at a concentrated load, far from the load, and at the support, were investigated.
The study showed that the presence of diaphragms improves the performance and reduces
the stresses in the skins but recommended the use of bonded diaphragms.
Peled and Frostig (1994) analyzed sandwich beams with flexible core and tapered
skins. The analysis showed that the displacement field in the core, through its thickness,
takes a nonlinear distribution. Also, the localized effects due to concentrated or
distributed loads as well as geometric and material changes were studied. The analysis
14
used a two-dimensional model for the core combined with a one-dimensional
longitudinal model for the skins.
Another general systematic high-order approach that allows a description of the
behaviour of a piece-wise uniform sandwich beam with a tapered transition zone between
two different depths and a flexible core was described by Peled and Frostig (1995).
According to the study, and as a result of the shear forces in the inclined skin, additional
internal concentrated forces at the ends of the transition zone were introduced, but they
were not dependent on the loading scheme. The study stated that the level of peeling
stresses, bending moments and shear stresses at the lower skin-core interface depend on
the core properties while the shear forces at the lower skin are independent of the core
type.
Thomsen and Frostig (1997) investigated the localized bending effects in
sandwich beams having soft core through an experimental investigation, using a
photoelastic set up, and an analytical investigation using the high-order theory of
sandwich panels. Narrow sandwich beams with skins made from transparent epoxy were
used for the photoelastic experiments, and with cores made from transparent and low
modulus polyurethane (PU) rubber. The sandwich beam was loaded in a three-point
bending scheme. The two principally different approaches that were used to analyze the
localized bending problem in sandwich panels were the elastic foundation concept and
the high-order sandwich panel theory. The study showed that the high-order theory of
sandwich panels provides accurate results compared with the experimental results.
The elastic foundation approach was explored by different authors. Frostig and
Baruch (1990) and Frostig et al. (1991) used the concept of elastic foundation approach
15
by means of superposition while Weissman-Berman et al. (1988) and Meyer-Piening
(1989) suggested the use of an elastic foundation model for the loaded face of the
sandwich panel. In the analyses, the core material modeling was presented by the
Winkler foundation model, basically through a continuously distributed tension-
compression springs. Thomsen (1992, 1993) investigated local bending of sandwich
beam skins by using elastic foundation approach. The elastic foundation approach
adopted to include the shear interaction between the skins and the core involved the use
of two-parameter elastic foundation models. However, these elastic foundation models
neglected interactions between the top and bottom skins.
The soft core nonlinearity was investigated by many researchers. Zhu et al (1997)
and (1998) determined the effect of nonlinear relations between the material type and its
density on the shape of the stress-strain curve. Shen et al (2004) used the high-order
sandwich panel theory (HSAPT), to predict the bending behaviour of soft core sandwich
beams subject to localized loads. For design purposes, the study proposed the use of
correction factors for both the bending and shear components of the classical deflection
formula for sandwich beams. A numerical study was performed to study the influence of
changes in the geometry and mechanical properties of soft core sandwich beams on the
proposed correction factors. The study showed that the most important practical
consequence was that the correction factors offered a simple way of calculating the
bending deflections in soft core sandwich beams subject to quasi-static concentrated
loads.
Frostig et al (2005) presented the governing equations of a sandwich panel, which
has a transversely flexible core with negligible flexural rigidity, including large
16
displacements. The study took into account the nonlinearity, not only in the core, but also
in the skins. As such, the resulting governing equations of the core were rather complex,
and the stress and displacement fields in the core were described through a set of
nonlinear partial differential equations without a general analytical solution. The study
adopted two simplified models due to their ability to analytically describe the stress and
the deformation fields in the core. The first model assumed that the core is undeformed
and its kinematic relations are linear while the second one takes into account the
deformed core and uses simplified non-linear kinematic relations. A comparison with a
FEA program results was performed in terms of load versus mid-span displacement of
the loaded skin of a sandwich panel loaded in a three-point bending configuration. The
FEA runs were associated with numerical difficulties and poor convergence because of
the large distortions of the core while approaching the panel failure load. According to
the study, at the failure load level, the sandwich panel loses its composite action because
of buckling of the compressed face sheet, and its flexural resistance drops to that of the
isolated face sheets in the case of the simplified models.
2.4 Sandwich Panels with Cellular Cores
The previous section addressed sandwich panels with solid flexible cores. This section
will discuss sandwich panels with cellular core structures, which are also flexible to a
much lesser degree than the solid foam cores. Cellular cores may be honeycombs,
corrugated structures, truss cores, Z-shaped cores, C-shaped cores, or I-shaped cores.
Allen (1969) and Plantema (1966) presented the fundamental models of sandwich
structures, assuming that the core is incompressible in the out-of-plane direction and does
not have any bending rigidity, the skins only have bending rigidity, and the core has only
17
shear rigidity. This approach is good for sandwich structures with incompressible cores,
which are Antiplane, like honeycomb cores.
Petras and Sutcliffe (1999) presented the failure mode maps for honeycomb
sandwich panels. They showed that indentation failures occur in GFRP/Nomex
honeycomb beams, at which the contact pressure is assumed to be transferred directly to
the core, leading to core failure when the core compressive strength was inadequate. The
high-order sandwich beam theory (HOSBT) was used to analyze the indentation
behaviour of sandwich beams with properties typical of Nomex honeycomb core and
composite skins to establish the displacement fields, as shown in Figure 2.3. According to
the study, classic calculations were appropriate for the failure analysis of these materials
and although there are some quantitative differences, the results were encouraging and
justify the use of the model in an analysis of indentation behaviour of such beams.
Meraghni et al (1999) determined the tubular and honeycomb cores rigidities by
several complementary investigations, namely numerical, analytical and experimental
approaches. The study showed that the influence of core wall thickness on the rigidity is
essential whereas the total thickness of the core is not very important on the equivalent
rigidities.
Failure of sandwich honeycomb structures under indentation loading was
considered by Petras and Sutcliffe (2000). A failure criterion for Nomex honeycombs
subjected to a combined compressive and shear stresses was determined using biaxial
tests. According to the study, the indentation failure load of the sandwich beam due to
core failure can be predicted by combining this work with a theoretical calculation of the
18
stress distribution in the core due to indentation loading, found from a high-order
sandwich beam theory (HOSBT).
According to a study by Kocher et al (2002), the advantages of a truss-reinforced
panel could include an enhanced delamination control in addition to face stability. Also
the skin local stability can be significantly improved by using a truss foam-filled core,
particularly if the angle between the pins (i.e. the truss members) and the facings is large.
Kocher et al (2002) stated that the highest predicted local stress concentration was
observed in the zone of the truss-facing junction of the truss-core sandwich panels. Also,
the high values of the material principal stresses of the lamina indicated that the initial
failure is more likely to occur in the form of pin penetration through the thickness, rather
than delamination between the layers of the facing, as shown in Figure 2.4. The study
showed that the buckling load for the pin occurs at a higher load than that corresponding
to the failure of facings. According to the study, the main advantage of the foam in the
cores was likely related to corrosion prevention, that is, the protection of the steel pins;
because the study showed that there was no significant benefit to the strength as a result
of adding the foam.
Markaki and Clyne (2003) evaluated the stiffness of a thin ultra-light stainless
steel sandwich beam with fibrous core. The study assumed straight fibres aligned
perpendicular to skins and embedded at both ends. The study shows that the beam
stiffness was significantly lower than expected from simple bending theory. This was
attributed to the low through-thickness stiffness of the core and also to a low resistance to
shear stresses. The study stated that increased fibre diameter and possibly an alteration to
the fibre sectional shape are the parameters that affect the core behaviour the most.
19
Kooistra et al (2004) studied different topologies of lattice materials, both experimentally
and theoretically. The lattice materials were basically sandwich systems consisting of a
three-dimensional (3D) network of fully triangulated solid (or hollow) rods. According to
the study, the compressive response of the tetrahedral truss core sandwich panels was
found to be very sensitive to the manufacturing process.
The optimal dimensions and the minimum weight of sandwich panels with
prismatic cores have been evaluated by Valdevit et al (2004). The study showed that the
corrugated core panel (Figure 2.5) performed best when loaded longitudinally because
the performance is limited by core plate buckling, rather than beam buckling. Because of
the greater buckling resistance, the corrugated core panel had its best performance in the
longitudinal orientation.
Zhou and Stronge (2005) studied the through-thickness Youngs modulus, initial
compressive yield stress and fully plastic compressive stress of fibrous core structure, as
shown in Figure 2.6. The calculation process was based on the geometrical properties of
the fibres, the volume fraction of the fibres, inclination angle of the fibres, the
dimensionless initial curvature of the fibre and its material properties. The study showed
that the fibre inclination angle had an effect on the core properties and a large angle of
inclination can increase the through-thickness Youngs modulus, but this is at the
expense of reducing the through-thickness shear modulus. According to the study, the
range of fibre angle of inclination depended on the specific application of the panel.
To optimize the sandwich panel dimensions, the minimum weight design of
sandwich plates comprised of truss cores faced with solid sheets of the same material was
investigated by Liu et al (2006) for various prescribed combinations of bending, shear
20
and compression, as shown in Figure 2.7. A homogenization approach was adopted, with
the periodic 3D truss cores replaced with 2D effective solid cores. The optimization was
subjected to the constraints that no failure mechanism was active, including overall
buckling, face sheet buckling/wrinkling, face sheet yielding, core member yielding and
buckling. The predictions using the homogenized 2D computational models were
checked against those obtained using a 3D FEA models. Four different sandwich panels
with different unit cell topologies were separately studied: (a) plagihedral pyramidal; (b)
tetrahedral; (c) pyramidal; (d) 4-rod. For the pyramidal core panel, the truss members
were hollow, whereas those in the rest of the panels were solid. The study showed that
the agreement between 2D and 3D model predictions for out-of-plane behaviour was not
good, due to the fact that the 2D homogenized model was based on the effective single-
layer sandwich approaches, which considers the in-plane deformations of the face sheets
but ignores their out-of-plane deformations. Also, to ensure the same shear resultants
between face sheets and truss core, closed form solutions should be formulated (Frostig,
2003).
A study on composite honeycomb sandwich panel structure was introduced by He
and Hu (2007). According to the study, the satisfying weight condition of the honeycomb
core was 5066.7% of the whole honeycomb panel weight, which achieved the maximum
flexural rigidity and bending strength of the sandwich panel. Fan et al (2007) investigated
carbon fibre reinforced Kagome lattice grids which were made by the interlocked
method. The grids were sandwiched by two carbon fibre reinforced laminates to assemble
sandwich panels. In-plane compression, out-of-plane compression and three-point
bending tests were carried out to study the mechanical behaviours of the grids and
21
sandwich panels. According to the experiments, a high specific strength and specific
stiffness can be achieved in comparison with other cellular materials, rendering the
carbon fibre reinforced grids weight efficient. Also, the detected failure modes, as shown
in Figure 2.8, showed that the failure process in the carbon fibre reinforced grids and the
assembled sandwich panels could be ductile to some extent, and if that is achieved, the
loads can be effectively transferred to other parts of the structure if one part is disabled.
2.5 Mechanical Properties of Sandwich Panels
Berthelot and Lolive (2002) presented a modeling procedure for evaluating the non-linear
behaviour of foam sandwich structures, subjected to mechanical loading. The non-linear
behaviour of PVC foams was characterized in shear, compressive and tensile tests.
According to the study, the shear and compressive properties were essential for
determining the overall flexural behaviour of foam sandwich structures, since the loads
applied were transmitted between the skins by transverse shear of the foam core. Also,
good properties under transverse compression decrease the indentation effects in the
vicinity of localized loads.
New approaches based on tri-axial weaving of wires to create ideal trusses (i.e.,
tetrahedral and Kagome trusses) have been presented by Lim and Kang (2006). The
mechanical properties of sandwich panels with the truss cores fabricated using the new
approaches under compression and bending loads were analyzed by elementary beam
theory and experiments. According to the study, regardless of loading types, failure will
be due to plastic buckling of the truss members (i.e., the struts of the core or the bars of
the face sheet) because of the high slenderness of the truss members and the ductility and
low yield stress of the material. The study showed that for compression load schemes,
22
Kagome truss core fabricated from wires exhibits double peaks on the loaddisplacement
curve which makes it possible to absorb much more energy during deformation.
Lascoup et al (2006) studied the mechanical effect of adding stitches to the
sandwich panel on the behaviour in terms of the ultimate strength and modulus.
According to the study, regardless of the type of mechanical testing considered, the gain
brought by the presence of the stitches was significant for the moduli and the maximum
stresses. Even if the addition of transverse reinforcement tends to increase the mass of the
panel, the specific properties (which take into account this increase) remain improved
compared to unstitched panels. The study also showed that the structural parameters
defining the stitches (stitching angle and step) and the components of the initial sandwich
(nature of the skin and the core) directly influence the mechanical performances.
According to the study, the parameter that had the most important influence is the
stitching step which determines the stitching density.
The mechanical properties balance in a novel X-Cor and K-Cor sandwich panels
was investigated by Marasco et al (2006). The panels tested had inserted fibres in the
panel core Z direction (Z-pinned). For the set of materials used in this study, Nomex
honeycomb sandwich panels were more efficient than the pinned-core sandwich panels in
terms of ultimate strength. However, when the sandwich panel stiffness is the key design
parameter, the choice of a pinned core would be appropriate. The study showed that if the
absolute weight is the overriding issue, for the configuration of the X-Cor core sandwich
panels tested, the Rohacell foam may be removed without a significant performance
penalty, except for situations where compressive loads are likely to occur.
23
The actual mechanical behaviour and the possible failure mechanisms of two
types of sandwich structures widely used in marine constructions (made of fibre-glass
laminate skins over PVC foam or polyester mat core) were studied by Russo and
Zuccarello (2007). The study showed that both types of sandwich structures exhibit a
significant non-linearity, with different behaviour under tensile, compressive and shear
loading. Also, in the presence of prevalent bending and shear loading, the theoretical
prediction of the possible failure modes (skin tension failure or core shear failure) was in
general not accurate. Also, the corresponding theoretical strength predictions were in
general not accurate. In particular, if the failure of the sandwich structure was caused by
the skin tension rupture, then the theoretical predictions underestimate the actual strength
with errors up to 15%. If the failure of the sandwich structure was caused by the core
shear rupture, then the predictions overestimate the actual strength with errors up to
100%.
The strain rate effects in phenolic composites and phenolic-impregnated
honeycomb structures were introduced by Heimbs et al (2007). The study tested,
dynamically, the phenolic GFRP and phenolic-impregnated aramid paper honeycomb
specimens in order to analyse the effect of loading rate on the mechanical behaviour. The
strain rate effect for the composite material was remarkable and lead to increasing
Youngs modulus, tensile strength, tensile failure strain, shear modulus, and shear
strength. Also, the study showed that for the honeycomb specimens, higher loading rates
lead to an increase of the stress-curves. In this study the plateau stress was primarily
analyzed and compared, since the densification point was not influenced by strain rate
and the peak stress could not finally be evaluated because of the strong influence of data
24
filters. The data presented in the study can be used for rate-dependent material modelling
for dynamic FE-simulations of phenolic GFRP/Nomex sandwich structures.
Reis and Rizkalla (2007) studied the material characteristics of 3-D FRP
sandwich panels with through-thickness fibres passing through the core. Based on the
experimental work presented in the study, the behaviour of the face sheets under tension
was bi-linear which could be caused by the presence of the fibres in the perpendicular
direction. Also, the behaviour was independent of the presence and the amount of
through-thickness fibres embedded in the face sheet. The reduction in the stiffness was
approximately 33% for all face sheets tested in this study. The study showed that
increasing the density of through-thickness fibres creates waviness among the fibres and
reduces the strength of the face sheets significantly. There was a 25% decrease in tensile
strength of the face sheet by increasing the amount of through-thickness fibre insertions
from 1.25 to 2.5 per cm2. The study showed that the thickness does not have any
significant effect on the initial core shear modulus, however, increasing the thickness
reduces the shear strength considerably.
2.6 Buckling and Skin Wrinkling of Sandwich Panels
The compression skins of sandwich panels are sometimes subjected to a particular kind
of instability, typically described as wrinkling, in which the compressive stresses induced
in the skins exceed the buckling stresses. Studying the wrinkling behaviour is very
important as it could control the ultimate strength of the panels.
Combined load buckling analysis was performed by Ko and Jackson (1991) on
truss core and honeycomb core sandwich panels having the same specific weight. The
combined load buckling interaction curves were generated for these two types of
25
sandwich panels with different aspect ratios and different core orientations. According to
the study, the infinite number of simultaneous characteristic equations to calculate the
shear buckling load components could be cut off at 12 (i.e., the order of matrix is 12) to
yield sufficiently accurate eigenvalue solutions for shear buckling loads. Also, for square
sandwich panels of both types, only the symmetric buckling will take place. The study
showed that for a panel aspect ratio greater than one, both symmetric and antisymmetric
buckling will take place, at which the antisymmetric buckling will occur when the
combined loading is shear dominated. Also the square-shaped sandwich panel (any type)
has the highest combined load buckling strength.
Another rigorous buckling analysis of sandwich beams with soft cores, with
flexibility effects of the core on the overall behaviour taken into account, was presented
by Frostig and Baruch (1993). The analysis was based on small deflections and moderate
rotations. The study included buckling analysis of a generally constructed simply
supported beam. The study determined two eigenvalues, a symmetrical one (i.e. the upper
and lower skins are oppositely displaced) and an asymmetric one in which the two skins
have identical displacements in the same direction. According to the study, the
symmetrical mode shape was usually associated with a high number of waves, leading to
a wrinkling type of buckling. The asymmetrical mode corresponds to a buckling mode
shape with one half wave shape.
Sleight and Wang (1995) used three numerical methods to analyze the buckling of
a debonded aluminum face skin of a sandwich panel with a foam core. The Rayleigh-Ritz
and finite-difference methods were used to study a 1-D beam-on-elastic-foundation
model that represents a strip of a sandwich panel with a through-the-width debond. The
26
results were compared with the buckling loads calculated from a 2-D finite-element
analysis using 4-noded quadrilateral plane-strain elements, as shown in Figure 2.9.
According to the study, two parametric studies were performed to evaluate the effects of
the elastic foundation stiffness and debond length. Results showed that the buckling load
increases with increasing core foundation stiffness and decreases with increasing debond
length. Both the Rayleigh-Ritz and finite-difference methods assumed a Winkler core
foundation and, thus, ignored the Poisson's ratio and longitudinal stiffness effects of the
core material.
Frostig (1998) presented a systematic stability analysis of sandwich panels with a
flexible core that uses the high-order theory. Pre-buckling and buckling governing
equations and the corresponding boundary conditions have been derived. According to
the study, the imperfection analysis equations were defined and were only used to
determine the magnitude of the stresses involved. The study showed that the geometrical
matrix is a diagonal matrix with two non-zero terms. Thus, only two eigenvalues and two
eigenvectors (mode shapes) exist for every mode number. In the case of a stiff core the
global mode shape with low mode number governs while for soft cores the critical load
usually correspond to the local (wrinkling) mode shape. A study using the same
analytical model on the boundary condition effects in buckling of soft core sandwich
panels was presented by Sokolinsky and Frostig (1999). The study showed that complex
responses such as wrinkling as well as the mode interaction that appears as local and
localized buckling patterns are caused by the soft core nonlinear displacement patterns
through its thickness.
27
Niu and Talreja (1999) presented an expression for the wrinkling mode of the
sandwich panel with continuous isotropic linear elastic core. The expression was
depending linearly on a case parameter which defined to specify the three cases of
wrinkling, single-sided skin wrinkling (case 1), antisymmetrical wrinkling, in-phase
(case 2), and symmetrical wrinkling, out-phase (case 3). It is shown that the stresses in
all three cases are almost identical for short wavelength wrinkling and can be expressed
by a single simplified analytical expression; however, they may differ significantly in
moderate and long wavelength wrinkling, and for these cases they are given by simplified
analytical expressions. According to the study, the critical stress for the in-phase
wrinkling was always the lowest among all cases; thus, only the in-phase wrinkling case
need to be considered for a sandwich panel with equal faces under in-plane compression.
Also, the study showed that the differences among wrinkling stresses for the three cases
were extremely small in short wavelength wrinkling, in which case there is no need to
classify wrinkling modes in the three different cases. Also, a simple unified wrinkling
stress expression has been derived, which has good accuracy in the short wavelength
range. The classical Winkler model and the two-parameter model are found to be
adequate when the core thickness ratio is relatively small.
Sokolinsky and Frostig (2000) presented a geometrically nonlinear analysis of
the longitudinally loaded sandwich panels with a flexible core, based on the high-order
sandwich panel theory (HSAPT). According to the study, the nonlinear analysis of the
various panels revealed that wrinkling of the skins does not necessarily mean that the
panel as a whole has buckled since the sandwich panel is a compound structure. Also, the
localized buckling modes were found in some cases to be the critical ones rather than the
28
usual sinusoidal buckling patterns. It is further shown that variations in the geometry,
boundary conditions and mechanical properties of the panel constituents can lead to a
qualitative shift in its nonlinear response from an imperfection-sensitive, shell-wise
response, to an imperfection-nonsensitive, plate-wise one.
Rectangular orthotropic FRP sandwich panels were tested for buckling in uni-
axial compression by Roberts et al (2002). The study showed that bifurcation in the load
versus engineering strain curve was noted in all cases. For all six sandwich panels tested
within the study using balsa core, the type of failure was easily identified as face sheet
delamination followed by core shear failure. Also, for all six PVC foam core sandwich
panels tested, the type of failure consisted of core shear failure with little or no face sheet
delamination. According to the study, in the failed balsa core panels there was little or no
evidence of balsa remaining on the FRP face sheet. On the other hand, in the PVC foam
core panels there were ample amounts of foam left on the FRP face sheet. The study
concluded that although the buckling loads for panels with foam core were not as high as
those for the balsa core panels, PVC foam core bonding to the FRP face sheets was
superior to balsa core bonding.
The qualitative features involved in the buckling response of sandwich panels
with a flexible core were analyzed numerically by Sokolinsky et al (2002), using the
linear buckling and geometrically nonlinear analysis based on a high-order sandwich
panel theory (HSAPT). According to the study, the ordinary (fully bonded) sandwich
panel behaved as a compound structure in which the localized, overall or interactive
forms of the response can take place and depend on the geometry, mechanical properties,
and boundary conditions of the structure. The study showed that the localized wrinkling
29
pattern that developed between the points of application of the concentrated loads was
not associated with the turning points on the equilibrium path.
Falzon and Hitchings (2003) investigated experimentally and numerically by FEA
the post-buckling blade-stiffened composite panel which was loaded in uni-axial
compression until failure. According to the study, in the post-buckling regime, the panel
was observed to undergo a sudden mode-switch to a higher mode-shape. Figure 2.10
shows the load-deflection responses for each model and the deformed shape of the FEA
model.
Frostig and Simitses (2004) derived the similarity conditions for the buckling
analysis of a compressed wide column which was symmetric, with and without geometric
imperfections. According to the study, both wrinkling and global buckling were
predicted. In the case of wrinkling, full similarity yields that the buckling modes for both
model and prototype were identical. The same was true for the case of global buckling
but the proof was only achieved numerically.
Mania (2005) studied the stability problem of a sandwich trapezoidal plate with a
rigid composite core. The study showed that the FEM solutions obtained for a large
number of models agreed to a large extent with the results obtained analytically and
verified the accuracy of the coordinate transformation method. Also, the buckling force
for trapezoidal laminates with the same lay-up increases with the increase of the skew
angle and the buckling load was sensitive to the value of the lamination angle.
2.7 Delamination and Skin Debonding of Sandwich Panels
The high-order theory of sandwich panels was used by Frostig (1992) to analyze the
bending behaviour of delaminated sandwich beams with flexible core, with or without
30
contact, at the upper or lower interfaces. According to the study, the normal stresses
developed at the crack (delamination) tips were finite but very large compared to those in
the loaded zone and at the support region. The study showed that these stresses may be in
tension as well as in compression, depending on crack tips, location of delamination, and
loading pattern. Also, the length of the delamination was one of the important parameters
which affect the panel behaviour. The deflections and the stresses in the various parts of
the sandwich beam increased as the length of the crack increased, which under some
circumstances may lead to the collapse of the structure.
Frostig and Thomsen (2005) studied the non-linear behaviour of delaminated or
debonded unidirectional sandwich panels with partial contact and a flexible core. The
analysis, which was based on the intermediate class of deformations in the face sheets,
took into consideration the partial contact between the debonded face sheet and the core.
The analysis was based on the high-order sandwich panel theory (HSAPT). The study
presented the governing equations for the various regions for isotropic face skins. The
study showed that a full contact type of delamination transforms into a partial contact
area with buckling of the compressed skins, as the load is increased and it is associated
with large displacements and stresses.
Schwarts-Givli et al (2007) studied the high-order nonlinear contact effects in
cyclic loading of delaminated sandwich panels with a soft core. Two types of
nonlinearities were considered. The first one was associated with geometrical nonlinear
effects of the face-sheets while the second type of nonlinearity was related to the contact
behaviour of the delaminated surfaces at the debonded region. A quadratic and a cubic
polynomial description of the vertical and longitudinal displacement fields through the
31
depth of the core, respectively, were adopted in the analysis. The study showed that this
formulation fully satisfied the static equilibrium equations through the thickness of the
core. The study also showed that the geometrical nonlinear effects of the skins amplified
the reduction in the stiffness that was associated with the formation of the delaminated
region.
2.8 Numerical Analysis of Sandwich Panels
Some of the previously used computational models are based on incompressible cores.
However, researchers as well as engineers in the industry use these models
indiscriminately for any type of sandwich panel, including the modern ones that are made
of a soft core. There are also computational models for sandwich panels with a
compressible soft core that consist of the spring models, with one or two parameters.
Recently, the high-order approach has been used for all types of sandwich panels,
compressible and incompressible. The extent to which theses classical approaches for
compressible and incompressible cores are valid for modern sandwich panels is one of
the main issues debated (Frostig, 2003).
Analytical and numerical comparisons between the high-order sandwich panel
theory (HSAPT) and the simplified models involved in the analysis of unidirectional
sandwich panels used for incompressible (splitted rigidity models) and compressible core
(elastic foundation models) were presented by Frostig (2003). According to the study, the
accurate models were rigorous formulations of the governing equations within the
assumption that the longitudinal rigidity of the core was neglected. The shear pattern was
assumed to take a specific shape based on assumptions that the section plane takes a
specific zigzag shape. In the model due to Allen (1969), this plane was vertical and
32
passes through the mid-planes of the face sheets and the core, thus yielding only bending
in the face sheets. In Frostig (1992) model, this zigzag line is assumed to be vertical
only within the core, thus causing bending as well as extension or contraction in the face
sheets.
Sokolinsky et al (2003) performed experimental and analytical study of nonlinear
bending response of sandwich beams. Four-point bending tests were carried out for
sandwich specimens with aluminum face sheets and a PVC foam core. The study
analyzed the mechanical response of the specimens with the aid of the classical sandwich
theory, and linear and geometrically nonlinear HSAPT. According to the study, the
analytical predictions for the vertical displacements of both linear and nonlinear HSAPT
were in excellent agreement with the experimental results, as shown in Figure 2.11. Also,
the classical sandwich theory underestimated the vertical displacements of the sandwich
beam specimens by more than 20%. The classical theory also failed to accurately predict
the bending deformation of soft core sandwich beams under concentrated loading
because of the inability to model the core indentation.
Demiray et al (2004) developed a finite plate element for sandwich panels with
transversely compressible core as shown in Figure 2.12. The element was based on an
effective multilayer approach where the Kirchhoff-Love hypothesis is adopted for the
face sheets whereas a second and third order power series expansion was used for the
core displacements. The finite element was triangular. According to the study, for the
determination of the element matrices, the sandwich element was decomposed into its
three principal layers. The face sheets were modelled using the standard formulations of
the discrete Kirchhoff triangular (DKT) element for the transverse deflections and the
33
constant strain triangular (CST) element for the in-plane deformation. For the core layer,
the three-dimensional displacements were expressed in terms of the face sheet
displacement fields as defined by the underlying shell theory. The study showed that the
local face wrinkling instability enabled by the transverse compressibility of the core has a
significant effect on the overall behaviour of sandwich plates.
A three-dimensional finite element buckling analysis of debonded sandwich
panels was presented by Avils and Carlsson (2006). The three-dimensional linear FE
modelling was conducted to examine the local buckling behaviour of foam-cored
sandwich panels containing a face-core debond. The study showed that in-plane
compressive loading produced local instability (buckling) failure of the debonded face
sheet. Also, it was found that the buckling load decreases with increased debond size and
reduced core modulus, as shown in Figure 2.13. Also, the sensitivity of the buckling load
to core modulus was found to be more pronounced for small debonds. According to the
FEA results, panels with square debonds buckled at lower loads than those with circular
debonds of the same characteristic dimension.
Another way of analysis of sandwich panel is to convert the 3D sandwich panel to
2D plate by determining the sandwich panel elastic constants. Libove and Hubka (1951)
determined the elastic constants for corrugated core sandwich panels consisting of
corrugated sheet fastened between two face sheets. Formulas and charts were presented
for the evaluation of these elastic constants. The study considered both the symmetrical
and unsymmetrical types of corrugated-core sandwich. Also, the extensions of the
existing sandwich-plate theory required to make it strictly applicable to the
unsymmetrical type are indicated. The study showed that the proposed expressions were
34
in a good agreement with the experimental result which were done to check these
formulas of the elastic constants.
Ha (1990) described the development of a range of possibilities of the FE analysis
of sandwich panels, varying from a simple element with only three degrees of freedom
per node to a complex general one with up to 3(n+1) degrees of freedom per node where
n is the number of layers through the element. According to the study, it is emphasized
that for a symmetric sandwich structure with flexible core and thin faces, a simplified
formulation with translation and rotation degrees of freedom should suffice.
Burton and Noor (1996) studied the continuum models for the sandwich panels
with honeycomb core. The panels had square-cell honeycomb core and simply supported
edges. According to the study, the sandwich core and face sheets were modeled by using
three-dimensional solid elements, and two-dimensional plate finite elements. The results
of the finite element models were compared with those obtained by using HSAPT, with
the core replaced by an effective (equivalent) continuum. The study showed that the
predictions of the continuum models with test-based equivalent properties were lower
than those of the finite element model.
Noor et al (1996) reviewed the recent developments in the computational
modeling of sandwich plates and shells. The study focused on the computational models,
characterization of sandwich properties, and sensitivity of the sandwich response to
variations in material and geometric parameters. The study also classified the various 2-D
FE approaches in the following three classes: The first class was global approximation
models at which the sandwich is replaced by an equivalent plate or shell element, with
approximated strains, displacements and stresses through the element thickness. The
35
second was discrete layers models at which the sandwich is divided into three or more
layers. For each layer, approximations are made for the behaviour in the thickness
direction. Finally the third was predictor-corrector approaches where iterative procedures
are utilized and the information gathered in the first phase of the analysis (predicting
phase) is used to correct the model and improve the response.
Bau-Madsen et al (1993) investigated the influence of geometrically non-linear
effects on the behaviour of a laterally loaded clamped sandwich plate. According to the
study, the load-strain behaviour became non-linear at a lower load than that in the load-
deflection behaviour. Also, there was a difference in the non-linear behaviour of strains
on the tension face and on the compression face.
Elastic stiffness properties and behaviour of truss-core sandwich panel were
presented by Lok and Cheng (2000). Transforming the 3D sandwich panel into an
equivalent orthotropic 2D thick plate continuum has been presented. Also, the equivalent
elastic constants relating to the transformation were presented. Figure 2.14 shows the
equivalent orthotropic thick plate for the truss-core sandwich panel. The study showed
that panels with vertical truss-core webs possess weak shear stiffness whereas shear
deformation can be ignored for panels with triangular core webs. The study did not
include buckling and local bending of the facing plates and core webs. A closed-form
solution was introduced using the derived stiffness constants to evaluate the maximum
deflection of the continuum. The calculated results were in good agreement with
numerical 3D finite-element results.
Chang et al (2005) studied the bending behaviour of corrugated-core sandwich
plates, as shown in Figure 2.15. The study included the effects of the geometric
36
parameters of the corrugated-core sandwich plates with various boundary conditions on
the plate behaviour and strength. The study has confirmed some experimental results
obtained by Lok and Cheng (2000) which lacked supported from 3D FEM and the other
analytical solutions. The analytical solution presented by Chang et al (2005) agreed with
all the experimental investigations, both for the deflections and the bending moments,
presented in Lok and Cheng (2000). The study developed recommendations for the
selection of the geometric parameters of corrugated core sandwich plates. It showed that
the plate became stronger when lower ratios of some geometric parameters such as (full
thickness / core thickness) and (core thickness / skin thickness) were used.
Boni at Mller (2008) studied laterally supported sandwich panels subjected to
large deflections. The study used the finite element (FE) models to predict the global
behaviour of simply supported sandwich panels fixed by bolts on the two longitudinal
edges or on all four edges, as shown in Figure 2.16. The tested panels were subjected to
distributed loads and developed large deflections. They were analyzed experimentally
and numerically. According to the study, a good correlation was observed between
experimental and numerical results for the displacements of the panels. Also, the
proposed two-dimensional FE approach provided a good trade-off between the level of
accuracy and computational efficiency.
37
(a)
(b)
Figure 2.1 Micrographs showing the structure of man-made cellular materials (a) flexible
open-cell polyurethane foam; (b) flexible closed-cell polyethylene foam.
[Gibson and Ashby, 1989]
38
Figure 2.2 (a) Geometry, (b) internal resultants, (c) sign convention,
and (d) deformations of sandwich panels
[Frostig et al, 1992]
39
All contour values are in mm and the scaling in both plots is the same.
Figure 2.3 Vertical displacement fields in the core produced by (a) Surface Displacement
Analysis (SDA) and (b) HOSBT model.
[Petras and Sutcliffe, 1999]
40
(a) Example of finite-element (b) Buckling of pin in model with = 60,

analysis model for = 60 with buckling pressure = 7.42 MPa
Figure 2.4 Buckling of truss core sandwich panel
[Kocher et al, 2002]
Figure 2.5 Corrugated-core cellular panels.

[Valdevit et al, 2004]
Figure 2.6 Fibrous sandwich panels.

[Zhou and Stronge, 2005]
41
Figure 2.7 Typical lattice truss topologies: (a) octet truss, (b) tetrahedral lattice
truss, (c) lattice block, (d) pyramidal lattice truss and (e) 3D kagome.
[Liu et al, 2006]
Figure 2.8 Bending behaviour of the sandwich panel:

(a) elastic deformation; (b) buckling and debonding; (c) core shear.
[Fan et al, 2007]
42
Figure 2.9 Comparisons of normalized buckling load results from different methods for
various debond ratios.
[Sleight and Wang, 1995]
(a) Load versus end-displacement (b) Panel deformation at the point of

curves. collapse (Load = 601 kN)
Figure 2.10 Finite element prediction and deformation of axially loaded panels
[Falzon and Hitchings, 2003]
43
Figure 2.11 Comparison of load-deflection predictions with the experimental results

for four-point bend test of sandwich beam with a soft foam core at mid-span
[Sokolinsky et al, 2003]
Figure 2.12 Sandwich element, geometry and decomposition.

[Demiray et al, 2004]
44
Figure 2.13 Local buckling loads determined using FEA and experimentally
for sandwich panels with circular debonds:
[Avils and Carlsson, 2006]
Figure 2.14 Equivalent orthotropic thick plate for truss-core sandwich panel
[Lok and Cheng, 2000]
Figure 2.15 Corrugated-core sandwich panel

[Chang et al, 2005]
45
Figure 2.16 Sandwich panel test setup and load-deflection response at mid-span
[Boni and Mller, 2008]
46
Chapter 3 Material Testing Program
CHAPTER 3: MATERIAL TESTING PROGRAM
3.1 Introduction
The composite panel studied in this thesis consists of a polyurethane foam core
sandwiched between two layers of glass fibre reinforced polymer (GFRP) skins, along
with GFRP ribs connecting the skins. Given the unconventional nature of these materials,
it was decided to carry out a comprehensive material testing program of the constituents.
In total, 40 coupon tests were carried out. The scope of the study was to develop the full
stress-strain responses of the GFRP skins and ribs as well as the polyurethane foam core
under different loading conditions, namely tension, compression and shear. ASTM
standards were used to determine all material properties. Also, the statistical methods
stated in the ASTM standards were used to analyze the results.
The sandwich panels used in the current study were fabricated by Comtek
Advanced Structures, Inc, Burlington, Canada. This chapter provides a summary of the
material tests on both the polyurethane foam core of two different densities and the
GFRP, including detailed description of the coupon specimens, test setups, procedures,
test results and failure modes. All GFRP coupons were cut from the skins of
prefabricated sandwich panels whereas all foam coupons were cut from the cores of the
same panels.
47
3.2 Experimental Program
3.2.1 Tests on Glass Fibre-Reinforced Polymer (GFRP)
The face skins were fabricated from 54 oz 3 weave E-glass 2022 silane sized and CoPoxy
4281A resin with Copoxy 4284 hardener, and had an average thickness of 1.6 mm. The
manufacturer data of the E-Glass fabric used (3tex, inc.) are provided in Table 3.1. A
schematic of the fibre reinforcement structure is shown in Figure 3.1. Also, the physical
properties of the epoxy resin used to fabricate the GFRP composite are listed in Table
3.2. The GFRP composite (cured lamina) material properties given by the manufacturer
are listed in Table 3.3.
3.2.1.1 Tension Coupons
Two main groups of GFRP coupons were prepared to examine the tensile properties of
GFRP skins, namely, five coupons cut from the longitudinal direction of the skin and five
cut in the transverse direction. ASTM D3039/D 3039M-00 recommends that the
minimum length of the specimen be taken as the gripping length at both ends plus two
times the coupon width plus a gage length. It also recommends that the width is to be
taken as needed. As such, the final coupon dimensions were taken as 250 mm long and
25 mm wide, except for one coupon in each direction that was 12.5 mm in width, to
check if width has any significant effect on the properties. No end tabs were used, which
is permitted by the standards. The coupon thickness was that of the skin. Figure 3.2(a)
shows the general dimensions of the tension coupons, whereas Table 3.6 shows the
dimensions of each specific coupon.
Tension tests were performed using an Instron Model 1350 testing machine with
wedge-type mechanical grips. The speed of the test was set to provide a constant strain
48
rate in the gage length of 0.01 min-1 as recommended by the standards, which is 1.5
mm/min. The longitudinal strain was measured using two electric resistance strain gages,
5 mm long, having a resistance of 119.8 0.2 , to measure the longitudinal strains, one
on each side of the coupon. Test data, including load and stroke of the Instron machine
were recorded using a Vishay System 5000 Data Acquisition System (DAS). The DAS
receives the load and stroke of the testing machine through a high-level input card with
an accuracy of 10 mV. Strain gages readings were transmitted through a strain card
with an accuracy of 5 mV. Test data were collected and stored using Strain Smart
computer software.
3.2.1.2 Compression Coupons
Two main groups of GFRP coupons were prepared to examine the material compressive
properties of GFRP, namely five coupons cut from the longitudinal direction of the skin
and five cut from the transverse direction. ASTM D3410/D 3410M-03, recommends that
minimum length of the specimen to be taken between 140 mm and 155 mm while the
width be from 6 mm to 25 mm and the gauge length ranges from 12 mm to 25 mm. As
such, the final coupon dimensions were taken 142 mm long and 24 mm wide, with a
gauge length of 12 mm. End taps were not used as recommended in the standards. The
coupon thickness was the same as that of the skin. Figure 3.2(b) shows the general
dimensions of the compression coupons, whereas table 3.7 shows the dimensions of each
coupon.
Compression tests were performed using an MTS testing machine with an MTS
685.60 hydraulic grip unit. The speed of the test was set to provide a constant strain rate
in the gage length as recommended by the standards, 0.01 min-1, which is 0.12 mm/min.
49
The longitudinal strain was measured using two electrical resistance strain gages, 5 mm
long, having a resistance of 119.8 0.2 , one on each side of the coupon. Strain, load
and stroke data were collected using the same procedure described for the tension
coupons.
3.2.1.3 Shear Coupons
One group of five GFRP coupons was prepared to examine the shear properties of GFRP
ribs. ASTM D3518/D 3518M-94 (reapproved 2001), recommends that minimum length
of the coupons be the gripping length at both ends plus two times the coupon width plus a
gage length. As such, the final coupon dimensions were taken 250 mm long and 24 mm
wide, and without end tabs. The coupon thickness was the same as the skin. ASTM
D3518/D 3518M-94 (reapproved 2001) is based on establishing the shear properties
using tension coupons cut at 45 degrees from the [0/90] cross-ply laminate. Therefore,
the coupons discussed here were cut in the diagonal direction of the panel. Figure 3.2(c)
shows the dimensions of the coupons, whereas Table 3.8 shows the exact dimensions of
the specific coupons.
Shear tests were performed using an MTS testing machine with an MTS 685.60
hydraulic grip unit. The speed of the test was set to provide a constant strain rate of 0.01
min-1 in the gage length, as recommended by the standards, which is 0.12 mm/min. The
longitudinal strain of the coupon was measured using two electrical resistance strain
gages, 5 mm long, having a resistance of 119.8 0.2 , one on each side of the coupon.
Two additional strain gauges were used to measure strain in the transverse direction of
the coupon. Strain, load and stroke data were collected as described earlier for the tension
coupons.
50
3.2.2 Polyurethane Foam Tests
Two types of Polyurethane foam of two different densities were examined in this study
and were manufactured by DUNA-USA Foams and Chemicals, Inc. The foam core was
fabricated from Corafoam U020 and U040. The two densities of this closed-cell
polyurethane foam were 32.04 kg/m3 (2 pcf) and 64.08 kg/m3 (4 pcf), respectively. The
polyurethane foam material properties provided by the manufacturer are summarized in
Tables 3.4 and 3.5.
3.2.2.1 Tension Coupons
Coupons were tested to determine the flatwise tensile properties that are in a direction
parallel to the skin. Five coupons were prepared from both foam types, namely the low
and high density polyurethane foam, in the form of a prism. The coupons had a constant
square cross section. As recommended by ASTM C 297-04, the minimum face area of
the specimen was 625 mm2. As such, the final coupon cross-section dimensions were
taken 70 x 70 mm, to provide a reasonable size. The coupon thickness was taken 50 mm.
The specimens were adhesively bonded to a specially prepared T-shape steel sections,
using Sika high bond epoxy matrix, as shown in Figure 3.3. The purpose of the steel
sections is to facilitate gripping during the tension test.
The tests were performed using a Zwick Rowell T1-FR020TH.A50 testing
machine with wedge-type mechanical grips. The speed of the test was set as suggested in
the standards to be 0.5 mm/min. The longitudinal strain was measured using the Zwick
Roell built-in extensometer. The strain data as well as the load and stroke of the Zwick
Rowell machine were recorded using a TestXpert Data Acquisition System (DAS).
51
3.2.2.2 Compression Coupons
Ten coupons were tested, five for each foam density, to determine the flatwise
compressive properties of the foam core. The coupons had a constant square cross-
section. As recommended by ASTM C 365-03, the minimum face area of the prism is
625 mm2 while the maximum face area should not exceed 10000 mm2. Therefore, the
final coupon cross-section dimensions were taken 70 x 70 mm square cross-section. The
coupon thickness was taken 50 mm, as shown in Figure 3.4. Testing was carried out by
direct bearing on the prism in compression, using the same machine, same
instrumentation, and same test procedure described earlier for the tension tests in section
3.2.2.1.
3.2.2.3 Shear Coupons
Ten coupons were tested, five for each foam density, to determine the shear properties of
the foam core. The coupons had a constant rectangular cross-section. The coupons were
scaled down from typical full scale sandwich panel dimensions, to fit into the testing
machine. Therefore, the final coupon dimensions were 240 mm long, 50 mm wide, and
20 mm thick. The length-to-thickness ratio was 12 as recommended by ASTM C273-00.
The specimens were adhesively bonded to specially prepared steel plates by Sika high
bond epoxy matrix, as shown in Figure 3.5. The steel plates were machined to provide
knife-edges for loading. Testing was carried out by applying compression on the two
knife edges, using the same machine and test procedure described earlier for tension and
compression in sections 3.2.2.1 and 3.2.2.2, respectively.
52
3.3 Test Results
3.3.1 GFRP Coupons
A summary of the test results for the GFRP coupons is provided in Tables 3.6, 3.7 and
3.8 for tension, compression and shear coupons, respectively. Figures 3.6 and 3.7 show
the tensile stress-strain responses for the GFRP coupons in both the longitudinal and
transverse directions, respectively. It can be seen from Figures 3.6 and 3.7 that the GFRP
tensile response is slightly nonlinear. Also, all the observed failure modes of tension
coupons were fibre rupture, away from the gripping region, as shown in Figure 3.8.
The GFRP compressive stress-strain responses are shown in Figures 3.9 and 3.10
for both directions, longitudinal and transverse, respectively. The GFRP compressive
response was even more nonlinear than in tension. The ultimate compressive strength of
the GFRP was almost one third of the ultimate tensile strength, for both directions. The
corresponding ultimate compressive strains in both directions were almost one fourth of
the tensile ultimate strains. Figure 3.11 shows the test setup and failure mode of the tested
GFRP coupons.
The GFRP tensile load-strain response of the 45o coupons is shown in Figure
3.12 and was highly nonlinear. The GFRP shear stress was calculated according to
Equation 3.1 while the shear strain was calculated according to Equation 3.2 (ASTM
C273-00) as follows:
P
12 (3.1)
2A
12 x y (3.2)
53
where 12 is the shear stress at any data point and P is the applied load while A is the
cross-sectional area of the tested coupon. 12 is the corresponding shear strain while x
and y are the longitudinal and lateral normal strain of the 45o tension coupon at any
data point, respectively.
Figure 3.13 shows the resulting shear stress-strain response for the GFRP coupons
while Figure 3.14 shows the test setup and the observed failure modes of the fibres, at 45
degrees. The ultimate shear strain of the GFRP under shear was almost the same as for
the polyurethane foam of low density, as will be shown later.
3.3.2 Polyurethane Foam Coupons
Tables 3.9, 3.10 and 3.11 present a summary of the test results of the polyurethane foam
core for tension, compression and shear coupons, respectively, for both foam densities.
Figure 3.15 shows the tensile stress-strain responses. It can be seen that the responses are
nonlinear, with a slight stiffening effect. This is due to the alignment of the polyurethane
foam cell walls with the applied loading direction, due to cell wall joints yielding as was
also reported by Gibson and Ashby (1988). The ultimate tensile strength for the high
density foam was almost double of that of the low density foam. It is clear that the
polyurethane foam tensile strength depends on the density. Also, the ultimate tensile
strain of the high density foam was almost 85% that of the low density foam. Figure 3.16
shows the observed failure mode for the tension coupons of the polyurethane foam
material. Failure occurred within a foam cross-section and not at the bond line to the steel
plate.
Figure 3.17 shows the compressive stress-strain responses of the polyurethane
foam coupons for both densities. The response was linear until the foam internal cell
54
walls start to fail under internal cell wall-bending as reported by Gibson and Ashby
(1988), forming what may be considered analogous to a yielding point. The yielding
point was followed by a long flat plateau, then a densification (hardening) regime, caused
by the cell wall stacking together due to wall yielding. As the applied force increases, this
results in stiffening and increase in the compressive strength, along with very large
corresponding compressive strain. As the material continued to compress, the test was
terminated at 80% compressive strain, as shown in Figure 3.18. Figure 3.17 shows the
significant effect of density on the strength and stiffness in compression.
Figure 3.19 shows the shear stress-strain responses for both polyurethane foam
densities. The shear stress was calculated according to Equation 3.3 while the shear strain
was calculated from Equation 3.4 according to ASTM C273-00 standards as follows:
P
(3.3)
Lb
u
(3.4)
t
where is the shear stress and P is the load applied to the coupon. L and b are the coupon
length and width, respectively. is the shear strain while u and t are the displacement of
the loading plate and the coupon thickness, respectively.
It can be seen from Figure 3.19 that the response of the polyurethane foam under
shear is also highly nonlinear. The ultimate shear strength for the polyurethane foam with
high density was more than double that of the polyurethane foam with low density. Also,
the ultimate shear strains were almost similar to that of the GFRP material under shear.
Figure 3.20 shows the failure mode of the polyurethane foam shear coupon. In all
55
coupons, it was observed that a very thin layer of foam remained adhered to the steel
plates, suggesting that adhesive bond was stronger than the shear strength of the foam
(i.e. failure was not premature).
56
Table 3.1 E-Glass fabric manufacturer data (3TEX, Inc)
Yarn Direction Warp (x) Weft (y) Z yarn

Yarn Type E-Glass Roving E-Glass-Roving E-Glass-Roving
Weight % 50.3 48.8 0.9
# of Layers 2 3 -------
Table 3.2 Epoxy resin manufacturer data (Epoxical, Inc)
Physical Properties Value

Tensile Strength (MPa) 59.3
Tensile Modulus (GPa) 2.4
Tensile Elongation, % 5.9
Compressive Strength (MPa) 129.6
Table 3.3 GFRP lamina manufacturer data (Comtek, Inc)
Physical Properties Value

Tensile Modulus, E 1 = E 2 (GPa) 20.7
Shear Modulus, G 12 (GPa) 2.07
Tensile Strength, F tu (MPa) 344.7
Ply thickness, t ply (mm) 1.6
Table 3.4 CORAFOAM U020 (low density polyurethane foam) manufacturer data
(DUNA-USA)
Physical Property Condition Value

Density (kg/m3) Core 32.04
Parallel to rise 0.28
Compressive Strength (MPa)
Perpendicular to rise 0.14
Tensile Strength (MPa)
Shear Strength (MPa) Average 0.16
Shear Modulus (MPa) Average 0.23
Table 3.5 CORAFOAM U040 (high density polyurethane foam) manufacturer data
(DUNA-USA)
Physical Property Condition Value

Density (kg/m3) Core 64.08
Compressive Strength (MPa)
Tensile Strength (MPa)
Shear Strength (MPa) Average 0.6
Shear Modulus (MPa) Average 0.41
57
Chapter 3
Table 3.6 GFRP material properties based on tension coupons tests
Elastic Ultimate Ultimate

Coupon Coupon width Thickness modulus Mean S.D.1 2 strength Mean S.D.1 2 strain
C.V. C.V. Mean S.D.1 C.V.2
direction number (mm) (mm) "intial slope" (GPa) (GPa) (MPa) (MPa) (MPa) (mm/mm)
E (GPa) ult. ult.
1 25 1.5 27.3 301.1 0.01643
2 25 1.6 25.6 289.3 0.01479
Transverse 3 25 1.5 26.4 26.32 0.97 3.68 308.8 295.1 10.27 3.48 0.01643 0.015 0.001 6.99
58
4 26 1.5 25.1 294.3 0.01395
5 12 1.5 27.2 282.3 0.01573
1 26 1.5 26.7 263.7 0.01307
2 25 1.5 26.1 275.4 0.01339
Longitudinal 3 25 1.6 25.8 25.58 1.54 6.02 286.5 275.9 14.63 5.3 0.01432 0.014 0.001 6.58
4 25 1.6 26.4 259.9 0.01251
5 12 1.6 22.9 294.3 0.01468

