Professional Documents
Culture Documents
Sharaf Tarek A 201008 PHD PDF
Sharaf Tarek A 201008 PHD PDF
By
Queens University
Kingston, Ontario, Canada
August, 2010
ABSTRACT
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.
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
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,
validated using experimental results. Possible failure modes, namely core shear failure,
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
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
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
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
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
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
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
iv
Table of Contents
TABLE OF CONTENTS
ABSTRACT. i
ACKNOWLEDGMENTS iii
TABLE OF CONTENTS v
LIST OF TABLES xi
NOTATIONS xxv
CHAPTER 1: INTRODUCTION. 1
1.1 Introduction..... 1
2.1 Introduction.... 9
v
Table of Contents
4.1 Introduction 73
4.3 Fabrication... 75
4.5 Instrumentation.. 76
vi
Table of Contents
4.7 Discussion 80
PANELS.. 98
5.1 Introduction..... 98
Pressure 102
5.5.2 Test II Half Full Scale Panel Under Discrete Line Loads........ 104
5.6.1 Test I - Full Scale Panel Under Uniform Air Pressure. 105
5.6.2 Test II Half Full Scale Panel Under Discrete Line Loads........ 106
vii
Table of Contents
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
viii
Table of Contents
Loadings. 216
ix
Table of Contents
Configurations 285
REFERENCES.. 291
x
List of Tables
LIST OF TABLES
Table 3.1 E-Glass fabric manufacturer data (3 Tex, Inc)... 57
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
xiii
List of Figures
xiv
List of Figures
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
the half-width panel in Test II. 130
Figure 5.37 Loaded skin strain distribution over cross section passing through
axis 1-2-3-4-5 in Test II.. 131
Figure 5.38 Loaded skin strain distribution over cross section passing through
axis 14-13-11-10-7 in Test II.. 131
Figure 5.39 Loaded skin strain distribution over cross section passing through
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
the half-width panel in Test II. 134
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
Figure 6.10 Convergence study results for the 3-D FE model (case of panels
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
xviii
List of Figures
xix
List of Figures
xx
List of Figures
xxi
List of Figures
xxii
List of Figures
xxiii
List of Figures
xxiv
Notations
NOTATIONS
a Distance from beam left end to the point of calculation
Ai Area at layer i
B1 Constant used to calculate critical wrinkling stress for the top skin
C a, C b Constant used to calculate shear forces due to elastic foundation effect for a
xxv
Notations
D x Constant used to calculate shear forces due to elastic foundation effect for a
i Layer number
l Segment number
xxvi
Notations
M` 1 Induced moments in the top skin at end 1 based on the fact that the top skin
M 1 Induced moments in the top skin at end 1 based on the fact that the top skin
M1, M2 Induced moments in the top skin at ends 1, 2 based on the fact that the top
P Concentrated load
pa Atmospheric pressure
xxvii
Notations
Q` 1 Induced shear force in the top skin at end 1 based on the fact that the top
Q 1 Induced moments in the top skin at end 1 based on the fact that the top skin
Q1, Q2 Induced shear force in the top skin at ends 1, 2 based on the fact that the top
Q t,l,i First moment of area for the transformed section at certain span segment l
for layer i
y bar Distance for the extreme tension fibres to the neutral axis
yi Layer location
elastic foundation
Shear strain
xxviii
Notations
v,l,bot Average shear deflection for layers below section neutral axis at segment l
v,l,top Average shear deflection for layers above section neutral axis at segment l
Strain level
xxix
Notations
Shear stress
Section curvature
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
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
Chapter 1 Introduction
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
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
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
2
Chapter 1 Introduction
materials of the sandwich panel, namely the polyurethane foam core and glass fibre
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
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
Assess the performance of the full scale panel relative to the building code requirements
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
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
Chapter 1 Introduction
considerable shear deformation of the core and the change in thickness of the panel due
wider range of geometric and mechanical property parameters, beyond the limitations of
The scope of this study consists of experimental investigations and numerical and
analytical modelling. The experimental programs are designed to address the general
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
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,
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
Chapter 1 Introduction
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,
5
Chapter 1 Introduction
Chapter 2 presents the literature review, including the background and fundamental
core types. A summary of related experimental and analytical work is also presented.
sandwich panels.
Chapter 4 describes Phase II of the experimental program, which examines the flexural
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
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
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
Chapter 1 Introduction
Figure 1.1 Typical precast reinforced concrete sandwich panels and proposed
composite sandwich panels
7
Chapter 1 Introduction
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
8
Chapter 2 Literature Review
2.1 Introduction
Sandwich panels are being used in many applications ranging from aerospace to
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
applications such as light-weight decking, flooring, or roofing and as cladding panels for
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
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
Chapter 2 Literature Review
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
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.
One of the commonly used materials as a core in sandwich panels is polyurethane foam.
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
Chapter 2 Literature Review
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
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
Solid flexible cores refer to cores typically made of foams and are compressible such that
sandwich panels in the 20th century have been in aircraft industry (Allen, 1969). This
11
Chapter 2 Literature Review
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
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).
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
Chapter 2 Literature Review
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
but without giving closed-form equations. A high-order bending theory based on virtual
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
Chapter 2 Literature Review
The presence of diaphragms causes localized effects in both the skins and the core,
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,
distributed loads as well as geometric and material changes were studied. The analysis
14
Chapter 2 Literature Review
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.
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
Chapter 2 Literature Review
(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
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
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
loads.
has a transversely flexible core with negligible flexural rigidity, including large
16
Chapter 2 Literature Review
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
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
Chapter 2 Literature Review
shear rigidity. This approach is good for sandwich structures with incompressible cores,
Petras and Sutcliffe (1999) presented the failure mode maps for honeycomb
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.
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.
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
Chapter 2 Literature Review
stress distribution in the core due to indentation loading, found from a high-order
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
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
Chapter 2 Literature Review
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
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.
sandwich plates comprised of truss cores faced with solid sheets of the same material was
20
Chapter 2 Literature Review
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
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).
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
bending tests were carried out to study the mechanical behaviours of the grids and
21
Chapter 2 Literature Review
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.
Berthelot and Lolive (2002) presented a modeling procedure for evaluating 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
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
Chapter 2 Literature Review
Kagome truss core fabricated from wires exhibits double peaks on the loaddisplacement
curve which makes it possible to absorb much more energy during deformation.
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
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
Chapter 2 Literature Review
The actual mechanical behaviour and the possible failure mechanisms of two
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%.
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
Chapter 2 Literature Review
filters. The data presented in the study can be used for rate-dependent material modelling
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
The compression skins of sandwich panels are sometimes subjected to a particular kind
in the skins exceed the buckling stresses. Studying the wrinkling behaviour is very
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
Chapter 2 Literature Review
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)
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
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
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
model that represents a strip of a sandwich panel with a through-the-width debond. The
26
Chapter 2 Literature Review
results were compared with the buckling loads calculated from a 2-D finite-element
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.
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
27
Chapter 2 Literature Review
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
(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
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
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
Chapter 2 Literature Review
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
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
with a flexible core were analyzed numerically by Sokolinsky et al (2002), using the
panel theory (HSAPT). According to the study, the ordinary (fully bonded) sandwich
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
Chapter 2 Literature Review
pattern that developed between the points of application of the concentrated loads was
compression until failure. According to the study, in the post-buckling regime, the panel
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
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.
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
Chapter 2 Literature Review
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
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
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
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
Chapter 2 Literature Review
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.
Some of the previously used computational models are based on incompressible cores.
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
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
Chapter 2 Literature Review
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.
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
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
Chapter 2 Literature Review
constant strain triangular (CST) element for the in-plane deformation. For the core layer,
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
panels was presented by Avils and Carlsson (2006). The three-dimensional linear FE
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
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
unsymmetrical type are indicated. The study showed that the proposed expressions were
34
Chapter 2 Literature Review
in a good agreement with the experimental result which were done to check these
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
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
modeling of sandwich plates and shells. The study focused on the computational models,
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
Chapter 2 Literature Review
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
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
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
plates, as shown in Figure 2.15. The study included the effects of the geometric
36
Chapter 2 Literature Review
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.
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
37
Chapter 2 Literature Review
(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
Chapter 2 Literature Review
Figure 2.2 (a) Geometry, (b) internal resultants, (c) sign convention,
and (d) deformations of sandwich panels
[Frostig et al, 1992]
39
Chapter 2 Literature Review
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
Chapter 2 Literature Review
41
Chapter 2 Literature Review
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]
42
Chapter 2 Literature Review
Figure 2.9 Comparisons of normalized buckling load results from different methods for
various debond ratios.