1
S.D. = Standard deviation
2
C.V. = Coeffecient of variation
Material Testing Program
Chapter 3
Table 3.7 GFRP material properties based on compression coupons tests
Elastic Ultimate Ultimate

E (GPa) ult. ult.
1 24 1.5 36.3 111.5 0.00442
2 24 1.6 33.1 117.2 0.0046
Transverse 3 24 1.5 38.7 35.1 2.34 6.66 107.5 109.9 4.55 4.14 0.00414 0.004 0.0002 4.06
59
4 24 1.5 33.4 106.6 0.00424
5 24 1.6 34.3 106.7 0.0044
1 24 1.5 34.6 80.9 0.00383
2 24 1.5 33.8 84.4 0.00414
Longitudinal 3 24 1.6 32.4 33.1 1.13 3.43 88.2 84.2 3.19 3.8 0.00326 0.004 0.0004 12.57
4 24 1.6 32.4 81.1 0.00301
5 24 1.6 31.9 86.3 0.00357

1
2
Chapter 3
Table 3.8 GFRP material properties based on shear coupons tests
Shear Ultimate Ultimate

G (GPa) ult. ult.
1 24 1.5 2.2 40.4 0.293
2 24 1.6 2.1 41.1 0.247
45o angle 3 24 1.5 2.1 1.96 0.28 14.25 42.6 41.7 1.44 3.44 0.3 0.272 0.024 8.603
60
4 24 1.5 1.9 40.7 0.263
5 24 1.6 1.5 43.8 0.256

1
2
Chapter 3
Table 3.9 Polyurethane foam material properties based on tension coupons tests
Elastic
Cross- Ultimate Ultimate
modulus
Foam Coupon sectional Thickness Mean S.D.1 2 strength Mean S.D.1 2 strain
"intial C.V. C.V. Mean S.D.1 C.V.2
type number dim. (mm) (GPa) (GPa) (MPa) (MPa) (MPa) (mm/mm)
slope" E
(mm) ult. ult.
(GPa)
1 2.17 0.267 0.0533
2 2.09 0.394 0.0709
U040 3 70x70 50 2.21 2.09 0.099 4.737 0.332 0.343 0.051 14.978 0.0639 0.06 0.007 10.722
4 2.01 0.333 0.0645
61
5 1.98 0.387 0.0694
1 1.28 0.148 0.0765
2 1.26 0.151 0.0759
U020 3 70x70 50 1.18 1.25 0.053 4.217 0.128 0.149 0.024 15.893 0.069 0.077 0.01 13.522
4 1.21 0.132 0.0676
5 1.31 0.188 0.0936

1
2
Chapter 3
Table 3.10 Polyurethane foam material properties based on compression coupons tests
Elastic
Cross- Ultimate Ultimate
modulus
Coupon Coupon sectional Thickness Mean S.D.1 strength Mean S.D.1 strain
"intial C.V.2 C.V.2 Mean S.D.1 C.V.2
direction number dim. (mm) (GPa) (GPa) (MPa) (MPa) (MPa) (mm/mm)
slope" E
(mm) ult. ult.
(GPa)
1 9.6 0.143 0.02
2 9.6 0.131 0.02
U040 3 70x70 50 9.7 9.52 0.239 2.508 0.135 0.137 0.007 5.3 0.02 0.02 0 0
4 9.1 0.131 0.02
62
5 9.6 0.147 0.02
1 2.2 0.016 0.02
2 2 0.015 0.02
U020 3 70x70 50 2.2 2.14 0.089 4.18 0.026 0.019 0.005 23.83 0.02 0.02 0 0
4 2.1 0.021 0.02
5 2.2 0.017 0.02

1
2
Chapter 3
Table 3.11 Polyurethane foam material properties based on shear coupons tests
Shear
Ultimate
modulus Ultimate
Coupon Coupon width Thickness Mean S.D.1 Mean S.D.1 strain
"intial C.V.2 strength C.V.2 Mean S.D.1 C.V.2
direction number (mm) (mm) (GPa) (GPa) (MPa) (MPa) (mm/mm)
slope" G (MPa) ult.
ult.
(GPa)
1 7.69 0.321 0.21
2 7.56 0.345 0.288
U040 3 7.7 8.01 0.49 6.07 0.324 0.33 0.015 4.51 0.233 0.243 0.032 12.865
4 8.49 0.314 0.26
63
5 8.57 0.347 0.223
240x50 20
1 2.24 0.182 0.387
2 2.1 0.185 0.316
U020 3 1.98 2.21 0.17 7.52 0.174 0.179 0.004 2.413 0.322 0.332 0.031 9.335
4 2.3 0.177 0.313
5 2.4 0.178 0.322

1
2
Figure 3.1 GFRP skin structure (3TEX, Inc)
Gripping Gauge length Gripping Gripping Gauge length Gripping

24 mm
25 mm
12 mm
50 mm 150 mm 50 mm 65 mm 65 mm
Wide coupon
12.5 mm
50 mm 150 mm 50 mm
Narrow coupon
a) Tension coupons b) Compression coupons
Gripping Gauge length Gripping

24 mm
50 mm 150 mm 50 mm
c) Shear coupons
Figure 3.2 GFRP Coupons dimensions
64
Steel section
70 mm Foam coupon
50 mm
Figure 3.3 Polyurethane foam tension coupons
70 mm
70 mm
50 mm
Figure 3.4 Polyurethane foam compression coupons
Steel plate Foam coupon
Knife edge
Figure 3.5 Polyurethane foam shear coupons
65
350
300
250
Stress (MPa)
200
150
100
50
Average
0
0.000 0.004 0.008 0.012 0.016 0.020
Strain
Figure 3.6 Tensile stress-strain responses in the longitudinal direction for
GFRP coupons
350
300
250
Stress (MPa)
200
150
100
50
Average
0
0.000 0.004 0.008 0.012 0.016 0.020
Strain
Figure 3.7 Tensile stress-strain responses in the transverse direction for
GFRP coupons
66
(a) Longitudinal direction (b) Transverse direction
Figure 3.8 Typical failure modes for GFRP coupons in tension
120
100
Compressive stress (MPa)
80
60
40
20
Average
0
0 0.001 0.002 0.003 0.004 0.005
Strain
Figure 3.9 Compressive stress-strain responses in the longitudinal direction for

GFRP coupons
67
120
100
Compressive stress (MPa)
80
60
40
20
Average
0
0 0.001 0.002 0.003 0.004 0.005
Strain
Figure 3.10 Compressive stress-strain responses in the transverse direction for

GFRP coupons
(a) Longitudinal direction
(b) Transverse direction
Figure 3.11 GFRP Compression coupons test setup and failure modes.
68
3.5
2.5
2
Load (kN)
1.5
0.5
0
-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2
Strain
Figure 3.12 Load-normal strain response of GFRP coupons in shear
50
45
Average
Fitting
40
35
Shear stress (MPa)
P
30 t
b
25
)
2
P/2 (b P/2
20
P

15 2bt
10
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Shear strain
Figure 3.13 Shear stress-strain responses of GFRP coupons
69
Failure mode
Figure 3.14 GFRP Shear coupons test setup and failure mode
0.45
0.40
Hard Foam
0.35 3
(64.08 kg/m )
0.30
Stress (MPa)
0.25
0.20 Soft Foam

3
(32.04 kg/m )
0.15
0.10
0.05 Average
0.00
0.00 0.02 0.04 0.06 0.08 0.10
Strain
Figure 3.15 Tensile stress-strain responses of polyurethane foam
70
(a) Low density polyurethane (b) High density polyurethane

foam foam
Figure 3.16 Tension failures of polyurethane foam coupons
2.50
Average
2.00
1.50 Hard foam

Stress (MPa)
3
(64.08 kg/m )
1.00
Soft foam
3
(32.04 kg/m )
0.50
0.00
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90
Strain
Figure 3.17 Compressive stress-strain responses of polyurethane foam
71
(a) Polyurethane foam (b) Polyurethane foam

before compression after compression
Figure 3.18 Compression failures of polyurethane foam coupons

0.50
Hard foam
0.45 3
(64.08 kg/m )
0.40
0.35
Shear stress (MPa)
0.30
0.25 Soft foam

3
(32.04 kg/m )
0.20
0.15
0.10
0.05 Average
0.00
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Shear strain
Figure 3.19 Shear stress-strain responses of polyurethane foam coupons
Figure 3.20 Shear failure of Polyurethane foam coupons
72
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
CHAPTER 4: EXPERIMENTAL STUDY ON SANDWICH
PANELS WITH DIFFERENT RIB
CONFIGURATIONS1
4.1 Introduction
A new generation of composite sandwich panels is being developed for cladding of
buildings, among other applications. An earlier study at Queens University (Shawkat,
2008) addressed the effects of polyurethane foam core density on the flexural behaviour
of sandwich panels without rib. The study also examined the level of distribution of the
applied load, ranging from a single load to eight loads simulating a uniform load. The
study showed the significant effect of core density on flexural strength and stiffness.
Also, the considerable contribution of shear deformations to deflection was demonstrated.
Although the high-density polyurethane core resulted in significant enhancement of
flexural strength and stiffness, it was deemed unsuitable for cladding from the insulation
stand point, where the low-density core was superior in this regard. As such, it is
recommended to use GFRP ribs to enhance the behaviour and address the shear
weakness. This chapter presents a study on sandwich panels with different GFRP rib
configurations, integrated with the two skins. This includes outer diaphragms and internal
longitudinal and transverse ribs.

1
This chapter combined with Chapter 3 have been published as the following journal paper:
Fam, A. and Sharaf, T. (2010) Sandwich Panels Comprising Polyurethane Core and GFRP Skins and
Ribs of Various Configurations: Materials and Flexural Load Tests Composite Structures, Vol (92),
No. 12, pp 2927-2935.
73
The panels were fabricated by Comtek Advanced Structures, Inc, Burlington,
Ontario, CANADA, using the low density soft (32.04 kg/m3) polyurethane foam core
and the GFRP materials discussed in Chapter 3. This chapter provides a summary of the
experimental program carried out on these sandwich panels. It includes detailed
description of the test specimens, fabrication, test parameters, test setup and procedures,
instrumentation and test results. The chapter also includes discussion of the test results,
evaluation of the flexural strength and stiffness of the sandwich panels in comparison to
conventional reinforced concrete panels of similar size. While rigorous numerical and
analytical models will later be presented in Chapters 5 and 6, in this chapter simple
analytical expressions based on the classical beam theory for flexure and shear have been
developed and their applicability is explored within the linear range of behaviour.
4.2 Description of Test Specimens and Fabrication
Six sandwich panels (S1 to S6) with different rib configurations, as shown in Figure 4.1,
were fabricated. Panels S1 to S3 were 2437x635x78 mm while S4 to S6 were
2537x685x78 mm. The slight difference in width and length was a result of the
fabrication process, where all panels were initially fabricated with an exterior GFRP rib
around the perimeter (a diaphragm). All panels included a 75 mm thick closed-cell
polyurethane foam core of a 31.6 kg/m3 density, along with two 1.6 mm thick GFRP face
skins. Material testing and mechanical properties of both materials were reported earlier
in Chapter 3. Panel S1 did not include any ribs, as a control specimen. Panels S2 and S3
included internal longitudinal ribs at mid-width and S3 also included additional internal
transverse rib at mid-length. Panel S4 included only exterior ribs at all four sides. Panels
S5 and S6 also included exterior ribs at all four sides. Additionally, they both included
74
internal longitudinal ribs at mid-width while S6 also included additional internal
transverse rib at mid-length.
As shown in Figure 4.1, the internal ribs consisted of two back-to-back C-shape
woven fabric of the same material as the skins (detail B), while the exterior (edge) ribs
consisted of one C-shape woven fabric (detail C). As such, the total GFRP skin
thickness at locations of the flanges of the ribs was doubled, to 3.2 mm. The flange width
of each of the C-shape ribs was 75 mm. The total web thickness of one internal rib was
3.2 mm and for an exterior rib was 1.6 mm. Table 4.1 presents a summary of details of
the test specimens, including dimensions, weight and rib configuration.
4.3 Fabrication
Figure 4.2 shows a summary of the fabrication process, using the Vacuum Assisted Resin
Transfer Moulding (VARTM) process. The glass fiber cloth for the ribs was cut to the
desired dimensions and laid up in the c-shape patterns around the edges of the
prefabricated foam blocks. This assembly was placed into the mold, above the cloth of
the bottom skin. The cloth for the top skin was then laid up on top of the foam blocks and
the mold was closed. This was followed by the resin infusion process, using the VARTM
process, in which resin is pulled into the mold by negative pressure, and impregnates the
fibers. The panel was allowed to cure and then was released. In fabricating panel S4, an
error occurred where the exterior rib did not include sufficient flanges overlapping with
the skins, which later affected the failure mode. Also, an inspection of panels S1 revealed
a very small and localized delamination between the compression skin and core at about
680 mm from the end, which slightly affected failure, in terms of location.
75
4.4 Test Setup
The panels were tested in one-way bending with a span of 2300 mm, under a simulated
uniform load, as shown in Figure 4.3. Loads were applied at four equally spaced
locations along the span, using steel spreader beams and rollers system. At each loading
point, a specially designed system has been developed to distribute each load over a large
surface area, to simulate as close as possible a uniform load. The system conforms to the
curved surface of the panel under loading when it deforms, in order to maintain a uniform
contact. Each load was applied over a 125 mm wide steel plate, which transferred the
load to another two 125 mm wide steel plates placed side-by-side. A 18 mm thick rubber
pad was placed between the two levels of steel plates. Additional layer of rubber pads
was also placed between the two steel plates and the surface of the panel, as shown in
Figure 4.3. This system ensured a wide distribution of loads while maintaining flexibility
and accommodation for the curvature of the panel surface. Loading and supporting
systems extended across the full width of the panel. The specimens were loaded to failure
using a 900 kN Riehle testing machine, at a stroke rate of 1 mm/minute.
4.5 Instrumentation
Strains in both top and bottom skins of specimen S1 were monitored using 24, 5 mm long
and 119.8 0.2 resistance, electric resistance strain gauges, 12 gauges for each skin.
The strain gauges were located along the longitudinal and transverse centre lines of the
specimen, within one half of the specimen, as shown in Figure 4.4(a). The goal was to
monitor the strain distribution along the longitudinal and transverse directions of the
specimen. For specimen S2, only six strain gauges were used to monitor the strain
behaviour on the top and bottom skins. They were all located at mid-span of the panel,
76
along the transverse direction, including three on the upper skin (compression) and three
on the lower skin (tension), as shown in Figure 4.4(b). Specimen S3 was instrumented
with eight strain gauges, four on each skin, two were located along the centre line and the
other two at the edge of the flange of the rib, as shown in Figure 4.4(c). Specimens S4, S5
and S6 were instrumented using six strain gauges, three for each skin, located at the panel
centre line at mid-span, as shown in Figure 4.4(b). Mid-span deflection was measured
using two Linear Potentiometers (LPs) as shown in Figure 4.4. Load was monitored using
load output analog channel of the Riehle testing machine.
4.6 Test Results and Failure Modes
A summary of key parameters of test results and failure modes is given in Table 4.2.
Figure 4.5 shows a comparison of the load-deflection responses of all test panels. The
panel widths of specimens S4 to S6, 685 mm, were slightly different from those of panels
S1 to S3, 635 mm. Therefore, additional normalized set of load-deflection curves with
respect to the width for panels S4 to S6 are also presented in Figure 4.5 (marked as
normalized), to facilitate a fair comparison with panels S1 to S3. Figure 4.6 shows the
load-longitudinal strain responses of all panels, at various locations of the top and bottom
GFRP skins. Figure 4.7 shows the distributions of strains along the two axes for panel S1.
A summary of various failure modes is shown in Figures 4.8 to 4.13.
Figure 4.5 clearly shows that the behavior of all panels is linear to failure,
regardless of the rib configuration. At the same time, flexural strength and stiffness of the
sandwich panels are greatly affected by the presence of the ribs and their configuration,
but the deflection at ultimate is not much affected. The flexural strength and stiffness
have generally increased by a percentage ranging from 51% to 154%, depending on the
77
rib configuration, relative to panel S1 without any ribs. Figure 4.6 shows that, in all
specimens, the maximum longitudinal tensile strain measured on the tension skins was
0.01, which is lower than the ultimate value from tension coupons, shown in Figure 3.6.
Panel S1 failed at 14.27 kN due to outwards skin wrinkling as shown in Figure
4.8, in a cylindrical pattern. Wrinkling occurred between the first and the second applied
loading points, where a small initial defect was observed earlier, in the form of a small
crack or delamination between skin and core. As will be shown in the theoretical
calculations, this failure occurred at a load very similar to the load that would have been
reached if the specimen had been free of defects. Figure 4.6(a) shows the longitudinal
strains along the transverse and longitudinal centerlines of the panel while Figure 4.7
shows strain distributions along the two axes.
Panel S2, with a longitudinal rib at mid width, failed at 27.28 kN, which is about
91% higher than panel S1. The top skin eventually suffered an outwards wrinkling at mid
span in a pattern that resembles double, back-to-back, half cones, since the skin was
restrained from wrinkling at mid width by the longitudinal rib, as shown in Figure 4.9.
Figure 4.6(b) shows the longitudinal strain response of panel S2. At location c on the
edge of the compression skin, the strain reverse direction at a high load level, reflecting
the onset of the outwards wrinkling behavior.
Panel S3, with a longitudinal rib at mid-width and a transverse rib at mid span,
failed at a 26.78 kN load, which is quite similar to panel S2, but showed a slightly
improved stiffness. This suggests the insignificant effect of the transverse rib at mid
span, which should be expected, given the one-way bending nature of the panel. Perhaps
the impact of transverse ribs would be rather more pronounced in a two-way bending
78
configuration. The transverse rib, however, slightly affected failure mode by providing a
continuous support for the top skin in the transverse direction at mid span, leading to an
outwards wrinkling that resembles double back-to-back smaller half cones, slightly
shifted away from mid-span, as shown in Figure 4.10. The sudden change in skin
thickness near the flange of the rib caused a stress concentration and failure at the edge of
the cross rib, in the form of a localized skin crushing.
Due to the fabrication error occurring in panel S4, where small flange and rib
cloth overlaps were provided for the longitudinal edge ribs, premature failure occurred at
a normalized load of 21.57 kN, which is just lower than those of S2 and S3. Similarly,
flexural stiffness was lower. Failure occurred at the junction of the compression skin and
the exterior rib, in an unzipping fashion as shown in Figure 4.11. Clearly, the exterior rib
failed to constrain the tendency for outwards wrinkling of the compression skin, due to
the lack of continuity (i.e. by means of overlapping) at this borderline that the flange of
the rib would have otherwise provided.
Panel S5 with exterior ribs and an internal longitudinal rib at mid-width failed at a
normalized ultimate load of 36.28 kN. This is 154% higher than panel S1 without any
ribs and 33% higher than panel S2 with only an internal longitudinal rib. Failure of panel
S5 occurred by crushing of the compression skin along a transverse line near mid-span,
as shown in Figure 4.12.
Panel S6 failed at a normalized load of 36.17 KN, which is almost identical to S5.
Similar to panels S2 versus S3, the transverse rib at mid-span of panel S6 clearly has
insignificant effect on strength. Similar to S5, panel S6 failed by crushing of the
79
compression skin at a transverse line at the edge of the flange of the transverse rib, where
the sudden change in skin thickness occurs, as shown in Figure 4.13.
4.7 Discussion
Figure 4.14 shows the variation of the ratio of experimental strength and stiffness of
panels S2 and S5 to those of panel S1, as a reference, with the percentage of GFRP ribs
cross-sectional area to that of the total core area. Panel S4 was excluded because of the
flaw in its ribs system that may have affected its strength. Figure 4.14 shows an increase
in strength and stiffness as the longitudinal ribs increase (the curves with circular and
triangular markers). The ribs distribution and locations across the panels width is also
crucial; however, it is difficult to separate this effect from the ribs cross-sectional area
based on the limited number of tests. Therefore, this aspect is studied further in a
parametric study in Chapter 8.
Shawkat (2008) studied the effect of foam core density in similar but smaller
(1500x300x76 mm) sandwich panels without any ribs2. Both, the 32.04 kg/m3 density,
which is used in panels S1 to S6 of this study, and the 64.08 kg/m3 density (both
described in detail in Chapter 3), were considered. Figure 4.14 shows the effect of core
density on strength (the vertical curve with square markers). It can be seen that the
pattern of ribs of panel S5 provides an increase in strength (about 151%) equivalent to
that resulting from doubling the density of the core (point a) when no ribs are used.
However, in some applications such as cladding walls and roofs, the lower density core
may be more desired due to its better thermal insulation effectiveness, and hence the use
2
The author contributed significantly to that work, which was published as the following journal paper:
Sharaf, T., Shawkat, W., and Fam, A. (2010) Structural Performance of Sandwich Wall Panels with
Different Foam Core Densities in One-Way Bending, Journal of Composite Materials, in press.
80
of ribs may be essential to enhance structural performance.
4.8 Analysis of Stiffness and Deflections at Service Load Levels
Sandwich panels generally have low stiffness compared to other structural materials,
which makes them vulnerable to excessive deformations; therefore, the following section
focuses on stiffness along, with a simple approach to predict deflections at service load.
Figure 4.5 shows the deflection control limits of (span/180) and (span/360), commonly
used in most design codes, for roofs and floors, respectively. It is clear from the figure
that the corresponding permissible service loads, according to those limits are well below
the ultimate capacity of the panels, leading to large safety factors for strength. As such,
design of the panels will likely be governed by stiffness. The equivalent flexural stiffness
(EI) equiv of panels S1, S2 and S5 have been estimated from the load-deflection responses,
within the service load range, assuming only flexure-dominated deformations and
uniform loading, based on Eq. 4.1:
5 PL3
( EI ) equiv (4.1)
384
where P, L and are the resultant of the uniformly distributed load, span and deflection,
respectively. The equation assumes that the loads are distributed enough to be
represented as uniformly distributed. The flexural stiffnesses of the same panels were
also estimated analytically using a transformed cross-section analysis. The ratio of the
experimental-to-analytical flexural stiffness is then plotted in Figure 4.14, versus the
percentage of GFRP ribs cross-sectional area (the curve with diamond shape markers,
along with the right vertical axis). The behaviour shows an increasing trend, from about
0.5 to 0.8, which appears to approach 1.0 (the case for only flexural deflection) as the ribs
81
increase. The fact that these ratios are significantly lower than 1.0, clearly emphasizes
the important contribution of shear deformations to mid-span deflection, particularly for
panel S1 with no ribs which was intentionally ignored in this demonstration. As the ribs
increase, the behavior becomes more governed by flexure and hence Eq. 4.1 becomes
more representative of the total deflection.
Equation 4.2 provides an expression to calculate the total deflection, accounting
for both flexural and shear deformations, for panel S1 without GFRP ribs, while Equation
4.3 provides a similar expression applicable to panels S2 and S5 with GFRP ribs. The
equations are established from basic linear mechanics of beam theory for flexural and
shear deformations. Given the significant differences between the Youngs and shear
modulii of the GFRP and those of core materials (n = E GFRP / E core 17300) and (m =
G GFRP / G core 880), transformed section analyses can be reduced to a consideration of
the GFRP system only for panels with GFRP ribs.
5 L3 L
Panels with no GFRP ribs: P (4.2)
384 EGFRP I GFRP 8 Gcore bcore t core
5 L3 L
Panels with GFRP ribs: P (4.3)
384 EGFRP I GFRP 8 GGFRP bGFRP tGFRP
where E GFRP is the longitudinal Youngs modulus and G GFRP is the shear modulus of the
GFRP material. G core is the shear modulus of the foam core. The values used for these
parameters are the slopes of the initial linear parts of the material curves presented in
Chapter 3. I GFRP is the cross-sectional moment of inertia based on the GFRP skins for
panel S1 or GFRP skins and ribs for panels S2 and S5. b core and t core are the width of the
panel and thickness of the core, respectively, when there are no GFRP ribs (panel S1).
82
b GFRP and t GFRP are the combined widths of the GFRP ribs and the height of the GFRP
ribs, respectively (panels S2 and S5).
Equations 4.2 and 4.3 have been used to predict the stiffness of the panels, as
shown in Figure 4.15 (dotted lines), with reasonable agreement with the measured
responses. Only initial parts of load-deflection responses are shown here, within the
design deflection limits. The relative contributions of flexure and shear to the total
deflection, as a percentage, are also shown in Figure 4.15 top left corner, based on the
two separate terms of Equations 4.2 and 4.3. Clearly, in panel S1 without ribs, shear
deformation of the soft core has contributed over 50% of the total deflection. In panels S2
and S5, the shear deformations of the GFRP ribs contributed about 15 to 20% of the total
deflection. The models presented in this section are quite simplified linear models that
represent service load level only and can not capture the strength or deflections at higher
load levels. More rigorous numerical and analytical models are presented in Chapters 6
and 7.
4.9 Comparison with Reinforced Concrete Panels
A comparison has been made between sandwich panels S1, S2 and S5, and single
reinforced conventional reinforced concrete (RC) panels of similar cross-sectional
dimensions. It was assumed that the concrete has a 30 MPa compressive strength and the
steel reinforcement has a 400 MPa and 550 MPa yield and ultimate strengths,
respectively. A 25 mm clear concrete cover was assumed. The objective of this analysis
was to establish the tensile steel reinforcement ratios in virtual RC panels RC1, RC2 and
RC5, required to provide similar strengths to those of sandwich panels S1, S2 and S5.
Conventional cracked section analyses using program RESPONSE-2000 were carried out
83
and showed that the steel reinforcement ratios required for the three RC panels to match
the strength of sandwich panels, were 0.6%, 1.4% and 2%, respectively. A comparison
of the moment-curvature responses of the test sandwich panels and the RC panels is
shown in Figure 4.16. Curvatures of the test panels were obtained from the longitudinal
strains measured at mid-span on both skins, as the slope of the strain profile at mid-span.
It is concluded from this exercise that the sandwich panels have equivalent flexural
strengths to 80 mm thick RC panels with moderate to heavy steel reinforcement ratios
(estimated as 3-10M/m to 10-10M/m). The sandwich panels, however, are 9 to 14 times
lighter in weight than the RC panels, which is a significant advantage in structural
applications involving shipping, lifting, craning and speed of installation. On the other
hand, as expected for any FRP-based structures, the behaviour is somewhat linear up to
failure and does not provide the ductility associated with mild steel reinforcement.
84
Table 4.1 Test Matrix

Dimension
Foam
Avg. Skin Total thickness Weight Rib
Specimen ID density Loading type
Length Core thickness Width (kg) configuration
3
(kg/m ) thickness (end-middle-end)
(mm) (mm) (mm)
(mm) (mm)
S1 21.2
77-78-77 75 - 76
S2 2437 635 26 L
S3 27 Simulated L T
32.04 1.6 79-78-79 77 -78 uniform by 4
S4 28.1 loads E
S5 2537 685 37.4 E L

77-78-77 75 -77
S6 38.2 E L T
L = Internal longitudinal rib

T = Internal transverse rib
E = Exterior rib (diaphragm)
Table 4.2 Summary of test results

u
Specimen ID Pu (KN) Mu (KN.m) u (mm) Failure Mode
Top Bottom
S1 14.27 4.1 67.82 0.00211 0.0024 Skin wrinkling
S2 27.28 7.9 69.72 0.00336 0.0042

Skin wrinkling and crushing
S3 26.78 7.7 63.75 0.0058 0.00911
S4 21.57 6.2 63.6 0.00364 0.00288 Delamination and wrinkling
S5 36.28 10.4 64.89 0.00365 0.00416

Skin crushing
S6 36.17 10.4 67.12 0.00365 0.00411
Pu = Ultimate load
Mu = Ultimate moment
u = Deflection at ultimate load
u = Longitudinal strain at ultimate load
85
No ribs Rib
Panel S1 Panel S2
Ribs Ribs
Panel S3 Panel S4
Ribs Ribs
Panel S5 Panel S6
Figure 4.1 Sandwich panels (S1 to S6) with different rib configurations
(top view and cross sections).
86
Figure 4.2 VARTM fabrication process for the sandwich panels
87
Figure 4.3 Test setup
88
(a) Specimen S1
(b) Specimens S2, S4, S5, S6
(c) Specimen S3
Figure 4.4 Sandwich panels instrumentation
89
40
Width of panels S4, S5 and S6 is 685mm. Their
normalized responses with respect to the 635mm width S5 S6
35 of S1, S2 and S3 are also shown.
S5 (normalized)
30
S3 S2
S6 (normalized)
25 Span Span
Load (kN)
360 180 S4
20
S4 (normalized)
15
S1
10
0
0 10 20 30 40 50 60 70 80
Deflection (mm)
Figure 4.5 Load-deflection responses for all specimens
90
(a) S1 (b) S2
(c) S3 (d) S4
(e) S5 (f) S6
Figure 4.6 Load-strain behaviour of sandwich panels
91
0.003
Bottom strain at 100% of ultimate load

0.002
Bottom strain at 50% of ultimate load

0.001
Strain
0.000
-0.001
Top strain at 50% of ultimate
l d
-0.002
Top strain at 100% of ultimate load
Panel support Panel centre line
-0.003
-1200 -1000 -800 -600 -400 -200 0
Distance (mm)
(a) Along the longitudinal direction
0.003
Bottom strain at 100% of ultimate

0.002
Bottom strain at 50% of ultimate

0.001
Strain
0.000
Top strain at 50% of ultimate

-0.001
-0.002 Top strain at 100% of ultimate
Panel centre line

-0.003
-300 -200 -100 0 100 200 300
Distance (mm)
(b) Along the transverse direction
Figure 4.7 Strain distributions along the two axes of panel S1
92
Outward wrinkling of skin,

far from mid-span
Figure 4.8 Failure mode of specimen S1
Crushing and outward wrinkling

of skin at mid-span
93
Crushing and outward wrinkling

of skin at mid-span
Delamination and wrinkling of top

skin from edge rib at mid-span
94
Crushing of top skin at mid-span
Crushing of top skin at mid-span
95
2.5 0.0
"a"Sharaf et al (2010)
Effect of foam core density

Ratio of Strength (or Stiffness) of panels S2 and
Density = 64.08 kg/m3 0.1
Ratio of experimental stiffness to calculated

2.0
on strength
S5 0.2
Strength (Exp.) 0.3

S2
S5 to that of panel S1
flexural stiffness
1.5
Stiffness (Exp.) 0.4
0.5
1.0
S1
0.6
S2
0.7
0.5 S5
0.8
0.0 0.9
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
% of GFRP ribs cross sectional area-to-core area
Figure 4.14 Effect of ribs and core density on strength and stiffness of
sandwich panels
10
S5
9
7 S2
6
Load (KN)
5
S1 S2 S5
4 S1
3 Experimental
2
Analytical
1 Span Span
360 180
0
0 3 6 9 12 15
Deflection (mm)
Figure 4.15 Experimental versus analytical load-deflection responses
96
12
10 S5
8 S2
Moment (kN.m)
4 S1
0
0 50 100 150 200 250 300 350 400
Curvature (1000/m)
Figure 4.16 Comparison between sandwich panels and conventional RC panels
97
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
CHAPTER 5: TESTING OF LARGE SCALE CLADDING
SANDWICH PANELS1
5.1 Introduction
This chapter presents an experimental investigation of large scale (9144 x 2438 x 78 mm)
sandwich panels intended to be used as a substitute for precast concrete cladding panels
for buildings. Similar to the panels described in Chapter 4, the sandwich panels described
in this chapter were manufactured using GFRP skins and ribs, and polyurethane foam
core. The panels included several internal ribs in the longitudinal and transverse
directions, in addition to an exterior rib around the perimeter. The panels tested before
and reported in Chapter 4 were intended to study the effect of different rib configurations
using a one-way bending test set up. The panels in this chapter represent a selected
specific design, in terms of the rib configuration and the supporting system of the panel at
six points. Two panels were tested. The first panel was tested under a realistic uniform air
pressure applied directly to the panel. This was preferred over the idea of loading using
an air bag which has to be in a direct contact with the panel and may affect the wrinkling
behaviour of the skin. To carry out this specialized test, the panel was tested at the
Insurance Research Lab for Better Homes (IRLBH) at London, Ontario. To simulate the
real life mounting condition, and to take advantage of the already existing reaction frame
columns at the lab, the panel was tested in a vertical position.

1
This chapter has been submitted for publication as the following journal paper:
Sharaf, T. and Fam, A. (2010) Experimental Investigation of Large Scale Cladding Sandwich Panels
under Out-of-Plane Transverse Loading for Building Applications ASCE Journal of Composites for
Construction, Under review.
98
A special test setup was constructed to create a sealed pressure box on one side of
the panel, and air pressure was applied using special Pressure Load Actuators. For the
second test, a second panel was cut longitudinally into two symmetrical parts, and was
tested at Queens University under a simulated uniform load by using conventional
mechanical loading at discrete points.
This chapter provides a summary of the experimental program carried out on the
full scale sandwich panels. It includes detailed description of the test specimens, test
setups and procedures, instrumentation and test results. The chapter also includes
discussion of the test results and an evaluation of the flexural strength and stiffness of the
panels, relative to of some of the design values of wind pressures in the National Building
Code of Canada (NBCC).
5.2 Description of the Wall Panels and Fabrication
The sandwich panels investigated are proposed for cladding applications of buildings and
are 9144 mm in height, 2438 mm wide, and 78 mm thick. The panel is designed to be
installed in the field in a vertical position, attached to the framing of the building at six
points, through adhesively bonded 305x305 mm plates, as shown in Figure 5.1. Figure
5.2 shows pictures of the panels during shipping and handling. The panel weighed 289
kg, which is about 13 times lighter than an equivalent precast concrete cladding panel of
similar dimensions.
The panel is composed of low-density polyurethane foam core (described in
Chapter 3 and also used in the panels of Chapter 4) sandwiched between two GFRP
skins. The panel also included internal I-shape ribs, two in the longitudinal direction and
three in the transverse direction, as well as an exterior C-shape rib around the perimeter
99
(details A and B in Figure 5.3). In order to fabricate the panel, twelve 75 mm thick
blocks of the soft (low-density) polyurethane foam were cut to the proper dimensions, to
form the cores between the ribs. Woven glass fibre cloth (described in Chapter 3 and also
used in the panels of Chapter 4) in the form of a C-shape of 75 mm wide flanges was
fitted around the edges of each block. The twelve blocks were then assembled in one
form and flat sheets of the same woven glass fibre cloth were placed on both sides to
form the skins (Figure 5.1). Because of the large size of the panel, three longitudinal
strips of fabric, running the full length of the panel, were used to form each skin. The
middle strip overlapped 200 mm on each side with the two side strips as shown in the
picture given in Figure 5.3. The center of the overlap was located 250 mm from the
panel centerline. Epoxy resin was applied to the assembly through a vacuum process and
was cured. In the final product, the GFRP skins are 1.6 mm thick and are 3.2 mm thick at
the overlap zones (nominal thickness). The internal ribs are composed of two back-to-
back C-shape ribs, forming the I-shape. The flanges of the ribs were integrated with the
skin, giving a total thickness of 3.2 mm. The thicknesses of the webs of the internal and
exterior ribs are 3.2 and 1.6 mm, respectively (details A and B in Figure 5.3).
5.3 Test Specimens
Two panels were fabricated specifically for this experimental study to carry out two
separate tests. In Test I, the full scale panel shown in Figures 5.1 and 5.3 and described
above, was tested under a uniform air pressure, until failure occurred in the panel at the
vicinity of the middle supports (i.e. negative moment region). The test was terminated
because of limitations of the setup in accommodating further deflection, in spite of the
fact that the statically indeterminate (two-span) panel could have sustained further
100
loading. Therefore, it was decided to carry out another test (Test II) on the second
specimen, using conventional mechanical loading, for the following reasons: (a) to
confirm the results of Test I, including the load at first failure, (b) measure the behaviour
beyond first failure to examine load stability at further deformations, and (c) observe the
subsequent failure modes. To fit within the test frame available at the Structures Lab. of
Queens University, one half of the second panel, 1220 mm wide, was cut along the
longitudinal axis of symmetry of the panel and used in Test II. Because of the aspect ratio
of the original panel, a one-way slab behaviour was more dominant any ways. As such,
the discontinuity along the cut-off edge of the skins in the transverse direction is believed
not to have a significant adverse effect on the behaviour, as will be demonstrated by test
results.
5.4 Materials
The panels were fabricated using the materials described in Chapter 3. The prefabricated
closed-cell polyurethane foam blocks of low density (32.04 kg/m3) were used due to their
superior insulation properties relative to the high density (64.08 kg/m3) foam. The
insulation properties are indicated by the value of the factor R which represents the ratio
of the temperature difference across an insulator (polyurethane foam in this case) and the
heat flow per unit area. For the low-density foam, the R-value is (5.26 hr.ft2.oF/BTU),
which is larger than that of the high-density foam (R-value = 4 hr.ft2.oF/BTU).
101
5.5 Test Setups and Instrumentation
5.5.1 Test I - Full Scale Panel Under Simulated Uniform Air Pressure
The full scale panel was tested in a vertical position, in a similar configuration to that
planned in real life mounting, as shown in Figure 5.1. A special self-reacting frame was
built, as shown in Figure 5.4. The frame comprises two W460 steel columns spaced 2.44
m apart, supporting a wooden box, essentially framed around the test panel as shown in
Figure 5.4. The back of the box was sealed with a heavy wooden panel comprising
wooden face plates and heavy studs (Figure 5.5), while the test panel was placed at the
front of the wooden box. Figure 5.6 is a top view cross-section showing the wooden back
panel, the front test panel, the 215 mm pressurized air gap in between, and the side skirts
used to seal this air gap. The top and bottom of the setup were also sealed with similar
wooden skirts. A continuous membrane of flexible Polycrepe was attached to all side
skirts from inside, to seal the joint between the skirts and both the back reaction panel
and front test panel, as shown in Figure 5.7. Sufficient slack was provided to the
membrane to allow for the deflection of the test panel.
Six supports were provided for the test panel, as shown in Figure 5.4. The
supports comprised 305x305x75 mm wooden blocks directly in contact with the test
panel. Three horizontal reaction steel beams were placed at the three elevations of
supports, to provide the reaction system (Figure 5.8). Special load cells were placed
between the support wooden blocks and the steel beam. Each load cell can measure the
reaction forces in three orthogonal directions, though the primary loading direction in this
case is normal to the panel. Steel tie rods were used to connect each reaction beam, at
both sides, to the vertical W460 steel columns (Figure 5.8), to complete the self-reaction
102
frame system. To avoid friction between the bottom edge of the test panel and the floor,
the panel was rested on a 50 mm diameter steel roller. This resulted in a slight change
from the design geometry shown in Figure 5.3, leading to a top and bottom overhangs of
510 mm and 410 mm, respectively, rather than 460 mm.
The air pressure was applied inside the box using Pressure Load Actuators (PLAs)
(Figure 5.9); each has the capacity of generating +/-11 kPa in about a 1.2 m3 volume,
with leakage of 1 m3/sec. It was decided to use ten actuators to provide and maintain the
necessary pressure to the panel throughout the test and be able to reach the failure load.
Air pressure was supplied by the ten PLAs through three access holes on each side skirt
near the bottom and four access holes in the back panel, as shown in Figure 5.10. The
pressure was monitored electronically using pressure sensors located at the panel side
skirts, as shown in Figure 5.10.
Four linear potentiometers (LPs) were used to measure deflections at the locations
of maximum expected deflection (Figure 5.11). Three LPs were installed in the lower
span on one line in the transverse direction. These included one LP at mid-width and one
on each edge of the panel. The fourth LP was installed in the upper span, at mid-width.
Additional LPs were used to monitor displacements of the supporting reaction beams, to
establish the net deflections of the panel. Electric resistance strain gauges were used to
measure the strains in the skins in the longitudinal and transverse directions at 19
locations. Figures 5.12 and 5.13 show the locations of the strain gauges on both skins,
namely the loaded and the opposite surfaces.
103
5.5.2 Test II Half Full Scale Panel Under Discrete Loads
The second test was carried out on a half full scale panel, as shown in Figure 5.14. The
panel was laid horizontally on top of three supports of similar configurations to those
used in Test I (i.e. 305x305x75 mm wooden blocks). Supports were placed at mid-width
of the panel, such that their locations relative to the outer edges of the panel are the same
as in the original mounting configuration shown in Figure 5.3.
In each of the two spans, loading was applied at two locations (middle third points
of the span), across the full width (Figure 5.14). A hydraulic actuator was used to apply
the load using stroke control at a 1.5 mm/minute rate of loading. Two levels of spreader
beams were used to apply the four loads, as shown in Figure 5.14. The loads were applied
through 38 mm diameter rigid steel pins at the four locations and were spread over 450
mm wide areas. This was accomplished using a set of alternating layers of steel and
plywood plates between the steel pin and the surface of the panel. The softer layer of
three side-by-side plywood plates provided a cushion between the panel and the two side-
by-side steel plates. The concept of multiple plates placed side-by-side, in lieu of one
wide plate, allows the system to accommodate the curvature within the 450 mm width.
Deflection was measured at various locations in the vicinities of mid-spans, using
seven LPs (Figure 5.15). In one span, deflection was measured at mid-span and at an
adjacent location of the maximum anticipated deflection values, at three points across the
width at each location. In the other span, deflection was measured at the location of
maximum anticipated deflection value, at mid-width of the panel. Several additional LPs
were used to measure any movements at the support locations. Electric resistance strain
gauges were used to measure the strains in the skins in the longitudinal direction at 32
104
locations. Figure 5.16 shows the locations of the strain gauges on both skins, namely the
loaded and opposite surfaces.
5.6 Results of the Experimental Program
5.6.1 Test I - Full Scale Panel Under Uniform Air Pressure
5.6.1.1 Load-deflection responses
Three tests were carried out in panel Test I. Figures 5.17 and 5.18 show the pressure
versus deflection responses near mid-spans of the panel, in two successive tests, which
are fairly linear. In the first test (Figure 5.17), the specimen was loaded up to 2.6 kPa
then unloaded, due to some leakage that resulted in air pressure loss. After the setup was
fixed, a second test was successfully carried out to failure, which occurred near the
middle support. The maximum pressure achieved was 7.5 kPa at a maximum deflection
of 82 mm, as shown in Figure 5.18. The deflections in the upper and lower spans varied
slightly. Because of the statically indeterminate nature of the specimen, being continuous
over two spans, this first failure at the middle support did not cause a complete loss of
load capacity. However, because of difficulties with controlling the pressure, along with
the deflection of the specimen approaching the setup limitation, further loading could not
be continued to explore the post-failure response or any subsequent secondary failure
modes. After unloading, a third testing was carried out by applying three loading and
unloading cycles to examine the stiffness of the panel after the first failure. Figure 5.19
shows that the stiffness has been reduced. Because of the deflection limitations of the
setup in Test I, it remained unknown how stable the residual strength is and whether
further loading could have caused pressure to rise beyond the first peak of 7.5 kPa or not.
For this reason it was decided to carry out Test II.
105
5.6.1.2 Load-strain responses
Figures 5.20 and 5.21 show the pressure versus longitudinal strains of the loaded and
opposite GFRP skins, respectively, while Figures 5.22 and 5.23 show similar responses
of the transverse strains. Tensile strains were generally well below the ultimate values
obtained from coupon tests, whereas compressive strains reached very close to the
ultimate values (Figure 3.9). Also, strains varied across the width of the panel, due to
some transverse bending but more importantly due to the change in skin thickness at the
rib location. This distribution is further studied in Test II.
5.6.1.3 Failure mode
Failure occurred by outward buckling (wrinkling) and localized crushing of the
GFRP skin in compression, adjacent to the middle supports, as shown in Figure 5.24.
Figures 5.21 and 5.23 show a combination of high levels of longitudinal and transverse
compressive strains near this location.
5.6.2 Test II Half Full Scale Panel Under Discrete Line Loads
5.6.2.1 Load-deflection responses
Figure 5.25 shows the load-deflection responses of the panel in Test II at various
locations in the vicinity of mid-span. Figures 5.26 and 5.27 show the load-deflection
responses near the edge and middle support locations, respectively, where small
displacements occurred as a result of the specimen slightly rotating around the edges of
the supporting pads. Figure 5.28 shows the deflected shape and rotation at the edge
support. It should be noted that in practice the panel will be adhesively bonded to the
plate. This practice was neglected in Tests I and II as a conservative approach. Bond will
provide further fixity and restraints and hence less deflections. Generally, linear
106
behaviour is observed until first failure (wrinkling near middle supports, exactly as in
Test I) occurred at 51.1 kN load and a maximum deflection of 104 mm (Figure 5.25). The
load dropped slightly to 49.2 kN, then further loading resulted in an increase in load up to
49.5 kN, when the second failure occurred (shear of the core near middle supports) and
the load dropped to 38.7 kN. Further loading resulted in an increase in load up to 43.2
kN, when the third and final failure occurred (outward wrinkling and crushing of the
compression skin near mid-span) and the system became unstable. Details of failure
modes are discussed later. This test has clearly provided an insight into the response after
first failure and demonstrated the progressive sequence of failure, and the associated
residual capacity. A comparison of the normalized flexural responses in Tests I and II is
presented later.
5.6.2.2 Deflection distribution across the width
Figures 5.29 to 5.32 show the deflection distribution at various locations across the width
of the panel, at different load levels. The half panel in Test II was supported at mid-
width (i.e. the same location relative to the outer edges, as in Test I). However, because
the half panel did not include a longitudinal GFRP rib at the cut-off edge, the shear center
of the cross-section was slightly off-set with respect to the center of the supports. Also,
while the applied loads were distributed across the full width (i.e. transverse line loads),
the supporting pads were only 305x305 mm. Because of those reasons combined, the
deflection was not uniform across the width. Figure 5.30 shows that deflection near mid
span was generally larger at the cut-off edge (the edge without the rib) than at the outer
rib, by about 23%. Figure 5.32 shows that, at the middle support, the outer edge (with
rib) moved slightly upwards, whereas the cut-off edge had a slight downwards deflection.
107
Figure 5.29 shows that at the outer support both edges moved slightly upwards, more
than at the mid-width point.
5.6.2.3 Load-strain responses
Figures 5.33 to 5.36 show the load-strain responses in the longitudinal direction at
various locations of the loaded and opposite GFRP skins. The maximum tensile strains
are generally lower than the ultimate values obtained from coupon tests, whereas the
compressive strains approached the ultimate values of coupons (Figure 3.9), particularly
where failure occurred.
5.6.2.4 Strain distribution across the width
Figures 5.37 to 5.41 show the longitudinal strain distributions at various load levels,
across the width of the half-panel, on the loaded and opposite skins. The distribution is
given at a section near mid-span (Figures 5.37 and 5.40) and a section adjacent to the
middle support (Figures 5.38 and 5.41). The figures show that the strains at the rib
locations could be higher than other locations. Also, the strains at the middle rib are
higher than the outer rib.
5.6.2.5 Failure modes
Failure occurred first by outward buckling (wrinkling) of the GFRP skin in compression,
adjacent to the middle support, as shown in Figure 5.42, exactly as occurred in the
specimen of Test I (Figure 5.24). It is important to note that the wrinkling was largely
within the area between the support and the outer edge that has the rib (Figure 5.42) and
did not extend all the way to the other cut-off edge of the panel. This means that
longitudinal cutting of the panel and the discontinuity of core and skins in the transverse
direction did not trigger a premature occurrence of this failure. At this point, the load
108
dropped slightly and further loading increased the strength again until shear failure
occurred in the core just next to the middle support (Figure 5.43). This was quite obvious
at the cut-off edge without a rib and was accompanied by localized peeling of the top and
bottom GFRP skins. It could be argued that discontinuity in the transverse direction may
have triggered a premature shear failure. However, as will be demonstrated numerically
in Chapter 6, this was not the case.
Because the panel was continuous over two spans and failure was close to the
middle support, the system remained generally stable as evident by the fact that the load
dropped by only 23% after failure. The system essentially lent itself analogous to two
simply supported spans. This is also analogues to the formation of a plastic hinge in
conventional reinforced concrete continuous beams, where redistribution takes place.
The load continued rising once again and recovered about one third of the lost load
capacity, when the final failure occurred at mid-span by outwards buckling and crushing
of the top compression skin (Figure 5.44). The successive failure modes and associated
load-deflection response in Test II provided a complete picture of the behaviour that was
missing in Test I beyond the first failure.
5.6.3 Comparison between the Responses in Tests I and II
Specimens in Tests I and II were different in loading conditions. Also, in Test II, the
specimen was of half the width. Therefore, normalized flexural responses can be used to
establish a comparison of strength and stiffness of the two panels. Figure 5.45 shows the
moment-curvature responses at a cross-section adjacent to the middle support, where the
first failure occurred in both specimens. The moment was calculated for each case based
on the loading configuration and span, as shown in the schematics in Figure 5.45, using
109
the SAP2000 program. To compare to the full-width specimen in Test I, the moment
obtained in Test II was multiplied by two. The curvature was calculated based on the
slope of the strain profile across the depth of the cross-section, using the average strains
at the loaded and opposite skins at the same cross-section. The figure shows that the
moments at first failure in Tests I and II are quite close with Test II showing a 7% higher
moment. Also, the average flexural stiffness (EI) across the width, which is the slope of
the responses, is quite similar.
5.6.4 Performance Relative to Design Wind Loads
In order to assess the performance of this new cladding panel with respect to design wind
loads for building applications, the maximum design wind pressures in the Province of
Ontario as well as the maximum design value in Canada were considered. The following
equation (NBCC, 2005) is used to estimate the wind pressure P:
P I w q Ce C g C p (5.1)
where q is the pressure load, depends on the geographic region. I w is an importance factor
equal to 1.15 and 0.75 for ultimate and service loads calculations, respectively. C e is an
exposure factor equal to (h/10)0.2, should not be taken less than 0.9 where h is the height
of the building. h in this case is taken the same as the height of the wall (9.2 m). C g is a
gust factor equals to 2.5 for cladding. C p is a pressure coefficient. The product (C e .C p )
depends on the surface area and is equal to 1.5 for the area of the panel used in this study.
The 50 years maximum pressure load q in Ontario, Canada, occurs in the
Cobourg-Newcastle-Port Hope region and is equal to 0.595 kPa, whereas the maximum
in Canada occurs in the North West Territories at Resolution Island, and is equal to 1.235
kPa. Using Eq. (5.1), the maximum design wind pressures P for serviceability limit state
110
are calculated as 0.66 and 1.37 kPa, for Ontario and Canada, respectively. For ultimate
limit state, the maximum P is calculated as 1.0 and 2.1 kPa for Ontario and Canada,
respectively, which are then multiplied by the wind load factor w of 1.4, giving design
values of 1.4 and 2.9 kPa, respectively.
Figure 5.18 shows that under maximum service loads (Ontario and Canada), the
maximum deflection of the panel near mid-span is generally less than or equal to the
(span/360) limit, which is a guide typically used for roof slabs in design codes. Also, the
maximum factored load pressure in Canada, 2.9 kPa, is well below the 7.5 kPa pressure
at failure, giving a 2.6 safety factor from a design stand point.
111
Back-to-back
GFRP woven cloth
GFRP skin
GFRP skin
Supports
d
n d Loa
Wi
Ribs
Polyurethane foam
Figure 5.1 General description and composition of the sandwich cladding panels
Figure 5.2 Panel during shipping and handling
112
Overlap zone
Centre line
Ribs Overlap zone
460 1674 mm 2438 mm 2438 mm 1674 mm 460
609 mm
609 mm
2436 mm
9144 mm
305 mm x 305 mm
support block
75 mm 150 mm C.L. 150 mm 75 mm
78 mm
A 609 mm B 1219 mm 609 mm
2438 mm
Detail A Detail B
Figure 5.3 Full scale sandwich panel dimensions
113
Reaction beams
Side skirt
Air pressure
inlets W460 steel
section
Air pressure
inlets
Figure 5.4 Overview of the apparatus used in Test I
Figure 5.5 Back and side forms of the pressure box
114
W460 columns
Back panel
Polycrepe
seal
Pressure
Load cell
Supporting pad
Sandwich panel
Reaction
beam
Figure 5.6 Cross-section through the sandwich panel and the test apparatus
Figure 5.7 Pressure box with the Polycrepe seal layer
115
Figure 5.8 Wooden supporting blocks (pads)
Figure 5.9 Pressure Load Actuators (PLAs)
116
Figure 5.10 Air pressure inlets and pressure reading sensors
460 1674 mm 2438 mm 2438 mm 1674 mm 460
LP#3 609 mm
130 130 609 mm
LP#1 LP#4
LP#2
Tensile stresses Compressive stresses
Span 4112 mm
Figure 5.11 Deflection measurements using Linear Potentiometers (LPs)
117
Longitudinal Strain Gauges (Loaded Skin)
230
T7
50
195
T3 T6
230 215
T2 T5
210 50
T1 T4
Compressive stresses Tensile stresses
Longitudinal Strain Gauges (Opposite Skin)

460 1674 mm 2438 mm 2438 mm 1674 mm 460
609 mm
50 609 mm
195
B3 B6
230 215
B2 B5
210 50
B1 B4
Span 4112 mm
Figure 5.12 Longitudinal strain measurements
Transverse Strain Gauges (Loaded Skin)
TT4 50
50 50
TT2 TT3
TT1 50
Transverse Strain Gauges (opposite Skin)

460 1674 mm 2438 mm 2438 mm 1674 mm 460
609 mm
609 mm
50
BT2
50
BT1
Span 4112 mm
Figure 5.13 Transverse strain measurements
118
Steel bar 38 mm
1 steel plate strip 6 mm
70 mm
2 ply wood strips 10 mm
2 steel plates strips 6 mm
3*150 mm
C.L. 3 ply wood strips 10 mm
Actuator
Main Spreader Beam Secondary Spreader Beam
78 mm
75 mm
500 mm
Half Sandwich Panel Supporting Pad
Lab Floor
1370 mm 1370 mm 1370 mm 460 mm

4572 mm
Figure 5.14 Half-width sandwich panel in Test II
119
460 1674 mm 2438 mm 2438 mm 1674 mm 460
130 385 130 609 mm
160 609 mm
150 Full deflection Full deflection Deflection check 150

measurements measurements
Span 4112 mm
Figure 5.15 Deflection measurements using Linear Potentiometers (LPs)
Panel bottom view (Strain Gauges)

460 1674 mm 2438 mm 2438 mm 1674 mm 460
85 75
90
130 385 160 130 609 mm
90
609 mm
Tensile stresses Compressive stresses 0.375 span 1541 mm 460 mm
Span 4112 mm
Panel top view (Strain Gauges)

85 75
90 90
130
90 90
130 385
Figure 5.16 Strain gauges distribution for top and bottom skins
120
2.5
2 Bottom Span (LP#1)

Bottom Span Top Span
Pressure (kPa)
1.5
0.5
0
0 5 10 15 20 25 30
Deflection (mm)
Figure 5.17 Load-deflection response for the first test on full scale panel
10
9 Compression skin
Compression skin wrinkling
wrinlking
8
Bottom Span (LP#1)
7
Pressure (kPa)
6
Span / 360
Top Span (LP#4)
5
First test
4
(Up to 2.7 kPa and
3 Max. factored load = 2.9 kPa

CANADA
Max. service load = 1.37 kPa
2
Max. factored load = 1.4 kPa
Ontario
1
Max. service load = 0.66 kPa
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
Figure 5.18 Load-deflection responses for the second test on full scale panel
121
8
Second test top
span (LP#4)
7
Second test bottom
span (LP#1)
6
Third test group bottom
span (LP#1)
5
Pressure (kPa)
4 Third test group top

span (LP#4)

0
0 10 20 30 40 50 60 70 80 90 100
Deflection (mm)
Figure 5.19 Load-deflection responses for the second and third tests for full scale panel
in Test I
8
T3
T2
7 T1 T7 T6 T4
T5
5
Pressure (kPa)
0
-0.005 -0.004 -0.003 -0.002 -0.001 0 0.001 0.002 0.003
Strain
Figure 5.20 Load-longitudinal strain responses for the loaded skin of the full scale
panel in Test I
122
7 B4 B3 B2
B6
B5 B1
6
5
Pressure (kPa)
0
-0.003 -0.002 -0.001 0 0.001 0.002
Strain
Figure 5.21 Load-longitudinal strain responses for the opposite skin of the full
scale panel in Test I
8
7
TT1
TT2 TT3
6
5
Pressure (kPa)
0
-0.0005 0 0.0005 0.001 0.0015 0.002
Strain
Figure 5.22 Load-transverse strain responses for the loaded skin of the full scale
panel in Test I
123
7
BT2 BT1
6
5
Pressure (kPa)
0
-0.004 -0.003 -0.002 -0.001 0 0.001
Strain
Figure 5.23 Load-transverse strain responses for the unloaded skin of the full scale
panel in Test I
Skin crushing near the panel

edge
Skin wrinkling over the middle

support
Figure 5.24 Skin wrinkling and crushing failure of the full scale panel in Test I
124
60
st LP#5 nd
1 failure 2 failure
50
LP#7 rd
3 failure
40 LP#13
Load (kN)
LP#9
30 LP#4
LP#6
st
1 failure = comp. skin wrinkling at middle support
LP#8
nd
20 2 failure = core shear failure near middle support
rd
3 failure = comp. skin crushing at mid-span
10
0
0 20 40 60 80 100 120 140 160 180
Deflection (mm)
Figure 5.25 Load-deflection responses for half-width panel in Test II
60
LP#14
LP#2
50 LP#3
LP#1
40
Load (KN)
30
20
-ve = Upward deflection
10
0
-20 -15 -10 -5 0 5 10
Deflection (mm)
Figure 5.26 Load-deflection responses of the half-width panel in Test II over the
edge support
125
60
LP#10 LP#12
50 LP#11
40
Load (kN)
30
20 -ve = Upward deflection
10
0
-25 -20 -15 -10 -5 0 5 10 15
Deflection (mm)
Figure 5.27 Load-deflection responses over the middle support for the half-width
panel in Test II
(a) Rotation over the edge support
(b) Deflection at mid-span
Figure 5.28 Sandwich panel deformations in Test II
126
9
3
8
+ve = Upward deflection
7
1 2 3
6 45 kN
Deflection (mm)
40 kN
5 1 25 kN
4 35 kN 20 kN
51 kN
3 30 kN
15 kN
2 2
10 kN
1
5 kN
0
0 200 400 600 800 1000 1200
Width (mm)
Test II
0
5 kN
-20 10 kN
15 kN
-40 20 kN
25 kN
Deflection (mm)
-60 30 kN
35 kN
-80 40 kN
4 45 kN
-100 5 51 kN
4 5 6 6
-120 -ve = Downward deflection
-140
-160
0 200 400 600 800 1000 1200
Width (mm)
Test II
127
0
5 kN
-20 10 kN
15 kN
-40 20 kN
25 kN
30 kN
Deflection (mm)
-60
35 kN
-80 40 kN
7 45 kN
-100 51 kN
8
7 8 9
9
-120 -ve = Downward deflection
-140
-160
0 200 400 600 800 1000 1200
Width (mm)
Test II
14
12
10
51 kN
10
10 11 12
45 kN
Deflection (mm)
8
35 kN 40 kN
6 30 kN
25 kN
4 20 kN +ve = Upward Deflection
15 kN
2 10 kN
5 kN 11
12
0
-2
0 200 400 600 800 1000 1200
Width (mm)
Test II
128
60
SG#2 SG#6
SG#17 SG#5
50
SG#3
40
Load (kN)
SG#4
30
SG#1
20
10
0
-0.004 -0.003 -0.002 -0.001 0 0.001 0.002
Strain
Figure 5.33 Load-strain responses for loaded skin at the mid-span regions for
the half-width panel in Test II
60
SG#15
SG#8 SG#13 SG#11
SG#7 SG#14
50 SG#10
40
SG#12
Load (kN)
30
SG#9
SG#16
20
10
0
-0.002 0 0.002 0.004 0.006 0.008 0.01
Strain
Figure 5.34 Load-strain responses for loaded skin at the middle support regions
for the half-width panel in Test II
129
60
SG#13 SG#7
SG#4
50
SG#8
40
SG#12 SG#5
Load (kN)
SG#6
30
SG#11
SG#9
20 SG#10
10
0
-0.01 -0.005 0 0.005 0.01 0.015
Strain
Figure 5.35 Load-strain responses for the opposite skin at the middle support
region for the half-width panel in Test II
60
SG#3 SG#15
50 SG#14
SG#2
SG#1
40
Load (kN)
30
20
10
0
-0.005 -0.004 -0.003 -0.002 -0.001 0 0.001 0.002 0.003 0.004
Strain
Figure 5.36 Load-strain responses for the opposite skin at the mid-span region for
the half-width panel in Test II
130
0.0005
5 kN
-0.0005 10 kN
20 kN 15 kN
30 kN 25 kN
-0.0015
3 5
40 kN 35 kN
Strain
-0.0025 51 kN
2 45 kN
1
-0.0035 1 4 5 4
2 3
-0.0045
-0.0055
0 200 400 600 800 1000 1200
Width (mm)
Figure 5.37 Loaded skin strain distribution over a cross-section passing through
axis 1-2-3-4-5 in Test II
0.009
14 10 7
13 11
0.008
0.007
0.006 10
45 kN
0.005
Strain
11
51 kN
0.004
13
14
0.003 40 kN 35 kN
0.002 30 kN 25 kN 7
20 kN 15 kN
0.001
10 kN
5 kN
0
0 200 400 600 800 1000 1200
Width (mm)
Figure 5.38 Loaded skin strain distribution over a cross-section passing
through axis 14-13-11-10-7 in Test II
131
0.009
0.008
16 12 9
0.007
0.006
12
0.005
51 kN 45 kN
Strain
0.004 16
40 kN
0.003 35 kN
30 kN
25 kN
0.002
20 kN
15 kN
9
0.001 10 kN
5 kN
0
-0.001
0 200 400 600 800 1000 1200
Width (mm)
Figure 5.39 Loaded skin strain distribution over a cross-section passing through
axis 16-12-9 in Test II
0.004
15 1 2
0.003
1
51 kN
Strain
0.002 15 45 kN
40 kN 2
35 kN
30 kN
25 kN
0.001
20 kN
15 kN
10 kN
5 kN
0
0 200 400 600 800 1000 1200
Width (mm)
Figure 5.40 Opposite skin strain distribution over a cross-section passing through
axis 15-1-2 in Test II
132
0.001
0
5 kN
10 kN
10 15 kN
-0.001 20 kN
11 30 kN 25 kN 4
Strain
-0.002 35 kN
40 kN
45 kN
-0.003 11 8 4 7 51 kN
10 7
8
-0.004
-0.005
0 200 400 600 800 1000 1200
Width (mm)
Figure 5.41 Opposite skin strain distribution over a cross-section passing through
axis 11-10-7-8-4 in Test II
Figure 5.42 Compression skin wrinkling over the middle support for the half-
width panel in Test II
133
Figure 5.43 Core shear failure mode near the middle support for the half-width
panel in Test II
Figure 5.44 Top skin crushing and wrinkling failure mode at the mid-span for the
half-width panel in Test II
134
35
M II 2 M II
30
t MI
25
Moment (kN.m)
20
Half Panel (with maximum load
carrying capacity of 51.09 kN)
MI
15
t
10
5 Full Panel (with maximum load

carrying capacity of 167.19 kN)
0
0 0.001 0.002 0.003 0.004 0.005
Curvature x t
Figure 5.45 Moment-curvature responses for both full and half-width sandwich
panels in Test I and II
135
Chapter 6 Finite Element Modeling of Sandwich Panels
CHAPTER 6: FINITE ELEMENT MODELING OF SANDWICH
PANELS
6.1 Introduction
In the previous chapters, the behaviour of composite sandwich panels fabricated from
polyurethane foam core sandwiched between GFRP skins and ribs was investigated
experimentally, along with a simulation of the wind load effect for cladding panels
application. In this chapter, nonlinear finite element analysis (FEA) is carried out on the
panels. Models are developed and used to predict the flexural behaviour and failure
modes of the sandwich panels. Currently, there are several commercial finite element
programs that can be used in the analysis and design of structures. These include SAP,
NASTRAN, ADINA, ANSYS and ABAQUS, among other programs, which are capable
of modeling very complex problems. In this study, program ANSYS was used to model
and analyze the sandwich panels because of its availability.
Many approaches for modeling sandwich panels are based on 2D plates and
shells, in which the sandwich structure is replaced by an equivalent plate or shell element,
with approximated equivalent strains, displacements and stresses (Noor et al, 1996).
Another approach for sandwich panels modeling is the discrete layer model in which the
sandwich panel is divided into three or more layers and for each layer, approximations
are made for the behaviour in the thickness direction (Noor et al, 1996). These models
may be based on the classical sandwich theory or on higher order theories.
136
Sandwich structures with more flexible cores, like foams or non-metallic
honeycombs, should be analyzed using more refined theories than those of Kirchhoff or
Mindlin (Sokolinsky et al., 2003). Factors like the non-uniform strain distribution through
the skin thickness and core depth, and the difference in deflection between the top and
bottom skins may be significant and should be accounted for.
Generally, 3-D FE models may either be partially 3-D models, where the core is
represented by solid elements and the skins by plate or shell elements, or fully detailed
3D models, where both the core and the skins are modeled with solid elements.
In this study, two modeling approaches were considered using the ANSYS
program, namely 2-D and 3-D models. The 2-D model is a simple approach, taking
advantage of the plane stress condition of the sandwich panel in one-way bending. The
second approach used is a partially 3-D model. This approach uses orthotropic shell
elements for the skins and continuum solid elements for the core. For convenience, this
approach will be referenced throughout this study as the 3-D model. Fully detailed 3-D
modeling was not used because the skins are very thin, almost 1.6 mm, compared to the
75 mm thick core. As such, modeling the skins with solid elements would cause problems
with the elements aspect ratio, and would require an extremely refined mesh which would
not be practical in terms of the cost and time of analysis. The numerical work developed
and presented in this study takes advantage of the versatility of the FE techniques and
establishes a methodology with a reasonable trade-off between the level of accuracy
desired and computational efficiency.
This chapter describes the finite element analysis carried out on the sandwich
panels to investigate their flexural behaviour and failure modes. The complete description
137
of the models, including the types of elements, meshing, boundary conditions, and
methods of analysis are presented. The correlations between experimental and numerical
results are also presented. The FE models were developed and verified for the sandwich
panels tested in this study as well as other panels tested in a previous study. As such, the
next section summarizes the other experimental study, before presenting the model.
6.2 Summary of Another Experimental Program (Shawkat, 2008)
Ten sandwich panels without ribs were tested as a part of a masters thesis (Shawkat,
2008), at Queens University, to examine their flexural behaviour. The sandwich panels
were loaded using different configurations, namely three-point bending, four-point
bending and a uniform load. The uniform load was simulated by means of eight
concentrated loads acting on the top skin of the sandwich panel.
The panels consisted of two types of polyurethane foam core, namely soft (low-
density) and hard (high-density) polyurethane foam, sandwiched between two GFRP
skins. A comprehensive material testing program of the different foams and GFRP was
carried out by the author of the current thesis, as presented in Chapter 3. The panels did
not include any ribs and were all tested in one-way bending.
Table 6.1 provides a summary of the test matrix and parameters for that study.
Specimens P1 to P6 were fabricated from the 32.04 kg/m3 (2 pcf) density foam while
specimens P7 to P10 were fabricated from the 64.08 kg/m3 (4 pcf) density foam.
Specimens P1 and P2 were tested in 3-point and 4-point bending, respectively, while all
other specimens were tested under uniformly distributed load simulated by eight discrete
loads. Figure 6.1 shows the different loading configurations. All specimens, except P6
and P10, were tested monotonically at a rate of 1 mm/min. Specimens P6 and P10 were
138
tested under low cyclic fatigue. Specimens (P3 to P5) and (P7 to P9) were identical in
each group, to confirm repeatability of performance and test results. Specimens P1, P2
and (P3 to P5) were compared to examine the effect of loading configuration. All the
panels had a span of 1400 mm with their full breadth (300 mm) supported. Also, loads
were applied across the full width. The line loads were applied using rigid steel strips
resting on a Teflon sheet, in order to protect the panel from stress concentration. Four
strain gauges were used to monitor the longitudinal strains, at the mid-span line, with two
on the upper skin (compression) and two on the lower skin (tension). Mid-span deflection
was monitored using two linear potentiometers. The results of this experimental study are
presented later, with the FE model calculations, in the model validation section.
6.3 Finite Element Model
The 2-D model consisted of beam elements to model the top and bottom skins while
plane stress elements were used to model the polyurethane foam core. For the 3-D model,
shell elements were used for modeling the skins and solid elements were used for the
foam core. Since the plane stress elements and the solid elements only have translational
degrees of freedom (TDOF), while their corresponding beam elements and shell
elements, respectively, have additional rotational degrees of freedom (RDOF), the
constraint DOFs between each corresponding node for both 2-D and 3-D models were
defined only for these TDOFs. However, when the independent node of the skin element
undergoes an angular displacement, the dependent node of the core element has an
equivalent translational displacement, allowing for the deformation compatibility
between the skins and the core elements. The displacement of the independent node of
the skin element is calculated from the element shape functions which take into account
139
the rotational degrees of freedom for that element. Therefore, the corresponding nodes at
the interconnecting line (in the case of 2-D model) or surface (in the case of 3-D model)
will have the same deflection values without violating the defined DOFs at those nodes.
6.3.1 The 2-D Model
The full scale model is an easy way of FE analysis as it uses the actual full dimensions
of the structure and does not require any parametric study to scale down the dimensions
of the model (Pokharel and Mahendran, 2004). However the drawback of using this
concept is a potential lower level of accuracy obtained due to the limited number of
elements used, relative to the model size. However, to overcome this problem, using
refined models will lead to uneconomical and large computational runtime.
To overcome these difficulties, a reduced size model (not scaled down) with
appropriately determined dimensions can be used, taking advantage of the symmetry of
the sandwich panel along both the longitudinal and transverse axes. In this model, only
one-half of the full panel was analyzed, using appropriate boundary conditions to
simulate the symmetry. This allowed the use of a mesh with a large number of elements
of smaller sizes to increase the level of accuracy of the numerical results. Figure 6.2
shows the geometry of the sandwich panels tested by Shawkat (2008), and the
dimensions used in the FEM.
The FEM was taken 3 mm in breadth (i.e. as a 1% of the total 300 mm breadth).
The GFRP skins were modeled using 2D beam element BEAM23 with a cubic
displacement function. BEAM23 is a uniaxial element with tension, compression, and
bending capabilities. The element has three degrees of freedom at each node, namely
translations in the nodal x and y directions and rotation about the nodal z-axis. Figure
140
6.3(a) shows the geometry, node locations, and the coordinate system for this element.
The element is defined by two nodes, the cross-sectional area, the moment of inertia, the
thickness (skin thickness), and the material properties.
The polyurethane foam core was modeled using plane stress element PLANE82
with quadratic displacement function and a capability of thickness effect. It is an eight
node element with a capability of triangular shapes with 6-nodes. The 6-node triangular
element is defined by three corner nodes and three mid-side nodes. Each node has two
degrees of freedom, which are translations in the nodal x and y directions. The thickness
effect is taken into consideration to simulate the accurate moment of inertia as well as the
overall bending stiffness of the sandwich panel. Figure 6.3(b) shows the geometry, node
locations, and the coordinate system for this element.
The interface between the GFRP skins and the polyurethane foam core was
considered in full contact state. In all the tested panels in this study, and those by
Shawkat (2008), even though they are without ribs, it was observed that no relative slip
ever occurred between the skins and the adjacent part of the core. The corresponding
nodes were coupled together using the translational degrees of freedom as mentioned
before. Both, the material nonlinearity for the GFRP and the polyurethane foam (results
were presented in Chapter 3) and geometric nonlinearity were taken into consideration.
Also, the material anisotropy was taken into account. The applied loads were modeled as
line loads to simulate as much as possible the exact loading condition in the experimental
study. The mesh was refined at the interface between the skins and the core to obtain an
accurate deformed profile of the skins as well as to obtain a converged solution. Mesh
density could indeed affect the accuracy of the finite element model, therefore, it is
141
necessary to have a fine enough mesh to obtain the appropriate solution, which is further
validated by comparing with the experimental results.
In order to determine the appropriate mesh density, a mesh refinement process
and a convergence study was conducted. Three preliminary numerical simulations of
different mesh densities, namely, coarse, intermediate, and fine, were first carried out on
the sandwich panel, as shown in Figure 6.4. Different sizes of the elements in the GFRP
skin and the polyurethane foam core are also shown in the figure. The total number of
elements was 1171 elements for the coarse mesh, 3278 elements for the intermediate
mesh and 8216 elements for the fine mesh. The calculated maximum load carrying
capacity for each mesh configuration is plotted in Figure 6.5. The figure shows almost
identical results with minor changes in the maximum load capacities when refining the
model beyond the coarse mesh. However, the computer run-time dramatically increases
with refining the mesh size. As such, to obtain a smooth deformed profile for the
sandwich panel, the intermediate mesh was deemed sufficient and was used in all the
analyses that followed. Figure 6.6 shows the final 2-D mesh and loading configuration.
6.3.2 The 3-D Model
6.3.2.1 The 3-D Model for Panels Tested by Shawkat (2008)
The 3-D FE model was used to model the behaviour of sandwich panels tested by
Shawkat (2008) as well as the sandwich panels tested in the present study. Because of the
symmetry in two directions, only one quarter of the panel is studied for most cases. The
dimensions of the geometric model used in the 3-D analysis of the panels tested by
Shawkat (2008) are shown in Figure 6.7.
142
A 3-D shell element SHELL99 was used to model the GFRP skins. The element
is defined by eight nodes and has both bending and membrane capabilities. It has six
degrees of freedom at each node, namely translations in the nodal x, y, and z directions,
and rotations about the nodal x, y, and z-axes. The element has a quadratic displacement
function. The geometry, node locations, and the coordinate system for this element are
shown in Figure 6.8(a). A 3-D SOLID186 element was used to model the polyurethane
foam core and the loading steel plates as well. The element is defined by twenty nodes
having three degrees of freedom at each node, namely translations in the nodal x, y, and z
directions. Also, it has a quadratic displacement function. The geometry, node locations,
and the coordinate system for this element are shown in Figure 6.8(b).
Full contact behaviour is considered between the two GFRP skins and the
polyurethane foam core, as indicated earlier. Full coupling constraints were applied
between the corresponding nodes at both the GFRP skin and the polyurethane foam core
for the translational degrees of freedom. The nonlinear material behaviour for both GFRP
and polyurethane foam was taken into account. Also, the geometric nonlinearity due to
the softness of the polyurethane foam core was considered. The applied load was in the
form of line loads acting on the steel plates as a simulation of the experimental
conditions.
A surface-to-surface contact pair CONTA174 and TARGE170 were used to
simulate the contact between the loading steel plates and the top skin and also between
the bottom skin and the support plates. CONTA174 is used to represent contact and
sliding between 3-D target surfaces (TARGE170) and a deformable surface, defined by
this element. The element is applicable to 3-D structural analyses. This element is located
143
on the surfaces of 3-D solid or shell elements that have mid-side nodes. The element has
eight nodes as the underling element, SHELL99 or SOLID186. It has the same geometric
characteristics as the solid or shell element face with which it is connected. Contact
occurs when the element surface penetrates one of the target segment elements
(TARGE170) on a specified target surface. The geometry and node locations for both
elements are shown in Figure 6.8 (c), and (d).
In order to determine the appropriate mesh density, a mesh refinement process
and a convergence study was conducted. Three preliminary numerical simulations of
different mesh densities, namely, coarse, intermediate, and fine, were first carried out on
the sandwich panel, as shown in Figure 6.9. Different sizes of the elements in GFRP skin
and the polyurethane foam core are also shown in the figure. The total number of
elements was 3808 elements for the coarse mesh, 18045 elements for the intermediate
mesh and 52676 elements for the fine mesh. The calculated maximum load carrying
capacity for each mesh configuration is plotted in Figure 6.10. The figure shows very
close results when refining the model beyond the coarse mesh. However, the computer
run-time dramatically increases with refining the mesh size. Therefore, to obtain a
smooth deformed profile for the sandwich panel, the intermediate mesh was deemed
sufficient and was used in all the analyses that followed. Figure 6.11 shows the final 3-D
mesh and loading configuration.
6.3.2.2 The 3-D Model for Panels with different Rib Configurations
(Chapter 4)
For the panels tested in the present study with different rib configurations (Chapter 4), the
FEM concept was the same as described in the previous section, except for the addition
144
of the GFRP ribs in this case. The ribs were also modelled using SHELL99 element as
the skins. Full contact behaviour was assumed between the rib and the adjacent core. The
corresponding nodes were coupled for the translational degree of freedom. Since the
loading areas on the panels consisted of three rubber strips, loading was applied directly
to the top skin in the form of a uniform pressure within an effective foot print of the
rubber strips (Figure 6.14). The contact element pairs were used only for the support
region because the flexible loading rubber strips were assumed to always be in full
contact with the skin.
A mesh refinement process and a convergence study were conducted in order to
determine the appropriate mesh density required for the FEM. The same technique used
for the previous 3-D model (Shawkat, 2008) was used. Three preliminary numerical
simulations of different mesh densities, namely, coarse, intermediate, and fine were
conducted. The three models were carried out on the control sandwich panel (S1). Figure
6.12 shows the three FEM with the corresponding different sizes of the elements in the
GFRP skin and the polyurethane foam core. The total number of elements was 12154
elements for the coarse mesh, 36834 elements for the intermediate mesh and 77170
elements for the fine mesh. The calculated maximum load carrying capacity for each
mesh configuration is plotted in Figure 6.13. The figure shows very close results in the
maximum load capacities when refining the model beyond the coarse mesh. It was
decided that the intermediate mesh was deemed sufficient and was used in all the
analyses that followed. Figure 6.14 shows the final 3-D mesh and loading configuration.
145
6.3.2.3 The 3-D Model for the Large Scale Cladding Panels (Chapter 5)
For the large scale sandwich panels tested and presented in Chapter 5, the same concept
for the FEM described in the previous section, was used with the proper loading
condition for the half-width panel that was tested under discrete loading (Figure 6.15(a)).
For the full size panel tested under uniform air pressure, the FE loading was also a
uniform pressure over the top skin area (Figure 6.15(b)). Also, in both cases, contact
element pairs were used to model the contact between the sandwich panel and the
wooden supports. Because of the slight difference in the length of the two over hanged
ends of the panel tested under air pressure, it was decided to model one half, rather than
one quarter of the panel. For the half-width panel tested under mechanical loading, the
same FE model was used, namely one half, with equal over hanged ends. Because of the
panel large dimensions of length and breadth, compared with the overall panel thickness,
the FE model was meshed according to the previous 3-D models with a larger element
surface area size, in the direction of the panel length and width, while making sure not to
violate the element aspect ratio that could be affected as a result of the mesh. The 3-D
finite element model mesh and loading configuration are shown in Figure 6.16.
In order to assess the effect of discontinuity of the cut-off edge of the half-width
panel, an additional hypothetical FEA model was done assuming the panel had the full
width (i.e. no cut-off edge) but taking advantage of symmetry, using the appropriate
boundary conditions.
6.3.3 Material Models
As was shown in Chapter 3, the constituent materials have a nonlinear behaviour,
especially the polyurethane foam. As such, it was important to take material nonlinearity
146
into account in the FEM model. The GFRP non-linearity, especially in compression, was
accounted for in the FEM by specifying a bi-linear isotropic hardening model for both,
tension and compression, as shown in Figure 6.17. Unidirectional elastic properties were
assigned, namely, Youngs modulus of 26.0 GPa with a tangent modulus of 16.2 GPa for
tension, and Youngs modulus of 33.6 GPa with a tangent modulus of 2.7 GPa for
compression. The bi-linear curve was based on the average of the experimental curves
obtained from the tested coupons in Chapter 3. The GFRP material shear behaviour was
also assumed to be bi-linear with shear modulus of 1.96 GPa and a tangent modulus of
0.059 GPa, as shown in Figure 6.18.
For the polyurethane foam material, a hyperelastic material polynomial curve
fitting was used to fit the material tension, compression and shear stress-strain responses
(ANSYS). The used polynomial was of a fifth order and it represents the strain energy
potential, and takes the following form:

N N
1 (6.1)
I2 3 J 1
i j 2k
W cij I 1 3
i j 1 k 1 d k
where W is the strain energy potential and I 1 and I 2 are the first and the second deviatoric
strain invariant, respectively. J is the determinate of the elastic deformation gradient and
c ij and d k are material constants based on the material stress-strain curves. The
polynomial order N was taken five to obtain better curve fitting results. The resulted
material curve fittings for soft and hard polyurethane foam under both tension and
compression are shown in Figure 6.19 while Figure 6.20 shows the shear curve fitting for
soft and hard polyurethane foam. This process is carried out by the program ANSYS,
147
where data files of all stress-strain curves are input to the program, and the fitting is then
carried out.
6.3.4 Failure Criteria
Two types of failure were considered, namely, a stability failure such as skin wrinkling
(local buckling, inward or outward), and material failure, such as foam core shear failure
or compression skin crushing. Stability failure was detected when the solution fails to
converge, suggesting very large displacement or sudden displacement that leads to a
highly distorted model and cause the run to terminate. Material failure was detected when
the ultimate strain (or stress) of any of the two materials used is reached in tension,
compression or in shear. As this limiting strain (or stress) is reached, the stress level is
locked. The model is incapable of eliminating the failed material elements from the
global stiffness matrix beyond this point. As such, the model can not predict the post-
peak behaviour (i.e. the behaviour beyond the first failure). To overcome this problem, if
desired, the failed elements should be removed from the model one-by-one in a step-by-
step solution procedure, which is quite tedious and will take a very long run time.
6.4 Evaluation of the Finite Element Model
6.4.1 Panels Tested by Shawkat (2008)
Figures 6.21 to 6.24 show the load-deflection responses for the sandwich panels tested by
Shawkat (2008), using both the 2-D and the 3-D FEA, compared with the experimental
results. It can be seen that in modeling the behaviour of panels without ribs, and tested in
one-way bending, the 2-D FEA gave very close results to the 3-D model (1.2%
difference). The load-strain responses of the panels are shown in Figures 6.25 to 6.28.
Figures 6.29 to 6.32 show the deformed shapes for the panels under different load
148
schemes. Panels P1 and P2 under three-point and four-point bending failed
experimentally and numerically by compression skin inwards wrinkling under the
concentrated loads due to a localized foam core compression failure, as shown in Figure
6.29 and 6.30. The failure mode for the panels loaded with a uniform eight loads is
shown in Figure 6.33 for both the soft foam (density of 32.04 kg/m3) and hard foam
(density of 64.08 kg/m3), respectively.
Panels P3 to P5 with soft core failed by excessive shear deformation of the core
(Fig. 6.33(a)), associated with excessive deflection (Figure 6.31). Panels P7 to P9 with
hard core failed by cracking of core under diagonal tension due to shear, which was
immediately followed by delamination from the skin (Fig. 6.33(b)). This failure is shown
numerically in Figure 6.33, where the shear stress reached the failure stress of the shear
coupons tested before. In general, the finite element model for those panels showed a
good agreement with the experiential results.
6.4.2 Sandwich Panels with different Rib Configurations (Chapter 4)
The load-deflection responses for the sandwich panels with different rib configurations,
S1 to S6, tested in this study and presented in Chapter 4, including the control specimen
without any ribs (S1), are shown in Figure 6.34. These panels were modeled using the 3-
D model. The FEA results were generally linear with a good agreement with the
experimental results, within a range of 5%. Figures 6.35 to 6.40 show load-strain
responses while Figures 6.41 to 6.46 show the obtained failure modes for the panels.
Panel S1 failed experimentally by outward wrinkling of the top skin between the first and
the second loading points, instead of failing at mid-span between the second and the third
loading points as predicated by the numerical 3-D model, and shown in Figure 6.41(a).
149
The calculated failure within the 3-D model was due to outward wrinkling (local
buckling) of the top skin at the panel centre span which was indicated by solution
divergence due to instability and excessive top skin deformation. Figure 6.41(b) shows
the Tsai-Wu failure criteria index values. It should be noticed that the Tsai-Wu values are
less than unity which indicates that failure was due to instability (top skin wrinkling) and
not due to material failure.
The failure mode for the panel that has one longitudinal rib at mid-width, S2, is
shown at Figure 6.42(a). The panel failed experimentally by top skin crushing at the edge
of the flange of the rib near the panel centre line and also by outward wrinkling of the top
skin at the panel edge. The calculated failure mode was the same as the experimental one
for the top skin crushing at the flange edge of the rib, which was indicated by solution
divergence due to GFRP material failure, as shown in Figure 6.42(b). The figure shows
the Tsai-Wu failure criteria index at failure of the panel reaching unity, indicating
material failure of GFRP, similar to the experimental observation. The model did not
predict the outward wrinkling failure mode near the panel edge, but instead, the model
showed the tendency for an inward wrinkling at the same position which can be shown
by the strain behaviour at this edge increasing at a higher rate near the end (Figure 6.36
points b and c).
For panel S3, with longitudinal and transverse ribs, the failure mode is shown at
Figure 6.43(a). The panel failed experimentally by top skin crushing at the transverse rib
edge, near the panel centre line. The calculated failure mode was also due to top skin
crushing at the intersection zone between both ribs, the longitudinal and the transverse
ribs, which was indicated by solution divergence due to GFRP material failure, as shown
150
in Figure 6.43(b). The figure shows the Tsai-Wu failure criteria index reaching 1.0 at
failure.
Figure 6.44(a) shows the experimental failure mode for panel S4, with an exterior
rib (diaphragm), which failed by outward wrinkling and splitting at the skin-rib joint. The
failure was due to the lake of flange in the rib which triggered the skin to separate from
the exterior rib at the edge line. The calculated failure model by the FE model was due to
excessive compression near the same position (near the panel edge at mid-span) as
indicated in Figure 6.44(b) by the Tsai-Wu failure index reaching 1.0. On the other hand,
the FE model did not show the skin-outer diaphragm separation because of the full bond
assumption between both the top skin and the exterior rib (diaphragm). Nonetheless, the
calculated failure load was very close to the experimental value (Figure 6.34).
The observed experimental failure mode for panel S5 (with longitudinal and
exterior ribs) was top skin crushing near the panel centre line as shown in Figure 6.45(a).
The calculated failure mode by the FE model was also top skin crushing near the panel
mid-span. Figure 6.45(b) shows the Tsai-Wu failure criteria index reaching 1.0.
For panel S6 with longitudinal and transverse ribs in addition to the exterior rib,
the experimental failure mode was due to top skin crushing at the transverse rib edge near
the panel centre line, as shown in Figure 6.46(a). The calculated failure mode by the FE
model was the same, as shown in Figure 6.46(b) by the value of Tsai-Wu failure criteria
index reaching 1.0.
6.4.3 Large Scale Cladding Panels (Chapter 5)
This section presents the results of the 3-D FEA models for the large scale panel tests
reported in Chapter 5. The panel load-deflection curves obtained from the FE models
151
were plotted along with the experimental results in Figure 6.47 for the full width panel
tested under air pressure, and in Figure 6.48 for the half-width panel tested under discrete
loads. Figure 6.49 shows additional load-deflection responses of the half-width panel
near the support points. It can be observed that the FEA prediction has a good agreement
with the experimental results. However, there is some tendency to overestimate (by 12%)
the ultimate pressure and deflection (Figure. 6.47). It is also noted that the FE model
captures the peak loads at first failure only but could not capture the response beyond this
point and the load drop. This would have required special techniques of removing
manually the failed elements, step-by-step. Figures 6.50 to 6.53 show the pressure-strain
responses for the large scale cladding panel tested under air pressure while Figures 6.54
to 6.57 show the load-strain responses for the half-width large scale panel tested under
discrete loading. Generally, reasonable agreement is observed.
The observed failure mode for the large scale cladding panel was due to
compression skin wrinkling at the middle support region with a skin crushing at the same
region and near the panel edge. The failure mode predicted by the FE model is shown in
Figure 6.58(a) for the full width panel in the experimental program along with the
corresponding FE Tsai-Wu failure criteria index value. The model predicted skin
crushing as shown by the Tsai-Wu index reaching 1.0. For the half-width large scale
panel, the predicted failure mode was compression skin wrinkling and crushing near the
middle support zone as was observed in the experimental program. Figure 6.58(b) shows
the panel failure along with the corresponding Tsai-Wu failure criteria index reaching
1.0.
152
In Chapter 5, it was shown that after the half-width panel failed by skin wrinkling
near middle support, the load slightly dropped then rose again to a very close value of the
first peak, when a second failure occurred by shear in the core. This shear failure
occurred very clearly at the cut-off edge, which raises the question whether the
discontinuity in the transverse direction have affected the behaviour and triggered a
premature shear failure or not. For this reason the additional FE analysis of a hypothetical
panel with full width was carried out for the test under discrete loads (test II, but
assuming full width). Figure 6.59 shows the load-deflection responses for both cases at
the point of maximum deflection (point LP9). The responses and failure load are quite
similar, suggesting that the cut-off edge did not have any negative effect on the first
failure load (first peak). To see if discontinuity had an effect on the second peak (shear
failure load), the shear stresses in the foam core and GFRP ribs are also examined in both
analyses. Figure 6.60(a and b) shows the shear stresses in the core just when first failure
occurred in the skin (i.e. at the last point the FE model was able to generate). Clearly the
shear stress values in both cases are very close, 0.151 and 0.157, respectively, suggesting
that the discontinuity did not have a significant effect. Furthermore, by plotting these
values on the stress-strain curve (Figure 6.60(c)), it can be seen that the values are very
close to the ultimate strength, suggesting that shear failure was imminent, which was
indeed the case after wrinkling. Similarly, Figure 6.61(a, b, and c) prove the same point
for the GFRP ribs, which may have also failed in the experiment but was not obvious. In
conclusion, it is safe then to say that the first failure load represents that maximum peak
strength.
153
Table 6.1 Summary of the experimental program by Shawkat (2008)
Chapter 6
Dimension (mm) ultimate

Specimen Density Loading Mu Pu u
3 Failure Mode Parameters
ID (kg/m ) Skin Total type (KN.mm) (KN) (mm)
Width Top Bottom
Thckness thicknes
3-Point
P1 890 2.54 27.8 -0.0036 0.00148 Wrinkling (inward) L
bending
4-Point
P2 1064 4.56 31 -0.0016 0.00144 Wrinkling (inward) L
bending
Excessive shear
P3 Uniform 1196 6.84 61 -0.0016 0.0012 D R
deformation of the core
32.04
Excessive shear
P4 Uniform 1295 7.4 65.9 -0.0017 0.00157 D R
Excessive shear
P5 Uniform 1275 7.3 64.9 -0.002 0.00134 L D R
154
1.6 78 300
Uniform
P6 1413 8.1 68.7 -0.0011 0.00175 Core shear crack C
(Cyclic)
P7 Uniform 3156 18.1 61.5 -0.0031 0.00495 Core shear crack L D R
P8 Uniform 3020 17.3 58.7 -0.0034 0.00406 Core shear crack D R

64.08
P9 Uniform 3402 19.44 65.9 -0.004 0.00417 Core shear crack D R
Uniform
P10 3069 17.54 40 -0.0014 0.00382 Wrinkling+Shear C
(Cyclic)
R = Repeatability Mu = Ultimate moment
C = Cyclic loading Pu = Ultimate load
D = Foam density effect u = Ultimate deflection
L = Loading configuration u = Longitudinal strain
Finite Element Modeling of Sandwich Panels
(a) Three-point bending (P1)
(b) Four-point bending (P2)
(c) Uniform Load (P3-P10)
Figure 6.1 Test setup of the panels tested by Shawkat (2008)
155
Symmetry axis Loading pads

(300 mm x 5 mm)
Load
300 mm
78 mm
700 mm 700 mm
50 mm 50 mm
Sandwich Panel
78 mm
50 mm
700 mm
2D FEA Model
50 mm
Figure 6.2 Sandwich panel geometry and the corresponding 2-D FEM
(for specimens tested by Shawkat (2008))
(a) 2-D BEAM23 element for skin
(b) 2-D PLANE82 stress/strain element for core
Figure 6.3 Elements used in the 2-D FEM (ANSYS)
156
(a) FE coarse mesh
(b) FE intermediate mesh
(c) FE fine mesh
Figure 6.4 Finite element mesh used in the convergence study for the
2-D FEA
157
7.75
Maximum predicted load carrying capacity (kN) 7.74
7.73
+ 4.88 % error
+ 4.61 % error
+ 4.58 % error
7.72
7.71
7.7
0 Coarse1 mesh 2 mesh
Intermediate Fine3mesh 4
tested under simulated uniform load Shawkat (2008))
(A)
Detail (B)
(B)
Node
Width of loading pad is 50 mm
Detail (A)
Figure 6.6 Final 2-D FEM mesh and loading configuration for
panels tested by Shawkat (2008)
158
Symmetry axis Loading pads

(300 mm x 5 mm)
Load
300 mm
78 mm
700 mm 700 mm
50 mm 50 mm
Sandwich Panel
78 mm 150 mm
50 mm
700 mm
3D FEA Model
50 mm
Figure 6.7 Sandwich panel geometry and the corresponding 3-D

FEM for specimens tested by Shawkat (2008)
(a) 3-D SHELL99 element (b) 3-D SOLID186 element

Associated target element
Contact element
Surface of underlying element Target element surface
(c) 3-D CONTA174 element (d) 3-D TARGE170 element
Figure 6.8 Elements used in the 3-D FEM (ANSYS)
159
(a) FE coarse mesh
(b) FE intermediate mesh
(c) FE fine mesh
Figure 6.9 Finite element mesh used in the convergence study for the
3-D FE model for the panels tested by Shawkat (2008)
160
7.75
Maximum predicted load carrying capacity (kN) 7.73
7.71
+ 4.61 % error
+ 4.14 % error
+ 4.07 % error
7.69
7.67
7.65
Figure 6.10 Convergence study results for the 3-D FE model (the case of
panels tested under simulated uniform load Shawkat (2008))
Line load
Steel plate
Top skin
50 mm
Foam core
Figure 6.11 Final 3-D FEM mesh for the panels tested by Shawkat (2008)
161
(a) Coarse mesh
(b) Intermediate mesh
(c) Fine mesh
Figure 6.12 Finite element mesh used in the convergence study for the 3-D
FE model for panels with different rib configuration (Chapter 4)
162
15.9
Maximum predicted load carrying capacity (kN)

15.7
15.5
15.3
15.1
+ 4.98 % error
+ 8.83 % error
+ 4.52% error
14.9
14.7
14.5
14.3
Figure 6.13 Convergence study results for the 3-D FE model for panels
tested in Chapter 4
Top skin Uniform pressure
Load
125 mm 125 mm
18 mm thick
rubber sheet
2*125 +2*18 = 286 mm

286 mm
3*125 mm
Figure 6.14 Final 3-D FEM mesh panels with different rib configurations
(Chapter 4)
163
Free boundary Four localized pressure loads

on 450 mm x 1219 mm area
Supporting pad
(a) Half width-panel tested

under mechanical loading 450 mm
1219 mm
Axis of symmetry Uniform pressure on

the entire surface
Supporting pad
Roller support
(b) One half of the full panel

tested under air pressure
Figure 6.15 FEA geometric models and assumptions of large scale cladding
panels (Chapter 5)
Internal rib Polyurethane foam core
Longitudinal Center line
GFRP skin
Supporting pad
Figure 6.16 Final 3-D FEM for the large scale sandwich panels (Chapter5)
164
350
288.7 MPa
300
250
200
Stress (MPa)
150 130 MPa
100
50
Tension
0
Compression
-50 80.7 MPa
-100 85.3 MPa

0.0041 0.0024 0.005 0.0148
-150
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
Strain
Figure 6.17 Bi-linear tension and compression stress-strain curves for GFRP
skins.
50
45 43.42 MPa
40
35 32 MPa
Shear stress (MPa)
30
25
20
15
10
5
0.0163 0.2114
0
0 0.05 0.1 0.15 0.2 0.25
Shear strain
Figure 6.18 Bi-linear Shear stress-strain curve for GFRP ribs
165
1.0
Tension
0.5
0.0
Compression
Stress (MPa)
-0.5
3
32.04 kg/m
density foam
-1.0
-1.5
3 Exp.
64.08 kg/m
-2.0 density foam Polynomial curve fitting
-2.5
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2
Strain
Figure 6.19 Polynomial curve fitting for soft and hard polyurethane foam,
tension and compression stress-strain curves
0.50
0.45
3
0.40 64.08 kg/m
density foam
0.35
Shear stress (MPa)
0.30
3
0.25 32.04 kg/m
density foam
0.20
0.15
0.10
Exp.
0.05 Polynomial curve fitting
0.00
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Shear strain
Figure 6.20 Polynomial curve fitting for shear stress-strain curves for both soft
and hard polyurethane foam
166
3
Nonconference due to excessive
inward deformation (wrinkling)
2.5
Skin wrinkling
2
Load kN
1.5
0.5 Experimental Results

2D FEA
3D FEA
0
0 5 10 15 20 25 30
Deflection mm
Figure 6.21 Load-deflection response for panel (P1) with soft polyurethane foam
core and loaded in three-point bending
6
Nonconference due to excessive
inward deformation (wrinkling)
5
4
Skin wrinkling
Load kN
Experimental Results
1 2D FEA
3D FEA
0
0 5 10 15 20 25 30 35
Deflection mm
foam core and loaded in four-point bending
167
9.0
Core shear failure
8.0
P4
7.0
P5
6.0 P3 Excessive
deformation of the
core
Load (kN)
5.0
4.0
3.0
Experimental (P3)
Expermintal
2.0 Experimental (P4)
Expermintal
Experimental (P5)
Expermintal
1.0 2D FEA
3D FEA
0.0
0 10 20 30 40 50 60 70 80
Deflection (mm)
Figure 6.23 Load-deflection responses for panels (P3, P4 and P5) with soft
polyurethane foam core and tested under a uniform load
25
Core shear failure
20 P9
P7
15 P8
Core shear failure
Load (kN)
10
Experimental
Expermintal (P7)
5 Experimental
Expermintal (P8)
Experimental
Expermintal (P9)
2D FEA
3D FEA
0
0 10 20 30 40 50 60 70
Deflection (mm)
Figure 6.24 Load-deflection response for panels (P7, P8 and P9) with hard
polyurethane foam core and tested under a uniform load
168
3.0
FEM FEM
SG#1
2.5
SG#1
Excessive strain is
localized at wrinkle SG#2
2.0 SG#2
Load (kN)
1.5
C.L.
inward
Load
1.0 wrinkling
Top skin
0.5 Bottom skin
0.0
-0.020 -0.015 -0.010 -0.005 0.000 0.005
Strain
Figure 6.25 Load-strain response for panel (P1) with soft polyurethane foam core and
loaded in three-point bending
5.0
SG#1
FEM
SG#1
FEM
4.0 SG#2
3.0
Load (kN)
SG#2
C.L.
2.0
1.0
0.0
-0.003 -0.002 -0.001 0.000 0.001 0.002
Strain
core and loaded in four-point bending
169
9.0
8.0 FEM
P4 FEM P5
P5 P4
7.0
P3
P3
6.0
Load (kN)
5.0
C.L.
4.0
3.0
Average
2.0
1.0
0.0
-0.004 -0.003 -0.002 -0.001 0.000 0.001 0.002 0.003
Strain
Figure 6.27 Load-strain response for panels (P3, P4 and P5) with soft polyurethane
foam core and tested under a uniform loading
25
20 FEM P9
P9 FEM
P7 P8
15 P8 P7
Load (kN)
10
Average
5
0
-0.008 -0.006 -0.004 -0.002 0.000 0.002 0.004 0.006
Strain
Figure 6.28 Load-strain response for panels (P7, P8 and P9) with hard
polyurethane foam core and tested under a uniform loading
170
(a) Deformed shape under

three-point bending
(b) FE model deformed shape

under three-point bending
(c) Vertical displacement
Units in mm
Figure 6.29 Deformed shape under three-point bending for panel P1
171
(a) Deformed shape under

four-point bending

under four-point bending
Units in mm
Figure 6.30 Deformed shape under four-point bending for panel P2
172
(a) Deformed shape under a

uniform loading

under a uniform loading
Units in mm
Figure 6.31 Deformed shape under uniform load for panels P3, P4 and P5
with soft polyurethane foam core
173
(a) Deformed shape under a

uniform loading

under a uniform loading
Units in mm
Figure 6.32 Deformed shape under uniform load for panels P7, P8 and P9
with hard polyurethane foam core
174
(a) Panel with soft polyurethane

foam core
Shear stress distribution
Units in MPa
(b) Panel with hard polyurethane

foam core
Shear stress distribution
Units in MPa
Figure 6.33 Shear failure modes within the soft and hard polyurethane foam
core under uniform loads
175
40
Top skin crushing at mid-span
35 Exp.
S6
FEA
S5 S2
30
S3
25 Top skin
Load (kN)
crushing at
mid-span
20
S4 Top skin wrinkling
at mid-span
15
10 S1
0
0 10 20 30 40 50 60 70 80
Deflection (mm)
Figure 6.34 Load-deflection responses for sandwich panels with different rib
configurations reported in Chapter 4
16
e b b d a
14
a c e
d a,b,c,d,e a,b,c,d,e
12 c
Load (kN)
10
4
Exp.
2 FEA
0
-0.004 -0.003 -0.002 -0.001 0.000 0.001 0.002 0.003 0.004
Strain
Figure 6.35 Load-strain responses for panel S1
176
30
b b
c
25
c
a a
20
Load (kN)
15
10
Exp.
5 FEA
0
-0.006 -0.004 -0.002 0.000 0.002 0.004 0.006
Strain
30
b c c b
d
25 d
a
20 a
Load (kN)
15
10
Exp.
5 FEA
0
-0.008 -0.006 -0.004 -0.002 0.000 0.002 0.004 0.006 0.008 0.010
Strain
177
25
c
b b
20
a c
Load (kN)
15
a
10
5 Exp.
FEA
0
-0.005 -0.003 -0.001 0.001 0.003 0.005
Strain
40
a b
35 a
b
30
25
Load (kN)
c c
20
15
10
Exp.
5 FEA
0
-0.005 -0.004 -0.003 -0.002 -0.001 0.000 0.001 0.002 0.003 0.004 0.005
Strain
178
40
b b
a
35 a
30
c c
25
Load (kN)
20
15
10
Exp.
5 FEA
0
-0.015 -0.010 -0.005 0.000 0.005 0.010 0.015
Strain
C.L. C.L.
Mid-span
Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
(a) Outward skin wrinkling in

the gap between first and
second loading areas
(b) Tsai-Wu failure criteria

index values less than 1.0
Figure 6.41 Failure mode of panel S1
179
C.L. C.L.
Mid-span
Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
(a) Top skin wrinkling and

crushing near the longitudinal rib

index values reached 1.0
C.L. C.L.
Mid-span
Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
(a) Top skin crushing and edge

wrinkling

180
C.L. C.L.
Mid-span
Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
(a) Wrinkling and splitting at

the edge
C.L. C.L.
Mid-span
Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
(a) Top skin crushing near the

panel centre line
181
C.L. C.L.
Mid-span
Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
(a) Top skin crushing at the

transverse rib edge


10
Compression skin wrinkling
9 and crushing
460 mm
7 LP#1
2134 mm
0.375 span 1541 mm

Pressure (kPa)
6 LP#4
Span 4112 mm
LP#4
5
2438 mm
Compressive stresses
4
2438 mm
3
Tensile stresses
2
Exp. LP#1
2134 mm
1 FEA
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Deflection (mm)
Figure 6.47 Pressure-deflection responses for the full width cladding panel tested
under air pressure
182
60
Compression skin wrinkling and
LP#5 crushing at the middle support
50
LP#4
40
Load (kN)
30
LP#6 Exp.
20
FEA
10
0
0 20 40 60 80 100 120 140 160 180
Deflection (mm)
(a) At axis passing through points 4-5-6
60
Compression skin wrinkling and
LP#7 crushing at the middle support
50
40
Load (kN)
30 LP#9
LP#8
Exp.
20 FEA
10
0
0 20 40 60 80 100 120 140 160 180
Deflection (mm)
(b) At axis passing through points 7-8-9
Figure 6.48 Load-deflection responses near mid-span for the half cladding
panel tested under discrete loading
183
60
LP#14
LP#3 LP#2
50
LP#1
40
Load (kN)
30
Exp.
20 FEA
10
0
-16 -14 -12 -10 -8 -6 -4 -2 0 2
Deflection (mm)
(a) At the end support edge points