[Sleight and Wang, 1995]
Figure 2.10 Finite element prediction and deformation of axially loaded panels
[Falzon and Hitchings, 2003]
43
Chapter 2 Literature Review
44
Chapter 2 Literature Review
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]
45
Chapter 2 Literature Review
Figure 2.16 Sandwich panel test setup and load-deflection response at mid-span
[Boni and Mller, 2008]
46
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
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
Chapter 3 Material Testing Program
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
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
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
Chapter 3 Material Testing Program
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.
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
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
Chapter 3 Material Testing Program
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.
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
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
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
Chapter 3 Material Testing Program
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
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
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
Chapter 3 Material Testing Program
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
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.
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
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
52
Chapter 3 Material Testing Program
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
Chapter 3 Material Testing Program
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
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
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.
foam coupons for both densities. The response was linear until the foam internal cell
54
Chapter 3 Material Testing Program
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
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
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
Chapter 3 Material Testing Program
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
56
Chapter 3 Material Testing Program
Table 3.4 CORAFOAM U020 (low density polyurethane foam) manufacturer data
(DUNA-USA)
Table 3.5 CORAFOAM U040 (high density polyurethane foam) manufacturer data
(DUNA-USA)
57
Chapter 3
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
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
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
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
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
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)
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
61
5 1.98 0.387 0.0694
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
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)
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
62
5 9.6 0.147 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
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)
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
63
5 8.57 0.347 0.223
240x50 20
1 2.24 0.182 0.387
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
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
50 mm 150 mm 50 mm
c) Shear coupons
64
Chapter 3 Material Testing Program
Steel section
70 mm Foam coupon
50 mm
70 mm
70 mm
50 mm
Knife edge
65
Chapter 3 Material Testing Program
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
Chapter 3 Material Testing Program
120
100
Compressive stress (MPa)
80
60
40
20
Average
0
0 0.001 0.002 0.003 0.004 0.005
Strain
67
Chapter 3 Material Testing Program
120
100
Compressive stress (MPa)
80
60
40
20
Average
0
0 0.001 0.002 0.003 0.004 0.005
Strain
Figure 3.11 GFRP Compression coupons test setup and failure modes.
68
Chapter 3 Material Testing Program
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
69
Chapter 3 Material Testing Program
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
0.10
0.05 Average
0.00
0.00 0.02 0.04 0.06 0.08 0.10
Strain
70
Chapter 3 Material Testing Program
2.50
Average
2.00
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
71
Chapter 3 Material Testing Program
Hard foam
0.45 3
(64.08 kg/m )
0.40
0.35
Shear stress (MPa)
0.30
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
72
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
CONFIGURATIONS1
4.1 Introduction
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.
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
73
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
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
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.
Six sandwich panels (S1 to S6) with different rib configurations, as shown in Figure 4.1,
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
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
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
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
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
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
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
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
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
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
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.
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
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
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.
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
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
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
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
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
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
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,
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
79
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
compression skin at a transverse line at the edge of the flange of the transverse rib, where
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
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
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
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
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
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
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
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
increase. The fact that these ratios are significantly lower than 1.0, clearly emphasizes
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
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 =
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
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
b GFRP and t GFRP are the combined widths of the GFRP ribs and the height of the GFRP
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.
A comparison has been made between sandwich panels S1, S2 and S5, and single
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
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
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
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
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
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
Pu = Ultimate load
Mu = Ultimate moment
u = Deflection at ultimate load
u = Longitudinal strain at ultimate load
85
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
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
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
87
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
88
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
(a) Specimen S1
(c) Specimen S3
89
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
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)
90
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
(a) S1 (b) S2
(c) S3 (d) S4
(e) S5 (f) S6
91
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
0.003
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)
0.003
0.000
92
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
93
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
94
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
95
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
2.5 0.0
"a"Sharaf et al (2010)
on strength
S5 0.2
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)
96
Chapter 4 Experimental Study on Sandwich Panels with different Rib Configurations
12
10 S5
8 S2
Moment (kN.m)
4 S1
0
0 50 100 150 200 250 300 350 400
Curvature (1000/m)
97
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
98
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
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
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.
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
(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).
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
fact that the statically indeterminate (two-span) panel could have sustained further
100
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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),
101
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
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
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,
103
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
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.
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
locations. Figure 5.16 shows the locations of the strain gauges on both skins, namely the
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
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.
105
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
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
5.6.2 Test II Half Full Scale Panel Under Discrete Line Loads
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
presented later.
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
Figure 5.29 shows that at the outer support both edges moved slightly upwards, more
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
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
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
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
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
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
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
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
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.
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
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
111
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
112
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
Overlap zone
Centre line
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
113
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
Reaction beams
Side skirt
Air pressure
inlets W460 steel
section
Air pressure
inlets
114
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
115
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
116
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
LP#3 609 mm
LP#1 LP#4
LP#2
Tensile stresses Compressive stresses
Span 4112 mm
117
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
230
T7
50
195
T3 T6
230 215
T2 T5
210 50
T1 T4
Compressive stresses Tensile stresses
609 mm
50 609 mm
195
B3 B6
230 215
B2 B5
210 50
B1 B4
Tensile stresses Compressive stresses
Span 4112 mm
TT4 50
50 50
TT2 TT3
TT1 50
609 mm
609 mm
50
BT2
50
BT1
Tensile stresses Compressive stresses
Span 4112 mm
118
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
Steel bar 38 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
78 mm
75 mm
500 mm
Half Sandwich Panel Supporting Pad
Lab Floor
119
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
160 609 mm
609 mm
Span 4112 mm
130 385
Figure 5.16 Strain gauges distribution for top and bottom skins
120
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
2.5
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
Bottom Span Top Span
Pressure (kPa)
6
Span / 360
Top Span (LP#4)
5
First test
4
(Up to 2.7 kPa and
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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)
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
Figure 5.24 Skin wrinkling and crushing failure of the full scale panel in Test I
124
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
60
LP#10 LP#12
50 LP#11
40
Load (kN)
30
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
126
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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)
Figure 5.29 Deflection distribution over an axis passing through points 1-2-3 in
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)
Figure 5.30 Deflection distribution over an axis passing through points 4-5-6 in
Test II
127
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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)
Figure 5.31 Deflection distribution over an axis passing through points 7-8-9 in
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)
Figure 5.32 Deflection distribution over an axis passing through points 10-11-12 in
Test II
128
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
Chapter 5 Testing of Large Scale Cladding Sandwich Panels
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
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
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
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
136
Chapter 6 Finite Element Modeling of Sandwich Panels
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
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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.
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
bending and a uniform load. The uniform load was simulated by means of eight
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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.
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
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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.
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
To overcome these difficulties, a reduced size model (not scaled down) with
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
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
The polyurethane foam core was modeled using plane stress element PLANE82
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
necessary to have a fine enough mesh to obtain the appropriate solution, which is further
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.
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
142
Chapter 6 Finite Element Modeling of Sandwich Panels
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.
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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.
especially the polyurethane foam. As such, it was important to take material nonlinearity
146
Chapter 6 Finite Element Modeling of Sandwich Panels
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
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
where data files of all stress-strain curves are input to the program, and the fitting is then
carried out.
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
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.
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
154
1.6 78 300
Uniform
P6 1413 8.1 68.7 -0.0011 0.00175 Core shear crack C
(Cyclic)
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
Chapter 6 Finite Element Modeling of Sandwich Panels
155
Chapter 6 Finite Element Modeling of Sandwich Panels
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))
156
Chapter 6 Finite Element Modeling of Sandwich Panels
Figure 6.4 Finite element mesh used in the convergence study for the
2-D FEA
157
Chapter 6 Finite Element Modeling of Sandwich Panels
7.75
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
Figure 6.5 Convergence study results for the 2-D FE model (case of panels
tested under simulated uniform load Shawkat (2008))
(A)
Detail (B)
(B)
Node
Detail (A)
Figure 6.6 Final 2-D FEM mesh and loading configuration for
panels tested by Shawkat (2008)
158
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Contact element
159
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Chapter 6 Finite Element Modeling of Sandwich Panels
7.75
7.71
+ 4.61 % error
+ 4.14 % error
+ 4.07 % error
7.69
7.67
7.65
0 Coarse1 mesh 2 mesh
Intermediate Fine3mesh 4
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Chapter 6 Finite Element Modeling of Sandwich Panels
15.9
15.5
15.3
15.1
+ 4.98 % error
+ 8.83 % error
+ 4.52% error
14.9
14.7
14.5
14.3
0 Coarse1 mesh 2 mesh
Intermediate Fine3mesh 4
Figure 6.13 Convergence study results for the 3-D FE model for panels
tested in Chapter 4
Load
125 mm 125 mm
18 mm thick
rubber sheet
Figure 6.14 Final 3-D FEM mesh panels with different rib configurations
(Chapter 4)
163
Chapter 6 Finite Element Modeling of Sandwich Panels
1219 mm
Figure 6.15 FEA geometric models and assumptions of large scale cladding
panels (Chapter 5)
GFRP skin
Supporting pad
Figure 6.16 Final 3-D FEM for the large scale sandwich panels (Chapter5)
164
Chapter 6 Finite Element Modeling of Sandwich Panels
350
288.7 MPa
300
250
200
Stress (MPa)
100
50
Tension
0
Compression
-50 80.7 MPa
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
165
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Chapter 6 Finite Element Modeling of Sandwich Panels
3
Nonconference due to excessive
inward deformation (wrinkling)
2.5
Skin wrinkling
2
Load kN
1.5
4
Skin wrinkling
Load kN
Experimental Results
1 2D FEA
3D FEA
0
0 5 10 15 20 25 30 35
Deflection mm
Figure 6.22 Load-deflection response for panel (P2) with soft polyurethane
foam core and loaded in four-point bending
167
Chapter 6 Finite Element Modeling of Sandwich Panels
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
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
0.0
-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
Figure 6.26 Load-strain response for panel (P2) with soft polyurethane foam
core and loaded in four-point bending
169
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Chapter 6 Finite Element Modeling of Sandwich Panels
Units in mm
171
Chapter 6 Finite Element Modeling of Sandwich Panels
Units in mm
172
Chapter 6 Finite Element Modeling of Sandwich Panels
Units in mm
Figure 6.31 Deformed shape under uniform load for panels P3, P4 and P5
with soft polyurethane foam core
173
Chapter 6 Finite Element Modeling of Sandwich Panels
Units in mm
Figure 6.32 Deformed shape under uniform load for panels P7, P8 and P9
with hard polyurethane foam core
174
Chapter 6 Finite Element Modeling of Sandwich Panels
Units in MPa
Units in MPa
Figure 6.33 Shear failure modes within the soft and hard polyurethane foam
core under uniform loads
175
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Figure 6.36 Load-strain responses for panel S2
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
Figure 6.37 Load-strain responses for panel S3
177
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Figure 6.39 Load-strain responses for panel S5
178
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Figure 6.40 Load-strain responses for panel S6
C.L. C.L.
Mid-span
Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
179
Chapter 6 Finite Element Modeling of Sandwich Panels
C.L. C.L.
Mid-span
Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
C.L. C.L.
Mid-span
Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
180
Chapter 6 Finite Element Modeling of Sandwich Panels
C.L. C.L.
Mid-span
Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
181
Chapter 6 Finite Element Modeling of Sandwich Panels
C.L. C.L.
Mid-span
Mid-width
Outer edge
Support
C.L. C.L.
Mid-span Mid-width
460 mm
7 LP#1
2134 mm
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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)
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
Chapter 6 Finite Element Modeling of Sandwich Panels
10
9
T5 T4
8 T7
T3 T6
7
Pressure (kPa)
6 T7
T2
5
T1
Tensile stresses
4
T6 T5 T4
Compressive stresses
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
Compressive stresses
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
Chapter 6 Finite Element Modeling of Sandwich Panels
10
9
B2
8 B4 B6 B3
460 mm
2134 mm
6 B1
B5
Span 4112 mm
5
2438 mm
Compressive stresses
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
6
Span 4112 mm
5
2438 mm
4
Compressive stresses
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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
Chapter 6 Finite Element Modeling of Sandwich Panels
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)
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
Chapter 6 Finite Element Modeling of Sandwich Panels
Cut-off edge
(C.L.)
(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)
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
190
Chapter 6 Finite Element Modeling of Sandwich Panels
(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)
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
PANELS
7.1 Introduction
This chapter presents an independent nonlinear strain compatibility model for the
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
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
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
192
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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
make the simplified assumption that the normal strain field varies linearly over the
193
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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
As a result of using soft polyurethane foam core for the sandwich panels, the material
strain of almost 80%. Figures 3.14 and 3.15 in Chapter 3 show the tensile and
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
fitting technique is based on a cubic spline function concept (De Boor, 2001). The cubic
(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)
Si ( xi ) yi Si 1 ( xi ) ,
S 'i ( xi ) S 'i 1 ( xi ) ,
(7.4)
S ' 'i ( xi ) S ' 'i 1 ( xi ) , and
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:
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
195
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
7. Set I o = 1 ; o = z o = 0.
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
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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)
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
The aforementioned concepts and geometric relationships have been used to establish the
code was written in FORTRAN90. The program can deal with any material stress-strain
follows:
1. Input the sandwich panel dimensions, the overall panel thickness, skin
2. Divide the core into n numbers of layers (in this study it was shown that n = 20
3. Define the stress-strain relationships for both GFRP and polyurethane foam
4. Assume a strain value at the top surface of the sandwich panel, c , (Figure 7.4)
5. Assume a neutral axis depth from bottom surface, y bar (Figure 7.4).
200
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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.
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
201
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
14. Return to Step 4 and assume a new strain. Repeat this process until the
The load-deflection response of the sandwich panel consists of two components, namely
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
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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.
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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-
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
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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)
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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
206
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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 ,
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
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
were neglected. The general solution of that equation takes the following form:
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:
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
q
w
2k
2 D a D b (7.30)
208
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
q
Aa Ab (7.31)
2k
M
q
B a B b (7.32)
4 2
Q
q
Ca Cb (7.33)
4
where:
B x e x sin x (7.35)
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
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
209
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
represents the skin ends, the moments and shear force have to be calculated based on the
where:
2
M o' 1 Q1' 1 C l 2 M 1' 1 D l (7.43)
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
1 e l
1 (7.46)
2 sinh l sin l
1 e l
2 (7.47)
2 sinh l sin l
210
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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
211
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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).
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.
212
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
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
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
elasticity and c in the core Poissons ratio, taken as 0.33 (Gibson and Ashby, 1988).
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
The model developed has several significant features, namely, accounting for the
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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)
215
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
7.4.1 Sandwich Panels with different Core Densities and different Loadings
panels P1 to P9 with two different foam core densities and tested under various loading
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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.
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
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
Figures 7.24 to 7.30 show the experimental and analytical load-deflection responses for
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
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
219
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
Load
GFRP Top Skin Foam Core
C.L.
x N.A.
H Hc
ybar i=k Part 2
t
i=1 Part 1
y
220
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
C.L.
Span / 2
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
80
GFRP
N.A. 250 70
100 30
0 0
0.40
Foam Hard Foam density
2.00
Foam
0.35 (64.08 kg/m3)
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)
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
End Do loop
NO
IF Force_sum 0.0
Curvature
YES
Increase value of c by a
small increment
End Do loop
222
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
(qL2/8)
(qL2/8)
i
i
Approximated Curvature
Approximated Slope
yi
Figure 7.6 Schematic for the process of determining deflection due to flexure
223
Section l
Chapter 7
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
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
C.L.
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
226
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
Do x = 1 to
Layers_num
strain profile
End Do loop
End Do loop
Figure 7.10 Flow chart of procedure used to obtain the load-deflection response
227
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
3.5
3.0
2.5
Failure
2.0 P1 (excessive compression
Load (kN)
1.5
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
-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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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)
2.0
Experimental P2
1.0 Flexural deflection only (Analytical)
Shear deflection only (Analytical)
Winkler deflection only (Analytical)
Total deflection (Analytical)
0.0
0 5 10 15 20 25 30 35
Deflection (mm)
230
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
5
Load
0
-5
-10
Deflection (mm)
-15
-20
-25
-30
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
231
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
Flexural deflection only (Analytical)
1.0 Shear deflection only (Analytical)
Winkler deflection only (Analytical)
Total deflection (Analytical)
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)
8.0
Experimental P7
Experimental P8
Experimental P9
4.0 Flexural deflection only (Analytical)
Shear deflection only (Analytical)
Winkler deflection only (Analytical)
Total deflection (Analytical)
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
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
Distance from support (mm)
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
Distance from support (mm)
Figure 7.22 Deflection responses along the span of panels P7 to P9 (Shawkat, 2008)
233
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
8.0
Model
P5 P4
P4
7.0 Model
P5 P3
P3
6.0
5.0
Load (kN)
4.0
3.0
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
234
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
16
14 S1
12
Failure
10 (core shear crack)
Load (kN)
4
Experimental S1
Flexural deflection only (Analytical)
Shear deflection only (Analytical)
2 Winkler deflection only (Analytical)
Total deflection (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
Flexural deflection only (Analytical)
Shear deflection only (Analytical)
5
Winkler deflection only (Analytical)
Total deflection (Analytical)
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
Figure 7.26 Load-deflection response for panel S2 (Chapter 4)
235
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
35
30
25 S3
Failure
(top skin crushing)
Load (kN)
20
15
10
Experimental S3
Flexural deflection only (Analytical)
5 Shear deflection only (Analytical)
Winkler deflection only (Analytical)
Total deflection (Analytical)
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
Figure 7.27 Load-deflection response for panel S3 (Chapter 4)
30
25
20 Failure
S4 (top skin crushing)
Load (kN)
15
10
Experimental S4
Flexural deflection only (Analytical)
5 Shear deflection only (Analytical)
Winkler deflection only (Analytical)
Total deflection (Analytical)
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
Figure 7.28 Load-deflection response for panel S4 (Chapter 4)
236
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
40
S5
35
30
Failure
(top skin crushing)
25
Load (kN)
20
15
10 Experimental S5
Flexural deflection only (Analytical)
Shear deflection only (Analytical)
5 Winkler deflection only (Analytical)
Total deflection (Analytical)
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
Figure 7.29 Load-deflection response for panel S5 (Chapter 4)
40
S6
35
30
Failure
(top skin crushing)
25
Load (kN)
20
15
10 Experimental S6
Total deflection (Analytical)
Flexural deflection only (Analytical)
5 Shear deflection only (Analytical)
Winkler deflection only (Analytical)
0
0 10 20 30 40 50 60 70 80 90
Deflection (mm)
Figure 7.30 Load-deflection response for panel S6 (Chapter 4)
237
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
Figure 7.32 Load-strain responses for panel S2 (Chapter 4)
238
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
Figure 7.34 Load-strain responses for panel S4 (Chapter 4)
239
Chapter 7 Semi-Analytical Modeling of Sandwich Panels
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
Figure 7.35 Load-strain responses for panel S5 (Chapter 4)
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
Figure 7.36 Load-strain responses for panel S6 (Chapter 4)
240
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,
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
materials has been proposed and validated by Youssef et al (2005). The actual
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
Chapter 8 Parametric Study on Sandwich Panels
(rectangular prism cubes) and relates analytically the elasticity and yield stress to the
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
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.