60
LP#10
50 LP#12
LP#11
40
Load (kN)
30
Exp.
20 FEA
10
0
-15 -10 -5 0 5
Deflection (mm)
(b) At the middle support points
Figure 6.49 Load-deflection responses at supports for the half-width panel
tested under discrete loading
184
10
9
T5 T4
8 T7
T3 T6
7
Pressure (kPa)
6 T7
T2
5
T1
Tensile stresses
4
T6 T5 T4
2 T3 T2 T1
Exp.
1 FEA
0
-0.004 -0.003 -0.002 -0.001 0 0.001 0.002 0.003 0.004 0.005
Strain
Figure 6.50 Pressure-longitudinal strain responses for the loaded skin of the
large scale panel tested under air pressure
10
9
TT3
8 TT1 TT2
7
Pressure (kPa)
4
Tensile stresses
TT3
TT2 TT1
3
2
Exp.
FEA
1
0
-0.0005 0 0.0005 0.001 0.0015 0.002 0.0025
Strain
Figure 6.51 Pressure-transverse strain responses for the loaded skin of the
large scale panel tested under uniform air pressure
185
10
9
B2
8 B4 B6 B3
460 mm
2134 mm
0.375 span 1541 mm

Pressure (kPa)
6 B1
B5
Span 4112 mm
5
2438 mm
4
B6 B5 B4
2438 mm
3
Tensile stresses
2 Exp.
B3 B2 B1
2134 mm
FEA
1
0
-0.004 -0.003 -0.002 -0.001 0 0.001 0.002
Strain
Figure 6.52 Pressure-longitudinal strain responses for the opposite skin of the large
scale cladding panel tested under uniform air pressure
10
9
BT2 BT1
8
7
460 mm
2134 mm
0.375 span 1541 mm

Pressure (kPa)
6
Span 4112 mm
5
2438 mm
4
BT1
BT2
3
2438 mm
Tensile stresses
2
Exp.
2134 mm
FEA
1
0
-0.004 -0.003 -0.002 -0.001 0 0.001
Strain
Figure 6.53 Pressure-transverse strain responses for the opposite skin of the
large scale cladding panel tested under uniform air pressure
186
60
SG#13 SG#16 SG#15 SG#11 SG#12
SG#8 SG#14
50 SG#7 SG#10
40 SG#9
Load (kN)
30
20 Exp.
FEA
10
0
-0.001 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008
Strain
Figure 6.54 Load-longitudinal strain responses of the loaded skin for the half-
width panel near middle support
60
SG#1
SG#2 SG#5
50
SG#6
SG#3
40 SG#4
SG#17
Load (kN)
30
Exp.
20 FEA
10
0
-0.004 -0.0035 -0.003 -0.0025 -0.002 -0.0015 -0.001 -0.0005 0
Strain
Figure 6.55 Load-longitudinal strain responses of the loaded skin of the half-
width panel near mid-span
187
60
SG#12 SG#13 SG#11
SG#7 SG#5 SG#4
50
SG#6
40 SG#10
SG#9
Load (kN)
30 SG#8
20 Exp.
FEA
10
0
-0.005 -0.004 -0.003 -0.002 -0.001 0 0.001
Strain
Figure 6.56 Load-longitudinal strain responses of the opposite skin for half-
width panel near the middle support
60
SG#15
SG#1
50 SG#2
SG#3 SG#14
40
Load (kN)
30
Exp.
20 FEA
10
0
0 0.001 0.002 0.003 0.004
Strain
Figure 6.57 Load-longitudinal strain responses of the opposite skin of the
half-width panel near the mid-span
188
(a) Full size panel
(b) Half-width panel
Figure 6.58 Compression skin wrinkling and crushing failure modes and the
corresponding Tsai-Wu failure criteria index value
60
50
Half panel without cut-off
edge and using symmetric
boundary conditions
40
Load (kN)
Half panel with a cut-off edge

30
20
10
9
385 mm
0
0 20 40 60 80 100 120
Deflection (mm)
Figure 6.59 Load-deflection response of panel Test II, as is, and assuming it was
of a full width
189
Cut-off edge
(C.L.)
= 0.151 MPa = 0.157 MPa
(a) Half panel with cut-off edge (b) Half panel with symmetric boundary
conditions assuming full width
0.20
3
Soft foam (32.04 kg/m )
0.18
0.16
Half panel with symmetric
boundary conditions
0.14
assuming full width
Shear stress (MPa)
Half panel with a cut-off edge ( = 0.157 Mpa)

0.12 ( = 0.151 Mpa)
0.10
0.08
0.06
0.04
0.02
0.00
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Shear strain
(c) Shear stress-strain curve for soft polyurethane foam

Figure 6.60 Shear stress distribution for panel Test II in the foam core at
the cut-off edge
190
Interior rib Cut-off edge

(C.L.)
= 41.636 MPa = 40.34 MPa
(a) Half panel with a cut-off edge (b) Half panel with symmetric boundary
conditions assuming full width
50
45
40
35
Shear stress (MPa)
Half panel with a cut-off edge

30 ( = 41.636 Mpa)
Half panel with symmetric
25 boundary conditions assuming
full width
20 ( = 40.34 Mpa)
15
10
0
0 0.05 0.1 0.15 0.2 0.25
Shear strain
(c) Shear stress-strain curve for GFRP
Figure 6.61 Shear stress distribution for panel Test II in the GFRP interior rib
191
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
CHAPTER 7: SEMI-ANALYTICAL MODELING OF SANDWICH
PANELS
7.1 Introduction
This chapter presents an independent nonlinear strain compatibility model for the
analysis of simply supported sandwich panels loaded in one-way bending, under
concentrated or uniform loads. The model takes into account the geometrical
nonlinearities in the foam core and the GFRP skins, in addition to the materials
nonlinearity. The model is used to determine the load-deflection and load-strain
responses of sandwich panels with GFRP skins and polyurethane foam cores of different
densities. The model can also predict the responses of sandwich panels with different rib
configurations. Since the polyurethane foam is generally a very soft material and can
sustain very large compressive strains, the Winkler hypothesis for beam on elastic
foundation was incorporated in the model to capture the top skin behaviour under
concentrated loads. A FORTRAN90 code was developed for this model. The model is
intended to provide an independent approach of capturing the behaviour of sandwich
panels and compare with the FE model described in Chapter 6. A key feature of the
model is the ability to account for the significant effect of shear deformation of the core,
which contributes considerably to the panel deflection. The model can further isolate this
shear contribution to deflection easily, and present it as a separate component, a feature
that was not available in the FE model.
192
7.2. The Semi-Analytical Model
A non-linear model has been developed, accounting for material and geometric
nonlinearities. The stress-strain curves of the GFRP skins and ribs and polyurethane foam
core, including nonlinearity, were considered. The total deflection is assumed to comprise
three main components, one due to flexure, one due to shear deformation, and one due to
the localized skin deflection under concentrated loads. An incremental approach is used,
where the concepts of force equilibrium and strain compatibility are satisfied in every
loading step. The normal strain profile through the panel thickness was assumed to have a
linear distribution. The numerical procedure was executed using FORTRAN90 code
programming, and incorporated failure criteria that consider the following possibilities:
(a) flexural tension or compression failure in the GFRP skin material, (b) flexural tension
or compression failure in the polyurethane core material, (c) core shear (diagonal tension)
failure, (d) GFRP ribs shear failure, and (e) skin wrinkling (local buckling). The model
establishes the moment-curvature responses of cross-sections, which are then terminated
at a point governed by one of the five failure criteria discussed above (i.e. the one
producing the minimum load capacity). The curvatures are then integrated along the span
to obtain the flexural deflection. The effects of shear deformation deflection and the
localized top skin deflection are then added as will be discussed. The following sections
provide a detailed description of different components of the model. Also, procedures of
using the model in organized steps are presented.
7.2.1 Strain Profile
The uncoupling of the flexural-induced and shear-induced deflections enables one to
make the simplified assumption that the normal strain field varies linearly over the
193
thickness. The two extreme fibre strains along with the zero strain point at neutral axis
level are assumed to follow a straight line. The relationship between the two extreme
fibre strains can be expressed by the following equation:
H ybar
c t (7.1)
ybar
where c and t are the extreme fibre compressive and tensile strains, respectively. H is
the overall cross-section height and y bar is the neutral axis distance measured from the
extreme tension fibre to the neutral axis. Figure 7.1 shows the assumed linear strain
profile through the sandwich panel thickness.
7.2.2 Nonlinear Material Properties
As a result of using soft polyurethane foam core for the sandwich panels, the material
nonlinearity has to be taken into consideration, especially under compressive stresses.
Polyurethane foam is a very compressible material and has a maximum compressive
strain of almost 80%. Figures 3.14 and 3.15 in Chapter 3 show the tensile and
compressive normal stress-strain responses for the polyurethane foam as established by
coupon tests. The shear behaviour of the polyurethane foam is characterized by a very
low shear rigidity and the resulting shear deformations are excessive and can not be
neglected, as shown in Figure 3.17 in Chapter 3, also based on coupon tests. The slight
nonlinearity of GFRP is also taken into consideration for tension (Figures 3.6 and 3.7)
and in compression (Figures 3.9 and 3.10) along with the significant nonlinearity of the
GFRP ribs in shear (Figure 3.13). In order to model the polyurethane foam and GFRP
material constitutive relationships, a curve fitting technique was developed to track the
average experimental stress-strain curve of a group of coupons for each case. The curve
194
fitting technique is based on a cubic spline function concept (De Boor, 2001). The cubic
spline is of the form:
(7.2)
S j ( x) a j b j ( x x j ) c j ( x x j ) d j ( x x j )
2 3
and for a given set of coordinates (stress versus strain values) [C]:
C [( x1 , y1 ), ( x2 , y2 ),..........., ( xn , yn )] , (7.3)
a sets of n-1 splines can be found and provide:
Si ( xi ) yi Si 1 ( xi ) ,
S 'i ( xi ) S 'i 1 ( xi ) ,
(7.4)
S ' 'i ( xi ) S ' 'i 1 ( xi ) , and
S ' '0 ( x0 ) S ' 'n ( xn ) 0
The following algorithm was used to calculate the required sets of cubic splines
that can predict the different values of stress, or strain, within a given stress-strain curve:
1. Input sets of coordinates, n, for the given curve points C.
2. Create new array a of size n+1, and for i = 0, 1, ., n-1, set a i = y i .
3. Create new arrays b, c and d each of size n.
4. Create new array h of size n-1 and for i =0, 1, .., n-2 set:
hi xi 1 xi (7.5)
5. Create new array i of size n-2 and for i =0, 1, ., n-2 set:
3 3
i (ai 1 ai ) (ai ai 1 ) (7.6)
h hi 1
6. Create new arrays I, , and z, each of size n.
195
7. Set I o = 1 ; o = z o = 0.
8. For i=1, 2, 3, , n-2 set:
I i 2( xi 1 xi 1 ) hi 1i 1 .
hi
i
Ii .
(7.7)
i hi 1 zi 1
zi
Ii .
I n1 1 ; z n1 cn1 0
9. For j = n-2, n-3, ,0 set:
c j z j j c j 1
a j 1 a j h j (c j 1 2c j )
bj
hj 3 (7.8)
c j 1 c j
dj
3h j
10. Create new set of splines which will be the output and for
I = 0, 1, 2, ., n-2 set:
Si , a ai (7.9)
S i ,b bi
S i ,c ci
196
S i ,d d i
Si , x xi
The resulting set will be the required cubic splines for the curve fitting of the
input stress-strain curve data [C] for any input material properties.
7.2.3 Meshing
A layer-by-layer approach is adopted to integrate stresses over the cross-sectional areas
of the GFRP skins and the polyurethane foam core. The cross-section is divided into
three main parts (Part 1 to Part 3), as shown in Figure 7.2. The model assumes a plain
stress problem where a constant strain occurs in the panel width direction. Therefore, all
layers extend the full width of the panel.
The sandwich panel problem is very sensitive to the shear effect and the through-
thickness compressibility. Both effects are mainly attributed to the core. Therefore, a
sensitivity study is carried out and focussed on the through-thickness mesh refinement of
the core. In the span direction, a very large number of segments (160) was used and kept
constant. Parts 1 and 3 are the GFRP skins and are represented by a single layer, each. In
order to establish the appropriate number of layers for the core (Part 2) that leads to a
converged solution, the convergence study was carried out using 2, 8, 12, 16, 20 and 24
layers within the cross-section depth of the core. The convergence study was carried out
on the sandwich panels tested by Shawkat (2008), taking into account three different
loading configurations, namely three-point bending (panel P1), four-point bending (panel
P2) and a uniform loading configuration (panels P3 to P5). The converged solution was
197
based on the maximum load carrying capacity, taking into consideration all the different
failure criteria mentioned earlier for both the skins and the core.
Details of the full procedure of the analysis are given later. Figure 7.3 shows the
variation of the predicted failure load with number of layers, for different number of
layers, taking into account the different failure criteria. The figure shows that
convergence depends slightly on the loading configuration used. Panel P1 loaded in
three-point bending showed a minimal total variation in the predicted failure load of
about 8% and the solution converged at 16 layers within the core. Panel P2, loaded in
four-point bending showed a variation of 10% in the predicted failure load, and the
solution converged also at 16 layers. For the panel loaded with a uniform load (P3 to P5),
the required number of layers to obtain convergence was 20 layers and the predicted
failure load variation was about 12%. As such, it was finally decided to use 20 layers
within the core and one layer for each skin in the rest of the study. In this chapter the 160
elements along the half span direction will be referred to as segments while the 22
element along the depth of the panel will be referred to as layers.
7.2.4 Force Equilibrium and Moments
Figure 7.4 shows a cross-section of the sandwich panel under a given strain distribution
induced in the sandwich panel at a given load. Only two independent parameters are
needed to establish the complete strain profile, namely the strain at any level, say at the
extreme bottom side t , and the neutral axis depth, from the bottom fibres, say y bar . From
these two parameters t , y bar , the extreme top strain can be established as follows:
H ybar
c t (7.10)
ybar
198
The strain i at any GFRP or polyurethane foam layer i, located at a distance y i
from the bottom extreme tension side, can then be determined from the linear strain
distribution as follows:
ybar yi
i t , If yi ybar (7.11)
ybar
yi ybar
i c , If yi ybar (7.12)
H ybar
where t and c are the extreme tensile and compressive strains, respectively, and H is the
overall thickness.
The normal stress in any element, either GFRP or polyurethane foam, i is then
calculated from the corresponding normal stress-strain curve, whether in tension or
compression, using the cubic spline fitting curve technique developed earlier. The total
cross-section force at a given stage of loading (i.e. for a given t and c ) can be obtained
by numerical integration of stresses over the cross-section, for both GFRP and
polyurethane foam elements, which must equal to zero in flexure to satisfy equilibrium,
as follows:
n n

GFRP ,i 1
( S i AS i )
PolyurethanFoam ,i 1
( C i AC i ) 0 (7.13)
The corresponding moment M is calculated as follows:
n n
M
GFRP ,i 1
( S i AS i yi )
PolyurethanFoam ,i 1
( C i AC i yi ) (7.14)
where Si and Ci are the stresses in GFRP skins or polyurethane foam core at layer i,
respectively, n represents the total number of layers used in the analysis. A Si and A Ci are
199
the cross-sectional areas of the GFRP or polyurethane layer i, respectively, and y i is the
distance between the centroid of the layer i and the bottom extreme fibre of the cross-
section.
The presence of longitudinal and transverse ribs was accounted for in the internal
forces. At each cross section, the specific width and thickness of the longitudinal or
transverse rib was considered with the section. Also, the contribution of the web of the
rib in each layer i is considered.
7.2.5 Moment-Curvature Response
The aforementioned concepts and geometric relationships have been used to establish the
moment-curvature response of a given cross-section in the sandwich panel. A computer
code was written in FORTRAN90. The program can deal with any material stress-strain
curve of any shape or function. The moment-curvature algorithm can be summarized as
follows:
1. Input the sandwich panel dimensions, the overall panel thickness, skin
thickness, loading span, and loading pattern.
2. Divide the core into n numbers of layers (in this study it was shown that n = 20
for the core is sufficient). Each skin counts as one layer.
3. Define the stress-strain relationships for both GFRP and polyurethane foam
materials in tension, compression and shear.
4. Assume a strain value at the top surface of the sandwich panel, c , (Figure 7.4)
less than the ultimate strain of the GFRP in compression, cu .
5. Assume a neutral axis depth from bottom surface, y bar (Figure 7.4).
200
6. Calculate the corresponding tensile strain t at the bottom surface of the cross-
section using Eq. 7.10. Check that this strain does not exceed the ultimate
tensile strain of the GFRP, tu , otherwise tension failure has occurred.
7. Construct the linear strain profile along the panel cross-section by calculating
i from Eqs. 7.11 and 7.12 at each layer i (Figure 7.4). It is worth nothing that
the ultimate tensile strain of the foam core, in tension or compression, are
significantly higher than those of the GFRP skin, as shown in the material
curves in Chapter 3. As such, it is not possible for the extreme layers of foam
core to fail in the longitudinal direction before the adjacent GFRP layers.
8. Calculate the corresponding stresses, Si and Ci , in the GFRP skins and
polyurethane foam core, respectively (Figure 7.1), according to the
corresponding material stress-strain relationship through the cubic spline
technique method described earlier.
9. For each layer i in the cross-section, calculate its cross-sectional area, A i ,
weather it is A Si (for the GFRP skins) or A Ci (for the polyurethane foam core).
10. Calculate the tensile and compressive forces induced in each layer of skin and
core, ( Si A Si ) or ( Ci A Ci ).
11. Check equilibrium by summing the tension and compression forces (Eq.
7.13). If equilibrium is not satisfied (i.e. the total force sum is not equal to
zero), return to Step 5 and assume a new neutral axis depth. Continue the
process and repeat until equilibrium is satisfied.
201
12. Determine the moments of the forces in all layers about the neutral axis. The
summation of all the moments is the total moment M (Eq. 7.14) in the member
for the strain c applied in Step 4.
13. Compute the curvature as = t / y bar .
14. Return to Step 4 and assume a new strain. Repeat this process until the
ultimate strain of GFRP skins is reached in tension or compression and the
complete moment-curvature response is developed.
A simplified flowchart illustrating the procedure for establishing the moment-
curvature response is provided in Figure 7.5.
7.2.6 Generation of Full Load-Deflection Response
The load-deflection response of the sandwich panel consists of two components, namely
a flexural component and a shear component. In typical structural components such as
steel and concrete beams, deflections are dominated by the flexural contribution and the
deflection due to shear is usually small and typically neglected. However, in sandwich
panels with soft cores, the shear contribution is quite significant and can not be neglected.
The following sections describe the methods used to compute deflection due to flexure
and shear, respectively.
7.2.6.1 Flexural Effect
Once the moment-curvature of the cross-section is obtained, the load-deflection response
of the panel can be estimated for a given loading scheme such as concentrated loads or a
uniform loading. The deflection ( m ) at any point along the span is calculated by
integrating the curvatures () along the span using the moment-area method (Ghali and
Neville, 1989), as shown in Figure 7.6, and given by the following equations:
202
d 2 m
(7.15)
dx 2
m ( x)dxdx (7.16)
To start the process, an initial load is assumed and one half of the span is divided
into several segments, each with a length of dx, as shown in Figure 7.6. The average
bending moment experienced within each of these segments (M i ) is calculated in
accordance with the loading scheme being used. The previously established moment-
curvature response (section 7.2.5) is then used to determine the average curvature
corresponding to M i within each segment ( i ). The product of the length of the segment
and the curvature within the segment ( i * dx) gives the change in slope ( i ) of the
deformed member within that segment. For symmetric geometry and loading, the slope of
the deformed member at mid-span is zero, and the slope of the member at midpoint of
each segment ( i ) is equal to the summation of the values of i for all segments between
mid-span and the point of interest. The product of the length of each segment and the
average slope within that segment ( i * dx) gives the change in deflection (y i ) of the
member within that segment. The summation of all the y i values for all segments
between mid-span and the support gives the total mid-span deflection of the panel ( m ).
This entire process is repeated at various load levels in order to establish the first
component of the load-deflection response of the member, which is due to flexure only.
7.2.6.2 Shear Effect
The second component of the load-deflection curve is the deflection due to shear
deformation as a result of the very low shear modulus of the polyurethane foam core. In
sandwich panels with soft polyurethane foam core, shear deflection can be quite
203
significant and should not be neglected. A method is proposed next to calculate this
deflection. The deflection of any segment l along the span, and at a layer i along the
depth, due to shear stress, v,l, i , in the interval dx along the span is equal to:
v ,l ,i l ,i dx (7.17)
where the shear strain l,i can be calculated from the shear stress l, i at a specific layer i at
a given segment l under a specific loading conditions, as shown in Figure 7.7. l varies
from 1 to 160 and i varies from 1 to 22. The shear stress l,i at any layer i at any segment l
along the span can be calculated as follows:
Vl Qt ,l ,i
l ,i (7.18)
I t ,l bt ,l ,i
where V l is the applied shear force at segment l of the panel. Q t,l,i is the first moment of
area for the transformed cross-section at specific layer i, at which shear deflection need to
be calculated about the neutral axis of the transformed section. I t,l is the moment of
inertia for the transformed cross-section and b t,l,i is the width of the transformed cross-
section at that layer i.
The transformed section is established by transforming the width b l,i of each skin
or core layer i at any segment l to a unified polyurethane foam material based on the
material stiffness of the foam in compression, using the following equation:
El ,i
bt ,l ,i bl ,i (7.19)
E fc
where b l,i is the original cross-section width at segment l for layer i and b t,l,i is the
transformed cross-section width at segment l for layer i. E l,i is the secant modulus of
elasticity of the normal stress-strain curve of the polyurethane foam or the GFRP, in
204
tension or compression, at segment l depending on the location of layer i relative to
neutral axis. E l,i is established from the material curve of the polyurethane foam or the
GFRP at the specific normal strain l,i of layer i at segment l at this particular loading
level. E fc is the reference modulus which is the initial modulus of the polyurethane foam
in compression. Figure 7.8 shows the original and the transformed cross-sections,
respectively.
After calculating the shear stress, the corresponding shear strain l,i can be
calculated using the core material shear stress-strain curve and used to compute the shear
deflection of that specific layer i. To calculate the total shear deflection of layer i at mid-
span of the panel (L = span/2), the shear deflection for each segment (l = 1 to 160)
should be summed in the longitudinal direction of the panel.
m 160 m 160
v ,i
l 1
l ,i dx
l 1
v ,l ,i (7.20)
where v,i is the total shear deflection of layer i specifically at mid-span, and m is the total
number of segment along the half span, which is 160 in this study.
As such, at any layer i, the shear deflection values will be different from one layer
to the other, which is obviously impossible because each layer is joined to the adjacent
layers and the whole cross-section must be continuous, without any gaps or overlaps
(Shanley, 1957). As a result, each layer will rotate clockwise (or counterclockwise) to
adjust the cross-section continuity at this segment (Figure 7.7). This rotation causes the
cross-section to warp, which means the cross-section will not remain plane. On the other
hand, the shear forces which produce the shear deflections were based on the beam
theory, which was based on the assumption that plane sections remain plane after
205
deformation. These shear forces cause shear deflections which require the cross-section
to be warped, thereby invalidating the bending theory assumptions. Also, the deflection
due to flexure calculated earlier was based on the same assumption. However, it has been
found that the assumption that plane cross-sections remain plane after deformation that
was originally used for pure bending only can be extended to include bending caused by
transverse loading (shear) with negligible errors in most cases (Shanley, 1957).
The top skin deflection due to shear forces at any segment, v,l,to p can be assumed
as the average deflection of all layers above the neutral axis, while the bottom skin
deflection at the same segment, v,l,bo t , is the average deflection of all layers below the
neutral axis, as follows:
ntop
v ,l , i
v ,l ,top i 1
ntop
nbot
(7.21)
v ,l ,i
v ,l ,bot i 1
nbot
where n top is the number of layers above the neutral axis and n bot is the number of layers
below the neutral axis. The total shear deflections at the panel mid-span, for both skins,
are:
m 160
v ,top
l 1
v ,l ,top
m 160
(7.22)
v ,bot
l 1
v ,l ,bot
where v,top and v,bot are the total top and bottom skin deflections due to shear, at the
panel mid-span, respectively.
206
In panels with GFRP ribs, the effects of longitudinal and transverse ribs on the
transformed section analysis were considered at each cross section. This was considered
in calculating b t,l,i in Eq. 7.19, in calculating I t,l used in Eq. 7.18 and in calculating Q t,l,i ,
also used in Eq. 7.18.
7.2.6.3 Winkler Model
Because of the very soft core, loads applied to the top skin will cause localized effects,
namely local bending and local deflection. To capture this effect the beam on elastic
foundation concept is employed. The concept is based on the assumption that the reaction
forces of the foundation at every point are proportional to the deflection of the beam at
that point. This assumption was introduced first by E. Winkler in 1867 (Hetenyi, 1946).
In sandwich panels, the top skin can be considered as a beam resting on elastic
foundation based on the flexibility and compressibility of the polyurethane foam core, as
shown in Figure 7.9. To overcome the discontinuity behaviour of Winkler model, the
two-parameter model developed by Hetenyi (1946) will be used instead. The general
differential equation for the deflection curve of a beam supported on elastic foundation
can be written as follows:
d 4w
EI 4 kw 0 (7.23)
dx
where w is the vertical deflection and EI is the flexural rigidity of the top skin. k
represents the elasticity modulus of the subgrade which is the polyurethane foam core
in this case. In panels with GFRP ribs, the ribs were considered rigid enough to prevent
the localized effect caused by the softness of the polyurethane foam core and then the
207
localized deflections caused by the elastic foundation concept in calculating deflections
were neglected. The general solution of that equation takes the following form:
w e x C1 cos x C2 sin x e x C3 cos x C4 sin x (7.24)
where:
k
4 (7.25)
4 EI
C 1 to C 4 are the integration constants and can be calculated by the applied boundary
conditions. As a special loading case of concentrated and uniform loads, and for infinite
length skin, the deflection w, slope , moment M, and shear Q induced in the skin will be
as follows:
For a concentrated load P:
P
w A x (7.26)
2k
P 2
B x (7.27)
k
P
M C x (7.28)
4
P
Q D x (7.29)
2
For a uniform load q:
q
w
2k
2 D a D b (7.30)
208
q
Aa Ab (7.31)
2k
M
q
B a B b (7.32)
4 2
Q
q
Ca Cb (7.33)
4
where:
A x e x cos x sin x (7.34)
B x e x sin x (7.35)
C x e x cos x sin x (7.36)
D x e x cos x (7.37)
The subscripts a and b refer to distance of the point of interest from both left and
right ends of the uniform load q, respectively, and x is the point at which these quantities
are being calculated.
Because of the fact that the skin does not have an infinite length but rather a finite
length, which is the panel span, the superposition method developed by Hetenyi (1946)
will be used. The superposition method depends on determining the end conditioning
forces (bending moments and shear forces at the skin ends) which will transform the
infinite length beam to a finite length beam with a specific span. To calculate the end
conditioning forces, M o 1 , P o 1 , M o 2 , and P o 2 , at locations 1 and 2 of the top skin, which
209
represents the skin ends, the moments and shear force have to be calculated based on the
fact that the skin has infinite length as follows:
Po1 Po' Po'' (7.38)
Po 2 Po' Po'' (7.39)
M o1 M o' M o'' (7.40)
M o 2 M o' M o'' (7.41)
where:
Po' 41 Q1' 1 D l M 1' 1 A l (7.42)
2
M o' 1 Q1' 1 C l 2 M 1' 1 D l (7.43)

Po'' 4 2 Q1'' 1 D l M 1'' 1 A l (7.44)
2
M o'' 2 Q1'' 1 C l 2 M 1'' 1 D l (7.45)

and A l , B l , C l and D l are the same as in Equations 7.34 to 7.37 with l equals the top
skin length and:
1 e l
1 (7.46)
2 sinh l sin l
1 e l
2 (7.47)
2 sinh l sin l
210
1
M 1' M1 M 2 (7.48)
2
1
M 1'' M1 M 2 (7.49)
2
1
Q1' Q1 Q2 (7.50)
2
1
Q1'' Q1 Q2 (7.51)
2
where l is the sandwich panel top skin span and Q 1 , Q 2 , M 1 , and M 2 are the induced
shear and moment at ends 1 and 2 when dealing with the skin as if it had an infinite
length.
Now, after calculating the end conditioning forces M o1 , P o1 , M o2 , and P o2 , the top
skin can be treated as a beam with a finite length loaded with any load scheme, in
addition to the end conditioning forces obtained before. The application of these end
conditioning forces in addition to the actual loads will adjust the boundary conditions at
points 1 and 2, which are the actual skin end points. To calculate deflection w, slope ,
moment M, and shear Q induced in the skin at any point throughout the span, equations
7.26 to 7.37 should be used depending on the load scheme used, in addition to the end
conditioning forces, and the location of the point of interest at which all these quantities
are needed to be calculated.
The addition of the compressive stresses resulting from the Winklers local
bending effect in the top skin, to the original flexural compressive stress, was considered.
The final skin deflection will be the sum of all previous deflections, namely due to
flexure, shear and Winklers effect, as follows:
211
tot ,l m,l v ,l ,top w,l (7.52)
where m,l , v,l and w,l are deflections at segment l due to flexure, shear and elastic
foundation effect, respectively. It should be noted that Winkler effect was neglected in
the bottom skin at the support regions.
7.2.6.4 Nonlinear Geometric Effects
As a result of shear stress variation across the sandwich panel thickness, the resulting
shear deflection is also variable through the thickness. Based on this result, each layer
will deform (skew) in a value different from the adjacent layers as indicated earlier. Also,
because of the different material behaviour in both tension and compression for the
polyurethane foam core and material nonlinearity, the layers under the neutral axis will
have different transformed widths when compared with the layers above the neutral axis.
This will result in different values for the top and bottom layers deflections (Eq. 7.22).
These different values of deflection will cause the cross-section to be squeezed
vertically (i.e. to have a smaller thickness under the next increment of loads), as shown in
Figure 7.7. In order to take this variation of the panel thickness into account, the neutral
axis location of the new transformed section has to be re-established in each load step, for
each segment along the span. After applying the first loading increment the resulting
deflection is calculated for both flexure and shear at each segment along half of the span.
Then, the new section height H new, l is calculated as shown in Equation 7.53, at each
segment and the new location of the neutral axis is then recalculated, also at each
segment.
H new,l H old ,l v ,top ,l v ,bot ,l w,l (7.53)
212
where H old,l is the cross-section thickness at the previous load increment
The new moment-curvature relationship at any segment is established for the
section with a new thickness, as explained earlier in section 7.2.5, using the developed
cubic spline material curves. Then, under any moment value at each segment along the
span, the corresponding curvature value is calculated. As the curvature values for every
segment at certain load increment is known, the deflection due to moment can be
calculated correctly at this load increment. Also, the moment-strain (tensile or
compressive) relationships at any segment, at a certain load increment, can be found.
That means, for a given moment value under a given load level, at any segment with a
certain cross-section thickness, the corresponding strains and curvatures are known and
used for both flexure and shear deflection calculations.
7.2.7 Failure Criteria
Seven main failure criteria were considered in the model, namely (1) a flexural tension
failure of GFRP skin, (2) a compression failure of GFRP skin by crushing, (3) a shear
failure of the foam core, (4) a shear failure of the GFRP rib, (5) a tension failure of the
foam core, (6) a compression failure of the foam core by excessive deformation, and (7) a
wrinkling failure (local buckling) of the GFRP compressive skin. The tension failure of
the GFRP skin is highly unlikely and never occurred in the experiments as the
compression skin or core shear failure usually governs. Six of the failure criteria are
material failures and are governed by the stress-strain curves established earlier. The
seventh failure criterion, namely the compression skin wrinkling under compressive
flexural stresses was based on the model by Allen (1969) as shown in Equation 7.55.
213
1 2
cr B1 E E
3
S
3
C and:
(7.55)
1

B1 3 12 3 c 1 c
2 2 3

where cr is the minimum critical wrinkling stress of the skin, E S is the skin compressive
modulus of elasticity in the longitudinal direction, E C is the core compressive modulus of
elasticity and c in the core Poissons ratio, taken as 0.33 (Gibson and Ashby, 1988).
Throughout the formulation of the moment-curvature response, the maximum
values of compressive and tensile strains in the skins are continuously monitored, to
detect any of the flexural or wrinkling failures. Also, shear failure is defined when the
shear stresses in the shear analysis algorithm exceed the failure values of the
polyurethane foam core or the GFRP ribs. Figure 7.10 shows a flow chart for the
procedure of obtaining the complete load-deflection response.
7.3 Illustration of Key Features of the Model
The model developed has several significant features, namely, accounting for the
geometric non-linearity, which is the change in thickness due to core compressibility,
material non-linearity of GFRP skins, which is very slight, material nonlinearity of
polyurethane foam core, which is significant, and a number of possible failure criteria of
GFRP and polyurethane. Also, the model is capable of displaying the different
components of deflection produced by flexure, shear, and localized loading effect on the
loaded skin according to the beam on elastic foundation principles. In order to illustrate
the significance of these individual features, the load-deflection responses of the test
specimens with two different densities (Shawkat, 2008) described in Chapter 6, have
been predicted under five different conditions. In case (1), the model neglected the
214
material nonlinearity of foam and GFRP as well as the geometric nonlinearity, and the
beam on elastic foundation effect of the loaded skin. In this case, the stiffnesses based on
the initial linear parts of the stress-strain curves of materials were used as constants
throughout the analysis. In case (2), only the material nonlinearity was considered for
both GFRP and Polyurethane foam, in tension, compression, and shear. In case (3), in
addition to material nonlinearity, the geometric nonlinearity was also taken into account.
In case (4), in addition to the material and geometric nonlinearities, the effect of core
compressibility under the loads is taken into account through the Winkler principle of
beam on elastic foundation. In all four cases so far, the load-deflection response was
continuous without termination by enforcing the failure criteria. In case (5), all the
features of the model employed in case (4) were applied, in addition to the failure criteria.
Figures 7.11 and 7.12 show the experimental and the analytical responses for the
five cases for specimens P3 to P5 with soft cores, and specimens P7 to P9 with hard
cores, respectively. The figures clearly show that ignoring the material nonlinearity, case
(1), would grossly under estimate deflection, especially at higher load levels. Considering
the material nonlinearity but ignoring the geometric nonlinearity, case (2), provides
significant improvements of the prediction throughout the full loading history. By taking
the geometric nonlinearities into account, case (3), and considering the effects of beam on
elastic foundation, case (4), slight refinement in the model prediction are observed. Case
(5), which enables the failure criteria, leads to the final prediction with the full
capabilities of the mode and it shows reasonable agreement with the experimental
responses. Clearly, the most important effect is the material nonlinearity (i.e. case (1)
versus case (2)).
215
7.4 Model Validation
The semi-analytical model was validated using the results obtained from the ten sandwich
panels of two different core densities tested using different loading configurations
(Shawkat, 2008), and the results of the six sandwich panels with different rib
configurations, which were described in details in Chapter 4.
7.4.1 Sandwich Panels with different Core Densities and different Loadings
In this section, the predicted load-deflection and load-longitudinal strain responses of
panels P1 to P9 with two different foam core densities and tested under various loading
configurations (Shawkat, 2008) are presented. The separate contributions of flexure,
shear and Winkler effect to the total deflection are presented. Also, the final deflection
profile of the loaded skin, including the Winkler effect is predicted along the span. Figure
7.13 shows the load-deflection responses of panel P1. The figure shows very good
agreement between measured and predicted responses, except for deflection at ultimate,
which was slightly overestimated. It is clear from Figure 7.13 that shear deflection is
significantly higher than the flexural deflection. Figure 7.14 shows the deflection of the
loaded skin along the span, where the Winkler effect at the loading point is quite clear.
Figure 7.15 shows the load-strain responses of panel P1. Although the model accounts for
the Winklers effect in terms of deflection and localized bending stresses of the skin, the
model could not capture the excessive compressive strain of the loaded skin at the
wrinkle location. The reason is that the strain at the point of maximum inward wrinkling
due to the localized bending was beyond the ultimate compressive strain obtained from
the GFRP compression coupons tested in Chapter 3. As such, the model can not capture
any compressive strain (or compressive stress) beyond the ultimate failure compressive
216
strain because of the termination of the material curve beyond this point. The model
predicted the correct failure mode at ultimate, namely compression failure of the foam
core under the load. This is detected by approaching the flat plateau of stress-strain curve
of the foam under the load, suggesting the excessive deformation. This in turn leads to
excessive thickness reduction of the cross-section, which triggered shear failure next to
the load. Although failure appears similar to a local buckling, it is actually an excessive
deformation of the core, as the critical stress cr (Eq. 7.55) was not reached in the
compression skin.
Figure 7.16 shows the load-deflection responses of panel P2. Reasonable
agreement of prediction with experimental responses is observed. It is to be noted that the
deformed shape of panel P2 during testing was not symmetric as deflection under one
load was higher than the other, and indeed triggered failure to occur at that loading point.
The model predicted correctly the failure mode, which was compression failure of the
foam core under loading point by excessive deformation, leading to shear failure as
indicated for panel P1. Also, the deflection at mid-span due to the Winkler effect is zero
because the loads are relatively far from mid-span. This is further illustrated in Figure
7.17. Figure 7.18 shows the load-strain responses which are in very good agreement.
Figures 7.19 and 7.20 show the load-deflection responses of panels P3 to P5 with
low-density foam and P7 to P9 with high-density foam, respectively. Very good
agreement is observed. The model also predicted the correct failure modes in both cases,
namely shear failure of the core. Figure 7.19 shows that the shear deflection is
significantly larger than the flexural deflection, because of the low-density core, whereas
Figure 7.20 shows that both flexural and shear deflections are similar in high-density
217
core. At the same load level, the shear deflection is significantly lower for high-density
core than for low-density core. Figures 7.21 and 7.22 show the deflection distribution of
the loaded skin along the span for panels P3 to P5 and P7 to P9, respectively. The
Winkler effect under the loads is quite obvious and agrees quite well with the test
specimen picture shown. Figures 7.23 and 7.24 show the load-strain responses of the two
groups of panels. Good agreement is observed in general.
7.4.2 Sandwich Panels with different Rib Configurations
Figures 7.24 to 7.30 show the experimental and analytical load-deflection responses for
specimens S1 to S6 presented in Chapter 4. Generally speaking, good agreement is
observed between the model and experimental results. The model predicted well the
failure modes of the panels, except for panel S1. All panels, except S1 failed in
compression by crushing of the GFRP akin, except S1, which failed experimentally by
outward wrinkling of the skin, at a point far from mid-span. The model predicted that S1
failed by shear failure of the foam core. It should be noted that FEA predicted top skin
wrinkling for this panel, nonetheless the maximum loads were so close. As indicated in
Chapter 4, inspection of specimen S1 revealed small defect in the form of localized
separation of the skin from the core at the failure region. Also, if wrinkling of the skin
was to occur it should have occurred at mid-span where maximum moment is. Figures
7.25 to 7.30 also show the individual components of deflection, resulting from flexure,
shear and Winkler effect. It is clear that as the longitudinal ribs were introduced, the
contribution of shear deflection relative to flexural deflection, and the Winkler effect both
became insignificant. Figures 7.31 to 7.36 show the load-longitudinal strain responses of
the six panels. Generally, good agreement is observed. However, as indicated in chapter 4
218
it was noticed that strain gauges at the centers of ribs tend to measure higher localized
strains than the skin can sustain, especially at locations of transverse ribs. This is likely
due to the higher stiffness in these areas.
219
Load
Top Skin
t c Foam, c GFRP, c
N.A. Hc H
Core ybar
t
t Foam, t GFRP, t
Bottom Skin
Normal strain Foam normal GFPR normal
profile stress stress
Figure 7.1 Normal stress and strain distributions
Load
GFRP Top Skin Foam Core
C.L.
A2 (m2xn) A2 (m2xn) A2 (m2xn)

A1 (m1xn) A1 (m1xn)
Part 3
GFRP Bottom Skin
t i=n
x N.A.
H Hc
ybar i=k Part 2
t
i=1 Part 1
y
Figure 7.2 Cross-section meshing
220
C.L.
Span / 2
(a) Mesh configuration

8600
7600 Panel P3 (failed by core shear)

7201 N
Convergence
6600
20 Layers
24 Layers
8 Layers
16 Layers
12 Layers
2 Layers
Failure load (N)
5600
Panel P2 (failed by excessive core compression under loading point)
4704 N
4600
Convergence
3600
Panel P1 (failed by excessive core compression under loading point)
2688 N
2600
Convergence
1600
0 5 10 15 20 25 30
No. of layers within core
(b) Variation of failure load with number of cross-section layers within core
Figure 7.3 Mesh configuration and convergence study results
Width b
y c
i=n
i yi ybar
i c
Zone I H ybar
x H N.A.
i=k yi
ybar i ybar yi
Zone II i t
yi x ybar
i=1 t
y
Figure 7.4 Normal strain profile
221
Input Sandwich Panel

Section geometry and
material stress-strain
relationship files for
tension, compression and
shear
Assume initial c at extreme
Material failure values for compression fibers
both GFRP skin and
Foam core
Do while c =< cu
And t =< tu
350 100
Assume elevation of 300 GFRP 90
80
GFRP
N.A. 250 70
Compressive stress MPa

Stress (MPa)
60
200
50
Do while Error in N.A.