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
Chapter 8 Parametric Study on Sandwich Panels
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
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
243
Chapter 8 Parametric Study on Sandwich Panels
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
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
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
Chapter 8 Parametric Study on Sandwich Panels
Ashby (1988) for the maximum compressive strain of the polyurethane foam which
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
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
Chapter 8 Parametric Study on Sandwich Panels
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
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
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
Chapter 8 Parametric Study on Sandwich Panels
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
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
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
247
Chapter 8 Parametric Study on Sandwich Panels
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
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
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
Chapter 8 Parametric Study on Sandwich Panels
polynomial the average of the two constants based on the two foam densities was used in
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.
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
Chapter 8 Parametric Study on Sandwich Panels
3 2 f
2
Gf f
s 1 s (8.16)
Es 8 s s
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
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
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
250
Chapter 8 Parametric Study on Sandwich Panels
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
3/2
pl f f
0.07445 3.6304 1 (8.20)
ys s s
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
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
shown in Figure 8.10, with the constants being m and D. In order to re-establish the
251
Chapter 8 Parametric Study on Sandwich Panels
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
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
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
Chapter 8 Parametric Study on Sandwich Panels
depth remained constant at 78 mm, which means that the core thickness reduces as the
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.
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,
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
Chapter 8 Parametric Study on Sandwich Panels
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
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
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).
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
Chapter 8 Parametric Study on Sandwich Panels
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
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
Chapter 8 Parametric Study on Sandwich Panels
GFRP material failure. Figure 8.44 shows the Tsai-Wu failure criteria index values at
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
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
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
256
Chapter 8 Parametric Study on Sandwich Panels
M1
3 3.2 M1t2 7.8 -2.5 0.293 11.85 47.25 11.30 CS
(32 kg/m )
M2
3 3.2 M2t2 26.4 230 0.806 207.23 62.61 -17.53 CS
(64 kg/m )
M3
3 3.2 M3t2 38.4 380 1.534 484.93 35.39 33.56 SC
(96 kg/m )
M4
3 3.2 M4t2 40.8 410 1.744 564.98 28.89 45.77 SC
(128 kg/m )
M5
3 3.2 M5t2 42.0 425 1.647 528.09 28.9 45.75 SC
(160 kg/m )
M6
3 3.2 M6t2 34.8 335 1.908 627.34 20.85 60.86 CS
(192 kg/m )
257
Chapter 8 Parametric Study on Sandwich Panels
Compressive Stress
Densification
c
Plc
Plateau (Elastic Buckling)
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
Chapter 8 Parametric Study on Sandwich Panels
64 kg/m3 polyurethane
foam (Predicted)
3
Stress (MPa)
64 kg/m3
2 polyurethane foam
(Experimental)
32 kg/m3
polyurethane foam
32 kg/m3 polyurethane
(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
Chapter 8 Parametric Study on Sandwich Panels
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
Chapter 8 Parametric Study on Sandwich Panels
0.45
0.40
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
Chapter 8 Parametric Study on Sandwich Panels
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
y = 0.1832x + 0.8466
1.6
1.4
32 kg/m3 polyurethane
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
Chapter 8 Parametric Study on Sandwich Panels
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
32 kg/m polyurethane
0.25 foam (Predicted)
0.20
0.15
3
32 kg/m polyurethane
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
Chapter 8 Parametric Study on Sandwich Panels
10.0
9.0
8.0
7.0
6.0
Load (kN)
5.0
4.0
3.0
Flexural deflection only (Analytical)
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
Flexural deflection only (Analytical)
2.0 Shear deflection only (Analytical)
0.0
0 10 20 30 40 50 60
Deflection (mm)
Figure 8.14 Load-deflection response of sandwich panel M1t2
264
Chapter 8 Parametric Study on Sandwich Panels
10.0
9.0
8.0
7.0
6.0
Load (kN)
5.0
4.0
3.0
Flexural deflection only (Analytical)
0.0
0 10 20 30 40 50 60
Deflection (mm)
Figure 8.15 Load-deflection response of sandwich panel M1t3
32.0
28.0
24.0
20.0
Load (kN)
16.0
12.0
0.0
0 10 20 30 40 50 60 70
Deflection (mm)
Figure 8.16 Load-deflection response of sandwich panel M2t1
265
Chapter 8 Parametric Study on Sandwich Panels
32.0
28.0
24.0
20.0
Load (kN)
16.0
12.0
0.0
0 10 20 30 40 50 60 70
Deflection (mm)
Figure 8.17 Load-deflection response of sandwich panel M2t2
32.0
28.0
24.0
20.0
Load (kN)
16.0
12.0
0.0
0 10 20 30 40 50 60 70
Deflection (mm)
Figure 8.18 Load-deflection response of sandwich panel M2t3
266
Chapter 8 Parametric Study on Sandwich Panels
60.0
50.0
40.0
Load (kN)
30.0
20.0
Flexural deflection only (Analytical)
0.0
0 5 10 15 20 25 30 35 40 45
Deflection (mm)
Figure 8.19 Load-deflection response of sandwich panel M3t1
60.0
50.0
40.0
Load (kN)
30.0
20.0
Flexural deflection only (Analytical)
0.0
0 5 10 15 20 25 30 35 40 45
Deflection (mm)
Figure 8.20 Load-deflection response of sandwich panel M3t2
267
Chapter 8 Parametric Study on Sandwich Panels
60.0
50.0
40.0
Load (kN)
30.0
20.0
Flexural deflection only (Analytical)
0.0
0 5 10 15 20 25 30 35 40 45
Deflection (mm)
Figure 8.21 Load-deflection response of sandwich panel M3t3
60.0
50.0
40.0
Load (kN)
30.0
20.0
Flexural deflection only (Analytical)
0.0
0 5 10 15 20 25 30 35 40
Deflection (mm)
Figure 8.22 Load-deflection response of sandwich panel M4t1
268
Chapter 8 Parametric Study on Sandwich Panels
60.0
50.0
40.0
Load (kN)
30.0
20.0
Flexural deflection only (Analytical)
0.0
0 5 10 15 20 25 30 35 40
Deflection (mm)
Figure 8.23 Load-deflection response of sandwich panel M4t2
60.0
50.0
40.0
Load (kN)
30.0
20.0
Flexural deflection only (Analytical)
0.0
0 5 10 15 20 25 30 35 40
Deflection (mm)
Figure 8.24 Load-deflection response of sandwich panel M4t3
269
Chapter 8 Parametric Study on Sandwich Panels
50.0
40.0
30.0
Load (kN)
20.0
0.0
0 5 10 15 20 25 30
Deflection (mm)
Figure 8.25 Load-deflection response of sandwich panel M5t1
50.0
40.0
30.0
Load (kN)
20.0
0.0
0 5 10 15 20 25 30
Deflection (mm)
Figure 8.26 Load-deflection response of sandwich panel M5t2
270
Chapter 8 Parametric Study on Sandwich Panels
50.0
40.0
30.0
Load (kN)
20.0
0.0
0 5 10 15 20 25 30
Deflection (mm)
Figure 8.27 Load-deflection response of sandwich panel M5t3
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)
Figure 8.28 Load-deflection response of sandwich panel M6t1
271
Chapter 8 Parametric Study on Sandwich Panels
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)
Figure 8.29 Load-deflection response of sandwich panel M6t2
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)
Figure 8.30 Load-deflection response of sandwich panel M6t3
272
Chapter 8 Parametric Study on Sandwich Panels
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)
Figure 8.32 Effect of skin thickness on load-deflection response of sandwich panels
with core density of 64 kg/m3
273
Chapter 8 Parametric Study on Sandwich Panels
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)
Figure 8.33 Effect of skin thickness on load-deflection response of sandwich panels
with core density of 96 kg/m3
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)
Figure 8.34 Effect of skin thickness on load-deflection response of sandwich panels
with core density of 128 kg/m3
274
Chapter 8 Parametric Study on Sandwich Panels
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)
Figure 8.35 Effect of skin thickness on load-deflection response of sandwich panels
with core density of 160 kg/m3
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)
Figure 8.36 Effect of skin thickness on load-deflection response of sandwich panels
with core density of 196 kg/m3
275
Chapter 8 Parametric Study on Sandwich Panels
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)
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)
276
Chapter 8 Parametric Study on Sandwich Panels
25
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
Figure 8.39 Effect of core density on load-deflection response of sandwich panels
with skin thickness = 3.2 mm
277
Chapter 8 Parametric Study on Sandwich Panels
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
Figure 8.40 Effect of core density on load-deflection response of sandwich panels
with skin thickness = 4.8 mm
278
Chapter 8 Parametric Study on Sandwich Panels
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 )
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 )
279
Chapter 8 Parametric Study on Sandwich Panels
685 mm
100 1150 mm 1150 mm 100
C.L. C.L.