150
40
100 30
calculation =< 0.0 20

50
10
0 0
Set Moment and the 0.000 0.004 0.008 0.012

Strain (mm/mm)
0.016 0.020 0 0.0005 0.001 0.0015 0.002 0.0025
Shear strain
0.003 0.0035 0.004 0.0045
Force_sum equal to 0.0 0.45 2.50
0.40
Foam Hard Foam density
2.00
Foam
0.35 (64.08 kg/m3)
Calculate value of t at the 0.30
extreme tension fiber 1.50 Hard foam density
Stress MPa
Stress MPa
0.25
(64.08 kg/m3)
0.20 Soft Foam density Soft foam
(32.04 kg/m3) 1.00
density
0.15
(32.04 kg/m3)
Construct the strain profile 0.10

0.50
over the cross-section depth 0.05
0.00 0.00
0.00 0.02 0.04 0.06 0.08 0.10 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90
Strain mm/mm
Strain mm/mm
Do I = 1 to Layers_num
Calculate strain in GFRP

and Foam in layer i
Tension Compression
Calculate stress in GFRP
and Foam in layer i
Calculate force in GFRP

and Foam in layer i
Calculate Moments and

Force sum
Moment M
End Do loop
NO
IF Force_sum 0.0
Curvature
YES
Output values of and M
Increase value of c by a
small increment
End Do loop
Figure 7.5 Flow chart of the procedure to obtain the moment-curvature

response of a cross-section
222
Panel divided Uniform Load q

into segments
Any Loading Scheme

dx
Span L
(qL2/8)
Actual Bending Moment
(qL2/8)
Approximated Bending Moment
i
i
Approximated Curvature
Approximated Slope
Approximated Change in Deflection
yi
Figure 7.6 Schematic for the process of determining deflection due to flexure
223
Section l
Chapter 7
Undeformed section Deformed section

I II I n, n
t i=n II
i=k x N.A. H-vI

H Hc
l,i i i l,i, l,i H-vII
ybar
t
224
i=1 y
1, 1
Shear stress (l )
dx
dx i
l,i
v,l,i
l,i
l,i
dx
l,i
Figure 7.7 Shear deflections in sandwich panel
Semi-Analytical Modeling of Sandwich Panels
Top Skin Foam, c GFRP, c
i=n t i=n
Chapter 7
N.A. Hc H
i=k ybar i=k

Core
bl,i i=1 Foam, t GFRP, t bt,l,i i=1

Bottom Skin
Actual geometry of Foam normal stress GFPR normal stress Transformed geometry of
cross-section at cross-section at segment l
segment l
0.40 350
Soft Polyurethane foam Tension GFRP skin behaviour in
(32.04 kg/m3) 300 the longitudinal direction
225
0.20
El,i 250
At any segment l and for any layer i
200
0.00
Compression 150
-0.20 100 E l ,i
Efc El,i
Stress (MPa)
50
Stress (MPa)
El,i
Tension
Transformed width, bt ,l ,i bl ,i
-0.40
0 E fc
El,i
-50
-0.60
-100
Compression
-0.80 -150
-1.00 -0.80 -0.60 -0.40 -0.20 0.00 0.20 -0.005 0.000 0.005 0.010 0.015 0.020
Strain Strain
Figure 7.8 Section transformation accounting for variable modulus in core

and skin in tension and compression
Semi-Analytical Modeling of Sandwich Panels
C.L.
Soft polyurethane foam core

d4y
GFRP skin
EI ky 0
dx 4
y e x C1 cos x C 2 sin x e x C3 cos x C 4 sin x
k
4 Flexural Defl. Shear Defl.
4EI
total m v w
Total Midspan Deflection Winkler Defl.
of Top Skin.
New sec. Height H new,l H old ,l v ,top ,l v ,bot ,l w,l
Figure 7.9 Winkler effect of polyurethane foam softness at loading points
226
Input Sandwich Panel

Section geometry and
material stress-strain
relationship files for Determine sandwich
tension, compression and panel N.A. Location
shear. Do x = 1 to no. of
panel segments
Material failure values for Generate the relationship curves
both GFRP skin and between M and t, c, and
Foam core Do x = 1 to
Do w = 0.0 to Layers_num
Any load
Calculate shear stress at each
Calculate Deflections and stresses in layer for each section
the top skin according to Winkler
YES
IF u STOP
Do x = 1 to no. of panel Failure Detected
NO
segments
Calculate the corresponding
Calculate moment and shear at shear strain from the material
each section. relationship files
End Do loop
Calculate t, c, and from the
generated curves End Do loop
YES
STOP IF t tu, c cu Calculate deflection due to shear
for top skin
Failure Detected
NO
Calculate total deflection for top
Calculate c of the top skin, from
skin
the material stress-strain curve
Output values for load, moment,
YES shear, and deflection
STOP IF c cr
Failure Detected Calculate the new section thickness
NO
because of section compressibility
Establish the strain profile along
the cross-section thickness Do x = 1 to no. of
panel segments
Calculate slope variation between
each two successive sections
Determine the new location of sandwich
End Do loop panel N.A. at each segment
Calculate the slope at each Calculate the new relationship curves

section between M and t, c, and at
each segment
Calculate deflection variation between
each tow successive sections End Do loop
Calculate deflection due to
End Do loop
moment
End
Do x = 1 to no. of
panel sections
Do x = 1 to
Layers_num
Calculate strain at each layer for

each section according to the
Load P
strain profile
Calculate corresponding stresses

according to material relationship files
Calculate secant modulus of elasticity

for each layer at each section
Deflection
Calculate modular ratio for each layer at

each section and the layer new width
End Do loop
End Do loop
Figure 7.10 Flow chart of procedure used to obtain the load-deflection response
227
8
P4
7
P5
Failure
6 P3 (Analytical)
5
Load (kN)
3
Experimental P3
Experimental P4
2 Experimental P5
Case (1) Ignoring material and geometric nonlinearity
Case (2) Accounts for material nonlinearity only
1 Case (3) Accounts for material and geometic nonlinearity only
Case (4) Accounts for material and geometic nonlinearity with Winkler effect only
Case (5) Similar to Case (4) but failure criteria was also applied
0
0 10 20 30 40 50 60 70 80
Defelction (mm)
Figure 7.11 Illustration of significance of various features of the model for panels with
soft cores, P3 to P5 (Shawkat, 2008)
20
P9
18
P7
16 P8 Failure
(Analytical)
14
12
Load (kN)
10
6
Experimental P7
Experimental P8
4 Experimental P9
Case (1) Ignoring material and geometric nonlinearity
Case (2) Accounts for material nonlinearity only
2 Case (3) Accounts for material and geometic nonlinearity only
Case (4) Accounts for material and geometic nonlinearity with Winkler effect only
Case (5) Similar to Case (4) but failure criteria was also applied
0
0 10 20 30 40 50 60 70 80
Defelction (mm)
Figure 7.12 Illustration of significance of various features of the model for panels with
hard cores, P7 to P9 (Shawkat, 2008)
228
3.5
3.0
2.5
Failure
2.0 P1 (excessive compression
Load (kN)
of core under loading)
1.5
Experimental and analytical

Load model location of deflection
1.0 measurements
Experimental P1
Flexural deflection only (Analytical)
0.5 Shear deflection only (Analytical)
Winkler deflection only (Analytical)
Total deflection (Analytical)
0.0
0 5 10 15 20 25 30 35 40
Deflection (mm)
Figure 7.13 Load-deflection response of panel P1 (Shawkat, 2008)
5
Load
0
-5
-10
Deflection (mm)
-15
-20
-25
-30 Moment contribution

Shear contribution
-35 Winkler contribution
At load = 2.688 kN
Total deflection
-40
0 100 200 300 400 500 600 700
Distance from support (mm)
Figure 7.14 Deflection responses along the span of specimen P1 (Shawkat, 2008)
229
3.0
Model Model SG#1
2.5
SG#1
SG#2
2.0 SG#2
Load (kN)
1.5
1.0
0.5
0.0
-0.0200 -0.0150 -0.0100 -0.0050 0.0000 0.0050
Strain
Figure 7.15 Load-longitudinal strain responses for specimen P1 (Shawkat, 2008)
6.0
Failure
(excessive compression
of core under loading)
5.0
4.0
Load (kN)
Experimental and analytical

model location of deflection
3.0 Load measurements
2.0
Experimental P2
1.0 Flexural deflection only (Analytical)
Shear deflection only (Analytical)
0.0
0 5 10 15 20 25 30 35
Deflection (mm)
Figure 7.16 Load-deflection response of panel P2 (Shawkat, 2008)
230
5
Load
0
-5
-10
Deflection (mm)
-15
-20
-25
-30
-35 Moment contribution

Shear contribution
-40 Winkler contribution At load = 4.704 kN
Total deflection
-45
0 100 200 300 400 500 600 700
Figure 7.17 Deflection responses along the span of specimen P2 (Shawkat, 2008)
5.0
Model SG#1
4.5
SG#2 Model
4.0 SG#1
3.5
SG#2
3.0
Load (kN)
2.5
2.0
1.5
1.0
0.5
0.0
-0.0030 -0.0020 -0.0010 0.0000 0.0010 0.0020
Strain
Figure 7.18 Load-longitudinal strain responses for specimen P2 (Shawkat, 2008)
231
8.0
P4
7.0 P5
6.0 P3
Failure
5.0 (core shear crack)
Load (KN)
4.0
3.0
2.0 Experimental P3
Experimental P4
Experimental P5
0.0
0 10 20 30 40 50 60 70
Deflection (mm)
Figure 7.19 Load-deflection responses of panels P3 to P5 (Shawkat, 2008)
24.0
P9
20.0
P7
16.0 P8
Failure
Load (KN)
(core shear crack)

12.0
8.0
Experimental P7
Experimental P8
Experimental P9
0.0
0 10 20 30 40 50 60 70
Deflection (mm)
Figure 7.20 Load-deflection responses of panels P7 to P9 (Shawkat, 2008)
232
10
-10
-20
Deflection (mm)
-30
-40
-50
Moment contribution
Shear contribution
-60
Winkler contribution
At Load = 7.2 kN
Total Deflection
-70
0 100 200 300 400 500 600 700
Figure 7.21 Deflection responses along the span of panels P3 to P5 (Shawkat, 2008)
20
-20
Deflection (mm)
-40
At Load = 19.2 kN
-60
Moment contribution
-80 Shear contribution
Winkler contribution
Total Deflection
-100
0 100 200 300 400 500 600 700
Figure 7.22 Deflection responses along the span of panels P7 to P9 (Shawkat, 2008)
233
8.0
Model
P5 P4
P4
7.0 Model
P5 P3
P3
6.0
5.0
Load (kN)
4.0
3.0
2.0 Average Experimental P3 (Avg.-ve)

Experimental P3 (Avg.+ve)
Experimental P4 (Avg.-ve)
1.0 Experimental P5 (Avg.-ve)
Analytical (Compression)
Analytical (tension)
0.0
-0.0030 -0.0020 -0.0010 0.0000 0.0010 0.0020 0.0030
Strain
Figure 7.23 Load-longitudinal strain responses for panel P3 to P5 (Shawkat, 2008)
22.0
Model Model P9
20.0 P9
18.0 P8
P8 P7
16.0
P7
14.0
Load (kN)
12.0
10.0
8.0

Average
Experimental P9 (Avg.-ve)
2.0 Experimental P9 (Avg.+ve)
Analytical (tension)
Analytical (compression)
0.0
-0.006 -0.004 -0.002 0.000 0.002 0.004 0.006
Strain
Figure 7.24 Load-longitudinal strain responses for panel P7 to P9 (Shawkat, 2008)
234
16
14 S1
12
Failure
10 (core shear crack)
Load (kN)
4
Experimental S1
2 Winkler deflection only (Analytical)
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
Figure 7.25 Load-deflection response for panel S1 (Chapter 4)
35
30
S2
25
Failure
(top skin crushing)
Load (kN)
20
15
10
Experimental S2
5
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
235
35
30
25 S3
Failure
(top skin crushing)
Load (kN)
20
15
10
Experimental S3
5 Shear deflection only (Analytical)
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
30
25
20 Failure
S4 (top skin crushing)
Load (kN)
15
10
Experimental S4
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
236
40
S5
35
30
Failure
(top skin crushing)
25
Load (kN)
20
15
10 Experimental S5
5 Winkler deflection only (Analytical)
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
40
S6
35
30
Failure
(top skin crushing)
25
Load (kN)
20
15
10 Experimental S6
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
237
16
b d
e b a
14 a c
e
Model c
12 d
Model
10
Load (kN)
0
-0.004 -0.003 -0.002 -0.001 0.000 0.001 0.002 0.003 0.004
Strain
Figure 7.31 Load-strain responses for panel S1 (Chapter 4)
35
30
c b
a b a
25 c
Model
Load (kN)
20 Model
15
10
0
-0.005 -0.004 -0.003 -0.002 -0.001 0.000 0.001 0.002 0.003 0.004 0.005
Strain
238
35
30 Model Model
b
b
25 c c a
a d
d
Load (kN)
20
15
10
0
-0.008 -0.006 -0.004 -0.002 0.000 0.002 0.004 0.006 0.008 0.010
Strain
30
Model
Model
25
b a
20 b
c
a
Load (kN)
15
c
10
0
-0.005 -0.004 -0.003 -0.002 -0.001 0.000 0.001 0.002 0.003 0.004 0.005
Strain
239
40
b c
a Model
35
b a
30 Model
c
25
Load (kN)
20
15
10
0
-0.005 -0.004 -0.003 -0.002 -0.001 0.000 0.001 0.002 0.003 0.004 0.005
Strain
40
a Model a
35 b Model
b c
c
30
25
Load (kN)
20
15
10
0
-0.012 -0.008 -0.004 0.000 0.004 0.008 0.012
Strain
240
Chapter 8 Parametric Study on Sandwich Panels
CHAPTER 8: PARAMETRIC STUDY ON SANDWICH PANELS
8.1 Introduction
Polyurethane foams used in the core of sandwich panels are made of a skeleton of open
or closed cells of a somewhat regular size. These cells are essentially air voids. By
controlling the volume ratio of these voids, the density of the polyurethane can be varied.
The material can then be used in several applications, including aircraft industry,
automobile, packaging, and other structural applications. The mechanical response of
polyurethane foam depends on its architecture, and on the intrinsic properties of the
polymer in the cell wall (Gibson and Ashby, 1988). The architecture is determined by
the cell wall thickness, the size distribution and the shape of the cells. Theoretical studies
on foam have addressed the behaviour of low density foams by simulating the structure
of this foam as a compact assembly of walls and struts. These models were either
analytical or numerical finite element models.
A comprehensive modelling methodology of the mechanical behaviour of cellular
materials has been proposed and validated by Youssef et al (2005). The actual
microstructure of the polyurethane foam has been obtained by processing of X-ray
topographic data. Then, the solid volume of the foam was meshed using tetrahedral finite
elements. The study simulated the compression test on the polyurethane foam by using
the constitutive equations of the bulk material. The model could account for open cell
materials as well as for closed cells. On the other hand, there are simpler models, which
largely simplify this microstructure, such as the model presented by Gibson and Ashby
(1988). This model was based on the assembly of geometrically symmetric cells
241
(rectangular prism cubes) and relates analytically the elasticity and yield stress to the
foam relative density.
This chapter presents, in section 8.3, a parametric study using two models
developed earlier in Chapters 6 and 7, namely the numerical and analytical models.
Therefore, the study is focussed on the global structural behaviour of the sandwich
panels, rather than the microstructure. It was shown by the studies on the microstructure
of the polyurethane foam as well as the experimental studies in this thesis and by others,
that in sandwich panels without ribs, the behaviour is almost completely governed by the
foam core, rather than the skins. As such, one of the key parameters studied here, is the
polyurethane foam core density. A special modeling technique by Gibson and Ashby
(1988) was adopted to fit the experimental stress-strain curves of the polyurethane foam
tested in this study (Chapter 3). This modeling technique then allows for extrapolating
beyond the two densities studied experimentally. As a result, a family of stress-strain
curves for a large range of densities was developed first in section 8.2 and used as input
for the models. Additional key parameters were also studied in section 8.3, namely the
skin thickness and the number of ribs per unit width of the panel.
8.2 Material Modeling of Polyurethane Core
8.2.1 Behaviour under Compression
The mechanical behaviour of the polyurethane foam can be described in three different
stages as shown in Figure 8.1. The first stage is an initial linear elastic domain
characterised by the Youngs modulus and Poissons ratio. The second stage is a plateau
domain that can be described by an elastic limit stress and a hardening slope. The third
stage is a densification domain, where the material exhibits the bulk behaviour. A
242
specific property of most cellular materials is the small degree of lateral expansion, which
takes place during the plateau domain. Thus, a specimen loaded in uniaxial compression
will exhibit a volume change when cells collapse (Gibson and Ashby, 1988). As a
consequence, foams can undergo large deformations while maintaining a near constant
applied stress prior to the densification stage.
8.2.1.1 Modelling of the Linear Domain
It is not possible to make polyurethane foam material without voids. That means the
density determined as the weight of the polyurethane foam divided by its volume ( f ) is
not the real (i.e. solid) density ( s ), but instead, it is the density related to polyurethane
foam including voids. The relative density is defined by the ratio of the foam density f
(weight / volume) divided by the solid foam density s (foam without voids). The s
value has been estimated by calculation by Gibson and Ashby (1988), considering the
density of the different components of the polyurethane, their proportion, and a mixing
law, and was equal to 1200 kg/m3, as suggested by Gibson and Ashby (1988).
In the Gibson and Ashby (1988) modelling, closed cell foams are described like
an arrangement of cubic cells composed of struts and wall, as shown in Figure 8.2. For
closed cells foam, the polymer is distributed between the cell struts and the cell walls.
The stiffness of the closed cells results from three components, namely the strut bending,
the stretching of the cell walls, and the contribution of the gas pressure inside the closed
cell. Taking into account all theses components, Gibson and Ashby (1988) expressed the
elastic modulus E fc for closed cell foam as follows:
243
po 1 2 f
2
E fc f f
1 s
2
s s
s
Es (8.1)
Es 1 f
s
where s is the polymer fraction in the struts which has a value of about 0.98 according
to Gibson and Ashby (1988), ( f / s ) is the relative density, E s is the modulus of the
constitutive material (solid foam) which takes a value of about 1.6 GPa (Gibson and
Ashby, 1988). Finally, f is the Poissons ratio of the foam and it has been assumed to
have a value of 0.33 (Gibson and Ashby, 1988). The low initial pressure of the gas p o (in
the present case, being the atmospheric pressure), makes the third term of the equation
negligible (Saint-Michel et al, 2006).
8.2.1.2 Modelling of the Nonlinear Domain
For the non linear domain prediction, Gibson and Ashby (1988) have linked the yield
stress obtained in compression tests ( plc ) with the yield stress of the solid foam ( ys ) in
the following equation:
3/2
f po pa
0.3 s f 1 s
Plc
(8.2)
ys s s ys
where ys has a value of 79 MPa according to Gibson and Ashby (1988). The cell
pressure is equal to the atmospheric pressure so the last term in Equation 8.2 can be
neglected.
As with elastic collapse, large plastic strains in compression causes the cell walls
to crush together, which makes the stress-strain curve rises steeply, approaching a
limiting strain Dc . Equation 8.3 presents the predicting equation provided by Gibson and
244
Ashby (1988) for the maximum compressive strain of the polyurethane foam which
forms the densification domain of the material stress-strain curve.
f
Dc 1 1.4 (8.3)
s
To determine the stress values corresponding to any strain value between the
proportional strain and the maximum strain, Gibson and Ashby (1988) divided the foam
behaviour into two main parts. The first is a plateau of a constant stress and the second
part is a nonlinear densification response. The behaviour can be presented by the
following Equations:
1
c Plc for Dc 1 (8.4)
D
m
1
c Plc Dc for Dc 1 (8.5)
D Dc D
where m and D are constants. For polyurethane foam, m equals 1.0 and D equals 1.55.
Equations 8.1 to 8.5 were applied to the polyurethane foam tested in this study
and reported in Chapter 3, for two densities, namely 32 kg/m3 to 64 kg/m3. The
experimental and predicted compressive stress-strain curves for the polyurethane foam
are shown in Figure 8.3, which shows a reasonable fit. The same method was then used
to establish the curves for densities up to 192 kg/m3. Figure 8.4 shows the stress-strain
curves for the polyurethane with different densities varying from 32 kg/m3 to 192 kg/m3.
245
8.2.2 Behaviour under Tension
The tensile stress-strain curve of the polyurethane foam can be described by two main
parts as shown in Figure 8.5. The first domain is linear and the second one is nonlinear.
Gibson and Ashby (1988) assumed that the polyurethane foam modulus of elasticity will
have the same value in tension as in compression. Also, they assumed that the stress at
which the cell walls will yield is the same in both tension and compression. The
experimental work done on the polyurethane foam coupons of this study (Chapter 3)
showed that the polyurethane foam has different modulus values and different yield
stresses in both tension and compression. As such, a technique is described next to
account for this fact.
Based on the same assumption taken while deriving the stress-strain relationships
in compression, the tensile modulus can be determined from the following equation:
po 1 2 f
2
E ft
C f C2 1 s f
2
s s
1 s
Es (8.6)
Es 1 f
s
where C 1 and C 2 are constants introduced and are to be determined from the
experimental data obtained from the coupon tests carried out on the 32 kg/m3 and 64
kg/m3 foams. Applying Equation 8.6 for both foam densities, C 1 was found to be -0.0227
and C 2 was 2.164, so Equation 8.6 can be written as:
po 1 2 f
2
E ft
0.0445 f 2.164 1 s f
2
s s
s
Es (8.7)
Es 1 f
s
246
Gibson and Ashby (1988) showed that at a small strain, the stiffness of polyurethane
foam increases and the cell edges which lie at an angle with respect to the loading axis
will rotate towards this axis and the acting bending moment on the cell will decrease.
Their model in tension assumed that the stretching of the polyurethane foam cell will be
proportional to the material relative density with a stretching ratio of:
EA f
(8.8)
Es s
where E A is the new modulus value after the proportional yielding limit. This assumption
was found to be not applicable for the tensile stress-strain relationship of the
polyurethane foam material tested in Chapter 3. They also assumed that the stress
increase beyond the proportional strain limit ( A ) is proportional to the relative density of
the polyurethane foam (Gibson and Ashby, 1988) as follows:
A f
(8.9)
ys s
Using the same assumptions used for the compressive stress-strain relationship,
one can write the expression for the plastic tensile stress as follows:
3/2
f f
C1 s C2 1 s
Plt
(8.10)
ys s s
where C 1 and C 2 are constants introduced and determined from the experimental data
obtained from the coupon tests carried out on the 32 kg/m3 and 64 kg/m3 foams.
Applying Equation 8.10 for both foam densities, C 1 was found to be -0.00667 and C 2
was 0.1599, so Equation 8.10 can be written as:
247
3/2
f f
0.00667 s 0.1599 1 s
Plt
(8.11)
ys s s
As the tension force increases, large plastic strains cause the cell walls to yield
and align together in the direction of the tension force. This makes the stress-strain curve
rises steeply up to a limiting strain Dt where rupture occurs. Equation 8.12 is analogous
to that proposed by Gibson and Ashby (1988) for the maximum compressive strain but in
this case it will be used for the maximum tensile strain of the polyurethane foam, which
forms the nonlinear domain of the material stress-strain curve.
f
Dt C1 C2 (8.12)
s
where C 1 and C 2 are constants introduced and are determined from the experimental data
obtained from the coupon tests carried out on the 32 kg/m3 and 64 kg/m3 foams.
Applying Equation 8.12 for both foam densities, C 1 was found to be 0.09 and C 2 was
0.375, so Equation 8.12 can be written as:
f
Dt 0.09 0.375 (8.13)
s
To determine the stress values corresponding to any strain value between the
proportional limit strain and the maximum strain D t , the values of ( D t / ( D t -)) were
plotted against the ( t / plt ) for both material densities, 32 kg/m3 and 64 kg/m3 foam, as
shown in Figure 8.6. It was found that the relationship between the ( t / plt ) and ( D t / ( D t
-)) can be simulated by a polynomial of the 5th order. For each of the six constants of the
248
polynomial the average of the two constants based on the two foam densities was used in
a general single polynomial given in the following equation:
5 4
t Dt Dt
0.00055 0.00505
plt
Dt t Dt t
3 2
Dt Dt
0.43275 5.58255 (8.14)
Dt t Dt t
Dt
29.667 24.7975
Dt t
This expression is then used to re-establish the stress-strain curves for the two
foam densities. The experimental and predicted curves using Equation 8.14 are shown in
Figure 8.7. Clearly the fit is not very accurate but for the sake of the parametric study,
Equation 8.14 was used to establish a family of curves. Figure 8.8 shows these curves for
polyurethane foam with different densities varying from 32 kg/m3 to 192 kg/m3.
8.2.3 Behaviour under Shear
Gibson and Ashy (1988) also gave an expression for the shear modulus (G f ) of the
polyurethane foam, based on the fact there is a relation between the elastic modulus and
shear modulus through Poissons ratio (Equation 8.15). It is also assumed that Poissons
ratio is constant and independent of the foam density ( f = 0.33), (Gibson and Ashy,
1988). Equation 8.16 presents the proposed expression for the shear modulus, using both
equations:
Ef
Gf (8.15)
2 1 c
249
3 2 f
2
Gf f
s 1 s (8.16)
Es 8 s s

However, applying this equation to the 32 kg/m3 and 64 kg/m3 polyurethane
foams densities did not show a good fit with the results from the coupon tests. As such, a
similar technique to that used to obtain the stress-strain curve in tension is adopted here.
From the coupon test results obtained before (Chapter 3), it can be seen that the shear
stress-strain curve for polyurethane foam can be divided into two domains. The first
domain is linear and the second is nonlinear, as shown in Figure 8.9.
Using the same assumptions used by Gibson and Ashy (1988), one can write the shear
modulus as follow:
2
Gf f f
C C2 1 s
2
(8.17)
s s
1 s
Es
where C 1 and C 2 are constants introduced and to be determined from the experimental
data obtained from the coupon tests carried out on the 32 kg/m3 and 64 kg/m3 foam
densities. Applying Equation 8.17 to both foam densities, C 1 was found to be 0.73044
and C 2 was 2.22401, so Equation 8.17 can be written as:
2
Gf
0.73044 f 2.224011 s f
2
(8.18)
s s
s
Es
The material plastic shear stress can be related to the material yielding stress in the same
manner the compressive yielding stress was related as follows:
250
3/2
pl f
C1 s f C2 1 s (8.19)
ys s s
where C 1 and C 2 are constants introduced as before and are determined from the
experimental data obtained from the coupon tests carried out on the 32 kg/m3 and 64
kg/m3 foam densities. Applying Equation 8.19 for both foam densities, C 1 was found to
be -0.07445 and C 2 was 3.6304, so Equation 8.19 can be written as:
3/2
pl f f
0.07445 3.6304 1 (8.20)
ys s s
Similarly, the maximum shear strain D can be expressed by:
f
D C1 C2 (8.21)
s
where C 1 and C 2 are constants determined as before and were found to be (0.365) and
(-2.24742), respectively. Equation 8.21 can be written as:
f
D 0.365 2.24742 (8.22)
s
To determine the shear stress values corresponding to any shear strain value
between the proportional shear strain and the maximum shear strain ( D ). A technique
similar to that used in tension is used here, with a slightly different strain normalization
expression, to suit the nature of the stress-strain curve. The values of (( - pl ) / ) were
plotted against the (/ pl ) for both foam densities, 32 kg/m3 and 64 kg/m3 foam, as shown
in Figure 8.10, where pl corresponds to pl . A simple linear expression was fitted as
shown in Figure 8.10, with the constants being m and D. In order to re-establish the
251
original stress-strain curve shape, it was found that the relationship between (/ pl ) and ((
- pl ) / ) can best be simulated by a polynomial in the form given by Equation 8.23. This
equation gives the shear stress corresponding to any shear strain beyond the
proportional shear strain pl . The equation is the same one proposed by Gibson and
Ashby (1988) for compression and given in Equation 8.5.
m
1 pl
(8.23)
pl D
where D and m equal to the average values obtained from the linear expressions in Figure
8.9 for both densities, and were, 0.9 and 0.185, respectively. The experimental and
predicted shear stress-strain curves for the polyurethane foam are shown in Figure 8.11.
Figure 8.12 shows the stress-strain curves for polyurethane foam with a large range of
densities varying from 32 kg/m3 to 192 kg/m3, using Equation 8.23.
8.3 Parametric Study
8.3.1 Skin Thickness and Core Density Effects
In this section, a parametric study was conducted using the polyurethane core material
curves for different densities, established in section 8.2, along with GFRP skins of
different thickness. The FORTRAN model developed in Chapter 7 was used to perform
this parametric study. The top and bottom skin thicknesses were 1.6 mm, 3.2 mm and 4.8
mm. The core densities were varied from 32 kg/m3 to 192 kg/m3 as discussed in section
8.2. The sandwich panel used in the parametric study had the same overall dimensions as
the ones that were tested in the experimental study by Shawkat (2008). The sandwich
panel dimensions are 1500 mm in length, 300 mm in width and 78 mm overall depth, and
the panel is loaded with a uniform pressure over a span of 1400 mm. The overall panel
252
depth remained constant at 78 mm, which means that the core thickness reduces as the
skin thickness increases.
Table 8.1 summarizes the parametric study structure and results. For each of the
six core densities used, the three skin thicknesses were used, giving a total of 18 cases.
Each case was given a specific ID. Figures 8.13 to 8.30 show the load-deflection
responses for all the 18 panels. The figures also show the individual contributions of
flexural deflection, shear deflection and the Winkler deflection to the total deflection. It
can be immediately seen from the figures that the shear deflection share of the total
deflection reduces significantly as the core density increases, while the flexural deflection
contribution increases.
8.3.1.1 Effect of Skin Thickness
The load-deflection responses for panels with different skin thicknesses are shown in
Figures 8.31 to 8.36. Figure 8.37 shows the effect of skin thickness on the ultimate load,
stiffness, maximum deflection, and contribution of flexural deflection. It can be seen
from Figure 8.37(a) that increasing the skin thickness does not always lead to a
significant increase of the ultimate load. For example increasing the skin thickness from
1.6 mm to 3.2 mm enhanced the ultimate strength for all core densities, with various
degrees, except for M1. On the other hand increasing the thickness from 3.2 mm to 4.8
mm enhanced the strength significantly for the M3 and M4 densities only. The reason is
that for those two foam densities the failure mode was skin compression failure not a core
shear failure. Overall, the panel stiffness was enhanced by using thicker skins, especially
as the core density increases, as shown in Figure 8.37(b). Figure 8.37(d) shows that the
contribution of flexural deflection to the total deflection increases as the core density
253
increase, and reduces as the skin thickness increases. In general, one can conclude that
increasing skin thickness becomes more effective, particularly for strength, as core
density increases to a certain extent, in this case it is the M4 density.
8.3.1.2 Effect of Core Density
Figures 8.38 to 8.40 show the load-deflection responses for different core densities, while
Figure 8.41 summarizes the effect of core density on ultimate strength, stiffness and
deflection. Increasing the density enhances flexural strength and stiffness, up to a certain
level, namely the M4 density. Beyond this, the strength may reduce again or stabilizes.
This behaviour is a result of changing failure mode from core shear failure to skin
compression failure and then core shear failure again. It is also very clear that the
contribution of shear deflection reduces dramatically as the core density increases.
It appears from this parametric study that perhaps the optimal core density for
strength is (M3-M4) range of 96 kg/m3 to 128 kg/m3. This range represents the lowest
density necessary to achieve the highest ultimate strength and stiffness. Furthermore, this
range of density combined with the largest skin thickness used in this study, 4.8 mm,
resulted in the highest level of strength (i.e. cases M3t3 and M4t3).
8.3.2 Rib Configuration Effect
A parametric study has been carried out to determine the effect of number of longitudinal
ribs, in sandwich panels, on their strength and stiffness. The dimensions of the panels
used in the parametric study were based on the tested sandwich panels reported in
Chapter 4, namely 2500x685x78 mm. Six different rib configurations, PS1 to PS6, were
modeled in this parametric study, including one sandwich panel PS1 without any ribs as
shown in Figure 8.42. The second panel, PS2, consisted of one longitudinal rib running at
254
mid width. The third panel, PS3, has two longitudinal edge ribs. The fourth panel, PS4,
has two longitudinal edge ribs and one longitudinal centre rib. The fifth panel, PS5, has
two longitudinal edge ribs in addition to two longitudinal inner ribs running at one third
points of the panel width. The sixth panel, PS6, has two edge ribs and three longitudinal
ribs running at quarter and mid points of the panel width. Panels PS3 to PS6 also
included end transverse ribs. All panels were loaded in the same manner as the
experimental specimens in Chapter 4. Also, panels had a skin thickness of 1.6 mm for the
top and bottom skins and ribs. Internal ribs were modeled as two back-to-back C-shape
ribs to form an I-shape rib. Each rib consists of two 1.6 mm webs (3.2 mm in total) and
two flanges, each 75 mm wide. Exterior ribs had a C-shape. The total skin thickness at
the overlap zones between the skin and the rib flange is 3.2 mm. The analytical model
developed in Chapter 7 is simplified in that it treats the panel width as a plain stress
problem, with no consideration of localized effects and stress concentrations at the rib
location. As such, it was decided to use the 3-D FE model described in Chapter 6 to
model these panels.
For panel PS1, the predicted failure mode was outward wrinkling (local buckling)
of the top skin at the panel mid span, exactly as in panel S1 in Chapter 6. This was
indicated by solution divergence due to instability and excessive top skin deformation.
Figure 8.43 shows the Tsai-Wu failure criteria index values which did not reach the unity
as an indication that top skin crushing failure did not occur. The predicted failure mode
for PS2 was due to top skin crushing at the rib flange edge, at mid span, essentially the
same as panel S2 in Chapter 6. This was indicated by solution divergence because of the
255
GFRP material failure. Figure 8.44 shows the Tsai-Wu failure criteria index values at
failure. The unity value in the figure indicates failure.
Figure 8.45 shows the predicted failure mode for panel PS3. The predicted failure
mode by the FE model was a compression failure near the panel transverse centre line
(near both panel edge and centre line) as shown in Figure 8.45 by the Tsai-Wu failure
indices value. The predicted failure mode for panel PS4 (provided with longitudinal and
exterior rib) was the top skin crushing at the panel centre line and also at a line near the
panel centre line, as shown in Figure 8.46.
For panels PS5 and PS6, the predicted failure modes are shown at Figures 8.47
and 8.48, which are due to top skin crushing also, at the mid span. The figures show the
Tsai-Wu failure criteria index reaching unity at failure.
Figure 8.49 shows the variation of the strength and stiffness of panels PS1 to PS6
with the percentage of the total rib cross sectional area to the core area. In calculating the
rib area, the exterior rib is a C-shape with a 75 mm wide flange and a 1.6 mm thickness
for the flange and web, while the interior rib is an I-shape with 150 mm wide flange, a
1.6 mm thick flange and a 3.2 mm thick web. Figure 8.49 shows a steep trend of increase
in strength and stiffness as the longitudinal ribs number increase. The figure shows a
stabilizing trend for strength beyond PS5 with four ribs but not for stiffness. It appears
that increasing the number of ribs in this study continued to show gain in stiffness. This is
likely because of the extra flanges that are associated with extra ribs. It is to be noted that
the width was only 685 mm and that loading and supporting conditions were uniform
across the width of the panel.
256
Table 8.1 Summary of the parametric study and results

Skin
Pu %age k %age %age Failure
Core density Thickness ID
(kN) Gain (kN/mm) Gain (mm) Reduced mode
(mm)
1.6 M1t1 8.0 ------ 0.263 ------ 53.27 ------ CS
M1
3 3.2 M1t2 7.8 -2.5 0.293 11.85 47.25 11.30 CS
(32 kg/m )
4.8 M1t3 7.6 -5.0 0.302 15.22 41.86 21.42 CS
1.6 M2t1 20.0 150 0.492 87.69 51.55 3.23 CS
M2
3 3.2 M2t2 26.4 230 0.806 207.23 62.61 -17.53 CS
(64 kg/m )
4.8 M2t3 25.6 220 0.907 245.87 46.51 12.69 CS
1.6 M3t1 21.6 170 0.916 249.24 27.82 47.78 SC
M3
3 3.2 M3t2 38.4 380 1.534 484.93 35.39 33.56 SC
(96 kg/m )
4.8 M3t3 56 600 1.975 653.09 42.28 20.63 SC
1.6 M4t1 21.6 170 0.998 280.30 25.23 52.64 SC
M4
3 3.2 M4t2 40.8 410 1.744 564.98 28.89 45.77 SC
(128 kg/m )
4.8 M4t3 57.6 620 2.308 779.80 34.17 35.86 SC
1.6 M5t1 22.4 180 0.998 280.46 25.1 52.88 SC
M5
3 3.2 M5t2 42.0 425 1.647 528.09 28.9 45.75 SC
(160 kg/m )
4.8 M5t3 44.4 455 2.368 802.96 21.81 59.06 CS
1.6 M6t1 22.4 180 1.034 294.14 24.67 53.69 SC
M6
3 3.2 M6t2 34.8 335 1.908 627.34 20.85 60.86 CS
(192 kg/m )
4.8 M6t3 33.6 320 2.618 898.18 13.91 73.89 CS
Pu = Ultimate load CS = Polyurethane foam core shear failure

K = Stiffness SC = GFRP top skin crushing failure
= Deflection at ultimate
257
Compressive Stress
Densification
c
Plc
Plateau (Elastic Buckling)
Efc Linear Elastic
Compressive Strain c Dc
Figure 8.1 Schematic for the compressive stress-strain curve for polyurethane
foam
Figure 8.2 Schematic for an idealized cubic cell model for closed-cell foam, showing
the edge thickness, t e , and the face thickness, t f . [Gibson and Ashby (1988)]
258
64 kg/m3 polyurethane
foam (Predicted)
3
Stress (MPa)
64 kg/m3
2 polyurethane foam
(Experimental)
32 kg/m3
polyurethane foam
(Predicted)
1 foam (Experimental)
0
0.0 0.2 0.4 0.6 0.8 1.0
Strain
Figure 8.3 Experimental and predicted compressive stress-strain relationship for
32 kg/m3 and 64 kg/m3 polyurethane foam
16
Densities are in kg/m3
14 Density =192
12
160
10
Stress (MPa)
128
8
6 96
64
4
32
2
0
0 0.2 0.4 0.6 0.8 1
Strain
Figure 8.4 Predicted compressive stress-strain curves for polyurethane foam
densities ranging from 32 kg/m3 to 192 kg/m3
259
Tensile Stress
t
Wall Alignment
Linear Elastic
Plt
Plastic Buckling
Eft
Tensile Strain Dt
Figure 8.5 Schematic for the tensile stress-strain curve for polyurethane foam
50
45 64 kg/m3
polyuretrhane foam y = -0.00153x5 + 0.01035x4 + 0.44160x3
40 - 6.87306x2 + 37.34553x - 31.72655
35 32 kg/m3
polyuretrhane foam
30
y = 0.0004x5 - 0.0205x4 + 0.4239x3 -
/plt
25
4.292x2 + 21.988x - 17.868
20
15
10
0
0 2 4 6 8 10 12 14 16 18
Dt / ( Dt -)
Figure 8.6 Normalized tensile stress-strain curve for polyurethane foams with
32 kg/m3 and 64 kg/m3 densities
260
0.45
0.40
0.35 64 kg/m3 polyurethane

3
foam (Experimental) 64 kg/m polyurethane
foam (predicted)
0.30
Stress (MPa)

foam (predicted)
0.20
0.15
0.10
3
32 kg/m polyurethane
0.05 foam (Experimental)
0.00
0.00 0.02 0.04 0.06 0.08 0.10
Strain
Figure 8.7 Experimental and predicted tensile stress-strain curve for
polyurethane foam with 32 kg/m3 and 64 kg/m3 densities
0.40
3
Densities are in kg/m 160
0.35 128
96
Density =192
0.30 64
0.25
Stress (MPa)
0.20 32
0.15
0.10
0.05
0.00
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Strain
Figure 8.8 Predicted tensile stress-strain curves for polyurethane foam with densities
ranging from 32 kg/m3 to 192 kg/m3
261
Shear Stress

Pl
Cell Shear Densification
Linear Elastic
Gf
Pl Shear Strain D
Figure 8.9 Schematic for the shear stress-strain curve of polyurethane foam
2.4
m D
2.2 y = 0.1853x + 0.9724

foam m D
/pl
y = 0.1832x + 0.8466
1.6
1.4
foam
1.2
1
0 1 2 3 4 5 6 7 8
( - pl)/
Figure 8.10 Normalized shear stress-strain curve for polyurethane foams with
32 kg/m3 and 64 kg/m3 densities
262
0.50
3
0.45 64 kg/m polyurethane
foam (Experimental)
0.40
3
0.35 64 kg/m polyurethane
Shear stress (MPa)
foam (Predicted)
0.30
3
0.25 foam (Predicted)
0.20
0.15
3
0.10 foam (Experimental)
0.05
0.00
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Shear Strain
Figure 8.11 Experimental and predicted shear stress-strain curve for
polyurethane foam with 32 kg/m3 and 64 kg/m3 densities
1.00
Density =192
Densities are in kg/m3
0.90 160
128
0.80
96
0.70
Shear stress (MPa)
0.60
0.50
64
0.40
0.30
32
0.20
0.10
0.00
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Shear strain
Figure 8.12 Predicted shear stress-strain curves for polyurethane foam with densities
ranging from 32 kg/m3 to 192 kg/m3
263
10.0
9.0
8.0
7.0
6.0
Load (kN)
5.0
4.0
3.0

1.0
0.0
0 10 20 30 40 50 60
Deflection (mm)
Figure 8.13 Load-deflection response of sandwich panel M1t1
10.0
9.0
8.0
7.0
6.0
Load (kN)
5.0
4.0
3.0

1.0
0.0
0 10 20 30 40 50 60
Deflection (mm)
264
10.0
9.0
8.0
7.0
6.0
Load (kN)
5.0
4.0
3.0

1.0 Total deflection (Analytical)
0.0
0 10 20 30 40 50 60
Deflection (mm)
32.0
28.0
24.0
20.0
Load (kN)
16.0
12.0

4.0
0.0
0 10 20 30 40 50 60 70
Deflection (mm)
265
32.0
28.0
24.0
20.0
Load (kN)
16.0
12.0

8.0

4.0
0.0
0 10 20 30 40 50 60 70
Deflection (mm)
32.0
28.0
24.0
20.0
Load (kN)
16.0
12.0

8.0
4.0 Winkler deflection only (Analytical)
0.0
0 10 20 30 40 50 60 70
Deflection (mm)
266
60.0
50.0
40.0
Load (kN)
30.0
20.0

10.0
0.0
0 5 10 15 20 25 30 35 40 45
Deflection (mm)
60.0
50.0
40.0
Load (kN)
30.0
20.0

10.0
0.0
0 5 10 15 20 25 30 35 40 45
Deflection (mm)
267
60.0
50.0
40.0
Load (kN)
30.0
20.0

10.0
0.0
0 5 10 15 20 25 30 35 40 45
Deflection (mm)
60.0
50.0
40.0
Load (kN)
30.0
20.0
Shear Ddeflection only (Analytical)

0.0
0 5 10 15 20 25 30 35 40
Deflection (mm)
268
60.0
50.0
40.0
Load (kN)
30.0
20.0

0.0
0 5 10 15 20 25 30 35 40
Deflection (mm)
60.0
50.0
40.0
Load (kN)
30.0
20.0

10.0
0.0
0 5 10 15 20 25 30 35 40
Deflection (mm)
269
50.0
40.0
30.0
Load (kN)
20.0
0.0
0 5 10 15 20 25 30
Deflection (mm)
50.0
40.0
30.0
Load (kN)
20.0
0.0
0 5 10 15 20 25 30
Deflection (mm)
270
50.0
40.0
30.0
Load (kN)
20.0
0.0
0 5 10 15 20 25 30
Deflection (mm)
40.0
35.0
30.0
25.0
Load (kN)
20.0
15.0
0.0
0 5 10 15 20 25 30
Deflection (mm)
271
40.0
35.0
30.0
25.0
Load (kN)
20.0
15.0
0.0
0 5 10 15 20 25 30
Deflection (mm)
40.0
35.0
30.0
25.0
Load (kN)
20.0
15.0
Shear Ddeflection only (Analytical)
0.0
0 5 10 15 20 25 30
Deflection (mm)
272
10.0
9.0
M1t1
8.0
M1t2
7.0 M1t3
6.0
Load (kN)
5.0
4.0
3.0
2.0
1.0
0.0
0 10 20 30 40 50 60
Deflection (mm)
Figure 8.31 Effect of skin thickness on load-deflection response of sandwich panels
with core density of 32 kg/m3
32.0
28.0 M2t2
M2t3
24.0
M2t1
20.0
Load (kN)
16.0
12.0
8.0
4.0
0.0
0 10 20 30 40 50 60 70
Deflection (mm)
273
60.0
M3t3
50.0
M3t2
40.0
Load (kN)
30.0
M3t1
20.0
10.0
0.0
0 5 10 15 20 25 30 35 40 45
Deflection (mm)
60.0
M4t3
50.0
M4t2
40.0
Load (kN)
30.0
M4t1
20.0
10.0
0.0
0 5 10 15 20 25 30 35 40
Deflection (mm)
274
50.0
M5t3
40.0 M5t2
30.0
Load (kN)
M5t1
20.0
10.0
0.0
0 5 10 15 20 25 30
Deflection (mm)
40.0
M6t3 M6t2
35.0
30.0
25.0
M6t1
Load (kN)
20.0
15.0
10.0
5.0
0.0
0 5 10 15 20 25 30
Deflection (mm)
275
70 3
M6
60 M4
2.5 M5
M3
50 M4
Ultimate load (kN)
Stiffness (kN/mm)
M5 M3
40
M6 1.5
30
M2
1
20 M2
10 0.5
M1 M1
0 0
0 2 4 6 0 2 4 6
Skin thickness (mm) Skin thickness (mm)
(a) Ultimate load (b) Stiffness
80 100
%age of flexural deflection from total deflection
70 M6
80
M5
Deflection at ultimate (mm)
60 M4
60 M3
50
M2
40 M1 40
M3 M4
30 M2
20
20 M5
M1
M6 0
10
0 -20
0 2 4 6 0 2 4 6
Skin thickness (mm) Skin thickness (mm)
(c) Ultimate deflection (d) Flexural deflection
Figure 8.37 Effect of skin thickness on behaviour of panels
276
25
Compression skin failure

M5t1
20
M6t1 M2t1
M3t1
15 M4t1
Core shear failure
Load kN
10
M1t1
0
0 10 20 30 40 50 60
Deflection mm
Figure 8.38 Effect of core density on load-deflection response of sandwich panels
with skin thickness = 1.6 mm
45
M5t2
40 Compression skin failure
Core shear failure
35
30 M6t2 M4t2
M3t2 M2t2
Load kN
25
20
Core shear failure
15
10 M1t2
0
0 10 20 30 40 50 60 70
Deflection mm
277
70
Compression skin failure
60
M4t3
Core shear failure
50
M5t3
40
Load kN
M3t3
M6t3
30
M2t3
20
Core shear failure
M1t3
10
0
0 10 20 30 40 50
Deflection mm
278
70 3
M6
M5
M1
M2
M3
M4
M5
M6
M4
M3
60
2.5
t3
M2
t2
M1
50
2
Ultimate load (kN)
Stiffness (kN/mm)
t2
t3 t3 t2
40
1.5
30
t1
t1 1
20
0.5
10
0 0
0 50 100 150 200 0 50 100 150 200
3 3
Core density (kg/m ) Core density (kg/m )
(a) Ultimate load (b) Stiffness
70 100
%age of flexural deflection from total deflection
90 t1
60
t2 80 t2
Deflection at ultimate (mm)
50
M2
70
M6
t3 60 t3
M1
40
50 M5
30
40
M1
M4
t1
30
M2
20
M3
M3
20
M4
10
M5
10
M6
0 0
0 50 100 150 200 0 50 100 150 200
3 3
Core density (kg/m ) Core density (kg/m )
(c) Ultimate deflection (a) Flexural deflection
Figure 8.41 Effect of skin thickness on behaviour of panels
279
685 mm
100 1150 mm 1150 mm 100
Panel PS1 Panel PS2
Panel PS3 Panel PS4
Panel PS5 Panel PS6
Figure 8.42 Rib configurations used in the parametric study
C.L. C.L.
Mid-span
Mid-width
C.L. C.L.
Mid-width Outer edge
Mid-span
Support
Tsai-Wu failure criteria index values 0.613 < 1.0
Figure 8.43 Predicted failure mode of panels PS1
280
C.L. C.L.
Mid-span
Mid-width
C.L. C.L.
Outer edge
Mid-span Mid-width
Support
Tsai-Wu failure criteria index values 1.03 > 1.0
Figure 8.44 Predicted failure mode of panel PS2

C.L. C.L.
Mid-span
Mid-width
C.L. C.L. Outer edge

Mid-width Support
Mid-span
281
C.L. C.L.
Mid-span
Mid-width
C.L. C.L.
Outer edge
Mid-span Mid-width
Support

C.L. C.L.
Mid-span
Mid-width
C.L. C.L.
Outer edge
Mid-span Mid-width
Support
282
C.L. C.L.
Mid-span
Mid-width
C.L. C.L.
Outer edge
Mid-span Mid-width
Support
1.0 45.0
PS5 PS6
0.9 Strength 40.0
PS4
0.8 35.0
PS2, PS3
0.7
Stiffness (kN/mm)
30.0
Strength (kN)
0.6
Stiffness 25.0
0.5 PS1
20.0
0.4
15.0
0.3
0.2 10.0
0.1 5.0
0.0 0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0
% of total GFRP ribs cross-sectional area-to-total area
Figure 8.49 Effect of rib configuration on strength and stiffness of sandwich

panel
283
Chapter 9 Summary and Conclusions
CHAPTER 9: SUMMARY AND CONCLUSIONS
9.1 Summary
The use of composite sandwich panels in civil structural engineering applications has not
gained yet a wide spread, as in other industries. Indeed, sandwich panels have the
potential to be effectively used in several applications such as cladding panels, light-
weight decking and roofing. Sandwich panels exhibit various types of failure modes,
depending on the loading type, and the constituent material properties, namely the GFRP
skins and ribs and the polyurethane foam core. These failure modes include flexural
wrinkling or crushing of the compressive skins, and shear failure of the polyurethane
foam core. It is likely that the design of sandwich panels will be governed by stiffness,
more so than strength. In order to achieve a safe and economical use of sandwich panels
in structural applications, it is important to gain a thorough understanding of the
behaviour of these panels, particularly in flexure.
The present study was motivated by a particular application of sandwich panels,
namely for cladding of buildings. The panels are composed of a polyurethane foam core
sandwiched between two layers of GFRP skins, adhered using epoxy resin. In this
research project, the flexural behaviour of sandwich panels with and without GFRP ribs
connecting the skins was investigated experimentally, numerically, and analytically. Most
of the studies in this thesis are carried out in a general-enough manner such that the
findings are not only applicable to cladding wall panels but also to several other potential
applications such as roofing and flooring panels and light-weight decking.
284
The experimental program was divided into three phases. Phase I investigates the
material properties of the polyurethane foam core and GFRP skins and ribs. In Phase II,
six medium size test specimens (2500x660x78 mm) were fabricated using different
configurations of GFRP ribs connecting the two skins to examine the effect of rib
configurations on flexural strength, stiffness, and failure modes. In Phase III, two full
scale (9150x2440x78 mm) sandwich panels were fabricated to represent what is
envisioned to be the actual building cladding system used in the field and were tested
under a realistic air pressure loading as well as discrete mechanical loading.
Two series of models were also developed. The first comprises FE models for the
panels tested in phases II and III of this study and for other sandwich panels tested by
Shawkat (2008) for different core densities. The models accounted for the material and
geometric nonlinearities. The second series of models was FORTRAN-coded analytical
models based on equilibrium and strain compatibility, accounting for the excessive shear
deformations. The models also capture the localized deformations of the loaded skin
using beam-on-elastic foundation principles. The FEA and analytical models were
successfully validated using experimental results and were used in a parametric study to
examine the effects of core density, skin thickness and rib spacing on flexural
performance of the panels.
9.2 Conclusions
9.2.1 Simply-Supported Panel Tests with different Rib Configurations
1. By adding GFRP ribs, flexural strength and stiffness of the sandwich panels have
generally increased substantially, depending on the rib configuration, relative to a
285
panel without any ribs. However, the deflection at ultimate was not much affected.
The load-deflection response of all panels was fairly linear up to failure.
2. Providing a single internal rib in the longitudinal direction at mid width resulted in a
91% increase in flexural strength and stiffness. On the other hand, providing exterior
ribs (a diaphragm) around the full perimeter resulted in about 51% increase in flexural
strength and stiffness. When both configurations were combined, the
increase was 154%.
3. The combined effect of internal and exterior ribs on strength, which is a 154%
increase, is equivalent to doubling the density of the foam core material in a panel
without any ribs.
4. The ultimate strength of the sandwich panel without ribs is equivalent to a
conventional reinforced concrete (RC) panel of similar cross-section, with a moderate
(0.6%) steel reinforcement ratio s . The strengths of panels with ribs are equivalent to
RC panels with heavy (1.4 to 2%) s . Sandwich panels, however, are 9 to 14 times
lighter in weight than RC panels.
5. In sandwich panels with a soft core (32 kg/m3 density) and without any ribs, shear has
contributed over 50% of mid-span deflection. By adding GFRP ribs, flexure became
more dominant and shear deformations of the ribs contributed 15 to 20 % of the total
deflection. Simple analytical expressions have been proposed to capture these effects
at the linear range and showed reasonable accuracy.
6. Design of sandwich panels is clearly governed by stiffness rather than strength. The
loads, at which deflection limits specified by common design codes were imposed,
such as span/(180-360), were only 10 to 20% of the ultimate loads.
286
7. Failure of the panel without ribs and those with internal ribs occurred due to outwards
wrinkling of the compression GFRP skin, whereas panels with both internal and
exterior ribs were restrained against wrinkling and failed by crushing of the
compression skin.
8. Providing a transverse rib at mid span had an insignificant effect on flexural strength
and stiffness, because of the one-way bending nature of these panels.
9.2.2 Two-Span Full Scale Cladding Panel Tests
1. The panels showed a linear flexural response and achieved a maximum pressure of 7.5
kPa, which is 2.6 times the maximum factored design wind pressure in Canada.
2. The deflection of the panel under the maximum design service wind pressure in
Canada did not exceed span/360.
3. At ultimate, failure occurred in a successive manner, at three stages:
(a) First failure occurred by outwards local buckling (wrinkling) of the GFRP skin in
compression, adjacent to the internal supports (negative moment region).
(b) After first failure, the load dropped very slightly then recovered, when the second
failure occurred by shear of the polyurethane core, adjacent to the internal supports.
The load dropped to a residual capacity of 76% of the peak value.
(c) Because the system is statically indeterminate, it did not suffer global collapse.
Further loading increased the capacity up to 85% of the peak load, when the final
failure occurred by outwards wrinkling and crushing of the GFRP compression skin
near mid-span (positive moment region).
287
9.2.3 Numerical and Analytical Models
1. The two independent models, namely the FE and analytical models provided very
good predictions of flexural responses and failure modes.
2. Sandwich panels, particularly those without ribs, are vulnerable to a change of
thickness due to core softness. Also, the loaded skin is vulnerable to localized bending
and deflections under concentrated loads. The FE and analytical models were very
successful in capturing these effects.
3. The analytical model had an advantage, over the FE model, of being able to isolate
and present separately, the individual contributions of flexure, shear, and localized
skin deformations, to the overall deflection of the panel.
4. Both the 2-D and 3-D FE models provided very good predictions, with insignificant
differences, of panels tested in one-way bending.
5. A sensitivity study using the analytical model showed that the most important features
in the model are accounting for material non-linearity of the core and enforcing the
failure criteria. The accuracy of predictions was less sensitive to neglecting geometric
non-linearities. Nonetheless, it is incorporated in the model.
6. The models have a limitation in that they can not predict the post-peak load behaviour
beyond first failure in statically indeterminate panels, where a progressive failure
continues to occur at lower load levels.
9.2.4 Parametric Study
1. As the density of polyurethane foam material increases, its tensile, compressive and
shear strengths and stiffnesses increase, and the strain at ultimate reduces.
288
2. As the core density increased from 32 to 192 kg/m3 the contribution of shear to the
overall deflection reduced from about 90 to 10 percent in panels without ribs.
3. For a very low density core (32 kg/m3), increasing the skin thickness in panels without
ribs has an insignificant effect on flexural strength and stiffness, as failure is
consistently governed by core shear failure.
4. As the core density of panels without ribs increases, failure mode changes from core
shear to compressive skin failure associated with an increase in strength and stiffness.
At large skin thicknesses, this trend could revert to core shear failure associated with
reduction in strength.
5. As the core density in panels without ribs increases, increasing the skin thickness
becomes more effective, leading to enhancement in strength and stiffness, but only up
to a certain core density.
6. It appears that the optimal core density of the sandwich panels without ribs is within
the 96 to 128 kg/m3 range. This represents the lowest density necessary to achieve the
highest ultimate strength and stiffness.
7. As the spacing of the GFRP ribs is reduced, the flexural strength and stiffness of the
panels increased. At a spacing of about 2.9 times the panel thickness, the strength
stabilized.
8. Failure of the GFRP tension skin is very unlikely in sandwich panels with or without
ribs. No panels in the entire study failed by fracture of the tension skin.
9.3 Recommendations for Future Work
This study provided an insight into the behaviour of sandwich panels, a rather uncommon
system in structural applications, which showed a great promise. A number of major
289
achievements have been accomplished in terms of thorough understanding of flexural
behaviour, failure modes, and analytical and numerical modeling. To further promote
structural engineering applications of this system, the following areas need further
investigations:
1. A thorough flexural investigation of panels in two-way bending. Although cladding
panels in this study were supported at discrete points, little two-way bending took
place because of the aspect ratio of the panel geometry. Further studies may focus on
the effect of two-way bending and the associated bi-axial stresses in the skins on its
strength and wrinkling failure modes. Also, the effect of two-way bending on shear
failure of the core needs to be explored.
2. Behaviour of panels under combined bending and axial compression loads. The panels
may have some limited capacity as load-bearing walls. The study may investigate
slenderness effects and overall buckling behaviour of the panels.
3. Behaviour of the panels under high-cycle fatigue to simulate repeated loading as in
cladding panels under wind pressure and suction or decking and flooring panels under
live load.
4. Develop a proper fire insulation system for the panels to be acceptable for building
applications.
5. Develop a practical supporting system of the panels for cladding applications.
6. Study the long term performance of the panels under sustained loading, in case of
flooring and roofing applications, and the durability of the panels under different
environmental conditions.
290
References
REFERENCES
1. Allen, H. G. (1969). Analysis and design of structural sandwich panels. Pergamon
Press, Oxford, London, England
2. ANSYS program revision 11 Users Manual, Swanson Analysis Systems.
3. ASTM D3039/D3039-93, Standard test method for tensile properties of polymer
matrix composite materials; 2000.
4. ASTM D3410/D 3410M-03, Standard test method for compressive properties of
polymer matrix composite materials with unsupported gage section by shear loading.
5. ASTM D3518/D 3518M-94 (Reapproved 2001) , Standard test method for in-plane
shear response of polymer matrix composite materials by tensile test of a 45o
laminate
6. ASTM C297-04, Standard test method for flatwise tensile strength of sandwich
constructions; 2004.
7. ASTM C365-03, Standard test method for flatwise compressive strength of sandwich
cores; 2004.
8. ASTM C273-00, Standard test method for shear properties for sandwich cores
materials; 2004.
9. Avils, F., and Carlsson, L. (2006), Three-dimensional Finite Element Buckling
Analysis of Debonded Sandwich Panels. Journal of Composite Materials, Vol. 40,
No. 11, 993-1008.
10. Awad, Z.K., Aravinthan, T. and Zhuge, Y., (2010), Cost optimum design of
structural fibre composite sandwich panel for flooring applications. CICE 2010,
291
References
Beijing, China.
11. Bau-Madsen NK, K.-H. Svendsena and A. Kildegaard. (1993), Large deflections of
sandwich platesan experimental investigation. Compos Struct; 23:4752.
12. Berthelot, J, and Lolive, E. (2002), Non-linear Behaviour of Foam Cores and
Sandwich Materials, Part 1: Materials and Modelling. Journal of Sandwich
structures and materials, Vol. 4, 219-247.
13. Boni, T. L., and Mller, S. F. (2008a), Laterally supported sandwich panels
subjected to large deflectionsPart 1: Test apparatus design and experimental
results. Thin-Walled Structures, Volume 46, Issue 4, April 2008, Pages 413-422.
14. Boni, T. L., and Mller, S. F. (2008b), Laterally supported sandwich panels
subjected to large deflections: Part 2: FE analyses and model validation. Thin-
Walled Structures, Volume 46, Issue 4, April 2008, Pages 423-434.
15. Burton WS, Noor AK.(1996), Assessment of continuum models for sandwich panel
honeycomb cores. Comput Meth Appl Mech ;145:34160.
16. Chandrashekhara, K., and Krishnamurthy (1990), Free vibration of composite beams
including rotary inertia and shear deformation." Composite Struct. 14, 269-279.
17. Chang, W. S., Ventsel, E., Krauthammer, T., and John, J. (2005), Bending
behaviour of corrugated-core sandwich plates. J. of Composite Structures 70, 8189.
18. De Boor, Carl (2001), A Practical Guide to Splines. Springer-Verlag, New York.
19. Demiray, S., Becker, W., and Hohe, J. (2004), A triangular v. Karman type finite
element for sandwich plates with transversely compressible core. J. of Comput.
Methods Appl. Mech. Engrg. 193, 22392260.
20. Falzon, B. G., and Hitchings, D. (2003), Capturing mode-switching in postbuckling
292
References
composite panels using a modified explicit procedure. J. of Composite Structures 60,
447453.
21. Fan, H.L., Meng, F.H., and Yang, W. (2007), Sandwich panels with Kagome lattice
cores reinforced by carbon fibers. J. Composite Structures 81, 533539.
22. Filon L. N. G. (1937), On antiplan stress in an elastic solid. Proc. Roy. Soc. A,
160, pp. 137-54
23. Freitas, G., Magee, C., Dardzinski, P., and Fusco, T. (1994), Fiber insertion process
for improved damage tolerance in aircraft laminates. J. Adv. Mater., 25, 3643.
24. Frostig Y, (1998), Inaccuracies and validity of simplified models in the theory of
sandwich structures. In: Proceedings of the Fourth International Conference on
Sandwich Construction, vol. 1. Stockholm, pp. 167-189.
25. Frostig, Y. and Baruch, M. (1990), Bending of sandwich beams with transversely
flexible core. AIAA J., 28 (11) 523-31.
26. Frostig, Y. (1992), Behavior of delaminated sandwich beam with transversely
flexible core -- high order theory. J. of Composite Structures 20, 1-16.
27. Frostig, Y. (1993a), High-order behaviour of sandwich beams with flexible core and
transverse diaphragms. J. of Eng. Mech., ASCE, 119(5), 955-972
28. Frostig, Y. (1993b), On stress concentration in the bending of sandwich beams with
a transversely flexible core. J. of Composite Structure. 24(2), 161-169.
29. Frostig, Y. (1998), Buckling of sandwich panels with a flexible core- High Order
theory. Int. J. Solid Structures Vol. 35, Nos. 3-4, pp. 183-204.
30. Frostig, Y. (2003), Classical and high-order computational models in the analysis of
modern sandwich panels. J. of Composites: Part B 34, 83100.
293
References
31. Frostig, Y. and Baruch, M. (1996), Localized load effects in high-order bending of
sandwich panels with transversely flexible core. J. ASCE, EM Div. 1996, 122(11),
1069- 1076.
32. Frostig, Y., and Baruch, M. (1993), High-Order buckling analysis of sandwich
beams with transversely flexible core. Journal of Engineering Mechanics, Vol. 119,
No. 3, pp 476-495.
33. Frostig, Y., and Thomsen, O. T. (2005), Non-linear behaviour of delaminated
unidirectional sandwich panels with partial contact and a transversely flexible core.
International Journal of Non-Linear Mechanics 40, 633 651.
34. Frostig, Y. and Simitses, G.J. (2004), Similitude of sandwich panels with a soft
core in buckling.J. of Composites: Part B 35, 599608.
35. Frostig, Y., Baruch, M., Vilnay, O. and Sheinman, I. (1991), Sandwich beams with
unsymmetrical skins and flexible core - bending behaviour. J. ASCE (EM Division),
117 (9) 1931-1952.
36. Frostig, Y., Baruch, M., Vilnay, O., and Sheinman, I. (1992), A high order theory for
the bending of sandwich beams with a flexible core. J. Engrg. Mech., ASCE, 118(5),
1026-1043.
37. Frostig, Y., Baurch, M. (1990), Bending of sandwich beams with transversely
flexible core. AIAA J., Vol. 28 (3) pp. 523-31.
38. Frostig, Y., Baurch, M., Vilnay, O. & Sheinman, I. (1992), A high order theory for
the behaviour of sandwich beams and a flexible core. ASCE J., Eng. Mech. Div.,
Vol. 118, No. 5, pp. 1-16
39. Frostig, Y., Thomsen, O.T., and Sheinman, I. (2005), On the non-linear high-order
294
References
theory of unidirectional sandwich panels with a transversely flexible core.
International Journal of Solids and Structures 42, 14431463.
40. Ghali, A. and Neville, A. M. (1989). Structural analysis: A unified classical
andmatrix approach, 3rd Ed., Chapman and Hall, London and New York.
41. Gibson LJ. (1989), Modelling the Mechanical Behavior of Cellular Materials
Journal of Materials Science and Engineering. Vol. 110 pp. 1-36.
42. Gibson L.J. and Ashby M.F. (1988), Cellular solids: structure and properties. Oxford:
Pergamon Press.
43. Gordaninejad, F., and Bert, C. W. (1989), A new theory for bending of thick
sandwich beams. Int. J. of Mech. Sci., 31,925-934.
44. Ha H.K. (1990), FE analysis of sandwich plates: an overview. International Journal
of Computational Structures ;37:397403.
45. He, M., and Hu, W. (2007), A study on composite honeycomb sandwich panel
structure. Materials and Design Vol. 29 (30) 709-713.
46. Heimbs, S., Schmeer, S., Middendorf, P., and Maier, M. (2007), Strain rate effects
in phenolic composites and phenolic-impregnated honeycomb structures.
Composites Science and Technology 67, 28272837.
47. Hetenyi, M.(1946), Beams on Elastic Foundation, Theory with Applications. Ann
Arbor: The University of Michigan press.
48. Hockman, l. E. (1973), Sandwich construction and design. Analysis and design of
flight vehicle structure, E. F. Bruhn, ed., S. R. Jacobs & Associates, Inc., Ind.
49. Holt, D. J., and Webber, J. P.H. (1982), Exact solution to some honeycomb
sandwich beam, plate and shell problems. J. Strain Anal., 17(1), 1-8.
295
References
50. Kant, T., and Mallikarjuna. (1989), A high-order theory for free vibration of
unsymmetrically laminated composite and sandwich plates-finite element
evaluation. Computers and Struct. 32(5), 1125-1132.
51. Kant, T., and Patil, H. S. (1991), Buckling load of sandwich columns with a higher
order theory. J. Reinforced Plastics and Composites, 10(1), 102-109.
52. Ko, W. L., and Jackson, R. H. (1991), Combined Compressive and Shear Buckling
Analysis of Hypersonic Aircraft Structural Sandwich Panels. NASA Technical
Memorandum 4290.
53. Kocher, C., Watson W., M. Gomez, M., Gonzalez I., and Birman, V. (2002),
Integrity of Sandwich Panels and Beams with Truss-Reinforced Cores. Journal of
Aerospace Engineering, Vol. 15, No. 3, pp 111-117.
54. Kooistra, G.W., Deshpande, V.S., Wadley, H.N.G. (2004), Compressive behaviour
of age hardenable tetrahedral lattice truss structures made from aluminum. Acta
Mater. 52, 42294237.
55. Lascoup, B., Aboura, Z., Khellil, K., and Benzeggagh, M. (2006), On the
mechanical effect of stitch addition in sandwich panel. Composites Science and
Technology 66, 13851398.
56. Libove, C., and Hubka, R. E. (1951), Elastic constants for corrugated-core sandwich
panels. National Advisory Committee for Aeronautics. Technical Note 2289.
57. Lim, J.H., and Kang, K.J. (2006), Mechanical behaviour of sandwich panels with
tetrahedral and Kagome truss cores fabricated from wires. International Journal of
Solids and Structures 43, 52285246.
58. Liu, T., Deng Z. C., and Lu T. J. (2006), Design optimization of truss-cored
296
References
sandwiches with homogenization. International Journal of Solids and Structures 43,
78917918.
59. Lok, S. T., and Cheng, Q. H. (2000), Elastic stiffness properties and behaviour of
truss-core sandwich panel. Journal of Structural Engineering, Vol. 126, No. 5, 552-
559.
60. Mania, R. (2005), Buckling analysis of trapezoidal composite sandwich plate
subjected to in-plane compression. J. of Composite Structures 69, 482490.
61. Marasco, A. I., Cartie, D. D. R., Partridge, I. K., and Rezai, A. (2006), Mechanical
properties balance in novel Z-pinned sandwich panels: Out-of-plane properties.
Composites: Part A 37, 295302.
62. Markaki, AE, Clyne TW. (2003a), Mechanics of thin ultra-light stainless steel
sandwich sheet material. Part 1. Stiffness. Acta Mater. 51, 13411350.
63. Markaki, AE, Clyne TW. (2003b), Mechanics of Thin Ultra-Light Stainless Steel
Sandwich Sheet Material: Part II - Resistance to Delamination, Acta Mater., 51(5),
pp. 1351-1357.
64. Meraghni, F., Desrumaux, F., and Benzeggagh, M.L. (1999), Mechanical behaviour
of cellular core for structural sandwich panels. J. of Composites: Part A 30 767
779.
65. Meyer-Piening, H.-R. (1989), Remarks on higher order sandwich stress and
deflection analyses. In Sandwich Construction - 1. Proc. First Int. Conf. on
Sandwich Construction, Royal Institute of Technology, Stockholm, Sweden, 19-21
June, eds K.-A. Olsson and R. P. Reichard, pp. 107-27.
66. National Building Code of Canada (NBCC), 2005
297
References
67. Niu, K., and Talreja, R. (1999), Modeling of wrinkling in sandwich panels under
compression. Journal of Engineering Mechanics, Vol. 125, No. 8, pp 875-883.
68. Noor AK, Burton WS, and Bert CW.(1996), Computational models for sandwich
panels and shells. Appl Mech Rev ;49(3):15599.
69. Ogorkiewicz, R. M., and Sayigh, A. A. M. (1973), Deflection of carbon
fibre/acrylic foam sandwich beams. Composites, 4, Nov., 254-257.
70. Ojalvo, I. V. (1977), Departures from elastic beam theory in laminated sandwich
and short beams. AIAA J., 15(10), 1518-1521.
71. Pamla, V. (2007), the pioneer phase of building with glass-fibre reinforced plastics
(GRP) 1942 to 1980. Thesis submitted to Architectural history institute, Design
Faculty, Germany.
72. Payder, N., and Libove, C. (1988), Bending of sandwich plates of variable
thickness. Journal of Applied Mechanics, 55,419-424.
73. Pearce, T. R. A. (1973), The stability of simply-supported panels with fibre
reinforced face plates. Thesis presented to the University of Bristol, at Bristol, U.K.,
in partial fulfillment of the requirements for the degree of Doctor of Philosophy.
74. Peled, D. and Frostig. Y. (1994), High-order bending of sandwich beams with
transversely flexible core and nonparallel skins. J. Engineering Mechanics Div.,
ASCE. 120 (6) 1255-1269.
75. Peled, D. and Frostig, Y. (1995), High-order bending of piecewise uniform
sandwich beams with a tapered transition zone and a transversely flexible core.
Composite Struct. 1995, 31, 151-162.
76. Petras, A., and Sutcliffe, M.P.F. (2000), Indentation failure analysis of sandwich
298
References
beams. Composite Structures 50, 311-318.
77. Petras, A., and Sutcliffe. M.P.F. (1999), Indentation resistance of sandwich beams.
Composite Structures 46, 413-424.
78. Plantema, F. J. (1966), Sandwich construction. John Wiley and Sons. New York,
N.Y.
79. Pokharel, N., and Mahendran, M. (2004), Finite element analysis and design of
sandwich panels subject to local buckling effects. J. of Thin-Walled Structures 42,
589611.
80. Reis, E. M., and Rizkalla, S. H. (2007), Material characteristics of 3-D FRP
sandwich panels. Construction and Building Materials, 22 (2008) 10091018.
81. Reissner, E. (1947), On bending of elastic plates. Q. J. Math., 5(4), 55-68.
82. Reissner, E. (1948), Finite deflections of sandwich paltes.J. Aerosp. Sci., 15(7),
435-440.
83. Roberts, J. C., Boyle, m. P., Wienhold, P. D., and White, G. J. (2002), Buckling,
collapse and failure analysis of FRP sandwich panels. J. of Composites: Part B 33,
315-324.
84. Russo, A., and Zuccarello, B. (2007), Experimental and numerical evaluation of the
mechanical behaviour of GFRP sandwich panels. J. of Composite Structures 81,
575586.
85. Saint-Michela, F., Chazeaua, L, Cavaill, JY, and Chabert, E., (2006), Mechanical
properties of high density polyurethane foams: I. Effect of the density Composites
part B, Vol 34, pp 83-100.
86. Schwarts-Givil H., Rabinovitch, O., and Frostig, Y. (2007), High-order nonlinear
299
References
contact effects in cyclic loading of delaminated sandwich panels. J. of Composites:
Part B 38, 86101.
87. Schwarts-Givli H. and Frostig, Y. (2001), High-order behaviour of sandwich panels
with a bilinear transverse flexible core. J. Composite structures 53, 87-106.
88. Senthilnathan, N. R., Lim, S. P., Lee, K. H., and Chow, S. T. (1988), Vibration of
laminated orthotropic plates using a simplified higher order deformation theory."
Composite Struct., 10,211-229.
89. Shanley, F.R. (1957), Strength of Materials. McGraw-Hill book company, Inc.
90. Shawkat, W. (2008), Hybrid Members Employing FRP Skins Reinforcement for
Beams and Cladding Wall Applications A Master Thesis, Queens University,
CANADA.
91. Shen, H., Sokolinsky, V. S. , and Nutt, S. R. (2004), Accurate predictions of
bending deflections for soft-core sandwich beams subject to concentrated loads. J.
Composite structures 64, 115-122.
92. Sleight, D. W., and Wang, J. T. (1995), Buckling Analysis of Debonded Sandwich
Panel Under Compression. NASA Technical Memorandum 4701.
93. Sokolinsky, V., and Frostig, Y. (1999), Boundary conditions effects in buckling of
soft core sandwich panels. Journal of Engineering Mechanics, Vol. 125, No. 8,
pp 865-874.
94. Sokolinsky, V. S., Shen, H., Vaikhanski, L., and Nutt, S. R. (2003), Experimental
and analytical study of nonlinear bending response of sandwich beams. J. of
Composite Structures 60. 219229.
95. Sokolinsky, V., and Frostig, Y. (2000), Branching behaviour in the nonlinear
300
References
response of sandwich panels with a transversely flexible core. International Journal
of Solids and Structures 37, 5745-5772.
96. Sokolinsky, V., Frostig, Y., and Nutt, S. R. (2002), Special behaviour of
unidirectional sandwich panels with transversely flexible core under statical loading.
International Journal of Non-Linear Mechanics 37,869895.
97. Swanson SR, Kim J. (2000), Comparison of a higher order theory for sandwich
beams with finite element and elasticity analyses. J Sandwich Struct. Mater 2(1):33
49.
98. Thomsen OT and Frostig Y. (1997), Localized bending effects in sandwich panels:
photoelastic investigation versus high-order sandwich theory results. Compos Struc.
37, 97108.
99. Thomsen, O. T., (1993), Analysis of local bending effects in sandwich plates with
orthotropic face layers subjected to localised loading. Composite Struct., 25, 5 11-
520.
100. Thomsen, O. T. (1992), Analysis of local bending effects in sandwich panels
subjected to concentrated loads. In Sandwich Construction - 2. Proc. 2nd Int. Conf.
on Sandwich Construction, University of Florida, Gainesville, Florida, USA, 9-12
March, eds K.-A. Olsson and D. Weissman-Berman.
101. Valdevit, L., Hutchinson, J.W., and Evans, A.G., (2004), Structurally optimized
sandwich panels with prismatic cores. International Journal of Solids and Structures
41 (1819), 51055124.
102. Weissman-Berman, D., Petrie, G. L. and Wang, M. H. (1988), Flexural Response
qf Foam-cored Sandwich Panels. Society of Naval Architects and Marine Engineers
301
References
(SNAME).
103. Youssef S, Maire E, Gaertner R. (2005), Finite element modelling of the actual
structure of cellular materials determined by X-ray tomography. Acta Mater, 53
(3):719.
104. Zhou, D, and Stronge W.J. (2005), Mechanical properties of fibrous core sandwich
panels. International Journal of Mechanical Sciences 47, 775798.
105. Zhu HX, Mills NJ, Knott JF. (1997), Analysis of high strain compression of open-
cell foams. J Mech. Phys. Solids. 45(11), 1875-1904.
106. Zhu HX, Mills NJ, Knott JF. (1998), The in-plane non-linear compression of
regular honeycomb. International Journal of Solids and Structures. 37, 1931-1949.
302
Appendix
APPENDIX A: FORTRAN CODE FOR FLEXURAL
ANALYSIS OF SANDWICH PANELS
This appendix provides the complete independent FORTRAN code that was
developed to predict the flexural behaviour of sandwich panels. The model accounts for
material nonlinearity for both GFRP skins and Polyurethane foam core. The program
accepts any material curve for tension, compression and shear behaviour. Also, the
program accounts for geometric nonlinearity because of material softness. Finally, the
program is applicable for sandwich panels loaded with different load schemes as shown
in Figure A.1. The analyzed sandwich panel could have one of six rib configurations with
any skin, core or ribs dimensions, as shown in Figure A.2.
303
Appendix
Three-point bending Four-point bending
Four equal loads Eight equal loads
Uniform load
Figure A.1 Sandwich panel load scheme
304
Appendix
No ribs Rib
Rib Configuration 1 Rib Configuration 2

Ribs Ribs
Ribs Ribs
St 0.5B f 0.5B f St B fe St
Wte
H ASt Wt ASt
Figure A.2 Rib configurations used in the FORTRAN program
305
Appendix
PROGRAM Sandwich_Panels
IMPLICIT NONE
REAL (KIND = 8):: T_s, C_s, Sahy, Ybar, D1, D2, Height, B, St, Moment,Shear, Mr, Curvature, Wt, Wte, ASt, Bf, Bfe, Rc, Mu, Error, Fs, Lt
REAL (KIND = 8):: Foam_stress, Skin_stress , Shear_strain, Shear_stress
REAL (KIND = 8):: Eo, F_T_s_f, F_C_s_f, F_T_s_s, F_C_s_s, S_m_t_s, S_m_c_s, F_m_t_s, F_m_c_s, W_s, Prf, A_s_s, A_f_s, Sm, Fm
REAL (KIND = 8):: S_s_f , Moment_fail, Shear_fail, Failure, Span, Load, Cal_Load , ybcbeg, ybcend, Xval, Yval, yppval, ypval
REAL (KIND = 8), DIMENSION (160,100):: Max, Vax, Csax, Tsax,Cax, Yax, Dsax, Sax, Ddax, It_at_x, Dvat, Dvab, Nhax, Dvax, Xval_k, Yval_k
REAL (KIND = 8), DIMENSION (22,160,100):: Sahyax, Sahyax, Enahyax, Neahyax, Bnahyax, Qahyax, Iahyax, Tahyax, Gahyax, Dvahyax
REAL (KIND = 8), DIMENSION (22,100)::Galyaw
REAL (KIND = 8), DIMENSION (100):: Dm, Dv, Dt, Db
REAL (KIND = 8):: Delta_Def_Max ,Ne, It, Slice, A_L, III, Mcp, Nft, Nfc, Nfs, NG, NGs, S_num, L_num, LS_num, Loads_num
REAL (KIND = 4):: w,x,I
REAL (KIND = 8), DIMENSION (401):: X_M, Y_M_cur, Y_M_t, Y_M_c, Ypp_M_cur, Ypp_M_t, Ypp_M_c
REAL (KIND = 8), DIMENSION (7):: X_F_t, Y_F_t, X_F_c, Y_F_c, Ypp_F_t, Ypp_F_c
REAL (KIND = 8), DIMENSION (37):: X_F_s, Y_F_s, Ypp_F_s
REAL (KIND = 8), DIMENSION (16):: X_GFRP, Y_GFRP, X_GFRP_s, Y_GFRP_s, X_GFRP_C, Y_GFRP_C, Ypp_GFRP, Ypp_GFRP_s
INTEGER ( KIND = 4 ):: ibcbeg

INTEGER ( KIND = 4 ):: ibcend
REAL (KIND = 8), DIMENSION (160,100):: ko, k, M1A, M2A, Q1A, Q2A, alpha1, alpha2, P1o, M1o, P2o, M2o, PoA, MoA, PoB, MoB
REAL (KIND = 8):: Load_Dist, Point_Dist , q, P, E_Skin, I_Skin, PP, SDP
REAL (KIND = 8), DIMENSION (160,100):: Beta, Abx, Bbx, Cbx, Dbx , Abx_a, Bbx_a, Cbx_a, Dbx_a , Abx_b, Bbx_b, Cbx_b, Dbx_b
REAL (KIND = 8), DIMENSION (160,100):: M_A, M_B, Q_A, Q_B, Wdax , Wtax, Wmax, Wsax, Wstax
REAL (KIND = 8), DIMENSION (92):: X_SDP, Y_SDP ,Ypp_SDP
OPEN(1,FILE='Foam-ten-strain.TXT', STATUS='Old')
OPEN(2,FILE='Foam-ten-stress.TXT', STATUS='Old')
OPEN(3,FILE='Foam-Com-strain.TXT', STATUS='Old')
OPEN(4,FILE='Foam-Com-stress.TXT', STATUS='Old')
OPEN(5,FILE='Foam-shear-strain.TXT', STATUS='Old')
OPEN(6,FILE='Foam-shear-stress.TXT', STATUS='Old')
OPEN(7,FILE='GFRP-strain.TXT', STATUS='Old')
OPEN(8,FILE='GFRP-stress.TXT', STATUS='Old')
OPEN(77,FILE='GFRP-shear-strain.TXT', STATUS='Old')
OPEN(88,FILE='GFRP-shear-stress.TXT', STATUS='Old')
OPEN(777,FILE='GFRP-strain-Comp.TXT', STATUS='Old')
OPEN(888,FILE='GFRP-stress-Comp.TXT', STATUS='Old')
OPEN(99,FILE='Output Window.TXT', STATUS='unknown')
OPEN(9,FILE='Moment.TXT', STATUS='unknown')
OPEN(10,FILE='Moment-Ten-Strain.TXT', STATUS='unknown')
OPEN(11,FILE='Moment-Comp-strain.TXT', STATUS='unknown')
OPEN(12,FILE='Moment-Curv.TXT', STATUS='unknown')
OPEN(13,FILE='ALL M-cur Analysis.TXT', STATUS='unknown')
OPEN(14,FILE='ALL Load-Def Analysis.TXT', STATUS='unknown')
OPEN(15,FILE='ALL Load-Def Analysis for (M) at Different Sections.TXT', STATUS='unknown')
OPEN(16,FILE='ALL Load-Def Analysis for (V) at Different Sections.TXT', STATUS='unknown')
OPEN(17,FILE='ALL Load-Strains Analysis.TXT', STATUS='unknown')
OPEN(229,FILE='Moment1.TXT', STATUS='unknown')
OPEN(2210,FILE='Moment-Ten-Strain1.TXT', STATUS='unknown')
OPEN(2211,FILE='Moment-Comp-strain1.TXT', STATUS='unknown')
OPEN(2212,FILE='Moment-Curv1.TXT', STATUS='unknown')
OPEN(2213,FILE='ALL M-cur Analysis1.TXT', STATUS='unknown')
OPEN(21,FILE='Winkler_Stress_points.TXT', STATUS='Old')
OPEN(22,FILE='Winkler_Defl_points.TXT', STATUS='Old')
OPEN(23,FILE='ALL Winkler_Load_Defl Analysis.TXT', STATUS='unknown')
OPEN(24,FILE='ALL Winkler_Load_Slope Analysis.TXT', STATUS='unknown')
OPEN(25,FILE='ALL Winkler_Load_Moment Analysis.TXT', STATUS='unknown')
OPEN(26,FILE='ALL Winkler_Load_Shear Analysis.TXT', STATUS='unknown')
OPEN(27,FILE='ALL Winkler_Load_Stress Analysis.TXT', STATUS='unknown')
OPEN(210,FILE='Winkler_ANY_OUTPUT.TXT', STATUS='unknown')
WRITE (*,*)''
WRITE (*,*) '>>>Enter the following values<<<'
WRITE (*,*) '1- Enter the Ribs Configurations as follows:'
WRITE (*,*) ' 1- NO ribs, enter 1:'
WRITE (*,*) ' 2- One long. rib, enter 2:'
WRITE (*,*) ' 3- One long. plus one trans., enter 3:'
WRITE (*,*) ' 4- Outer diaphragm, enter 4:'
WRITE (*,*) ' 5- outer diaphragm plus one long., enter 5:'
WRITE (*,*) ' 6- ALL, enter 6:'
READ (*,*) Rc
IF(Rc == 1)THEN
WRITE (*,*) 'You Entered >>> NO Ribs'
ELSEIF(Rc == 2)THEN
WRITE (*,*) 'You Entered >>> ONE Longitudinal Rib'
ELSEIF(Rc == 3)THEN
WRITE (*,*) 'You Entered >>> ONE Long. Rib plus Trans. Rib'
ELSEIF(Rc == 4)THEN
WRITE (*,*) 'You Entered >>> Outer Diaphragm'
ELSEIF(Rc == 5)THEN
306
Appendix
WRITE (*,*) 'You Entered >>> Outer Diaphragm plus One Long. Rib'
ELSEIF(Rc == 5)THEN
WRITE (*,*) 'You Entered >>> ALL Rib Configurations'
ENDIF
WRITE (*,*)''
WRITE (*,*) '2- Enter the total number of sections:'
READ (*,*) S_num
WRITE (*,*) '>>>The input value is:', S_num,' Sections'
WRITE (*,*)''
WRITE (*,*) '3- Enter the total number of FOAM layers ONLY:'
WRITE (*,*) ' "Without the Skin or the Additional skin thick."'
WRITE (*,*) ' "ALL layers (GFRP and FOAM) should be <= 22 Layers."'
READ (*,*) L_num
WRITE (*,*) '>>>The input value is:', L_num,' Layers'
WRITE (*,*)''
WRITE (*,*) '4- Enter the total applied load (N):'
READ (*,*) Load
WRITE (*,*) '>>>The input value is:', Load,' N'
WRITE (*,*)''
WRITE (*,*) '5- Enter the total number of load Substeps:'
READ (*,*) LS_num
WRITE (*,*) '>>>The input value is:', LS_num,' Substeps'
WRITE (*,*)''
WRITE (*,*) '6- Enter the total number of the applied loads'
WRITE (*,*) ' as follows:'
WRITE (*,*) ' 1- Three-Point bending, enter 1:'
WRITE (*,*) ' 2- Four-Point bending, enter 2:'
WRITE (*,*) ' 3- Six Loads, enter 3:'
WRITE (*,*) ' 4- Eight Loads, enter 4:'
WRITE (*,*) ' 5- Uniform Load, enter 5:'
READ (*,*) Loads_num
IF(Loads_num == 1)THEN
WRITE (*,*) 'You Entered >>> Three-Point bending'
ELSEIF(Loads_num == 2)THEN
WRITE (*,*) 'You Entered >>> Four-Point bending'
WRITE (*,*) 'You Entered >>> Six Loads'
WRITE (*,*) 'You Entered >>> Eight Loads'
WRITE (*,*) 'You Entered >>> Uniform Load'
ENDIF
WRITE (*,*)
WRITE (*,*)
Height = 78
B = 300
Span = 1400
St = 1.5
ASt = 0
Bf = 0
Bfe = 0
Wt = 0
Wte = 0
Moment_fail = 0
Shear_fail = 0
Eo = 2.14
E_Skin = 31000
I_Skin = (B*(St**3))/12
S_m_t_s = 275.9
S_m_c_s = 84.2
F_m_t_s = 0.16
F_m_c_s = 0.55
F_T_s_s = 0.014
F_C_s_s = 0.004
F_T_s_f = 0.076
F_C_s_f = 0.8
Prf = 0.1
Error = 0.00001
IF(Rc == 1)THEN
Lt = (Height-(2*St)) / L_num
ELSEIF(Rc /= 1)THEN
Lt = (Height-(2*St + 2*ASt)) / L_num
ENDIF
C_s = 0.0
T_s = 0.0
Shear_strain = 0.0
NG = 16
NGs = 16
Nft = 7
Nfc = 7
Nfs = 37
Mcp = 401
SDP = 92
Fs = 0
Moment = 0
READ (7,*) X_GFRP
READ (8,*) Y_GFRP
READ (88,*) X_GFRP_s
READ (777,*) X_GFRP_C
READ (888,*) Y_GFRP_C
READ (77,*) Y_GFRP_s
READ (1,*) X_F_t
307
Appendix
READ (2,*) Y_F_t

READ (3,*) X_F_c
READ (4,*) Y_F_c
READ (5,*) Y_F_s
READ (6,*) X_F_s
READ (22,*) X_SDP
READ (21,*) Y_SDP
WRITE (13,"(12A25)") "Fs","LOAD(Uniform)", "SHEAR", "MOMENT", "CURVATURE","TENSILE STRAIN","COMPRESSIVE STRAIN", "D1", "D2", "YBAR",
"P1(Tension Part)","P2(Comp. Part)"
WRITE (2213,"(12A25)") "Fs","LOAD(Uniform)", "SHEAR", "MOMENT", "CURVATURE","TENSILE STRAIN","COMPRESSIVE STRAIN", "D1", "D2", "YBAR",
"P1(Tension Part)","P2(Comp. Part)"
WRITE (*,*) '>>>N.A. Location Prediction Started'

WRITE (*,*) ' '
WRITE (99,*) '>>>N.A. Location Prediction Started'
WRITE (99,*) ' '
CALL Non_Geometry (Height,L_num,Lt,Ybar,Rc,X_F_t,Y_F_t,X_F_c,Y_F_c,X_GFRP,Y_GFRP,X_GFRP_C,Y_GFRP_C)
CLOSE (9)
CLOSE (10)
CLOSE (11)
CLOSE (12)
CLOSE (229)
CLOSE (2210)
CLOSE (2211)
CLOSE (2212)
WRITE (*,*) '>>>N.A. Location Prediction Finished'

WRITE (*,*) ' '
WRITE (*,*) '>>>Load Deflection Calculations Started'
WRITE (99,*) '>>>N.A. Location Prediction Finished'

WRITE (99,*) ' '
WRITE (99,*) '>>>Load Deflection Calculations Started'
WRITE (14,"(12A25)") "Total Load", "Shear", "Moment", "Curvature", "Flexural Defl.,mm", "Shear Defl. Top,mm", "Shear Defl. Bottom,mm", "Winkler Defl.,mm", "Total Defl.
Top,mm", "Total Defl. Bottom,mm", "Tension Strain", "Compression Strain"
WRITE (15,"(4A25)") "Load", "Section No.","Delta-Defl", "Central-Defl"
WRITE (16,"(3A25)") "Load", "Section No.","Shear Defl"
WRITE (17,"(3A25)") "Load", "Tensile Strain","Compressive Strain"
ko= 0.011302204
k = ko * B
Slice = (0.5 * Span) / (S_num-1)
IF (Loads_num == 1)THEN
w=1
WRITE (*,*) '>>>Load Deflection Calculations For Load: W =',(Load/LS_num)* w,'N'
WRITE (*,*) ' '
WRITE (*,*) ' Deflection Calculations due to Moment started'
WRITE (99,*) '>>>Load Deflection Calculations For Load: W =',(Load/LS_num)* w,'N'

WRITE (99,*) ' '
WRITE (99,*) ' Deflection Calculations due to Moment started'
IF(Rc == 1)THEN
P = (Load/LS_num)* w
Load_Dist = 0.5 * Span
DO x = 1,S_num
M_A(x, w)=0
Q_A(x, w)=0
M_B(x, w)=0
Q_B(x, w)=0
Wdax (x, w) = 0
Wtax (x, w)= 0
Wmax (x, w)= 0
Wsax (x, w)= 0
Wstax (x, w)= 0
Beta(x, w) = (k(x, w)/(4*E_Skin*I_Skin))**(0.25)
Abx(x, w) = exp(-Beta(x, w)*Load_Dist)*(cos(Beta(x, w)*Load_Dist) + sin(Beta(x, w)*Load_Dist))

Bbx(x, w) = exp(-Beta(x, w)*Load_Dist)* sin(Beta(x, w)*Load_Dist)
Cbx(x, w) = exp(-Beta(x, w)*Load_Dist)*(cos(Beta(x, w)*Load_Dist) - sin(Beta(x, w)*Load_Dist))
Dbx(x, w) = exp(-Beta(x, w)*Load_Dist)* cos(Beta(x, w)*Load_Dist)
M_A(x, w) = M_A(x, w) + (P/(4*Beta(x, w))) * Cbx(x, w)

Q_A(x, w) = Q_A(x, w) + (P/2) * Dbx(x, w)
M_B(x, w) = M_A(x, w)
Q_B(x, w) = - Q_A(x, w)
M1A(x, w) = 0.5*(M_A(x, w) + M_B(x, w))

M2A(x, w) = 0.5*(M_A(x, w) - M_B(x, w))
Q1A(x, w) = 0.5*(Q_A(x, w) - Q_B(x, w))
Q2A(x, w) = 0.5*(Q_A(x, w) + Q_B(x, w))
alpha1(x, w) = 0.5*((exp(Beta(x, w)*Span))/(sinh(Beta(x, w)*Span) + sin(Beta(x, w)*Span)))
Abx(x, w) = exp(-Beta(x, w)*Span)*(cos(Beta(x, w)*Span) + sin(Beta(x, w)*Span))

Bbx(x, w) = exp(-Beta(x, w)*Span)* sin(Beta(x, w)*Span)
Cbx(x, w) = exp(-Beta(x, w)*Span)*(cos(Beta(x, w)*Span) - sin(Beta(x, w)*Span))
Dbx(x, w) = exp(-Beta(x, w)*Span)* cos(Beta(x, w)*Span)
P1o(x, w) = 4*alpha1(x, w)*(Q1A(x, w)*(1+Dbx(x, w))+(Beta(x, w)*M1A(x, w)*(1-Abx(x, w))))

M1o(x, w) = (-2/Beta(x, w))*alpha1(x, w)*(Q1A(x, w)*(1+Cbx(x, w))+(2*Beta(x, w)*M1A(x, w)*(1-Dbx(x, w))))
308
Appendix
alpha2(x, w) = 0.5*(exp(Beta(x, w)*Span)/(sinh(Beta(x, w)*Span) - sin(Beta(x, w)*Span)))
P2o(x, w) = 4*alpha2(x, w)*(Q2A(x, w)*(1-Dbx(x, w))+(Beta(x, w)*M2A(x, w)*(1+Abx(x, w))))

M2o(x, w) = (-2/Beta(x, w))*alpha2(x, w)*(Q2A(x, w)*(1-Cbx(x, w))+(2*Beta(x, w)*M2A(x, w)*(1+Dbx(x, w))))
PoA(x, w) = P1o(x, w) + P2o(x, w)

MoA(x, w) = M1o(x, w) + M2o(x, w)
PoB(x, w) = P1o(x, w) - P2o(x, w)
MoB(x, w) = M1o(x, w) - M2o(x, w)
Point_Dist = (x-1) * Slice

IF((Load_Dist - Point_Dist) > 0.0) THEN
Beta(x, w) =(k(x, w)/(4*E_Skin*I_Skin))**(0.25)
Abx(x, w) = exp(-Beta(x, w)*abs(Load_Dist - Point_Dist))*(cos(Beta(x, w)*abs(Load_Dist - Point_Dist))+sin(Beta(x, w)*abs(Load_Dist -

Point_Dist)))
Bbx(x, w) = exp(-Beta(x, w)*abs(Load_Dist - Point_Dist))* sin(Beta(x, w)*abs(Load_Dist - Point_Dist))
Cbx(x, w) = exp(-Beta(x, w)*abs(Load_Dist - Point_Dist))*(cos(Beta(x, w)*abs(Load_Dist - Point_Dist))-sin(Beta(x, w)*abs(Load_Dist -
Point_Dist)))
Dbx(x, w) = exp(-Beta(x, w)*abs(Load_Dist - Point_Dist))* cos(Beta(x, w)*abs(Load_Dist - Point_Dist))
Wdax (x, w) = ((P*Beta(x, w)/(2*k(x, w)))*Abx(x, w))

Wtax (x, w) = ((P*(Beta(x, w)**2)/k(x, w))*Bbx(x, w))
Wmax (x, w) = ((P/(4*Beta(x, w)))*Cbx(x, w))
Wsax (x, w) = ((P/2)*Dbx(x, w))
Wstax (x, w) = ko (x, w)*Wdax (x, w)
ELSEIF((Load_Dist - Point_Dist) <= 0.0) THEN

Point_Dist)))
Point_Dist)))

Wtax (x, w) = - ((P*(Beta(x, w)**2)/k(x, w))*Bbx(x, w))
Wsax (x, w) = - ((P/2)*Dbx(x, w))
ENDIF
Abx(x, w) = exp(-Beta(x, w)*Point_Dist)*(cos(Beta(x, w)*Point_Dist)+sin(Beta(x, w)*Point_Dist))

Bbx(x, w) = exp(-Beta(x, w)*Point_Dist)* sin(Beta(x, w)*Point_Dist)
Cbx(x, w) = exp(-Beta(x, w)*Point_Dist)*(cos(Beta(x, w)*Point_Dist)-sin(Beta(x, w)*Point_Dist))
Dbx(x, w) = exp(-Beta(x, w)*Point_Dist)* cos(Beta(x, w)*Point_Dist)
Wdax (x, w) = Wdax (x, w) + ((PoA(x, w)*Beta(x, w)/(2*k(x, w)))*Abx(x, w)) + ((MoA(x, w)*(Beta(x, w)**2)/k(x, w))*Bbx(x, w))
Wtax (x, w) = Wtax (x, w) - ((PoA(x, w)*(Beta(x, w)**2)/k(x, w))*Bbx(x, w)) + ((MoA(x, w)*(Beta(x, w)**3)/k(x, w))*Cbx(x, w))
Wmax (x, w) = Wmax (x, w) + ((PoA(x, w)/(4*Beta(x, w)))*Cbx(x, w)) + ((MoA(x, w)/2)*Dbx(x, w))
Wsax (x, w) = Wsax (x, w) - ((PoA(x, w)/2)*Dbx(x, w)) - ((MoA(x, w)*Beta(x, w)/2)*Abx(x, w))
Wstax (x, w) = Wstax (x, w) + ko (x, w)*Wdax (x, w)
Abx(x, w) = exp(-Beta(x, w)*(Span-Point_Dist))*(cos(Beta(x, w)*(Span-Point_Dist))+sin(Beta(x, w)*(Span-Point_Dist)))

Bbx(x, w) = exp(-Beta(x, w)*(Span-Point_Dist))* sin(Beta(x, w)*(Span-Point_Dist))
Cbx(x, w) = exp(-Beta(x, w)*(Span-Point_Dist))*(cos(Beta(x, w)*(Span-Point_Dist))-sin(Beta(x, w)*(Span-Point_Dist)))
Dbx(x, w) = exp(-Beta(x, w)*(Span-Point_Dist))* cos(Beta(x, w)*(Span-Point_Dist))
Wdax (x, w) = Wdax (x, w) + ((PoB(x, w)*Beta(x, w)/(2*k(x, w)))*Abx(x, w)) + ((MoB(x, w)*(Beta(x, w)**2)/k(x, w))*Bbx(x, w))
Wtax (x, w) = Wtax (x, w) + ((PoB(x, w)*(Beta(x, w)**2)/k(x, w))*Bbx(x, w)) + ((MoB(x, w)*(Beta(x, w)**3)/k(x, w))*Cbx(x, w))
Wmax (x, w) = Wmax (x, w) + ((PoB(x, w)/(4*Beta(x, w)))*Cbx(x, w)) + ((MoB(x, w)/2)*Dbx(x, w))
Wsax (x, w) = Wsax (x, w) + ((PoB(x, w)/2)*Dbx(x, w)) - ((MoB(x, w)*Beta(x, w)/2)*Abx(x, w))
ENDDO
ELSEIF(Rc /= 1)THEN
DO x = 1,S_num
Wdax (x, w) = 0.0
Wtax (x, w) = 0.0
Wmax (x, w) = 0.0
Wsax (x, w) = 0.0
Wstax (x, w) = 0.0
ENDDO
ENDIF
DO x = 1, S_num
Max (x, w) = 0.5*((Load/LS_num)* w) * ((x-1)*Slice)
Vax (x, w) = ((Load/LS_num)* w) / 2
Xval = Max (x, w)
CALL Ten_Comp_Curv (Load,w,x,W_s,Prf,Span,Bf,Bfe,Max, Vax, Xval,Height,L_num,Lt,Slice,Ybar,Rc,X_M, Y_M_cur,Y_M_t,Y_M_c,&
Tsax,Csax,Cax,Yax,Dsax )
ENDDO
Yax (1, w) = 0.5*Height
CALL Mom_Defl (Slice,Dsax,Sax,Ddax,Dm,Delta_Def_Max )
WRITE (*,*) '>>>Deflection due to Moment =',Dm (w),'mm <<<'
WRITE (*,*) ' Deflection Calculations due to Shear started'
DO x = 1, S_num, 1
CALL Transformed (w,x,Bnahyax, Qahyax,Iahyax,Height,L_num,Lt,Ybar,Rc,&
X_M, Y_M_cur,Y_M_t,Y_M_c,Tsax,Csax,Cax,Yax,X_F_t,Y_F_t,X_F_c,Y_F_c,X_F_s,Y_F_s,X_GFRP,Y_GFRP,X_GFRP_C,Y_GFRP_C)
309
Appendix
ENDDO
DO x = 1, S_num
CALL Shear_Defl (w,x,Load,Iahyax,It_at_x,Lt,L_num,St,Slice,Yax,Qahyax,Rc,Tahyax,Gahyax,Vax,Bnahyax,Xval,Yval,Dvax,Galyaw,&
Dvat,Dvab,Dvahyax,Nfs,X_F_s,Y_F_s)
ENDDO
CALL Total_Defl (Dvax,S_num,Rc,Gahyax,L_num,Slice,Load,LS_num,w,Galyaw,Dvat,Dvab,Dt,Dm,Wdax,Db,Max,Vax,Tsax,&
Csax,Cax,Yax,Nhax,Height,Lt,St)
CALL New_Height_1 (w, Height, Slice, S_num, L_num, Rc, Gahyax, Wdax, Nhax)
Xval_k = 0.0
Yval_k = 0.0
DO w = 2, LS_num, 1
WRITE (*,*) ' '
WRITE (99,*) ' '

IF(Rc == 1)THEN
P = (Load/LS_num)* w
DO x = 1,S_num
Xval_k (x, w) = Wdax (x, w-1)
Xval = -Xval_k (x, w)
IF(Xval == 0.0) THEN
ko(x, w) = 0.011302204
k(x, w) = ko(x, w) * B
ELSEIF(Xval /= 0.0) THEN
CALL GetValue (SDP, X_SDP, Y_SDP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_SDP, Xval, Yval, Ypval, Yppval)
Yval_k (x, w) = Yval
ko(x, w) = (abs(Yval_k(x, w))-abs(Yval_k(x, (w-1)))) / (abs(Xval_k(x, w)) - abs(Xval_k(x, (w-1))))
k(x, w) = ko(x, w) * B
ENDIF
M_A(x, w)=0
Q_A(x, w)=0
M_B(x, w)=0
Q_B(x, w)=0
Wdax (x, w) = 0
Wtax (x, w)= 0
Wmax (x, w)= 0
Wsax (x, w)= 0
Wstax (x, w)= 0


Q_A(x, w) = Q_A(x, w) + (P/2) * Dbx(x, w)
M_B(x, w) = M_A(x, w)
Q_B(x, w) = - Q_A(x, w)
M1A(x, w) = 0.5*(M_A(x, w) + M_B(x, w))

M2A(x, w) = 0.5*(M_A(x, w) - M_B(x, w))
Q1A(x, w) = 0.5*(Q_A(x, w) - Q_B(x, w))
Q2A(x, w) = 0.5*(Q_A(x, w) + Q_B(x, w))



PoA(x, w) = P1o(x, w) + P2o(x, w)

MoA(x, w) = M1o(x, w) + M2o(x, w)
PoB(x, w) = P1o(x, w) - P2o(x, w)
MoB(x, w) = M1o(x, w) - M2o(x, w)

Point_Dist)))
Point_Dist)))
310
Appendix

Wtax (x, w) = ((P*(Beta(x, w)**2)/k(x, w))*Bbx(x, w))
Wsax (x, w) = ((P/2)*Dbx(x, w))

Point_Dist)))
Point_Dist)))

Wtax (x, w) = - ((P*(Beta(x, w)**2)/k(x, w))*Bbx(x, w))
Wsax (x, w) = - ((P/2)*Dbx(x, w))
ENDIF


ENDDO
ELSEIF(Rc /= 1)THEN
DO x = 1,S_num
Wdax (x, w) = 0.0
Wtax (x, w) = 0.0
Wmax (x, w) = 0.0
Wsax (x, w) = 0.0
Wstax (x, w) = 0.0
ENDDO
ENDIF
DO x = 1, S_num
WRITE (*,*) ' '
WRITE (*,*) '>>>Start Non-Geometry Calculations for Load: W=',(Load/LS_num)* w ,'N'
WRITE (99,*) ' '

WRITE (99,*) '>>>Start Non-Geometry Calculations for Load: W=',(Load/LS_num)* w ,'N'
Height = Nhax (x, (w-1))
WRITE (*,*) ' New Section Height at section',x ,'is',Height,'mm'

WRITE (99,*) ' New Section Height at section',x ,'is',Height,'mm'
WRITE (*,*) ' Nonlinear Geometry Calculations Have Been Finished'