Mid-span
Mid-width
C.L. C.L.
Mid-width Outer edge
Mid-span
Support
280
Chapter 8 Parametric Study on Sandwich Panels
C.L. C.L.
Mid-span
Mid-width
C.L. C.L.
Outer edge
Mid-span Mid-width
Support
281
Chapter 8 Parametric Study on Sandwich Panels
C.L. C.L.
Mid-span
Mid-width
C.L. C.L.
Outer edge
Mid-span Mid-width
Support
C.L. C.L.
Outer edge
Mid-span Mid-width
Support
282
Chapter 8 Parametric Study on Sandwich Panels
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
283
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
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
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
284
Chapter 9 Summary and Conclusions
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
envisioned to be the actual building cladding system used in the field and were tested
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
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
9.2 Conclusions
1. By adding GFRP ribs, flexural strength and stiffness of the sandwich panels have
285
Chapter 9 Summary and Conclusions
panel without any ribs. However, the deflection at ultimate was not much affected.
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
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
(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
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
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,
286
Chapter 9 Summary and Conclusions
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
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
(a) First failure occurred by outwards local buckling (wrinkling) of the GFRP skin in
(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.
(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
287
Chapter 9 Summary and Conclusions
1. The two independent models, namely the FE and analytical models provided very
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
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
4. Both the 2-D and 3-D FE models provided very good predictions, with insignificant
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
6. The models have a limitation in that they can not predict the post-peak load behaviour
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
Chapter 9 Summary and Conclusions
2. As the core density increased from 32 to 192 kg/m3 the contribution of shear to the
3. For a very low density core (32 kg/m3), increasing the skin thickness in panels without
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
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
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.
This study provided an insight into the behaviour of sandwich panels, a rather uncommon
289
Chapter 9 Summary and Conclusions
behaviour, failure modes, and analytical and numerical modeling. To further promote
structural engineering applications of this system, the following areas need further
investigations:
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
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
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.
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
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
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.
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
12. Berthelot, J, and Lolive, E. (2002), Non-linear Behaviour of Foam Cores and
13. Boni, T. L., and Mller, S. F. (2008a), Laterally supported sandwich panels
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
15. Burton WS, Noor AK.(1996), Assessment of continuum models for sandwich panel
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
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
292
References
447453.
21. Fan, H.L., Meng, F.H., and Yang, W. (2007), Sandwich panels with Kagome lattice
22. Filon L. N. G. (1937), On antiplan stress in an elastic solid. Proc. Roy. Soc. A,
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
25. Frostig, Y. and Baruch, M. (1990), Bending of sandwich beams with transversely
27. Frostig, Y. (1993a), High-order behaviour of sandwich beams with flexible core and
28. Frostig, Y. (1993b), On stress concentration in the bending of sandwich beams with
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
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.
unidirectional sandwich panels with partial contact and a transversely flexible core.
34. Frostig, Y. and Simitses, G.J. (2004), Similitude of sandwich panels with a soft
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),
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
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.,
39. Frostig, Y., Thomsen, O.T., and Sheinman, I. (2005), On the non-linear high-order
294
References
andmatrix approach, 3rd Ed., Chapman and Hall, London and New York.
41. Gibson LJ. (1989), Modelling the Mechanical Behavior of Cellular Materials
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
45. He, M., and Hu, W. (2007), A study on composite honeycomb sandwich panel
46. Heimbs, S., Schmeer, S., Middendorf, P., and Maier, M. (2007), Strain rate effects
47. Hetenyi, M.(1946), Beams on Elastic Foundation, Theory with Applications. Ann
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
51. Kant, T., and Patil, H. S. (1991), Buckling load of sandwich columns with a higher
52. Ko, W. L., and Jackson, R. H. (1991), Combined Compressive and Shear Buckling
Memorandum 4290.
53. Kocher, C., Watson W., M. Gomez, M., Gonzalez I., and Birman, V. (2002),
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
55. Lascoup, B., Aboura, Z., Khellil, K., and Benzeggagh, M. (2006), On the
56. Libove, C., and Hubka, R. E. (1951), Elastic constants for corrugated-core sandwich
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
58. Liu, T., Deng Z. C., and Lu T. J. (2006), Design optimization of truss-cored
296
References
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.
61. Marasco, A. I., Cartie, D. D. R., Partridge, I. K., and Rezai, A. (2006), Mechanical
62. Markaki, AE, Clyne TW. (2003a), Mechanics of thin ultra-light stainless steel
63. Markaki, AE, Clyne TW. (2003b), Mechanics of Thin Ultra-Light Stainless Steel
pp. 1351-1357.
64. Meraghni, F., Desrumaux, F., and Benzeggagh, M.L. (1999), Mechanical behaviour
779.
65. Meyer-Piening, H.-R. (1989), Remarks on higher order sandwich stress and
297
References
67. Niu, K., and Talreja, R. (1999), Modeling of wrinkling in sandwich panels under
68. Noor AK, Burton WS, and Bert CW.(1996), Computational models for sandwich
70. Ojalvo, I. V. (1977), Departures from elastic beam theory in laminated sandwich
71. Pamla, V. (2007), the pioneer phase of building with glass-fibre reinforced plastics
Faculty, Germany.
72. Payder, N., and Libove, C. (1988), Bending of sandwich plates of variable
reinforced face plates. Thesis presented to the University of Bristol, at Bristol, U.K.,
74. Peled, D. and Frostig. Y. (1994), High-order bending of sandwich beams with
sandwich beams with a tapered transition zone and a transversely flexible core.
76. Petras, A., and Sutcliffe, M.P.F. (2000), Indentation failure analysis of sandwich
298
References
77. Petras, A., and Sutcliffe. M.P.F. (1999), Indentation resistance of sandwich beams.
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
589611.
80. Reis, E. M., and Rizkalla, S. H. (2007), Material characteristics of 3-D FRP
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
575586.
85. Saint-Michela, F., Chazeaua, L, Cavaill, JY, and Chabert, E., (2006), Mechanical
86. Schwarts-Givil H., Rabinovitch, O., and Frostig, Y. (2007), High-order nonlinear
299
References
88. Senthilnathan, N. R., Lim, S. P., Lee, K. H., and Chow, S. T. (1988), Vibration of
89. Shanley, F.R. (1957), Strength of Materials. McGraw-Hill book company, Inc.
90. Shawkat, W. (2008), Hybrid Members Employing FRP Skins Reinforcement for
CANADA.
92. Sleight, D. W., and Wang, J. T. (1995), Buckling Analysis of Debonded Sandwich
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
95. Sokolinsky, V., and Frostig, Y. (2000), Branching behaviour in the nonlinear
300
References
96. Sokolinsky, V., Frostig, Y., and Nutt, S. R. (2002), Special behaviour of
unidirectional sandwich panels with transversely flexible core under statical loading.
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:
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.
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.
301
References
(SNAME).
103. Youssef S, Maire E, Gaertner R. (2005), Finite element modelling of the actual
(3):719.
104. Zhou, D, and Stronge W.J. (2005), Mechanical properties of fibrous core sandwich
105. Zhu HX, Mills NJ, Knott JF. (1997), Analysis of high strain compression of open-
106. Zhu HX, Mills NJ, Knott JF. (1998), The in-plane non-linear compression of
302
Appendix
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
303
Appendix
Uniform load
304
Appendix
No ribs Rib
Ribs Ribs
St 0.5B f 0.5B f St B fe St
Wte
H ASt Wt ASt
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
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'
ELSEIF(Loads_num == 3)THEN
WRITE (*,*) 'You Entered >>> Six Loads'
ELSEIF(Loads_num == 4)THEN
WRITE (*,*) 'You Entered >>> Eight Loads'
ELSEIF(Loads_num == 5)THEN
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
CLOSE (229)
CLOSE (2210)
CLOSE (2211)
CLOSE (2212)
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'
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)
308
Appendix
ENDIF
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)
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))
Wstax (x, w) = Wstax (x, w) + ko (x, w)*Wdax (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 (*,*) '>>>Load Deflection Calculations For Load: W =',(Load/LS_num)* w,'N'
WRITE (*,*) ' 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
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
310
Appendix
ENDIF
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)
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))
Wstax (x, w) = Wstax (x, w) + ko (x, w)*Wdax (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 (*,*) '>>>Start Non-Geometry Calculations for Load: W=',(Load/LS_num)* w ,'N'
311
Appendix
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
ENDDO
Load_Dist= Span/3
DO PP = 1,2
P = ((Load/LS_num)*w)/2
312
Appendix
ENDIF
Load_Dist = Load_Dist + (Span/3)
ENDDO
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)
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))
Wstax (x, w) = Wstax (x, w) + ko (x, w)*Wdax (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
IF (((x-1) * Slice) <= (Span / 3))THEN
Max (x, w) = (((Load/LS_num)* w)/2) * ((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 )
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)
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 )
ENDIF
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'
WRITE (99,*) '>>>Deflection due to Moment =',Dm (w),'mm <<<'
WRITE (99,*) ' 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)
313
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 (*,*) '>>>Load Deflection Calculations For Load: W =',(Load/LS_num)* w,'N'
WRITE (*,*) ' Deflection Calculations due to Moment started'
WRITE (99,*) ' '
WRITE (99,*) '>>>Load Deflection Calculations For Load: W =',(Load/LS_num)* w,'N'
WRITE (99,*) ' Deflection Calculations due to Moment started'
IF(Rc == 1)THEN
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=0
Q_A=0
M_B=0
Q_B=0
Load_Dist= Span/3
DO PP = 1,2
P = ((Load/LS_num)*w)/2
ENDDO
314
Appendix
Load_Dist= Span/3
DO PP = 1,2
P = ((Load/LS_num)*w)/2
ENDIF
Load_Dist = Load_Dist + (Span/3)
ENDDO
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)
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))
Wstax (x, w) = Wstax (x, w) + ko (x, w)*Wdax (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 (*,*) '>>>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))
315
Appendix
316
Appendix
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
q = (((Load/LS_num)*w)/4)/(span/11)
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)
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 )
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)
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 )
ELSEIF (((x-1) * Slice) > (15*Span / 88) .and. ((x-1) * Slice) <= (29*Span / 88))THEN
Max (x, w) = q * ((((x-1) * Slice)*Span/11) + ((Span**2)/88))
Vax (x, w) = q * Span /11
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 )
ELSEIF (((x-1) * Slice) > (29*Span / 88) .and. ((x-1) * Slice) <= (37*Span / 88))THEN
Max (x, w) = q * ((-0.5*((x-1) * Slice)**2) + (37*Span/88)*((x-1) * Slice) - (665*(Span**2)/15488))
Vax (x, w) = q * ((37*Span/88) - ((x-1) * Slice))
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 )
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)
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 )
ENDIF
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'
WRITE (99,*) '>>>Deflection due to Moment =',Dm (w),'mm <<<'
WRITE (99,*) ' 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)
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 (*,*) '>>>Load Deflection Calculations For Load: W =',(Load/LS_num)* w,'N'
WRITE (*,*) ' Deflection Calculations due to Moment started'
WRITE (99,*) ' '
WRITE (99,*) '>>>Load Deflection Calculations For Load: W =',(Load/LS_num)* w,'N'
WRITE (99,*) ' Deflection Calculations due to Moment started'
IF(Rc == 1)THEN
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
Point_Dist = (x-1) * Slice
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
Load_Dist= 7*Span/88
DO PP = 1,4
q = (((Load/LS_num)*w)/4)/(span/11)
Beta(x, w) = (k(x, w)/(4*E_Skin*I_Skin))**(0.25)
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,
w)*(Load_Dist+Span/11)))
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-
Load_Dist-Span/11)))
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))
Q_A(x, w) = Q_A(x, w) + (q/(4*Beta(x, w))) * (Cbx_a(x, w) - Cbx_b(x, w))
Load_Dist = Load_Dist + (Span/4)
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))
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))))
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)
Load_Dist= 7*Span/88
DO PP = 1,4
q = (((Load/LS_num)*w)/4)/(span/11)
IF(Point_Dist < Load_Dist) THEN
Beta(x, w) =(k(x, w)/(4*E_Skin*I_Skin))**(0.25)
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))
Wstax (x, w) = Wstax (x, w) + ko (x, w)*Wdax (x, w)
ELSEIF(Point_Dist > Load_Dist .and. Point_Dist < (Load_Dist+Span/11)) THEN
Beta(x, w) =(k(x, w)/(4*E_Skin*I_Skin))**(0.25)
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,
w)*((Span/11)+Load_Dist-Point_Dist)))
Dbx_b(x, w) = exp(-Beta(x, w)*((Span/11)+Load_Dist-Point_Dist))* cos(Beta(x, w)*((Span/11)+Load_Dist-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))
Wstax (x, w) = Wstax (x, w) + ko (x, w)*Wdax (x, w)
ELSEIF(Point_Dist > (Load_Dist+Span/11)) THEN
Beta(x, w) =(k(x, w)/(4*E_Skin*I_Skin))**(0.25)
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))
319
Appendix
320
Appendix
321
Appendix
DO x = 1, S_num
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
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)
322
Appendix
Xval_k = 0.0
Yval_k = 0.0
DO w = 2, LS_num, 1
WRITE (*,*) ' '
WRITE (*,*) '>>>Load Deflection Calculations For Load: W =',(Load/LS_num)* w,'N'
WRITE (*,*) ' Deflection Calculations due to Moment started'
WRITE (99,*) ' '
WRITE (99,*) '>>>Load Deflection Calculations For Load: W =',(Load/LS_num)* w,'N'
WRITE (99,*) ' Deflection Calculations due to Moment started'
IF(Rc == 1)THEN
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
Point_Dist = (x-1) * Slice
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
P = ((Load/LS_num)*w)/8
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)
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/8)
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))
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))))
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)
Load_Dist= Span/16
DO PP = 1,8
P = ((Load/LS_num)*w)/8
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) = 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))
Wstax (x, w) = Wstax (x, w) + ko (x, w)*Wdax (x, w)
ELSEIF((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))
323
Appendix
324
Appendix
325
Appendix
Wsax (x, w) = Wsax (x, w) + ((PoB(x, w)/2)*Dbx(x, w)) - ((MoB(x, w)*Beta(x, w)/2)*Abx(x, w))
Wstax (x, w) = Wstax (x, w) + ko (x, w)*Wdax (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/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)
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'
WRITE (99,*) '>>>Deflection due to Moment =',Dm (w),'mm <<<'
WRITE (99,*) ' 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)
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 (*,*) '>>>Load Deflection Calculations For Load: W =',(Load/LS_num)* w,'N'
WRITE (*,*) ' Deflection Calculations due to Moment started'
WRITE (99,*) ' '
WRITE (99,*) '>>>Load Deflection Calculations For Load: W =',(Load/LS_num)* w,'N'
WRITE (99,*) ' Deflection Calculations due to Moment started'
IF(Rc == 1)THEN
P = ((Load/LS_num)* w)/Span
Load_Dist = 0.5 * Span
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
Beta(x, w) = (k(x, w)/(4*E_Skin*I_Skin))**(0.25)
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))
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))))
326
Appendix
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
ENDSUBROUTINE GetValue
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 ), INTENT(INOUT) :: Ibcbeg, Ibcend
INTEGER ( KIND = 4 ) :: II
328
Appendix
ELSE
CALL Spline4 ( N_num, a, Ypp, Ypp )
ENDIF
ENDSUBROUTINE Spline2
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
DO II = 1, N_num
IF ( a(2,II) == 0.0D+00 ) THEN
WRITE ( *, '(a)' ) ' '
WRITE ( *, '(a)' ) 'Spline4 - Fatal error!'