WRITE (*,*) ' New (Ybar) for Section',x,'is',Ybar,'mm'
WRITE (99,*) ' Nonlinear Geometry Calculations Have Been Finished'
WRITE (99,*) ' New (Ybar) for Section',x,'is',Ybar,'mm'
Max (x, w) = 0.5*((Load/LS_num)* w) * ((x-1)*Slice)
Xval = Max (x, w)
CALL Transformed (w,x,Bnahyax, Qahyax,Iahyax,Height,L_num,Lt,Ybar,Rc,X_M, Y_M_cur,Y_M_t,Y_M_c,Tsax,Csax,Cax,Yax,X_F_t,&
Y_F_t,X_F_c,Y_F_c,X_F_s,Y_F_s,X_GFRP,Y_GFRP,X_GFRP_C,Y_GFRP_C)
ENDDO
WRITE (99,*) '>>>Deflection due to Moment =',Dm (w),'mm <<<'
WRITE (99,*) ' Deflection Calculations due to Shear started'
311
Appendix

CALL New_Height_2 (w, Slice, S_num, L_num, Rc, Gahyax, Wdax, Nhax)
10 WRITE (23,"(100e20.10)") ((Wdax (x, p), p=1,LS_num), x=1,S_num)
ENDDO
WRITE (24,"(100e20.10)") ((Wtax (x, w), w=1,LS_num), x=1,S_num)
WRITE (25,"(100e20.10)") ((Wmax (x, w), w=1,LS_num), x=1,S_num)
WRITE (26,"(100e20.10)") ((Wsax (x, w), w=1,LS_num), x=1,S_num)
WRITE (27,"(100e20.10)") ((Wstax (x, w), w=1,LS_num), x=1,S_num)
ELSEIF (Loads_num == 2)THEN

w=1
WRITE (*,*) ' '

WRITE (99,*) ' '
IF(Rc == 1)THEN
DO x = 1, S_num
M_A=0
Q_A=0
M_B=0
Q_B=0
Wdax (x, w) = 0
Wtax (x, w)= 0
Wmax (x, w)= 0
Wsax (x, w)= 0
Wstax (x, w) = 0
Load_Dist= Span/3
DO PP = 1,2
P = ((Load/LS_num)*w)/2


Q_A(x, w) = Q_A(x, w) + (P/2) * Dbx(x, w)
Abx(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))*(cos(Beta(x, w)*(Span-Load_Dist)) + sin(Beta(x, w)*(Span-Load_Dist)))

Bbx(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))* sin(Beta(x, w)*(Span-Load_Dist))
Cbx(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))*(cos(Beta(x, w)*(Span-Load_Dist)) - sin(Beta(x, w)*(Span-Load_Dist)))
Dbx(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))* cos(Beta(x, w)*(Span-Load_Dist))
M_B(x, w) = M_B(x, w) + (P/(4*Beta(x, w))) * Cbx(x, w)

Q_B(x, w) = Q_B(x, w) - (P/2) * Dbx(x, w)
Load_Dist = Load_Dist + (Span/3)
ENDDO
M1A(x, w) = 0.5*(M_A(x, w) + M_B(x, w))

M2A(x, w) = 0.5*(M_A(x, w) - M_B(x, w))
Q1A(x, w) = 0.5*(Q_A(x, w) - Q_B(x, w))
Q2A(x, w) = 0.5*(Q_A(x, w) + Q_B(x, w))



PoA(x, w) = P1o(x, w) + P2o(x, w)

MoA(x, w) = M1o(x, w) + M2o(x, w)
PoB(x, w) = P1o(x, w) - P2o(x, w)
MoB(x, w) = M1o(x, w) - M2o(x, w)
Load_Dist= Span/3
DO PP = 1,2
312
Appendix
Abx(x, w) = exp(-Beta(x, w)*abs(Load_Dist - Point_Dist))*(cos(Beta(x, w)*abs(Load_Dist - Point_Dist)) + sin(Beta(x,

w)*abs(Load_Dist - Point_Dist)))
Cbx(x, w) = exp(-Beta(x, w)*abs(Load_Dist - Point_Dist))*(cos(Beta(x, w)*abs(Load_Dist - Point_Dist))-sin(Beta(x,
Wdax (x, w) = Wdax (x, w) + ((P*Beta(x, w)/(2*k(x, w)))*Abx(x, w))

Wtax (x, w) = Wtax (x, w) + ((P*(Beta(x, w)**2)/k(x, w))*Bbx(x, w))
Wmax (x, w) = Wmax (x, w) + ((P/(4*Beta(x, w)))*Cbx(x, w))
Wsax (x, w) = Wsax (x, w) + ((P/2)*Dbx(x, w))
Abx(x, w) = exp(-Beta(x, w)*abs(Load_Dist - Point_Dist))*(cos(Beta(x, w)*abs(Load_Dist - Point_Dist))+sin(Beta(x,


Wtax (x, w) = Wtax (x, w) - ((P*(Beta(x, w)**2)/k(x, w))*Bbx(x, w))
Wsax (x, w) = Wsax (x, w) - ((P/2)*Dbx(x, w))
ENDIF
ENDDO


ENDDO
ELSEIF(Rc /= 1)THEN
DO x = 1,S_num
Wdax (x, w) = 0.0
Wtax (x, w) = 0.0
Wmax (x, w) = 0.0
Wsax (x, w) = 0.0
Wstax (x, w) = 0.0
ENDDO
ENDIF
DO x = 1, S_num
IF (((x-1) * Slice) <= (Span / 3))THEN
Max (x, w) = (((Load/LS_num)* w)/2) * ((x-1)*Slice)
Xval = Max (x, w)
ELSEIF (((x-1) * Slice) > (Span / 3) .and. ((x-1) * Slice) <= (Span / 2))THEN
Max (x, w) = (((Load/LS_num)* w)/2) * (Span / 3)
Vax (x, w) = 0.0
Xval = Max (x, w)
ENDIF
ENDDO
DO x = 1, S_num, 1
313
Appendix
ENDDO
DO x = 1, S_num
ENDDO
Xval_k = 0.0
Yval_k = 0.0
DO w = 2, LS_num, 1
WRITE (*,*) ' '
WRITE (99,*) ' '
IF(Rc == 1)THEN
DO x = 1,S_num
ko(x, w) = 0.011302204
k(x, w) = ko(x, w) * B
k(x, w) = ko(x, w) * B
ENDIF
M_A=0
Q_A=0
M_B=0
Q_B=0
Wdax (x, w)= 0

Wtax (x, w)= 0
Wmax (x, w)= 0
Wsax (x, w)= 0
Wstax (x, w) = 0
Load_Dist= Span/3
DO PP = 1,2


Q_A(x, w) = Q_A(x, w) + (P/2) * Dbx(x, w)


Q_B(x, w) = Q_B(x, w) - (P/2) * Dbx(x, w)
ENDDO
M1A(x, w) = 0.5*(M_A(x, w) + M_B(x, w))

M2A(x, w) = 0.5*(M_A(x, w) - M_B(x, w))
Q1A(x, w) = 0.5*(Q_A(x, w) - Q_B(x, w))
Q2A(x, w) = 0.5*(Q_A(x, w) + Q_B(x, w))



314
Appendix
PoA(x, w) = P1o(x, w) + P2o(x, w)

MoA(x, w) = M1o(x, w) + M2o(x, w)
PoB(x, w) = P1o(x, w) - P2o(x, w)
MoB(x, w) = M1o(x, w) - M2o(x, w)
Load_Dist= Span/3
DO PP = 1,2

Cbx(x, w) = exp(-Beta(x, w)*abs(Load_Dist - Point_Dist))*(cos(Beta(x, w)*abs(Load_Dist - Point_Dist)) - sin(Beta(x,



ENDIF
ENDDO
Abx(x, w) = exp(-Beta(x, w)*Point_Dist)*(cos(Beta(x, w)*Point_Dist) + sin(Beta(x, w)*Point_Dist))

Cbx(x, w) = exp(-Beta(x, w)*Point_Dist)*(cos(Beta(x, w)*Point_Dist) - sin(Beta(x, w)*Point_Dist))
Abx(x, w) = exp(-Beta(x, w)*(Span-Point_Dist))*(cos(Beta(x, w)*(Span-Point_Dist)) + sin(Beta(x, w)*(Span-Point_Dist)))

Cbx(x, w) = exp(-Beta(x, w)*(Span-Point_Dist))*(cos(Beta(x, w)*(Span-Point_Dist)) - sin(Beta(x, w)*(Span-Point_Dist)))
ENDDO
ELSEIF(Rc /= 1)THEN
DO x = 1,S_num
Wdax (x, w) = 0.0
Wtax (x, w) = 0.0
Wmax (x, w) = 0.0
Wsax (x, w) = 0.0
Wstax (x, w) = 0.0
ENDDO
ENDIF
DO x = 1, S_num
WRITE (*,*) ' '
WRITE (99,*) ' '

315
Appendix

Max (x, w) = (((Load/LS_num)* w)/2) * ((x-1)*Slice)
Xval = Max (x, w)
ELSEIF (((x-1) * Slice) > (Span / 3) .and. ((x-1) * Slice) <= (Span / 2))THEN
Max (x, w) = (((Load/LS_num)* w)/2) * (Span / 3)
Vax (x, w) = 0.0
Xval = Max (x, w)
ENDIF
ENDDO
WRITE (23,"(100e20.10)") ((Wdax (x, PP), PP=1,100), x=1,160)
ENDDO
20 WRITE (18,"(100e20.10)") ((Nhax (x, PP), PP=1,100), x=1,160)
WRITE (24,"(100e20.10)") ((Wtax (x, PP), PP=1,100), x=1,160)
WRITE (25,"(100e20.10)") ((Wmax (x, PP), PP=1,100), x=1,160)
WRITE (26,"(100e20.10)") ((Wsax (x, PP), PP=1,100), x=1,160)
WRITE (27,"(100e20.10)") ((Wstax (x, PP), PP=1,100), x=1,160)
w=1
WRITE (*,*) ' '
WRITE (99,*) ' '
IF(Rc == 1)THEN
DO x = 1, S_num
M_A=0
Q_A=0
M_B=0
Q_B=0
Wdax (x, w) = 0
Wtax (x, w)= 0
Wmax (x, w)= 0
Wsax (x, w)= 0
Wstax (x, w) = 0
Load_Dist= 7*Span/88
DO PP = 1,4
q = (((Load/LS_num)*w)/4)/(span/11)
Abx_a(x, w) = exp(-Beta(x, w)*Load_Dist)*(cos(Beta(x, w)*Load_Dist)+sin(Beta(x, w)*Load_Dist))
Bbx_a(x, w) = exp(-Beta(x, w)*Load_Dist)* sin(Beta(x, w)*Load_Dist)
Cbx_a(x, w) = exp(-Beta(x, w)*Load_Dist)*(cos(Beta(x, w)*Load_Dist)-sin(Beta(x, w)*Load_Dist))
Dbx_a(x, w) = exp(-Beta(x, w)*Load_Dist)* cos(Beta(x, w)*Load_Dist)
Abx_b(x, w) = exp(-Beta(x, w)*(Load_Dist+Span/11))*(cos(Beta(x, w)*(Load_Dist+Span/11))+sin(Beta(x,
w)*(Load_Dist+Span/11)))
Bbx_b(x, w) = exp(-Beta(x, w)*(Load_Dist+Span/11))* sin(Beta(x, w)*(Load_Dist+Span/11))
Cbx_b(x, w) = exp(-Beta(x, w)*(Load_Dist+Span/11))*(cos(Beta(x, w)*(Load_Dist+Span/11))-sin(Beta(x,
Dbx_b(x, w) = exp(-Beta(x, w)*(Load_Dist+Span/11))* cos(Beta(x, w)*(Load_Dist+Span/11))
M_A(x, w) = M_A(x, w) - (q/(4*Beta(x, w))**2) * (Bbx_a(x, w) - Bbx_b(x, w))
Q_A(x, w) = Q_A(x, w) + (q/(4*Beta(x, w))) * (Cbx_a(x, w) - Cbx_b(x, w))
Abx_a(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))*(cos(Beta(x, w)*(Span-Load_Dist))+sin(Beta(x, w)*(Span-Load_Dist)))
Bbx_a(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))* sin(Beta(x, w)*(Span-Load_Dist))
Cbx_a(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))*(cos(Beta(x, w)*(Span-Load_Dist))-sin(Beta(x, w)*(Span-Load_Dist)))
Dbx_a(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))* cos(Beta(x, w)*(Span-Load_Dist))
Abx_b(x, w) = exp(-Beta(x, w)*(Span-Load_Dist-Span/11))*(cos(Beta(x, w)*(Span-Load_Dist-Span/11))+sin(Beta(x, w)*(Span-
Load_Dist-Span/11)))
Bbx_b(x, w) = exp(-Beta(x, w)*(Span-Load_Dist-Span/11))* sin(Beta(x, w)*(Span-Load_Dist-Span/11))
Cbx_b(x, w) = exp(-Beta(x, w)*(Span-Load_Dist-Span/11))*(cos(Beta(x, w)*(Span-Load_Dist-Span/11))-sin(Beta(x, w)*(Span-
Dbx_b(x, w) = exp(-Beta(x, w)*(Span-Load_Dist-Span/11))* cos(Beta(x, w)*(Span-Load_Dist-Span/11))
M_A(x, w) = M_A(x, w) + (q/(4*Beta(x, w))**2) * (Bbx_a(x, w) - Bbx_b(x, w))
ENDDO
M1A(x, w) = 0.5*(M_A(x, w) + M_B(x, w))
M2A(x, w) = 0.5*(M_A(x, w) - M_B(x, w))
Q1A(x, w) = 0.5*(Q_A(x, w) - Q_B(x, w))
Q2A(x, w) = 0.5*(Q_A(x, w) + Q_B(x, w))
316
Appendix

PoA(x, w) = P1o(x, w) + P2o(x, w)
MoA(x, w) = M1o(x, w) + M2o(x, w)
PoB(x, w) = P1o(x, w) - P2o(x, w)
MoB(x, w) = M1o(x, w) - M2o(x, w)
DO PP = 1,4
IF(Point_Dist < Load_Dist) THEN
Abx_a(x, w) = exp(-Beta(x, w)*(Load_Dist-Point_Dist))*(cos(Beta(x, w)*(Load_Dist-Point_Dist))+sin(Beta(x, w)*(Load_Dist-
Point_Dist)))
Bbx_a(x, w) = exp(-Beta(x, w)*(Load_Dist-Point_Dist))* sin(Beta(x, w)*(Load_Dist-Point_Dist))
Cbx_a(x, w) = exp(-Beta(x, w)*(Load_Dist-Point_Dist))*(cos(Beta(x, w)*(Load_Dist-Point_Dist))-sin(Beta(x, w)*(Load_Dist-
Point_Dist)))
Dbx_a(x, w) = exp(-Beta(x, w)*(Load_Dist-Point_Dist))* cos(Beta(x, w)*(Load_Dist-Point_Dist))
Abx_b(x, w) = exp(-Beta(x, w)*(Span - Point_Dist))*(cos(Beta(x, w)*(Span - Point_Dist))+sin(Beta(x, w)*(Span - Point_Dist)))
Bbx_b(x, w) = exp(-Beta(x, w)*(Span - Point_Dist))* sin(Beta(x, w)*(Span - Point_Dist))
Cbx_b(x, w) = exp(-Beta(x, w)*(Span - Point_Dist))*(cos(Beta(x, w)*(Span - Point_Dist))-sin(Beta(x, w)*(Span - Point_Dist)))
Dbx_b(x, w) = exp(-Beta(x, w)*(Span - Point_Dist))* cos(Beta(x, w)*(Span - Point_Dist))
Wdax (x, w) = Wdax (x, w) + (q/(2*k(x, w)))*(Dbx_a(x, w) - Dbx_b(x, w))
Wtax (x, w) = Wtax (x, w) + ((q*Beta(x, w))/(2*k(x, w)))*(Abx_a(x, w) - Abx_b(x, w))
Wmax (x, w) = Wmax (x, w) - (q/(4*(Beta(x, w))**2))*(Bbx_a(x, w) - Bbx_b(x, w))
Wsax (x, w) = Wsax (x, w) + (q/(4*Beta(x, w)))*(Cbx_a(x, w) - Cbx_b(x, w))
ELSEIF(Point_Dist > Load_Dist .and. Point_Dist < (Load_Dist+Span/11)) THEN
Abx_a(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist))*(cos(Beta(x, w)*(Point_Dist-Load_Dist))+sin(Beta(x, w)*(Point_Dist-
Load_Dist)))
Bbx_a(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist))* sin(Beta(x, w)*(Point_Dist-Load_Dist))
Cbx_a(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist))*(cos(Beta(x, w)*(Point_Dist-Load_Dist))-sin(Beta(x, w)*(Point_Dist-
Load_Dist)))
Dbx_a(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist))* cos(Beta(x, w)*(Point_Dist-Load_Dist))
Abx_b(x, w) = exp(-Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))*(cos(Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))+sin(Beta(x,
w)*((Span/11)+Load_Dist-Point_Dist)))
Bbx_b(x, w) = exp(-Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))* sin(Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))
Cbx_b(x, w) = exp(-Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))*(cos(Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))-sin(Beta(x,
Dbx_b(x, w) = exp(-Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))* cos(Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))
ELSEIF(Point_Dist > (Load_Dist+Span/11)) THEN
Load_Dist)))
Load_Dist)))
Abx_b(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))*(cos(Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))+sin(Beta(x,
w)*(Point_Dist-Load_Dist-(Span/11))))
Bbx_b(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))* sin(Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))
Cbx_b(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))*(cos(Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))-sin(Beta(x,
Dbx_b(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))* cos(Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))
Wdax (x, w) = Wdax (x, w) - (q/(2*k(x, w)))*(Dbx_a(x, w) - Dbx_b(x, w))
Wmax (x, w) = Wmax (x, w) + (q/(4*(Beta(x, w))**2))*(Bbx_a(x, w) - Bbx_b(x, w))
ENDIF
ENDDO
ENDDO
317
Appendix
ELSEIF(Rc /= 1)THEN
DO x = 1,S_num
Wdax (x, w) = 0.0
Wtax (x, w) = 0.0
Wmax (x, w) = 0.0
Wsax (x, w) = 0.0
Wstax (x, w) = 0.0
ENDDO
ENDIF
DO x = 1, S_num
IF (((x-1) * Slice) <= (7*Span / 88))THEN
Max (x, w) = (2*q/11)* ((x-1)*Slice) * Span
Vax (x, w) = (2*q/11)* Span
Xval = Max (x, w)
ELSEIF (((x-1) * Slice) > (7*Span / 88) .and. ((x-1) * Slice) <= (15*Span / 88))THEN
Max (x, w) = q * ((-0.5*((x-1) * Slice)**2) + (23*Span/88)*((x-1) * Slice) - (49*Span**2/15488))
Vax (x, w) = q * ((23*Span/88) - ((x-1) * Slice))
Xval = Max (x, w)
Max (x, w) = q * ((((x-1) * Slice)*Span/11) + ((Span**2)/88))
Vax (x, w) = q * Span /11
Xval = Max (x, w)
Max (x, w) = q * ((-0.5*((x-1) * Slice)**2) + (37*Span/88)*((x-1) * Slice) - (665*(Span**2)/15488))
Xval = Max (x, w)
ELSEIF (((x-1) * Slice) > (37*Span / 88) .and. ((x-1) * Slice) <= (Span / 2))THEN
Max (x, w) = ((q /22)*(Span**2))
Vax (x, w) = 0.0
Xval = Max (x, w)
ENDIF
ENDDO
DO x = 1, S_num, 1
CALL Transformed (w,x,Bnahyax, Qahyax,Iahyax,Height,L_num,Lt,Ybar,Rc,&
X_M, Y_M_cur,Y_M_t,Y_M_c,Tsax,Csax,Cax,Yax,X_F_t,&
ENDDO
DO x = 1, S_num
ENDDO
Xval_k = 0.0
Yval_k = 0.0
DO w = 2, LS_num, 1
WRITE (*,*) ' '
WRITE (99,*) ' '
IF(Rc == 1)THEN
DO x = 1,S_num
ko(x, w) = 0.011302204
k(x, w) = ko(x, w) * B
k(x, w) = ko(x, w) * B
ENDIF
M_A=0
Q_A=0
M_B=0
Q_B=0
Wdax (x, w)= 0
Wtax (x, w)= 0
Wmax (x, w)= 0
Wsax (x, w)= 0
Wstax (x, w) = 0
318
Appendix
DO PP = 1,4
Abx_a(x, w) = exp(-Beta(x, w)*Load_Dist)*(cos(Beta(x, w)*Load_Dist)+sin(Beta(x, w)*Load_Dist))
Bbx_a(x, w) = exp(-Beta(x, w)*Load_Dist)* sin(Beta(x, w)*Load_Dist)
Cbx_a(x, w) = exp(-Beta(x, w)*Load_Dist)*(cos(Beta(x, w)*Load_Dist)-sin(Beta(x, w)*Load_Dist))
Dbx_a(x, w) = exp(-Beta(x, w)*Load_Dist)* cos(Beta(x, w)*Load_Dist)
Abx_b(x, w) = exp(-Beta(x, w)*(Load_Dist+Span/11))*(cos(Beta(x, w)*(Load_Dist+Span/11))+sin(Beta(x,
Bbx_b(x, w) = exp(-Beta(x, w)*(Load_Dist+Span/11))* sin(Beta(x, w)*(Load_Dist+Span/11))
Cbx_b(x, w) = exp(-Beta(x, w)*(Load_Dist+Span/11))*(cos(Beta(x, w)*(Load_Dist+Span/11))-sin(Beta(x,
Dbx_b(x, w) = exp(-Beta(x, w)*(Load_Dist+Span/11))* cos(Beta(x, w)*(Load_Dist+Span/11))
M_A(x, w) = M_A(x, w) - (q/(4*Beta(x, w))**2) * (Bbx_a(x, w) - Bbx_b(x, w))
Abx_a(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))*(cos(Beta(x, w)*(Span-Load_Dist))+sin(Beta(x, w)*(Span-Load_Dist)))
Bbx_a(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))* sin(Beta(x, w)*(Span-Load_Dist))
Cbx_a(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))*(cos(Beta(x, w)*(Span-Load_Dist))-sin(Beta(x, w)*(Span-Load_Dist)))
Dbx_a(x, w) = exp(-Beta(x, w)*(Span-Load_Dist))* cos(Beta(x, w)*(Span-Load_Dist))
Abx_b(x, w) = exp(-Beta(x, w)*(Span-Load_Dist-Span/11))*(cos(Beta(x, w)*(Span-Load_Dist-Span/11))+sin(Beta(x, w)*(Span-
Bbx_b(x, w) = exp(-Beta(x, w)*(Span-Load_Dist-Span/11))* sin(Beta(x, w)*(Span-Load_Dist-Span/11))
Cbx_b(x, w) = exp(-Beta(x, w)*(Span-Load_Dist-Span/11))*(cos(Beta(x, w)*(Span-Load_Dist-Span/11))-sin(Beta(x, w)*(Span-
Dbx_b(x, w) = exp(-Beta(x, w)*(Span-Load_Dist-Span/11))* cos(Beta(x, w)*(Span-Load_Dist-Span/11))
M_A(x, w) = M_A(x, w) + (q/(4*Beta(x, w))**2) * (Bbx_a(x, w) - Bbx_b(x, w))
ENDDO
M1A(x, w) = 0.5*(M_A(x, w) + M_B(x, w))
M2A(x, w) = 0.5*(M_A(x, w) - M_B(x, w))
Q1A(x, w) = 0.5*(Q_A(x, w) - Q_B(x, w))
Q2A(x, w) = 0.5*(Q_A(x, w) + Q_B(x, w))
PoA(x, w) = P1o(x, w) + P2o(x, w)
MoA(x, w) = M1o(x, w) + M2o(x, w)
PoB(x, w) = P1o(x, w) - P2o(x, w)
MoB(x, w) = M1o(x, w) - M2o(x, w)
DO PP = 1,4
IF(Point_Dist < Load_Dist) THEN
Abx_a(x, w) = exp(-Beta(x, w)*(Load_Dist-Point_Dist))*(cos(Beta(x, w)*(Load_Dist-Point_Dist))+sin(Beta(x, w)*(Load_Dist-
Point_Dist)))
Bbx_a(x, w) = exp(-Beta(x, w)*(Load_Dist-Point_Dist))* sin(Beta(x, w)*(Load_Dist-Point_Dist))
Cbx_a(x, w) = exp(-Beta(x, w)*(Load_Dist-Point_Dist))*(cos(Beta(x, w)*(Load_Dist-Point_Dist))-sin(Beta(x, w)*(Load_Dist-
Point_Dist)))
Dbx_a(x, w) = exp(-Beta(x, w)*(Load_Dist-Point_Dist))* cos(Beta(x, w)*(Load_Dist-Point_Dist))
ELSEIF(Point_Dist > Load_Dist .and. Point_Dist < (Load_Dist+Span/11)) THEN
Load_Dist)))
Load_Dist)))
Abx_b(x, w) = exp(-Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))*(cos(Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))+sin(Beta(x,
Bbx_b(x, w) = exp(-Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))* sin(Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))
Cbx_b(x, w) = exp(-Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))*(cos(Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))-sin(Beta(x,
Dbx_b(x, w) = exp(-Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))* cos(Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))
ELSEIF(Point_Dist > (Load_Dist+Span/11)) THEN
Load_Dist)))
319
Appendix

Load_Dist)))
Abx_b(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))*(cos(Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))+sin(Beta(x,
Bbx_b(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))* sin(Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))
Cbx_b(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))*(cos(Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))-sin(Beta(x,
Dbx_b(x, w) = exp(-Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))* cos(Beta(x, w)*(Point_Dist-Load_Dist-(Span/11)))
Wdax (x, w) = Wdax (x, w) - (q/(2*k(x, w)))*(Dbx_a(x, w) - Dbx_b(x, w))
Wmax (x, w) = Wmax (x, w) + (q/(4*(Beta(x, w))**2))*(Bbx_a(x, w) - Bbx_b(x, w))
ENDIF
ENDDO
ENDDO
ELSEIF(Rc /= 1)THEN
DO x = 1,S_num
Wdax (x, w) = 0.0
Wtax (x, w) = 0.0
Wmax (x, w) = 0.0
Wsax (x, w) = 0.0
Wstax (x, w) = 0.0
ENDDO
ENDIF
DO x = 1, S_num
WRITE (*,*) ' '
WRITE (99,*) ' '
IF (((x-1) * Slice) <= (7*Span / 88))THEN
Max (x, w) = (2*q/11)* ((x-1)*Slice) * Span
Vax (x, w) = (2*q/11)* Span
Xval = Max (x, w)
Max (x, w) = q * ((-0.5*((x-1) * Slice)**2) + (23*Span/88)*((x-1) * Slice) - (49*Span**2/15488))
Xval = Max (x, w)
Max (x, w) = q * ((((x-1) * Slice)*Span/11) + ((Span**2)/88))
Vax (x, w) = q * Span /11
Xval = Max (x, w)
Max (x, w) = q * ((-0.5*((x-1) * Slice)**2) + (37*Span/88)*((x-1) * Slice) - (665*(Span**2)/15488))
Xval = Max (x, w)
Max (x, w) = ((q /22)*(Span**2))
Vax (x, w) = 0.0
Xval = Max (x, w)
ENDIF
320
Appendix

ENDDO
ENDDO
30 WRITE (18,"(100e20.10)") ((Nhax (x, PP), PP=1,100), x=1,160)
w=1
WRITE (*,*) ' '
WRITE (99,*) ' '
IF(Rc == 1)THEN
DO x = 1, S_num
M_A=0
Q_A=0
M_B=0
Q_B=0
Wdax (x, w) = 0
Wtax (x, w)= 0
Wmax (x, w)= 0
Wsax (x, w)= 0
Wstax (x, w) = 0
Load_Dist= Span/16
DO PP = 1,8
Q_A(x, w) = Q_A(x, w) + (P/2) * Dbx(x, w)
Q_B(x, w) = Q_B(x, w) - (P/2) * Dbx(x, w)
ENDDO
M1A(x, w) = 0.5*(M_A(x, w) + M_B(x, w))
M2A(x, w) = 0.5*(M_A(x, w) - M_B(x, w))
Q1A(x, w) = 0.5*(Q_A(x, w) - Q_B(x, w))
Q2A(x, w) = 0.5*(Q_A(x, w) + Q_B(x, w))
PoA(x, w) = P1o(x, w) + P2o(x, w)
MoA(x, w) = M1o(x, w) + M2o(x, w)
PoB(x, w) = P1o(x, w) - P2o(x, w)
MoB(x, w) = M1o(x, w) - M2o(x, w)
Load_Dist= Span/16
DO PP = 1,8
321
Appendix
Abx(x, w) = exp(-Beta(x, w)*abs(Load_Dist - Point_Dist))*(cos(Beta(x, w)*abs(Load_Dist - Point_Dist))+sin(Beta(x,

ENDIF
ENDDO
ENDDO
ELSEIF(Rc /= 1)THEN
DO x = 1,S_num
Wdax (x, w) = 0.0
Wtax (x, w) = 0.0
Wmax (x, w) = 0.0
Wsax (x, w) = 0.0
Wstax (x, w) = 0.0
ENDDO
ENDIF
DO x = 1, S_num
Max (x, w) = (((Load/LS_num)* w) * ((x-1)*Slice)) / 2
Xval = Max (x, w)
ELSEIF (((x-1) * Slice) > (Span / 16) .and. ((x-1) * Slice) <= (3*Span / 16))THEN
Max (x, w) = (((Load/LS_num)* w)/64) * ((0.5*Span)+(24*(x-1)*Slice))
Vax (x, w) = (3*((Load/LS_num)* w)) / 8
Xval = Max (x, w)
Max (x, w) = (((Load/LS_num)* w)/64) * ((2*Span)+(16*(x-1)*Slice))
Xval = Max (x, w)
Xval = Max (x, w)
Max (x, w) = (((Load/LS_num)* w)/8) * Span
Vax (x, w) = 0.0
Xval = Max (x, w)
ENDIF
ENDDO
DO x = 1, S_num
ENDDO
DO x = 1, S_num
ENDDO
322
Appendix
Xval_k = 0.0
Yval_k = 0.0
DO w = 2, LS_num, 1
WRITE (*,*) ' '
WRITE (99,*) ' '
IF(Rc == 1)THEN
DO x = 1,S_num
ko(x, w) = 0.011302204
k(x, w) = ko(x, w) * B
k(x, w) = ko(x, w) * B
ENDIF
M_A=0
Q_A=0
M_B=0
Q_B=0
Wdax (x, w)= 0
Wtax (x, w)= 0
Wmax (x, w)= 0
Wsax (x, w)= 0
Wstax (x, w) = 0
Load_Dist= Span/16
DO PP = 1,8
Q_A(x, w) = Q_A(x, w) + (P/2) * Dbx(x, w)
Q_B(x, w) = Q_B(x, w) - (P/2) * Dbx(x, w)
ENDDO
M1A(x, w) = 0.5*(M_A(x, w) + M_B(x, w))
M2A(x, w) = 0.5*(M_A(x, w) - M_B(x, w))
Q1A(x, w) = 0.5*(Q_A(x, w) - Q_B(x, w))
Q2A(x, w) = 0.5*(Q_A(x, w) + Q_B(x, w))
PoA(x, w) = P1o(x, w) + P2o(x, w)
MoA(x, w) = M1o(x, w) + M2o(x, w)
PoB(x, w) = P1o(x, w) - P2o(x, w)
MoB(x, w) = M1o(x, w) - M2o(x, w)
Load_Dist= Span/16
DO PP = 1,8
323
Appendix

ENDIF
ENDDO
Abx(x, w) = exp(-Beta(x, w)*Point_Dist)*(cos(Beta(x, w)*Point_Dist) + sin(Beta(x, w)*Point_Dist))
Cbx(x, w) = exp(-Beta(x, w)*Point_Dist)*(cos(Beta(x, w)*Point_Dist) - sin(Beta(x, w)*Point_Dist))
Abx(x, w) = exp(-Beta(x, w)*(Span-Point_Dist))*(cos(Beta(x, w)*(Span-Point_Dist)) + sin(Beta(x, w)*(Span-Point_Dist)))
Cbx(x, w) = exp(-Beta(x, w)*(Span-Point_Dist))*(cos(Beta(x, w)*(Span-Point_Dist)) - sin(Beta(x, w)*(Span-Point_Dist)))
ENDDO
ELSEIF(Rc /= 1)THEN
DO x = 1,S_num
Wdax (x, w) = 0.0
Wtax (x, w) = 0.0
Wmax (x, w) = 0.0
Wsax (x, w) = 0.0
Wstax (x, w) = 0.0
ENDDO
ENDIF
DO x = 1, S_num
WRITE (*,*) ' '
WRITE (99,*) ' '

Max (x, w) = (((Load/LS_num)* w) * ((x-1)*Slice)) / 2
Xval = Max (x, w)
ELSEIF (((x-1) * Slice) > (Span / 16) .and. ((x-1) * Slice) <= (3*Span / 16))THEN
Vax (x, w) = (3*((Load/LS_num)* w)) / 8
Xval = Max (x, w)
Max (x, w) = (((Load/LS_num)* w)/64) * ((2*Span)+(16*(x-1)*Slice))
Xval = Max (x, w)
Xval = Max (x, w)
Max (x, w) = (((Load/LS_num)* w)/8) * Span
Vax (x, w) = 0.0
Xval = Max (x, w)
ENDIF


ENDDO
324
Appendix

ENDDO
40 WRITE (23,"(100e20.10)") ((Wdax (x, PP), PP=1,100), x=1,160)
ELSEIF (Loads_num ==5)THEN
w=1
WRITE (*,*) ' '
WRITE (99,*) ' '
IF(Rc == 1)THEN
P = ((Load/LS_num)* w)/Span
DO x = 1,S_num
M_A(x, w)=0
Q_A(x, w)=0
M_B(x, w)=0
Q_B(x, w)=0
Wdax (x, w) = 0
Wtax (x, w)= 0
Wmax (x, w)= 0
Wsax (x, w)= 0
Wstax (x, w)= 0
Abx_a(x, w) = exp(-Beta(x, w)*0.0)*(cos(Beta(x, w)*0.0)+sin(Beta(x, w)*0.0))
Bbx_a(x, w) = exp(-Beta(x, w)*0.0)* sin(Beta(x, w)*0.0)
Cbx_a(x, w) = exp(-Beta(x, w)*0.0)*(cos(Beta(x, w)*0.0)-sin(Beta(x, w)*0.0))
Dbx_a(x, w) = exp(-Beta(x, w)*0.0)* cos(Beta(x, w)*0.0)
Abx_b(x, w) = exp(-Beta(x, w)*Span)*(cos(Beta(x, w)*Span)+sin(Beta(x, w)*Span))
Bbx_b(x, w) = exp(-Beta(x, w)*Span)* sin(Beta(x, w)*Span)
Cbx_b(x, w) = exp(-Beta(x, w)*Span)*(cos(Beta(x, w)*Span)-sin(Beta(x, w)*Span))
Dbx_b(x, w) = exp(-Beta(x, w)*Span)* cos(Beta(x, w)*Span)
M_A(x, w) = (P/(4*(Beta(x, w))**2))*(Bbx_a(x, w) + Bbx_b(x, w))
Q_A(x, w) = (P/(4*Beta(x, w)))*(Cbx_a(x, w) - Cbx_b(x, w))
M_B(x, w) = M_A(x, w)
Q_B(x, w) = - Q_A(x, w)
M1A(x, w) = 0.5*(M_A(x, w) + M_B(x, w))
M2A(x, w) = 0.5*(M_A(x, w) - M_B(x, w))
Q1A(x, w) = 0.5*(Q_A(x, w) - Q_B(x, w))
Q2A(x, w) = 0.5*(Q_A(x, w) + Q_B(x, w))
PoA(x, w) = P1o(x, w) + P2o(x, w)
MoA(x, w) = M1o(x, w) + M2o(x, w)
PoB(x, w) = P1o(x, w) - P2o(x, w)
MoB(x, w) = M1o(x, w) - M2o(x, w)
Abx_a(x, w) = exp(-Beta(x, w)*Point_Dist)*(cos(Beta(x, w)*Point_Dist)+sin(Beta(x, w)*Point_Dist))
Bbx_a(x, w) = exp(-Beta(x, w)*Point_Dist)* sin(Beta(x, w)*Point_Dist)
Cbx_a(x, w) = exp(-Beta(x, w)*Point_Dist)*(cos(Beta(x, w)*Point_Dist)-sin(Beta(x, w)*Point_Dist))
Dbx_a(x, w) = exp(-Beta(x, w)*Point_Dist)* cos(Beta(x, w)*Point_Dist)
Wdax (x, w) = (P/(2*k(x, w)))*(2 - Dbx_a(x, w) - Dbx_b(x, w))
Wtax (x, w) = ((P*Beta(x, w))/(2*k(x, w)))*(Abx_a(x, w) - Abx_b(x, w))
Wmax (x, w) = (P/(4*(Beta(x, w))**2))*(Bbx_a(x, w) + Bbx_b(x, w))
Wsax (x, w) = (P/(4*Beta(x, w)))*(Cbx_a(x, w) - Cbx_b(x, w))
325
Appendix
ENDDO
ELSEIF(Rc /= 1)THEN
DO x = 1,S_num
Wdax (x, w) = 0.0
Wtax (x, w) = 0.0
Wmax (x, w) = 0.0
Wsax (x, w) = 0.0
Wstax (x, w) = 0.0
ENDDO
ENDIF
DO x = 1, S_num
Max (x, w) = (0.5*(Load/LS_num)* w/Span) * (Span*((x-1)*Slice) - ((x-1)*Slice)**2)
Vax (x, w) = ((Load/LS_num)* w/Span) *(0.5*Span - ((x-1)*Slice))
Xval = Max (x, w)
ENDDO
DO x = 1, S_num, 1
ENDDO
DO x = 1, S_num
ENDDO
Xval_k = 0.0
Yval_k = 0.0
DO w = 2, LS_num, 1
WRITE (*,*) ' '
WRITE (99,*) ' '
IF(Rc == 1)THEN
P = ((Load/LS_num)* w)/Span
DO x = 1,S_num
ko(x, w) = 0.011302204
k(x, w) = ko(x, w) * B
k(x, w) = ko(x, w) * B
ENDIF
M_A(x, w)=0
Q_A(x, w)=0
M_B(x, w)=0
Q_B(x, w)=0
Wdax (x, w) = 0
Wtax (x, w)= 0
Wmax (x, w)= 0
Wsax (x, w)= 0
Wstax (x, w)= 0
Abx_a(x, w) = exp(-Beta(x, w)*0.0)*(cos(Beta(x, w)*0.0)+sin(Beta(x, w)*0.0))
Bbx_a(x, w) = exp(-Beta(x, w)*0.0)* sin(Beta(x, w)*0.0)
Cbx_a(x, w) = exp(-Beta(x, w)*0.0)*(cos(Beta(x, w)*0.0)-sin(Beta(x, w)*0.0))
Dbx_a(x, w) = exp(-Beta(x, w)*0.0)* cos(Beta(x, w)*0.0)
Abx_b(x, w) = exp(-Beta(x, w)*Span)*(cos(Beta(x, w)*Span)+sin(Beta(x, w)*Span))
Bbx_b(x, w) = exp(-Beta(x, w)*Span)* sin(Beta(x, w)*Span)
Cbx_b(x, w) = exp(-Beta(x, w)*Span)*(cos(Beta(x, w)*Span)-sin(Beta(x, w)*Span))
Dbx_b(x, w) = exp(-Beta(x, w)*Span)* cos(Beta(x, w)*Span)
M_A(x, w) = (P/(4*(Beta(x, w))**2))*(Bbx_a(x, w) + Bbx_b(x, w))
Q_A(x, w) = (P/(4*Beta(x, w)))*(Cbx_a(x, w) - Cbx_b(x, w))
M_B(x, w) = M_A(x, w)
Q_B(x, w) = - Q_A(x, w)
M1A(x, w) = 0.5*(M_A(x, w) + M_B(x, w))
M2A(x, w) = 0.5*(M_A(x, w) - M_B(x, w))
Q1A(x, w) = 0.5*(Q_A(x, w) - Q_B(x, w))
Q2A(x, w) = 0.5*(Q_A(x, w) + Q_B(x, w))
326
Appendix

PoA(x, w) = P1o(x, w) + P2o(x, w)
MoA(x, w) = M1o(x, w) + M2o(x, w)
PoB(x, w) = P1o(x, w) - P2o(x, w)
MoB(x, w) = M1o(x, w) - M2o(x, w)
Abx_a(x, w) = exp(-Beta(x, w)*Point_Dist)*(cos(Beta(x, w)*Point_Dist)+sin(Beta(x, w)*Point_Dist))
Bbx_a(x, w) = exp(-Beta(x, w)*Point_Dist)* sin(Beta(x, w)*Point_Dist)
Cbx_a(x, w) = exp(-Beta(x, w)*Point_Dist)*(cos(Beta(x, w)*Point_Dist)-sin(Beta(x, w)*Point_Dist))
Dbx_a(x, w) = exp(-Beta(x, w)*Point_Dist)* cos(Beta(x, w)*Point_Dist)
Wdax (x, w) = (P/(2*k(x, w)))*(2 - Dbx_a(x, w) - Dbx_b(x, w))
Wtax (x, w) = ((P*Beta(x, w))/(2*k(x, w)))*(Abx_a(x, w) - Abx_b(x, w))
Wmax (x, w) = (P/(4*(Beta(x, w))**2))*(Bbx_a(x, w) + Bbx_b(x, w))
Wsax (x, w) = (P/(4*Beta(x, w)))*(Cbx_a(x, w) - Cbx_b(x, w))
ENDDO
ELSEIF(Rc /= 1)THEN
DO x = 1,S_num
Wdax (x, w) = 0.0
Wtax (x, w) = 0.0
Wmax (x, w) = 0.0
Wsax (x, w) = 0.0
Wstax (x, w) = 0.0
ENDDO
ENDIF
DO x = 1, S_num
WRITE (*,*) ' '
WRITE (99,*) ' '
Max (x, w) = (0.5*(Load/LS_num)* w/Span) * (Span*((x-1)*Slice) - ((x-1)*Slice)**2)
Vax (x, w) = ((Load/LS_num)* w/Span) *(0.5*Span - ((x-1)*Slice))
Xval = Max (x, w)
ENDDO
ENDDO
50 WRITE (23,"(100e20.10)") ((Wdax (x, w), w=1,LS_num), x=1,S_num)
WRITE (24,"(100e20.10)") ((Wtax (x, w), w=1,LS_num), x=1,S_num)
WRITE (25,"(100e20.10)") ((Wmax (x, w), w=1,LS_num), x=1,S_num)
WRITE (26,"(100e20.10)") ((Wsax (x, w), w=1,LS_num), x=1,S_num)
WRITE (27,"(100e20.10)") ((Wstax (x, w), w=1,LS_num), x=1,S_num)
ENDIF
CONTAINS
SUBROUTINE GetValue (N_num,X_cord,Y_cord,Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp, Xval, Yval, Ypval, Yppval)
327
Appendix
IMPLICIT NONE
REAL ( KIND = 8 ), INTENT(INOUT) :: N_num, X_cord(N_num), Y_cord(N_num), Ybcbeg, Ybcend, Ypp(N_num), Xval, Yval, Ypval, Yppval
INTEGER ( KIND = 4 ), INTENT(INOUT) :: Ibcbeg, Ibcend
CALL Spline1 (N_num,X_cord,Y_cord,Ibcbeg, Ybcbeg, Ibcend, Ybcend)

CALL Spline2 (N_num, X_cord, Y_cord, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp)
CALL Spline3 (N_num, X_cord, Y_cord, Ypp, Xval, Yval, Ypval, Yppval)
ENDSUBROUTINE GetValue
SUBROUTINE Spline1 (N_num,X_cord,Y_cord,Ibcbeg, Ybcbeg, Ibcend, Ybcend)

IMPLICIT NONE
REAL ( KIND = 8 ), INTENT(INOUT) :: N_num, X_cord(N_num), Y_cord(N_num), Ybcbeg, Ybcend
Ibcbeg = 1
Ybcbeg = (Y_cord(2)- Y_cord(1))/(X_cord (2)- X_cord(1))
Ibcend = 1
Ybcend = (Y_cord(N_num)- Y_cord(N_num-1))/(X_cord(N_num)- X_cord(N_num-1))
ENDSUBROUTINE Spline1
SUBROUTINE Spline2 (N_num, X_cord, Y_cord, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp)
IMPLICIT NONE
REAL ( KIND = 8 ) :: a(3,N_num)
REAL ( KIND = 8 ), INTENT(INOUT) :: N_num, X_cord(N_num), Y_cord(N_num), Ybcbeg, Ybcend, Ypp(N_num)
INTEGER ( KIND = 4 ) :: II
IF ( N_num <= 1 ) THEN

WRITE ( *, '(a)' ) ' '
WRITE ( *, '(a)' ) 'Spline2 - Fatal error!'
WRITE ( *, '(a)' ) ' The number of knots must be at least 2.'
WRITE ( *, '(a,i8)' ) ' The input value of N_num = ', N_num
STOP
ENDIF
DO II = 1, N_num-1
IF ( X_cord(II+1) <= X_cord(II) ) THEN
WRITE ( *, '(a)' ) ' '
WRITE ( *, '(a)' ) ' The knots must be strictly increasing, but'
WRITE ( *, '(a,i8,a,g14.6)' ) ' X_cord(', II,') = ', X_cord(II)
WRITE ( *, '(a,i8,a,g14.6)' ) ' X_cord(',II+1,') = ', X_cord(II+1)
STOP
ENDIF
ENDDO
IF ( Ibcbeg == 0 ) THEN
Ypp(1) = 0.0D+00
a(2,1) = 1.0D+00
a(1,2) = -1.0D+00
ELSEIF ( Ibcbeg == 1 ) THEN
Ypp(1) = ( Y_cord(2) - Y_cord(1) ) / ( X_cord(2) - X_cord(1) ) - Ybcbeg
a(2,1) = ( X_cord(2) - X_cord(1) ) / 3.0D+00
a(1,2) = ( X_cord(2) - X_cord(1) ) / 6.0D+00
ELSEIF ( Ibcbeg == 2 ) THEN
Ypp(1) = Ybcbeg
a(2,1) = 1.0D+00
a(1,2) = 0.0D+00
ELSE
WRITE ( *, '(a)' ) ' '
WRITE ( *, '(a)' ) ' The boundary flag IBCBEG must be 0, 1 or 2.'
WRITE ( *, '(a,i8)' ) ' The input value is IBCBEG = ', Ibcbeg
STOP
ENDIF
DO II = 2, N_num-1
Ypp(II) = ( Y_cord(II+1) - Y_cord(II) ) / ( X_cord(II+1) - X_cord(II) ) - ( Y_cord(II) - Y_cord(II-1) ) / ( X_cord(II) - X_cord(II-1) )
a(3,II-1) = ( X_cord(II) - X_cord(II-1) ) / 6.0D+00
a(2,II) = ( X_cord(II+1) - X_cord(II-1) ) / 3.0D+00
a(1,II+1) = ( X_cord(II+1) - X_cord(II) ) / 6.0D+00
ENDDO
IF ( Ibcend == 0 ) THEN
Ypp(N_num) = 0.0D+00
a(3,N_num-1) = -1.0D+00
a(2,N_num) = 1.0D+00
ELSEIF ( Ibcend == 1 ) THEN
Ypp(N_num) = Ybcend - ( Y_cord(N_num) - Y_cord(N_num-1) ) / ( X_cord(N_num) - X_cord(N_num-1) )
a(3,N_num-1) = ( X_cord(N_num) - X_cord(N_num-1) ) / 6.0D+00
a(2,N_num) = ( X_cord(N_num) - X_cord(N_num-1) ) / 3.0D+00
ELSEIF ( Ibcend == 2 ) THEN
Ypp(N_num) = Ybcend
a(3,N_num-1) = 0.0D+00
a(2,N_num) = 1.0D+00
ELSE
WRITE ( *, '(a)' ) ' '
WRITE ( *, '(a)' ) ' The boundary flag IBCEND must be 0, 1 or 2.'
WRITE ( *, '(a,i8)' ) ' The input value is IBCEND = ', Ibcend
STOP
ENDIF
IF ( N_num == 2 .and. Ibcbeg == 0 .and. Ibcend == 0 ) THEN
Ypp(1) = 0.0D+00
Ypp(2) = 0.0D+00
328
Appendix
ELSE
CALL Spline4 ( N_num, a, Ypp, Ypp )
ENDIF
SUBROUTINE Spline3 (N_num, X_cord, Y_cord, Ypp, Xval, Yval, Ypval, Yppval)
IMPLICIT NONE
REAL ( KIND = 8 ) :: dt, HH
REAL ( KIND = 8 ) ::
REAL ( KIND = 8 ), INTENT(INOUT) :: N_num, X_cord(N_num), Xval, Y_cord(N_num), Ypp(N_num), Yppval, Ypval, Yval
INTEGER ( KIND = 4 ) :: Left, Right
CALL Spline5 ( N_num, X_cord, Xval, Left, Right )

dt = Xval - X_cord(Left)
HH = X_cord(Right) - X_cord(Left)
Yval = Y_cord(Left) + dt * ( ( Y_cord(Right) - Y_cord(Left) ) / HH - ( Ypp(right) / 6.0D+00 + Ypp(left) / 3.0D+00 ) * HH + dt * ( 0.5D+00 * Ypp(Left) &
+ dt * ( ( Ypp(right) - Ypp(Left) ) / ( 6.0D+00 * HH ) ) ) )
Ypval = ( Y_cord(Right) - Y_cord(Left) ) / HH - ( Ypp(Right) / 6.0D+00 + Ypp(Left) / 3.0D+00 ) * HH + dt * ( Ypp(Left) &
+ dt * ( 0.5D+00 * ( Ypp(Right) - Ypp(Left) ) / HH ) )
Yppval = Ypp(Left) + dt * ( Ypp(Right) - Ypp(Left) ) / HH
SUBROUTINE Spline4 (N_num, a, BB, XX)

IMPLICIT NONE
REAL ( KIND = 8 ), INTENT(INOUT) :: N_num, a(3,N_num), BB(N_num), XX(N_num)
REAL ( KIND = 8 ) :: Xmult
DO II = 1, N_num
IF ( a(2,II) == 0.0D+00 ) THEN
WRITE ( *, '(a)' ) ' '
WRITE ( *, '(a,i8,a)' ) ' A(2,', II, ') = 0.'
RETURN
ENDIF
ENDDO
XX(1:N_num) = BB(1:N_num)
DO II = 2, N_num
xmult = a(3,II-1) / a(2,II-1)
a(2,II) = a(2,II) - xmult * a(1,II)
XX(II) = XX(II) - xmult * XX(II-1)
ENDDO
XX(N_num) = XX(N_num) / a(2,N_num)
DO II = N_num-1, 1, -1
XX(II) = ( XX(II) - a(1,II+1) * XX(II+1) ) / a(2,II)
ENDDO
SUBROUTINE Spline5 (N_num, XX, Xvalue, Left, Right)

IMPLICIT NONE
REAL ( KIND = 8 ) :: Xvalue
REAL ( KIND = 8 ), INTENT(INOUT) :: N_num, XX(N_num)
INTEGER ( KIND = 4 ), INTENT(INOUT) :: Left, Right
DO II = 2, N_num - 1
IF ( Xvalue < XX(II) ) THEN
Left = II - 1
Right = II
RETURN
ENDIF
ENDDO
Left = N_num - 1
Right = N_num
SUBROUTINE Non_Geometry (Height,L_num,Lt,Ybar,Rc,X_F_t,Y_F_t,X_F_c,Y_F_c,X_GFRP,Y_GFRP,X_GFRP_C,Y_GFRP_C)

IMPLICIT NONE
REAL (KIND = 8):: T_s, C_s , Sahy, D1, D2, Moment, Curvature, B, St, Wt, Wte, ASt, Bf, Bfe, Error, Fs, Foam_stress, Skin_stress, Eo, F_T_s_s, F_C_s_s, Span, Load
REAL (KIND = 8), INTENT(INOUT):: Ybar, Height , Lt
REAL (KIND = 8), DIMENSION (7), INTENT(INOUT):: X_F_t , Y_F_t, X_F_c, Y_F_c
REAL (KIND = 8), DIMENSION (16), INTENT(INOUT):: X_GFRP, Y_GFRP, X_GFRP_C, Y_GFRP_C
REAL (KIND = 8), INTENT(INOUT):: L_num, Rc
REAL (KIND = 4):: I
B = 300
Span = 1400
St = 1.5
ASt = 0.0
Bf = 0.0
Bfe = 0.0
Wt = 0.0
Wte = 0.0
F_T_s_s = 0.014
F_C_s_s = 0.004
Error = 0.00001
IF(Rc == 1)THEN
ELSEIF(Rc /= 1)THEN
Lt = (Height-(2*St + 2*ASt)) / L_num
ENDIF
C_s = 0.0
329
Appendix
T_s = 0.0
Fs = 0.0
Moment = 0.0
NG = 16
Nft = 7
Nfc = 7
CLOSE (9)
CLOSE (10)
CLOSE (11)
CLOSE (12)
CLOSE (229)
CLOSE (2210)
CLOSE (2211)
CLOSE (2212)
OPEN(9,FILE='Moment.TXT', STATUS='replace')
OPEN(10,FILE='Moment-Ten-Strain.TXT', STATUS='replace')
OPEN(11,FILE='Moment-Comp-strain.TXT', STATUS='replace')
OPEN(12,FILE='Moment-Curv.TXT', STATUS='replace')
DO WHILE (Abs(T_s) <= F_T_s_s .and. Abs(C_s) <= F_C_s_s)

D1 = 0
D2 = Height
Ybar = (D1 + D2) / 2
DO WHILE ((D2 - D1) > Error)
Fs = 0
Moment = 0
T_s = - ((Ybar * C_s) / (Height - Ybar))
Foam_stress = 0
Sahy = T_s * ((Ybar-(0.5*St))/Ybar)
Xval = Sahy
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Skin_stress = Yval
Fs = Fs + ((B * St)*(Skin_stress + Foam_stress))
Moment = Moment + (Abs (0.5*St - Ybar) * Abs ((B * St)*(Skin_stress + Foam_stress)))
Sahy = T_s * ((Ybar-(St + 0.5*ASt))/Ybar)
Xval = Sahy
Skin_stress = Yval
Xval = Sahy
CALL GetValue (Nft, X_F_t, Y_F_t, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_t, Xval, Yval, Ypval, Yppval)
Foam_stress = Yval
Fs = Fs + ((Bf+2*Bfe) * ASt * Skin_stress) + ((B-(Bf+2*Bfe)) * ASt * Foam_stress)
Moment = Moment + (Abs ((St + 0.5*ASt) - Ybar) * Abs (((Bf+2*Bfe) * ASt * Skin_stress) + ((B-(Bf+2*Bfe)) * ASt *
Foam_stress)))
Foam_stress = 0
Sahy = C_s * ((L_num*Lt+1.5*St+2*ASt- Ybar))/(Height-Ybar)
Xval = -Sahy
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Skin_stress = -Yval
Moment = Moment + (Abs (L_num*Lt+1.5*St+2*ASt - Ybar) * Abs ((B * St)*(Skin_stress + Foam_stress)))
Sahy = C_s * ((L_num*Lt + St + 1.5*ASt- Ybar))/(Height-Ybar)
Xval = -Sahy
Skin_stress = -Yval
Xval = -Sahy
CALL GetValue (Nfc, X_F_c, Y_F_c, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_c, Xval, Yval, Ypval, Yppval)
Foam_stress = -Yval
Fs = Fs + ((Bf+2*Bfe) * ASt * Skin_stress) + ((B-(Bf+2*Bfe)) * ASt * Foam_stress)
Moment = Moment + (Abs (L_num*Lt + St + 1.5*ASt - Ybar) * Abs (((Bf+2*Bfe) * ASt * Skin_stress) + ((B-(Bf+2*Bfe)) * ASt *
Foam_stress)))
DO I = 3, L_num+2
IF (((I-2)-0.5)*Lt+St+ASt > St .and. ((I-2)-0.5)*Lt+St+ASt <= Ybar) THEN
Sahy = T_s * ((Ybar-(((I-2)-0.5)*Lt+St+ASt))/Ybar)
Xval = Sahy
Foam_stress = Yval
Xval = Sahy
Skin_stress = Yval
ELSEIF (((I-2)-0.5)*Lt+St+ASt> Ybar .and. ((I-2)-0.5)*Lt+St+ASt <= (Height - St)) THEN
Sahy = C_s * ((((I-2)-0.5)*Lt+St+ASt- Ybar))/(Height-Ybar)
Xval = -Sahy
Foam_stress = -Yval
Xval = -Sahy
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval,
Yppval)
Skin_stress = -Yval
ENDIF
Fs = Fs + ((Wt+2*Wte) * Lt * Skin_stress) + ((B-(Wt+2*Wte)) * Lt * Foam_stress)
Moment = Moment + (Abs (((I-2)-0.5)*Lt+St+ASt - Ybar) * Abs (((Wt+2*Wte) * Lt * Skin_stress) + ((B-
(Wt+2*Wte)) * Lt * Foam_stress)))
ENDDO
Curvature = T_s / Ybar
IF (Fs < 0) THEN
D1 = Ybar
Ybar = (D1 + D2) / 2
ELSEIF (Fs > 0) THEN
D2 = Ybar
Ybar = (D1 + D2) / 2
330
Appendix
ELSEIF (Fs == 0) THEN

EXIT
ENDIF
ENDDO
Cal_Load = (8*Moment) / (Span**2)
Shear = Cal_Load*Span /2
WRITE (9,"(f25.13)") Moment
WRITE (10,"(f25.13)") T_s
WRITE (11,"(f25.13)") C_s
WRITE (12,"(f25.13)") Curvature
C_s = C_s - 0.00001
ENDDO
CLOSE (9)
CLOSE (10)
CLOSE (11)
CLOSE (12)
IF(Rc ==3.or.Rc ==4.or.Rc ==5.or.Rc ==6)THEN
C_s = 0.0
T_s = 0.0
OPEN(229,FILE='Moment1.TXT', STATUS='replace')
OPEN(2210,FILE='Moment-Ten-Strain1.TXT', STATUS='replace')
OPEN(2211,FILE='Moment-Comp-strain1.TXT', STATUS='replace')
OPEN(2212,FILE='Moment-Curv1.TXT', STATUS='replace')
DO WHILE (Abs(T_s) <= F_T_s_s .and. Abs(C_s) <= F_C_s_s)

D1 = 0
D2 = Height
Ybar = (D1 + D2) / 2
DO WHILE ((D2 - D1) > Error)
Fs = 0
Moment = 0
T_s = - ((Ybar * C_s) / (Height - Ybar))
Foam_stress = 0
Sahy = T_s * ((Ybar-(0.5*St))/Ybar)
Xval = Sahy
Skin_stress = Yval
Moment = Moment + (Abs (0.5*St - Ybar) * Abs ((B * St)*(Skin_stress + Foam_stress)))
Sahy = T_s * ((Ybar-(St + 0.5*ASt))/Ybar)
Xval = Sahy
Skin_stress = Yval
Xval = Sahy
Foam_stress = Yval
Fs = Fs + ((B) * ASt * Skin_stress) + ((B-(B)) * ASt * Foam_stress)
Moment = Moment + (Abs ((St + 0.5*ASt) - Ybar) * Abs (((B) * ASt * Skin_stress) + ((B-(B)) * ASt * Foam_stress)))
Foam_stress = 0
Sahy = C_s * ((L_num*Lt+1.5*St+2*ASt- Ybar))/(Height-Ybar)
Xval = -Sahy
Skin_stress = -Yval
Moment = Moment + (Abs (L_num*Lt+1.5*St+2*ASt - Ybar) * Abs ((B * St)*(Skin_stress + Foam_stress)))
Sahy = C_s * ((L_num*Lt + St + 1.5*ASt- Ybar))/(Height-Ybar)
Xval = -Sahy
Skin_stress = -Yval
Xval = -Sahy
Foam_stress = -Yval
Fs = Fs + ((B) * ASt * Skin_stress) + ((B-(B)) * ASt * Foam_stress)
Moment = Moment + (Abs (L_num*Lt + St + 1.5*ASt - Ybar) * Abs (((B) * ASt * Skin_stress) + ((B-(B)) * ASt * Foam_stress)))
DO I = 3, L_num+2
IF (((I-2)-0.5)*Lt+St+ASt > St .and. ((I-2)-0.5)*Lt+St+ASt <= Ybar) THEN
Sahy = T_s * ((Ybar-(((I-2)-0.5)*Lt+St+ASt))/Ybar)
Xval = Sahy
Foam_stress = Yval
Xval = Sahy
Skin_stress = Yval
ELSEIF (((I-2)-0.5)*Lt+St+ASt> Ybar .and. ((I-2)-0.5)*Lt+St+ASt <= (Height - St)) THEN
Sahy = C_s * ((((I-2)-0.5)*Lt+St+ASt- Ybar))/(Height-Ybar)
Xval = -Sahy
Foam_stress = -Yval
Xval = -Sahy
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval,
Yppval)
Skin_stress = -Yval
ENDIF
Fs = Fs + ((Wt+2*Wte) * Lt * Skin_stress) + ((B-(Wt+2*Wte)) * Lt * Foam_stress)
Moment = Moment + (Abs (((I-2)-0.5)*Lt+St+ASt - Ybar) * Abs (((Wt+2*Wte) * Lt * Skin_stress) + ((B-
(Wt+2*Wte)) * Lt * Foam_stress)))
ENDDO
Curvature = T_s / Ybar
IF (Fs < 0) THEN
D1 = Ybar
Ybar = (D1 + D2) / 2
ELSEIF (Fs > 0) THEN
D2 = Ybar
331
Appendix
Ybar = (D1 + D2) / 2

ELSEIF (Fs == 0) THEN
EXIT
ENDIF
ENDDO
Cal_Load = (8*Moment) / (Span**2)
Shear = Cal_Load*Span /2
WRITE (229,"(f25.13)") Moment
WRITE (2210,"(f25.13)") T_s
WRITE (2211,"(f25.13)") C_s
WRITE (2212,"(f25.13)") Curvature
C_s = C_s - 0.00001
ENDDO
CLOSE (229)
CLOSE (2210)
CLOSE (2211)
CLOSE (2212)
ENDIF
ENDSUBROUTINE Non_Geometry
SUBROUTINE Ten_Comp_Curv (Load,w,x,W_s,Prf,Span,Bf,Bfe,Max, Vax, Xval,Height,L_num,Lt,Slice,Ybar,Rc,X_M, Y_M_cur,Y_M_t,Y_M_c,Tsax,Csax,Cax,Yax,Dsax )

IMPLICIT NONE
REAL (KIND = 8):: T_s, C_s, Sahy, D1, D2, Moment, Curvature, B, St , Wt, Wte, ASt, Bf, Bfe, Error , Fs, Foam_stress, Skin_stress, Eo, F_T_s_s, F_C_s_s
REAL (KIND = 8), INTENT(INOUT):: Ybar, Height , Slice, Lt, Xval , Load, w ,x, L_num , Rc, W_s , Prf
REAL (KIND = 8), DIMENSION (401), INTENT(INOUT):: X_M, Y_M_cur , Y_M_t, Y_M_c, Ypp_M_cur, Ypp_M_t, Ypp_M_c
REAL (KIND = 8), DIMENSION (160,100), INTENT(INOUT):: Max, Vax, Csax, Tsax , Cax, Yax, Dsax
INTEGER ( KIND = 4 ):: ibcbeg, ibcend
REAL ( KIND = 8 ):: ybcbeg, ybcend, Yval, yppval, ypval, Span, Mcp, A_s_s, A_f_s, Sm, Fm
REAL (KIND = 8), DIMENSION (7):: Ypp_F_t , Ypp_F_c
REAL (KIND = 8), DIMENSION (37):: Ypp_F_s
REAL (KIND = 8), DIMENSION (16):: Ypp_GFRP
REAL (KIND = 4):: I
Mcp = 401
Bfe = 0.0
Bf = 0.0
IF((x-1)*slice <= Bfe)THEN
IF(Rc == 1 .or.Rc == 2.or.Rc == 3.or.Rc == 4)THEN
READ (9,*) X_M
READ (10,*) Y_M_t
READ (11,*) Y_M_c
READ (12,*) Y_M_cur
ELSEIF (Rc == 5.or.Rc == 6)THEN
READ (229,*) X_M
READ (2210,*) Y_M_t
READ (2211,*) Y_M_c
READ (2212,*) Y_M_cur
ENDIF
ELSEIF(Bfe < (x-1)*Slice .and. (x-1)*Slice < (0.5*Span) - 0.5*Bf)THEN
IF(Rc == 1.or.Rc == 2.or.Rc == 3.or.Rc == 5.or.Rc == 6)THEN
READ (9,*) X_M
READ (10,*) Y_M_t
READ (11,*) Y_M_c
READ (12,*) Y_M_cur
ELSEIF(Rc == 4)THEN
READ (229,*) X_M
READ (2210,*) Y_M_t
READ (2211,*) Y_M_c
ENDIF
ELSEIF((x-1)*Slice >= (0.5*Span) - 0.5*Bf)THEN
IF(Rc == 1.or.Rc == 2.or.Rc == 5)THEN
READ (9,*) X_M
READ (10,*) Y_M_t
READ (11,*) Y_M_c
READ (12,*) Y_M_cur
ELSEIF(Rc == 3.or.Rc == 4.or.Rc == 6)THEN
READ (229,*) X_M
READ (2210,*) Y_M_t
332
Appendix
READ (2211,*) Y_M_c

ENDIF
ENDIF
CALL GetValue (Mcp, X_M, Y_M_t, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_M_t, Xval, Yval, Ypval, Yppval)
Tsax (x, w) = Yval
Tsax (1, w) = 1.1605*0.000001
CALL GetValue (Mcp, X_M, Y_M_c, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_M_c, Xval, Yval, Ypval, Yppval)
Csax (x, w) = -Yval
Csax (1, w) = -0.000001
CALL GetValue (Mcp, X_M, Y_M_cur, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_M_cur, Xval, Yval, Ypval, Yppval)
Cax (x, w)= Yval
Cax (1, w) = (Tsax (1, w) + Abs(Csax (1, w)))/Height
Xval = -Csax (x, w)
A_s_s = -Yval
Xval = -Csax (x, w)
A_f_s = -Yval
Yax (x, w) = (Height * Tsax (x, w))/((Abs(Csax (x, w))) +Tsax (x, w))
Dsax (x, w) = Slice * Cax (x, w)
Sm = A_s_s / Csax (x, w)

Fm = A_f_s / Csax (x, w)
Fm = 2.14
Sm = 31000
Prf = 0.1
W_s = -(3*(3*((12*((3-Prf)**2)*((1+Prf)**2))**(-0.160160160))*((Sm)**(0.160160160))*((Fm)**(0.666666))))
WRITE(*,*) ' Compressive strain =',Csax (x, w), ' mm/mm'
WRITE(*,*) ' Actual skin stress =',A_s_s, 'MPa'
WRITE(*,*) ' Wrinkling stress =',W_s, 'MPa'
WRITE(99,*) ' Compressive strain =',Csax (x, w), ' mm/mm'
WRITE(99,*) ' Actual skin stress =',A_s_s, 'MPa'
WRITE(99,*) ' Wrinkling stress =',W_s, 'MPa'
IF(Tsax (x, w) >= 0.014)THEN
WRITE (*,*) ' '
WRITE (*,*) '>>>FAILURE OCCURRED DUE TO ECESSIVE GFRP TENSILE STRESS'
WRITE (*,*) ' Tensil Strain = ',Tsax (x, w),' mm/mm'
WRITE (*,*) ' Compressive Strain = ',Csax (x, w),' mm/mm'
WRITE (*,*) ' Curvature = ',Cax (x, w),' 1/mm'
WRITE (*,*) ' LOAD =',(Load / 100) * w, 'N'
WRITE (*,*) ' SECTION =',x
WRITE (99,*) ' '
WRITE (99,*) '>>>FAILURE OCCURRED DUE TO ECESSIVE GFRP TENSILE STRESS'
WRITE (99,*) ' Tensil Strain = ',Tsax (x, w),' mm/mm'
WRITE (99,*) ' Compressive Strain = ',Csax (x, w),' mm/mm'
WRITE (99,*) ' Curvature = ',Cax (x, w),' 1/mm'
WRITE (99,*) ' LOAD =',(Load / 100) * w, 'N'
WRITE (99,*) ' SECTION =',x
GOTO 60
ENDIF
IF(ABS(Csax (x, w)) >= 0.004)THEN
WRITE (*,*) ' '
WRITE (*,*) '>>>FAILURE OCCURRED DUE TO ECESSIVE GFRP COMPRESSIVE STRESS'
WRITE (99,*) ' '
WRITE (99,*) '>>>FAILURE OCCURRED DUE TO ECESSIVE GFRP COMPRESSIVE STRESS'
GOTO 60
ENDIF
IF(Abs(A_s_s) >= Abs(W_s))THEN
WRITE (*,*) ' '
WRITE (*,*) '>>>FAILURE OCCURRED DUE TO GFRP WRINKLING STRESS'
WRITE (*,*) ' Compressive Stress = ',A_s_s,' MPa'
WRITE (*,*) ' W_s = ',W_s,' MPa'
WRITE (99,*) ' '
WRITE (99,*) '>>>FAILURE OCCURRED DUE TO GFRP WRINKLING STRESS'
WRITE (99,*) ' Compressive Stress = ',A_s_s,' MPa'
WRITE (99,*) ' W_s = ',W_s,' MPa'
GOTO 60
ENDIF
CLOSE (9)
CLOSE (10)
333
Appendix
CLOSE (11)
CLOSE (12)
CLOSE (229)
CLOSE (2210)
CLOSE (2211)
CLOSE (2212)
60 RETURN
ENDSUBROUTINE Ten_Comp_Curv
SUBROUTINE Transformed (w,x,Bnahyax, Qahyax,Iahyax,Height,L_num,Lt,Ybar,Rc,X_M, Y_M_cur,Y_M_t,Y_M_c,Tsax,Csax,Cax,Yax,X_F_t,&

IMPLICIT NONE
REAL (KIND = 8):: T_s, C_s , Sahy , D1, D2, Moment, Curvature, B, St , Wt, Wte, ASt, Bf, Bfe, Error, Fs, Foam_stress, Skin_stress , Eo, F_T_s_s, F_C_s_s
REAL (KIND = 8), INTENT(INOUT):: Ybar, Height, Lt , L_num
REAL (KIND = 8), DIMENSION (401), INTENT(INOUT):: X_M, Y_M_cur, Y_M_t , Y_M_c, Ypp_M_cur, Ypp_M_t, Ypp_M_c
REAL (KIND = 8), DIMENSION (7), INTENT(INOUT):: X_F_t, Y_F_t, X_F_c, Y_F_c , Ypp_F_t , Ypp_F_c
REAL (KIND = 8), DIMENSION (37), INTENT(INOUT):: X_F_s, Y_F_s, Ypp_F_s
REAL (KIND = 8), DIMENSION (16), INTENT(INOUT):: X_GFRP, Y_GFRP, X_GFRP_C, Y_GFRP_C , Ypp_GFRP
REAL (KIND = 8), DIMENSION (160,100), INTENT(INOUT):: Csax, Tsax, Cax, Yax
REAL (KIND = 8), DIMENSION (22,160,100):: Sahyax , Enahyax , Neahyax
REAL (KIND = 8), DIMENSION (22,160,100), INTENT(INOUT):: Bnahyax, Qahyax, Iahyax
INTEGER ( KIND = 4 ):: ibcbeg, ibcend, I
REAL ( KIND = 8 ):: ybcbeg, ybcend, Xval, Yval, yppval, ypval, Span, Load
REAL (KIND = 4), INTENT(INOUT):: x,w, Rc
B = 300
Span = 1400
St = 1.5
ASt = 0
Bf = 0
Bfe = 0
Wt = 0
Wte = 0
Eo = 2.14
E_Skin = 31000
I_Skin = (B*(St**3))/12
NG = 16
Nft = 7
Nfc = 7
Nfs = 37
Mcp = 401
IF(Rc == 1)THEN
Sahyax (1, x, w) = Tsax (x, w)* ((Yax (x, w)-(0.5*St))/Yax (x, w))
Xval = Sahyax (1, x, w)
Sahyax (1, x, w) = Yval
Enahyax (1, x, w) = Abs(Sahyax (1, x, w) / Sahyax (1, x, w))
Neahyax (1, x, w) = Enahyax (1, x, w) / Eo
Bnahyax (1, x, w) = Neahyax (1, x, w) * B
Qahyax (1, x, w) = (Bnahyax (1, x, w) * St) * Abs(Yax (x, w) - 0.5*St)
Iahyax (1, x, w) = (Bnahyax (1, x, w) * St) * (Yax (x, w) - 0.5*St)**2
DO I = 2, L_num+1
IF (((I-1)-0.5)*Lt + St <= Yax (x, w)) THEN
Sahyax (I, x, w) = Tsax (x, w)* (((Yax (x, w)-((I-1)-0.5)*Lt+St))/Yax (x, w))
Xval = Sahyax (I, x, w)
Sahyax (I, x, w) = Yval
ELSEIF (((I-1)-0.5)*Lt + St > Yax (x, w)) THEN
Sahyax (I, x, w) = Csax (x, w) * ((((I-1)-0.5)*Lt+St)- Yax (x, w))/(Height-Yax (x, w))
Xval = -Sahyax (I, x, w)
Sahyax (I, x, w) = -Yval
ENDIF
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / Eo
Bnahyax (I, x, w) = Neahyax (I, x, w) * B
Qahyax (I, x, w) = ((Bnahyax (I, x, w) * Lt) * Abs(Yax (x, w) - ((I-0.5)*Lt)))
Iahyax (I, x, w) = (Bnahyax (I, x, w) * Lt) * (Yax (x, w) - ((I-0.5)*Lt))**2
ENDDO
Sahyax (L_num+2, x, w) = Csax (x, w) * (((L_num)*Lt+1.5*St)- Yax (x, w))/(Height-Yax (x, w))
Xval = -Sahyax (L_num+2, x, w)
Sahyax (L_num+2, x, w) = -Yval
Enahyax (L_num+2, x, w) = Abs(Sahyax (L_num+2, x, w) / Sahyax (L_num+2, x, w))
Neahyax (L_num+2, x, w) = Enahyax (L_num+2, x, w) / Eo
Bnahyax (L_num+2, x, w) = Neahyax (L_num+2, x, w) * B
Qahyax (L_num+2, x, w) = (Bnahyax (L_num+2, x, w) * St) * Abs(Yax (x, w) - (Height-0.5*St))
Iahyax (L_num+2, x, w) = (Bnahyax (L_num+2, x, w) * St) * (Yax (x, w) - (Height-0.5*St))**2
ELSEIF(Rc == 2)THEN
Sahyax (1, x, w) = Tsax (x, w) * ((Yax (x, w)-(0.5*St))/Yax (x, w))
Neahyax (1, x, w) = Enahyax (1, x, w) / E_Skin
Sahyax (2, x, w) = Tsax (x, w) * ((Yax (x, w)-(St+0.5*ASt))/Yax (x, w))
334
Appendix

Bnahyax (2, x, w) = Neahyax (2, x, w) * Bf
Bnahyax (2, x, w) = Bnahyax (2, x, w) + Neahyax (2, x, w) * (B-Bf)
Qahyax (2, x, w) = (Bnahyax (2, x, w) * ASt) * Abs(Yax (x, w) - (St + 0.5*ASt))
Iahyax (2, x, w) = (Bnahyax (2, x, w) * ASt) * (Yax (x, w) - (St + 0.5*ASt))**2
DO I = 3, L_num+2
IF (((I-2)-0.5)*Lt + St + ASt <= Yax (x, w)) THEN
Sahyax (I, x, w) = Tsax (x, w) * ((Yax (x, w)-(((I-2)-0.5)*Lt+St+ ASt))/Yax (x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
Bnahyax (I, x, w) = Neahyax (I, x, w) * Wt
Bnahyax (I, x, w) = Bnahyax (I, x, w) + Neahyax (I, x, w) * (B-Wt)
Qahyax (I, x, w) = (Bnahyax (I, x, w) * Lt) * Abs(Yax (x, w) - (St + ASt + 0.5*Lt))
Iahyax (I, x, w) = (Bnahyax (I, x, w) * Lt) * (Yax (x, w) - (St + ASt + 0.5*Lt))**2
ELSEIF (((I-2)-0.5)*Lt + St + ASt> Yax (x, w)) THEN
Sahyax (I, x, w) = Csax (x, w) * (((((I-2)-0.5)*Lt+St+ ASt)- Yax (x, w)))/(Height-Yax (x, w))
ENDIF
ENDDO
Sahyax (L_num+4, x, w) = Csax (x, w) * ((((L_num)*Lt+1.5*St+2*ASt)- Yax (x, w)))/(Height-Yax (x, w))
Neahyax (L_num+4, x, w) = Enahyax (L_num+4, x, w) / E_Skin
Sahyax (L_num+3, x, w) = Csax (x, w) * ((((L_num)*Lt+St+1.5*ASt)- Yax (x, w))) /(Height-Yax (x, w))
Xval = Sahyax (L_num+3, x, w)
Sahyax (L_num+3, x, w) = Yval
Bnahyax (L_num+3, x, w) = Neahyax (L_num+3, x, w) * Bf
Bnahyax (L_num+3, x, w) = Bnahyax (L_num+3, x, w) + Neahyax (L_num+3, x, w) * (B-Bf)
Qahyax (L_num+3, x, w) = (Bnahyax (L_num+3, x, w) * ASt) * Abs(Yax (x, w) - (St + 0.5*ASt))
Iahyax (L_num+3, x, w) = (Bnahyax (L_num+3, x, w) * ASt) * (Yax (x, w) - (St + 0.5*ASt))**2
ELSEIF(Rc == 3)THEN
IF((x-1)*Slice >= (0.5*Span-0.5*Bf))THEN
Sahyax (1, x, w) = Tsax (x, w)* ((Yax (x, w)-((0.5)*St))/Yax (x, w))
Sahyax (2, x, w) = Tsax (x, w)* ((Yax (x, w)-(St+0.5*ASt))/Yax (x, w))
ELSEIF((x-1)*Slice < (0.5*Span-0.5*Bf))THEN
335
Appendix
Sahyax (1, x, w) = Tsax (x, w)* ((Yax (x, w)-((0.5)*St))/Yax (x, w))
Sahyax (2, x, w) = Tsax (x, w)* ((Yax (x, w)-(St+0.5*ASt))/Yax (x, w))
Bnahyax (2, x, w) = Neahyax (2, x, w) * Bf
Bnahyax (2, x, w) = Bnahyax (2, x, w) + Neahyax (2, x, w) * (B-Bf)
ENDIF
DO I = 3, L_num+2
Sahyax (I, x, w) = Tsax (x, w)* ((Yax (x, w)-(((I-2)-0.5)*Lt+St+ ASt))/Yax (x, w))
Sahyax (I, x, w) = Csax (x, w) * ((((I-2)-0.5)*Lt+St+ ASt)- Yax (x, w))/(Height-Yax (x, w))
ENDIF
ENDDO
IF((x-1)*Slice >= (0.5*Span-0.5*Bf))THEN
Sahyax (L_num+4, x, w) = Csax (x, w) * (((L_num)*Lt+1.5*St+2*ASt)- Yax (x, w))/(Height-Yax (x, w))
Sahyax (L_num+3, x, w) = Csax (x, w) * (((L_num)*Lt+St+1.5*ASt)- Yax (x, w))/(Height-Yax (x, w))
ELSEIF((x-1)*Slice < (0.5*Span-0.5*Bf))THEN
Sahyax (L_num+4, x, w) = Csax (x, w) * (((L_num)*Lt+1.5*St+2*ASt)- Yax (x, w))/(Height-Yax (x, w))
Sahyax (L_num+3, x, w) = Csax (x, w) * (((L_num)*Lt+St+1.5*ASt)- Yax (x, w))/(Height-Yax (x, w))
336
Appendix

Bnahyax (L_num+3, x, w) = Neahyax (L_num+3, x, w) * Bf
Bnahyax (L_num+3, x, w) = Bnahyax (L_num+3, x, w) + Neahyax (L_num+3, x, w) * (B-Bf)
ENDIF
ELSEIF(Rc == 4)THEN
IF((x-1)*Slice >= (Bfe))THEN
Sahyax (1, x, w) = Tsax (x, w) * ((Yax (x, w)-((0.5)*St))/Yax (x, w))
Bnahyax (2, x, w) = Neahyax (2, x, w) * 2*Bfe
Bnahyax (2, x, w) = Bnahyax (2, x, w) + Neahyax (2, x, w) * (B-2*Bfe)
ELSEIF((x-1)*Slice < (Bfe))THEN
ENDIF
DO I = 3, L_num+2
Bnahyax (I, x, w) = Neahyax (I, x, w) * 2*Wte
Bnahyax (I, x, w) = Bnahyax (I, x, w) + Neahyax (I, x, w) * (B-2*Wte)
Bnahyax (I, x, w) = Neahyax (I, x, w) * 2*Wte
Bnahyax (I, x, w) = Bnahyax (I, x, w) + Neahyax (I, x, w) * (B-2*Wte)
ENDIF
337
Appendix
ENDDO
Sahyax (L_num+3, x, w) = Csax (x, w) * ((((L_num)*Lt+St+1.5*ASt)- Yax (x, w)))/(Height-Yax (x, w))
Bnahyax (L_num+3, x, w) = Neahyax (L_num+3, x, w) * 2*Bfe
Bnahyax (L_num+3, x, w) = Bnahyax (L_num+3, x, w) + Neahyax (L_num+3, x, w) * (B-2*Bfe)
Sahyax (L_num+3, x, w) = Csax (x, w) * ((((L_num)*Lt+St+1.5*ASt)-Yax (x, w)))/(Height-Yax (x, w))
ENDIF
ELSEIF(Rc == 5)THEN
Bnahyax (2, x, w) = Neahyax (2, x, w) * (Bf + 2*Bfe)
Bnahyax (2, x, w) = Bnahyax (2, x, w) + Neahyax (2, x, w) * (B - Bf - 2*Bfe)
ENDIF
338
Appendix
DO I = 3, L_num+2
Bnahyax (I, x, w) = Neahyax (I, x, w) * (Wt + 2*Wte)
Bnahyax (I, x, w) = Bnahyax (I, x, w) + Neahyax (I, x, w) * (B - Wt - 2*Wte)
ENDIF
ENDDO
Bnahyax (L_num+3, x, w) = Neahyax (L_num+3, x, w) * (Bf + 2*Bfe)
Bnahyax (L_num+3, x, w) = Bnahyax (L_num+3, x, w) + Neahyax (L_num+3, x, w) * (B - Bf - 2*Bfe)
ENDIF
ELSEIF(Rc == 6)THEN
IF((x-1)*Slice <= (Bfe) .or. (x-1)*Slice >= (0.5*Span-Bfe))THEN
Sahyax (1, x, w) =Tsax (x, w) * ((Yax (x, w)-((0.5)*St))/Yax (x, w))
339
Appendix

ELSEIF((x-1)*Slice > (Bfe) .and. (x-1)*Slice < (0.5*Span-Bfe))THEN
Bnahyax (2, x, w) = Neahyax (2, x, w) * (Bf + 2*Bfe)
Bnahyax (2, x, w) = Bnahyax (2, x, w) + Neahyax (2, x, w) * (B - Bf - 2*Bfe)
ENDIF
DO I = 3, L_num+2
ENDIF
ENDDO
IF((x-1)*Slice <= (Bfe) .or. (x-1)*Slice >= (0.5*Span-Bfe))THEN
ELSEIF((x-1)*Slice > (Bfe) .and. (x-1)*Slice < (0.5*Span-Bfe))THEN
340
Appendix

Bnahyax (L_num+3, x, w) = Neahyax (L_num+3, x, w) * (Bf + 2*Bfe)
Bnahyax (L_num+3, x, w) = Bnahyax (L_num+3, x, w) + Neahyax (L_num+3, x, w) * (B - Bf - 2*Bfe)
ENDIF
ENDIF
ENDSUBROUTINE Transformed
SUBROUTINE Mom_Defl (Slice,Dsax,Sax,Ddax,Dm,Delta_Def_Max )
IMPLICIT NONE
REAL (KIND = 8), INTENT(INOUT):: Slice , Delta_Def_Max
REAL (KIND = 8), DIMENSION (160,100), INTENT(INOUT):: Dsax, Sax, Ddax
REAL (KIND = 8), DIMENSION (100), INTENT(INOUT):: Dm
REAL (KIND = 4):: I
Sax (S_num, w) = Dsax (S_num, w)
DO x = 1, S_num-1, 1
Sax ((S_num-x), w) = Sax (((S_num+1)-x), w) + Dsax ((S_num-x), w)
ENDDO
DO x = 1, S_num, 1
Ddax (x, w) = Dsax (x, w)* (((x-1)-0.5) * Slice)
ENDDO
Dm = Sum (Ddax , Dim = 1)
DO x = 1, S_num, 1
DO I = x, S_num, 1
Ddax (I, w) = Dsax (I, w)* (((I-x)-0.5) * Slice)
ENDDO
Delta_Def_Max = Sum (Ddax)
DO I = 1, S_num, 1
Ddax (I, w) = 0.0
ENDDO
WRITE (15,"(4f25.13)") (Load/LS_num)* w, x, Delta_Def_Max, Dm (w)
ENDDO
ENDSUBROUTINE Mom_Defl
SUBROUTINE Shear_Defl (w,x,Load,Iahyax,It_at_x,Lt,L_num,St,Slice,Yax,Qahyax,Rc,Tahyax,Gahyax,Vax,Bnahyax,Xval,Yval,Dvax,Galyaw,&

IMPLICIT NONE
REAL (KIND = 8), INTENT(INOUT):: Slice, St , Lt, Nfs , L_num, Rc
REAL (KIND = 8), DIMENSION (22,160,100), INTENT(INOUT):: Bnahyax, Qahyax, Iahyax , Tahyax, Gahyax, Dvahyax
REAL (KIND = 8), DIMENSION (160,100), INTENT(INOUT):: It_at_x , Dvat, Dvab, Vax, Yax, Dvax
REAL (KIND = 8), DIMENSION (22,100), INTENT(INOUT)::Galyaw
REAL (KIND = 8), DIMENSION (37), INTENT(INOUT):: X_F_s, Y_F_s
INTEGER ( KIND = 4 ):: ibcbeg, ibcend
REAL ( KIND = 8 ):: ybcbeg, ybcend, Xval, Yval, yppval, ypval, Mcp
REAL (KIND = 8), DIMENSION (401):: Ypp_M_cur, Ypp_M_t , Ypp_M_c
REAL (KIND = 8), DIMENSION (7):: Ypp_F_t, Ypp_F_c
REAL (KIND = 8), DIMENSION (37):: Ypp_F_s
REAL (KIND = 8), DIMENSION (16):: Ypp_GFRP, Ypp_GFRP_s
REAL (KIND = 4), INTENT(INOUT):: Load ,x ,w
REAL (KIND = 4):: I
It_at_x = Sum (Iahyax, Dim = 1)
DO I = 2, L_num
IF ((((I-1)-0.100)*Lt + St) <= Yax (x, w)) THEN
Qahyax (I, x, w) = Qahyax ((I-1), x, w) + Qahyax (I, x, w)
ENDIF
ENDDO
DO I = L_num, 1, -1
IF((((I-1)+0.100)*Lt + St) > Yax (x, w)) THEN
Qahyax (I+1, x, w) = Qahyax ((I+2), x, w) + Qahyax (I+1, x, w)
ENDIF
ENDDO
IF(Rc == 1)THEN
DO I = 1, L_num+2
Tahyax (I, x, w) = (Vax (x, w) * Qahyax (I, x, w)) / (It_at_x (x, w) * Bnahyax (I, x, w))
IF ( Tahyax (I, x, w) <= 0.18298) THEN
Xval = Tahyax (I, x, w)
CALL GetValue (Nfs, X_F_s, Y_F_s, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_s, Xval, Yval, Ypval, Yppval)
Gahyax (I, x, w) = Yval
ELSE
WRITE (*,*) ' '
WRITE (*,*) '>>> '
WRITE (*,*) '>>>FAILURE OCCURRED DUE TO ECESSIVE SHEAR STRESS'
WRITE (*,*) '>>> '
WRITE (*,*) ' Q = ',Qahyax (I, x, w),' mm3'
WRITE (*,*) ' B = ',Bnahyax (I, x, w),' mm'
WRITE (*,*) ' Taw = ',Tahyax (I, x, w),' MPa'
341
Appendix
WRITE (*,*) ' Gamma = ',Gahyax (I, x, w),' radians'

WRITE (*,*) ' LAYER =',I
WRITE (99,*) ' '
WRITE (99,*) '>>> '
WRITE (99,*) '>>>FAILURE OCCURRED DUE TO ECESSIVE SHEAR STRESS'
WRITE (99,*) '>>> '
WRITE (99,*) ' Q = ',Qahyax (I, x, w),' mm3'
WRITE (99,*) ' B = ',Bnahyax (I, x, w),' mm'
WRITE (99,*) ' Taw = ',Tahyax (I, x, w),' MPa'
WRITE (99,*) ' Gamma = ',Gahyax (I, x, w),' radians'
WRITE (99,*) ' LAYER =',I
!STOP
GOTO 100
ENDIF
ENDDO
ELSEIF(Rc /= 1)THEN
DO I = 1, L_num+4
Tahyax (I, x, w) = (Vax (x, w) * Qahyax (I, x, w)) / (It_at_x (x, w) * Bnahyax (I, x, w))
IF ( Tahyax (I, x, w) <= 41.7) THEN
Xval = Tahyax (I, x, w)
CALL GetValue (NGs, X_GFRP_s, Y_GFRP_s, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP_s, Xval, Yval, Ypval, Yppval)
Gahyax (I, x, w) = Yval
ELSE
WRITE (*,*) ' '
WRITE (*,*) '>>> '
WRITE (*,*) '>>>FAILURE OCCURRED DUE TO ECESSIVE SHEAR STRESS'
WRITE (*,*) '>>> '
WRITE (*,*) ' Q = ',Qahyax (I, x, w),' mm3'
WRITE (*,*) ' B = ',Bnahyax (I, x, w),' mm'
WRITE (*,*) ' Taw = ',Tahyax (I, x, w),' MPa'
WRITE (*,*) ' Gamma = ',Gahyax (I, x, w),' radians'
WRITE (*,*) ' LAYER =',I
WRITE (99,*) ' '
WRITE (99,*) '>>> '
WRITE (99,*) '>>>FAILURE OCCURRED DUE TO ECESSIVE SHEAR STRESS'
WRITE (99,*) '>>> '
WRITE (99,*) ' Q = ',Qahyax (I, x, w),' mm3'
WRITE (99,*) ' B = ',Bnahyax (I, x, w),' mm'
WRITE (99,*) ' Taw = ',Tahyax (I, x, w),' MPa'
WRITE (99,*) ' Gamma = ',Gahyax (I, x, w),' radians'
WRITE (99,*) ' LAYER =',I
!STOP
GOTO 100
ENDIF
ENDDO
ENDIF
100 RETURN
ENDSUBROUTINE Shear_Defl
SUBROUTINE Total_Defl (Dvax,S_num,Rc,Gahyax,L_num,Slice,Load,LS_num,w,Galyaw,Dvat,Dvab,Dt,Dm,Wdax,Db,Max,Vax,Tsax,Csax,Cax,Yax,Nhax,Height,Lt,St)

IMPLICIT NONE
REAL (KIND = 8), DIMENSION (22,160,100), INTENT(INOUT):: Gahyax
REAL (KIND = 8), DIMENSION (22,100), INTENT(INOUT)::Galyaw
REAL (KIND = 8), DIMENSION (160,100), INTENT(INOUT):: Dvat, Dvab, Vax , Nhax , Dvax, Wdax , Csax , Tsax, Cax, Yax
REAL (KIND = 8), DIMENSION (100), INTENT(INOUT):: Dt, Db , Dm
REAL (KIND = 8), DIMENSION (100):: Dv_Top, Dv_Bot
REAL (KIND = 8), INTENT(INOUT):: Slice, S_num, LS_num, Height , L_num,Lt,St , Rc
REAL (KIND = 8):: Moment,Shear, Curvature,T_s, C_s, n1, n2
REAL (KIND = 4), INTENT(INOUT):: Load,w
REAL (KIND = 4):: I,x
Dvat = 0.0
Dvab = 0.0
DO x = 1, S_num
n1 = 1.0
n2 = 1.0
IF(Rc ==1)THEN
Dvab (x, w) = Dvab (x, w)+ Gahyax (1, x, w) * Slice
Dvat(x, w) = Dvat(x, w) + Gahyax (L_num+2, x, w) * Slice
DO I = 2, L_num
IF(((I-1)*Lt + St) <= Yax (x, w)) THEN
n1 = n1 + 1
Dvab (x, w)= Dvab (x, w)+ Gahyax (I, x, w) * Slice
ELSEIF(((I-1)*Lt + St) >= Yax (x, w)) THEN
n2 = n2 + 1
Dvat (x, w)= Dvat (x, w)+ Gahyax (I, x, w) * Slice
ENDIF
ENDDO
Dvax (x, w)= (Dvab (x, w)+ Dvat (x, w))/(L_num+2)
Dvab (x, w)= Dvab (x, w)/n1
Dvat (x, w)= Dvat (x, w)/n2
ELSEIF(Rc /=1)THEN
Dvab (x, w)= Dvab (x, w)+ Gahyax (1, x, w) * Slice
Dvab (x, w)= Dvab (x, w)+ Gahyax (2, x, w) * Slice
Dvat (x, w)= Dvat (x, w)+ Gahyax (L_num+4, x, w) * Slice
342
Appendix
Dvat (x, w)= Dvat (x, w)+ Gahyax (L_num+3, x, w) * Slice

DO I = 3, L_num+2
IF(((I-1)*Lt + St) <= Yax (x, w)) THEN
n1 = n1 + 1
Dvab (x, w)= Dvab (x, w)+ Gahyax (I, x, w) * Slice
ELSEIF(((I-1)*Lt + St) >= Yax (x, w)) THEN
n2 = n2 + 1
Dvat (x, w)= Dvat (x, w)+ Gahyax (I, x, w) * Slice
ENDIF
ENDDO
Dvax (x, w)=(Dvab (x, w)+ Dvat (x, w))/(L_num+4)
Dvab (x, w)= Dvab(x, w)/n1
Dvat (x, w)= Dvat (x, w)/n2
ENDIF
WRITE (16,"(3f25.13)") (Load/LS_num)* w, x, Dvax
ENDDO
WRITE (*,*) '>>>Deflection due to Shear has been calculated for midspan'
WRITE (99,*) '>>>Deflection due to Shear has been calculated for midspan'
Dv_Top = Sum(Dvat, Dim = 1)

Dv_Bot = Sum(Dvab, Dim = 1)
Dt (w) = Dm (w) + Dv_Top (w) + Wdax (S_num, w)
Db (w) = Dm (w) + Dv_Bot (w)
WRITE (*,*)''
WRITE (*,*)'---***---'
WRITE (*,*) '>>>TOTAL Deflection Calculations are Finished at Midspan of the Panel'
WRITE (*,*) '>>>For Applied Load Equals =',(Load/LS_num)* w,'N'
WRITE (*,*) '>>>TOTAL Top Deflection at midspan =',Dt( w),'mm <<<'
WRITE (*,*) '>>>TOTAL Bottom Deflection at midspan =',Db( w),'mm <<<'
WRITE (*,*)'---***---'
WRITE (*,*)''
WRITE (99,*)''
WRITE (99,*)'---***---'
WRITE (99,*) '>>>TOTAL Deflection Calculations are Finished at Midspan of the Panel'
WRITE (99,*) '>>>For Applied Load Equals =',(Load/LS_num)* w,'N'
WRITE (99,*) '>>>TOTAL Top Deflection at midspan =',Dt( w),'mm <<<'
WRITE (99,*) '>>>TOTAL Bottom Deflection at midspan =',Db( w),'mm <<<'
WRITE (99,*)'---***---'
WRITE (99,*)''
Moment = 0.001*Max (S_num, w)
Shear = Vax (S_num, w)
T_s = Tsax (S_num, w)
C_s = -1*Csax (S_num, w)
Curvature = Cax (S_num, w)
A_L = 0.001*(Load / LS_num) * w
WRITE (14,"(12f25.13)") A_L, Shear, Moment, Curvature, Dm (w), Dv_Top (w), Dv_Bot (w), Wdax (S_num, w), Dt (w), Db (w), T_s, C_s
WRITE (17,"(3f25.13)") A_L, T_s, C_s
ENDSUBROUTINE Total_Defl
SUBROUTINE New_Height_1 (w, Height, Slice, S_num, L_num, Rc, Gahyax, Wdax, Nhax)
IMPLICIT NONE
REAL (KIND = 8), DIMENSION (160,100), INTENT(INOUT):: Nhax, Wdax
REAL (KIND = 8), INTENT(INOUT):: Slice , S_num, Height, L_num, Rc
REAL (KIND = 4), INTENT(INOUT):: w
DO x = 1, S_num
Nhax (x, w) = Height - Abs(Dvab (x, w) - Dvat (x, w))- Wdax (x, w)
ENDDO
ENDSUBROUTINE New_Height_1
SUBROUTINE New_Height_2 (w, Slice, S_num, L_num, Rc, Gahyax, Wdax, Nhax)
IMPLICIT NONE
REAL (KIND = 8), DIMENSION (160,100), INTENT(INOUT):: Nhax, Wdax
REAL (KIND = 8), INTENT(INOUT):: Slice, S_num, L_num, Rc
REAL (KIND = 4), INTENT(INOUT):: w
DO x = 1, S_num
Nhax (x, w) = Nhax (x, (w-1)) - Abs(Dvab (x, w) - Dvat (x, w))- Wdax (x, w)
ENDDO
ENDSUBROUTINE New_Height_2
ENDPROGRAM Sandwich_Panels
343

Sharaf Tarek A 201008 PHD PDF

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Sharaf Tarek A 201008 PHD PDF

Diunggah oleh

Hak Cipta:

Format Tersedia

FLEXURAL BEHAVIOUR OF SANDWICH PANELS

COMPOSED OF POLYURETHANE CORE AND

GFRP SKINS AND RIBS

Tarek Abdel Moneim Sharaf

A thesis submitted to the Department of Civil Engineering

Copyright Tarek A. Sharaf, 2010

This study addresses the flexural performance of sandwich panels composed of a

The experimental program included three phases. Phase I is a comprehensive

50 to 150%, depending on the rib configuration, compared to a panel without ribs. In

using the principals of beam-on-elastic foundation. Both models were successfully

and compression skin crushing or wrinkling were successfully predicted.

stabilized at a spacing of 2.9 times the panel thickness.

past four years unforgettable.

guidance. He will never be forgotten. I would also like to acknowledge my fellow

experimental phase of this study.

I wish to acknowledge the financial support provided by the Egyptian Government

support of this project.

encouragement throughout my entire life.

your love, patience and dedication, I am truly grateful.

LIST OF FIGURES... xii

1.2 Research Objectives..... 2

1.3 Scope and Contents.. 4

CHAPTER 2: LITERATURE REVIEW... 9

2.2 Polyurethane Foam Material Composition and Structure. 10

2.3 Sandwich Panels with Solid Cores.. 11

2.4 Sandwich Panels with Cellular Cores. 17

2.5 Mechanical Properties of Sandwich Panels . 22

2.6 Buckling and Skin Wrinkling of Sandwich Panels . 25

2.7 Delamination and Skin Debonding of Sandwich Panels 30

2.8 Numerical Analysis of Sandwich Panels .... 32

CHAPTER 3: MATERIAL TEST PROGRAM...... 47

3.1 Introduction ... 47

3.2 Experimental Program. 48

3.2.1 Tests on Glass Fibre-Reinforced Polymer (GFRP).. 48

3.2.1.1 Tension Coupons 48

3.2.1.2 Compression Coupons............. 49

3.2.1.3 Shear Coupons.............. 50

3.2.2 Polyurethane Foam Tests.... 51

3.2.2.1 Tension Coupons 51

3.2.2.2 Compression Coupons............. 51

3.2.2.3 Shear Coupons.............. 52

3.3 Test Results 53

3.3.1 GFRP Coupons. 53

3.3.2 Polyurethane Foam Coupons......... 54

CHAPTER 4: EXPERIMENTAL STUDY ON SANDWICH PANELS WITH

DIFFERENT RIB CONFIGURATIONS. 73

4.2 Description of Test Specimens and Fabrication 74

4.4 Test Setup.... 76

4.6 Test Results, Failure Modes ... 77

4.8 Analysis of Stiffness and Deflections at Service Load Levels.. 81

4.9 Comparison with Reinforced Concrete Panels . 83

CHAPTER 5: TESTING OF LARGE SCALE CLADDING SANDWICH

5.2 Description of the Wall Panels and Fabrication 99

5.3 Test Specimens.. 100

5.4 Materials 101

5.5 Test Setups and Instrumentation.. 102

5.5.1 Test I - Full Scale Panel Under Simulated Uniform Air

5.6 Results of the Experimental Program.. 105

5.6.1.1 Load-deflection responses ...... 105

5.6.1.2 Load-strain responses ............. 106

5.6.1.3 Failure mode.............. 106

5.6.2.1 Load-deflection responses 106

5.6.2.2 Deflection distribution across the width . 107