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
ENDSUBROUTINE Spline4
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
ENDSUBROUTINE Spline5
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
Lt = (Height-(2*St)) / L_num
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')
330
Appendix
331
Appendix
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
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')
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
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')
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
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')
READ (9,*) X_M
READ (10,*) Y_M_t
READ (11,*) Y_M_c
READ (12,*) Y_M_cur
ELSEIF(Rc == 4)THEN
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')
READ (229,*) X_M
READ (2210,*) Y_M_t
READ (2211,*) Y_M_c
READ (2212,*) Y_M_cur
ENDIF
ELSEIF((x-1)*Slice >= (0.5*Span) - 0.5*Bf)THEN
IF(Rc == 1.or.Rc == 2.or.Rc == 5)THEN
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')
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
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')
READ (229,*) X_M
READ (2210,*) Y_M_t
332
Appendix
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)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
A_s_s = -Yval
Xval = -Csax (x, w)
CALL GetValue (Nfc, X_F_c, Y_F_c, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_c, Xval, Yval, Ypval, Yppval)
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)
333
Appendix
CLOSE (11)
CLOSE (12)
CLOSE (229)
CLOSE (2210)
CLOSE (2211)
CLOSE (2212)
60 RETURN
ENDSUBROUTINE Ten_Comp_Curv
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)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
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)
CALL GetValue (Nft, X_F_t, Y_F_t, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_t, Xval, Yval, Ypval, Yppval)
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)
CALL GetValue (Nfc, X_F_c, Y_F_c, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_c, Xval, Yval, Ypval, Yppval)
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)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
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))
Xval = Sahyax (1, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
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) / E_Skin
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
Sahyax (2, x, w) = Tsax (x, w) * ((Yax (x, w)-(St+0.5*ASt))/Yax (x, w))
Xval = Sahyax (2, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
334
Appendix
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))
Xval = Sahyax (1, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
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) / E_Skin
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
Sahyax (2, x, w) = Tsax (x, w)* ((Yax (x, w)-(St+0.5*ASt))/Yax (x, w))
Xval = Sahyax (2, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (2, x, w) = Yval
Enahyax (2, x, w) = Abs(Sahyax (2, x, w) / Sahyax (2, x, w))
Neahyax (2, x, w) = Enahyax (2, x, w) / E_Skin
Bnahyax (2, x, w) = Neahyax (2, x, w) * B
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
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))
Xval = Sahyax (1, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
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) / E_Skin
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
Sahyax (2, x, w) = Tsax (x, w)* ((Yax (x, w)-(St+0.5*ASt))/Yax (x, w))
Xval = Sahyax (2, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (2, x, w) = Yval
Enahyax (2, x, w) = Abs(Sahyax (2, x, w) / Sahyax (2, x, w))
Neahyax (2, x, w) = Enahyax (2, x, w) / E_Skin
Bnahyax (2, x, w) = Neahyax (2, x, w) * Bf
Xval = Sahyax (2, x, w)
CALL GetValue (Nft, X_F_t, Y_F_t, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_t, Xval, Yval, Ypval, Yppval)
Sahyax (2, x, w) = Yval
Enahyax (2, x, w) = Abs(Sahyax (2, x, w) / Sahyax (2, x, w))
Neahyax (2, x, w) = Enahyax (2, x, w) / E_Skin
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
ENDIF
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))
Xval = Sahyax (I, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
Bnahyax (I, x, w) = Neahyax (I, x, w) * Wt
Xval = Sahyax (I, x, w)
CALL GetValue (Nft, X_F_t, Y_F_t, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_t, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
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))
Xval = -Sahyax (I, x, w)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = -Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
Bnahyax (I, x, w) = Neahyax (I, x, w) * Wt
Xval = -Sahyax (I, x, w)
CALL GetValue (Nfc, X_F_c, Y_F_c, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_c, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = -Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
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
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))
Xval = -Sahyax (L_num+4, x, w)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+4, x, w) = -Yval
Enahyax (L_num+4, x, w) = Abs(Sahyax (L_num+4, x, w) / Sahyax (L_num+4, x, w))
Neahyax (L_num+4, x, w) = Enahyax (L_num+4, x, w) / E_Skin
Bnahyax (L_num+4, x, w) = Neahyax (L_num+4, x, w) * B
Qahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * Abs(Yax (x, w) - (Height-0.5*St))
Iahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * (Yax (x, w) - (Height-0.5*St))**2
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)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+3, x, w) = Yval
Enahyax (L_num+3, x, w) = Abs(Sahyax (L_num+3, x, w) / Sahyax (L_num+3, x, w))
Neahyax (L_num+3, x, w) = Enahyax (L_num+3, x, w) / E_Skin
Bnahyax (L_num+3, x, w) = Neahyax (L_num+3, x, w) * B
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((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))
Xval = -Sahyax (L_num+4, x, w)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+4, x, w) = -Yval
Enahyax (L_num+4, x, w) = Abs(Sahyax (L_num+4, x, w) / Sahyax (L_num+4, x, w))
Neahyax (L_num+4, x, w) = Enahyax (L_num+4, x, w) / E_Skin
Bnahyax (L_num+4, x, w) = Neahyax (L_num+4, x, w) * B
Qahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * Abs(Yax (x, w) - (Height-0.5*St))
Iahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * (Yax (x, w) - (Height-0.5*St))**2
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)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+3, x, w) = Yval
336
Appendix
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))
Xval = Sahyax (1, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
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) / E_Skin
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
Sahyax (2, x, w) = Tsax (x, w) * ((Yax (x, w)-(St+0.5*ASt))/Yax (x, w))
Xval = Sahyax (2, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (2, x, w) = Yval
Enahyax (2, x, w) = Abs(Sahyax (2, x, w) / Sahyax (2, x, w))
Neahyax (2, x, w) = Enahyax (2, x, w) / E_Skin
Bnahyax (2, x, w) = Neahyax (2, x, w) * 2*Bfe
Xval = Sahyax (2, x, w)
CALL GetValue (Nft, X_F_t, Y_F_t, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_t, Xval, Yval, Ypval, Yppval)
Sahyax (2, x, w) = Yval
Enahyax (2, x, w) = Abs(Sahyax (2, x, w) / Sahyax (2, x, w))
Neahyax (2, x, w) = Enahyax (2, x, w) / E_Skin
Bnahyax (2, x, w) = Bnahyax (2, x, w) + Neahyax (2, x, w) * (B-2*Bfe)
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
ELSEIF((x-1)*Slice < (Bfe))THEN
Sahyax (1, x, w) = Tsax (x, w) * ((Yax (x, w)-((0.5)*St))/Yax (x, w))
Xval = Sahyax (1, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
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) / E_Skin
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
Sahyax (2, x, w) = Tsax (x, w) * ((Yax (x, w)-(St+0.5*ASt))/Yax (x, w))
Xval = Sahyax (2, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (2, x, w) = Yval
Enahyax (2, x, w) = Abs(Sahyax (2, x, w) / Sahyax (2, x, w))
Neahyax (2, x, w) = Enahyax (2, x, w) / E_Skin
Bnahyax (2, x, w) = Neahyax (2, x, w) * B
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
ENDIF
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))
Xval = Sahyax (I, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
Bnahyax (I, x, w) = Neahyax (I, x, w) * 2*Wte
Xval = Sahyax (I, x, w)
CALL GetValue (Nft, X_F_t, Y_F_t, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_t, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
Bnahyax (I, x, w) = Bnahyax (I, x, w) + Neahyax (I, x, w) * (B-2*Wte)
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))
Xval = -Sahyax (I, x, w)
CALL GetValue (Nfc, X_F_c, Y_F_c, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_c, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = -Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
Bnahyax (I, x, w) = Neahyax (I, x, w) * 2*Wte
Xval = -Sahyax (I, x, w)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = -Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
Bnahyax (I, x, w) = Bnahyax (I, x, w) + Neahyax (I, x, w) * (B-2*Wte)
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
ENDIF
337
Appendix
ENDDO
IF((x-1)*Slice >= (Bfe))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))
Xval = -Sahyax (L_num+4, x, w)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+4, x, w) = -Yval
Enahyax (L_num+4, x, w) = Abs(Sahyax (L_num+4, x, w) / Sahyax (L_num+4, x, w))
Neahyax (L_num+4, x, w) = Enahyax (L_num+4, x, w) / E_Skin
Bnahyax (L_num+4, x, w) = Neahyax (L_num+4, x, w) * B
Qahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * Abs(Yax (x, w) - (Height-0.5*St))
Iahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * (Yax (x, w) - (Height-0.5*St))**2
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)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+3, x, w) = Yval
Enahyax (L_num+3, x, w) = Abs(Sahyax (L_num+3, x, w) / Sahyax (L_num+3, x, w))
Neahyax (L_num+3, x, w) = Enahyax (L_num+3, x, w) / E_Skin
Bnahyax (L_num+3, x, w) = Neahyax (L_num+3, x, w) * 2*Bfe
Xval = Sahyax (L_num+3, x, w)
CALL GetValue (Nfc, X_F_c, Y_F_c, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_c, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+3, x, w) = Yval
Enahyax (L_num+3, x, w) = Abs(Sahyax (L_num+3, x, w) / Sahyax (L_num+3, x, w))
Neahyax (L_num+3, x, w) = Enahyax (L_num+3, x, w) / E_Skin
Bnahyax (L_num+3, x, w) = Bnahyax (L_num+3, x, w) + Neahyax (L_num+3, x, w) * (B-2*Bfe)
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((x-1)*Slice < (Bfe))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))
Xval = -Sahyax (L_num+4, x, w)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+4, x, w) = -Yval
Enahyax (L_num+4, x, w) = Abs(Sahyax (L_num+4, x, w) / Sahyax (L_num+4, x, w))
Neahyax (L_num+4, x, w) = Enahyax (L_num+4, x, w) / E_Skin
Bnahyax (L_num+4, x, w) = Neahyax (L_num+4, x, w) * B
Qahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * Abs(Yax (x, w) - (Height-0.5*St))
Iahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * (Yax (x, w) - (Height-0.5*St))**2
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)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+3, x, w) = Yval
Enahyax (L_num+3, x, w) = Abs(Sahyax (L_num+3, x, w) / Sahyax (L_num+3, x, w))
Neahyax (L_num+3, x, w) = Enahyax (L_num+3, x, w) / E_Skin
Bnahyax (L_num+3, x, w) = Neahyax (L_num+3, x, w) * B
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
ENDIF
ELSEIF(Rc == 5)THEN
IF((x-1)*Slice >= (Bfe))THEN
Sahyax (1, x, w) = Tsax (x, w) * ((Yax (x, w)-((0.5)*St))/Yax (x, w))
Xval = Sahyax (1, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
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) / E_Skin
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
Sahyax (2, x, w) = Tsax (x, w) * ((Yax (x, w)-(St+0.5*ASt))/Yax (x, w))
Xval = Sahyax (2, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (2, x, w) = Yval
Enahyax (2, x, w) = Abs(Sahyax (2, x, w) / Sahyax (2, x, w))
Neahyax (2, x, w) = Enahyax (2, x, w) / E_Skin
Bnahyax (2, x, w) = Neahyax (2, x, w) * (Bf + 2*Bfe)
Xval = Sahyax (2, x, w)
CALL GetValue (Nft, X_F_t, Y_F_t, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_t, Xval, Yval, Ypval, Yppval)
Sahyax (2, x, w) = Yval
Enahyax (2, x, w) = Abs(Sahyax (2, x, w) / Sahyax (2, x, w))
Neahyax (2, x, w) = Enahyax (2, x, w) / E_Skin
Bnahyax (2, x, w) = Bnahyax (2, x, w) + Neahyax (2, x, w) * (B - Bf - 2*Bfe)
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
ELSEIF((x-1)*Slice < (Bfe))THEN
Sahyax (1, x, w) = Tsax (x, w) * ((Yax (x, w)-((0.5)*St))/Yax (x, w))
Xval = Sahyax (1, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
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) / E_Skin
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
Sahyax (2, x, w) = Tsax (x, w) * ((Yax (x, w)-(St+0.5*ASt))/Yax (x, w))
Xval = Sahyax (2, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (2, x, w) = Yval
Enahyax (2, x, w) = Abs(Sahyax (2, x, w) / Sahyax (2, x, w))
Neahyax (2, x, w) = Enahyax (2, x, w) / E_Skin
Bnahyax (2, x, w) = Neahyax (2, x, w) * B
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
ENDIF
338
Appendix
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))
Xval = Sahyax (I, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
Bnahyax (I, x, w) = Neahyax (I, x, w) * (Wt + 2*Wte)
Xval = Sahyax (I, x, w)
CALL GetValue (Nft, X_F_t, Y_F_t, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_t, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
Bnahyax (I, x, w) = Bnahyax (I, x, w) + Neahyax (I, x, w) * (B - Wt - 2*Wte)
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))
Xval = -Sahyax (I, x, w)
CALL GetValue (Nfc, X_F_c, Y_F_c, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_c, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = -Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
Bnahyax (I, x, w) = Neahyax (I, x, w) * (Wt + 2*Wte)
Xval = -Sahyax (I, x, w)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (I, x, w) = -Yval
Enahyax (I, x, w) = Abs(Sahyax (I, x, w) / Sahyax (I, x, w))
Neahyax (I, x, w) = Enahyax (I, x, w) / E_Skin
Bnahyax (I, x, w) = Bnahyax (I, x, w) + Neahyax (I, x, w) * (B - Wt - 2*Wte)
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
ENDIF
ENDDO
IF((x-1)*Slice >= (Bfe))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))
Xval = -Sahyax (L_num+4, x, w)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+4, x, w) = -Yval
Enahyax (L_num+4, x, w) = Abs(Sahyax (L_num+4, x, w) / Sahyax (L_num+4, x, w))
Neahyax (L_num+4, x, w) = Enahyax (L_num+4, x, w) / E_Skin
Bnahyax (L_num+4, x, w) = Neahyax (L_num+4, x, w) * B
Qahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * Abs(Yax (x, w) - (Height-0.5*St))
Iahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * (Yax (x, w) - (Height-0.5*St))**2
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)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+3, x, w) = Yval
Enahyax (L_num+3, x, w) = Abs(Sahyax (L_num+3, x, w) / Sahyax (L_num+3, x, w))
Neahyax (L_num+3, x, w) = Enahyax (L_num+3, x, w) / E_Skin
Bnahyax (L_num+3, x, w) = Neahyax (L_num+3, x, w) * (Bf + 2*Bfe)
Xval = Sahyax (L_num+3, x, w)
CALL GetValue (Nfc, X_F_c, Y_F_c, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_F_c, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+3, x, w) = Yval
Enahyax (L_num+3, x, w) = Abs(Sahyax (L_num+3, x, w) / Sahyax (L_num+3, x, w))
Neahyax (L_num+3, x, w) = Enahyax (L_num+3, x, w) / E_Skin
Bnahyax (L_num+3, x, w) = Bnahyax (L_num+3, x, w) + Neahyax (L_num+3, x, w) * (B - Bf - 2*Bfe)
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((x-1)*Slice < (Bfe))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))
Xval = -Sahyax (L_num+4, x, w)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+4, x, w) = -Yval
Enahyax (L_num+4, x, w) = Abs(Sahyax (L_num+4, x, w) / Sahyax (L_num+4, x, w))
Neahyax (L_num+4, x, w) = Enahyax (L_num+4, x, w) / E_Skin
Bnahyax (L_num+4, x, w) = Neahyax (L_num+4, x, w) * B
Qahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * Abs(Yax (x, w) - (Height-0.5*St))
Iahyax (L_num+4, x, w) = (Bnahyax (L_num+4, x, w) * St) * (Yax (x, w) - (Height-0.5*St))**2
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)
CALL GetValue (NG, X_GFRP_C, Y_GFRP_C, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
Sahyax (L_num+3, x, w) = Yval
Enahyax (L_num+3, x, w) = Abs(Sahyax (L_num+3, x, w) / Sahyax (L_num+3, x, w))
Neahyax (L_num+3, x, w) = Enahyax (L_num+3, x, w) / E_Skin
Bnahyax (L_num+3, x, w) = Neahyax (L_num+3, x, w) * B
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
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))
Xval = Sahyax (1, x, w)
CALL GetValue (NG, X_GFRP, Y_GFRP, Ibcbeg, Ybcbeg, Ibcend, Ybcend, Ypp_GFRP, Xval, Yval, Ypval, Yppval)
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) / E_Skin
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
Sahyax (2, x, w) = Tsax (x, w) * ((Yax (x, w)-(St+0.5*ASt))/Yax (x, w))
339
Appendix
340
Appendix
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
341
Appendix
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
WRITE (*,*) '>>>Deflection due to Shear has been calculated for midspan'
WRITE (99,*) '>>>Deflection due to Shear has been calculated for midspan'
ENDSUBROUTINE Total_Defl
SUBROUTINE New_Height_1 (w, Height, Slice, S_num, L_num, Rc, Gahyax, Wdax, Nhax)
IMPLICIT NONE
REAL (KIND = 8), DIMENSION (22,160,100), INTENT(INOUT):: Gahyax
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
REAL (KIND = 4):: I,x
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 (22,160,100), INTENT(INOUT):: Gahyax
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
REAL (KIND = 4):: I,x
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