Web 5192

DESIGN OF A
GLASS FLOOR STRUCTURE
PONTUS DUFVENBERG and FREDRIK JNSSON
Structural
Masters Dissertation
Mechanics
DEPARTMENT OF CONSTRUCTION SCIENCES
DIVISION OF STRUCTURAL MECHANICS

ISRN LUTVDG/TVSM--14/5192--SE (1-65) | ISSN 0281-6679
MASTERS DISSERTATION
DESIGN OF A
GLASS FLOOR STRUCTURE
PONTUS DUFVENBERG and FREDRIK JNSSON
Supervisor: PER-ERIK AUSTRELL, Assoc. Professor; Div. of Structural Mechanics, LTH, Lund.
Examiner: KENT PERSSON, PhD; Div. of Structural Engineering, LTH, Lund.
Copyright 2014 Division of Structural Mechanics

Faculty of Engineering (LTH), Lund University, Sweden.
Printed by Media-Tryck LU, Lund, Sweden, March 2014 (Pl).
For information, address:

Div. of Structural Mechanics, LTH, Lund University, Box 118, SE-221 00 Lund, Sweden.
Homepage: http://www.byggmek.lth.se
Preface
The work presented in this master thesis was carried out at The Division of Structural
Mechanics, Department of Construction Sciences, Lund University. This report is the
end stage of several years of studies at The Faculty of Engineering (LTH) at Lund
University, which finally ends up in a Masters Degree in Civil Engineering.
We would like to express our gratitude to Kent Persson for sharing his knowledge
concerning finite element modelling and the behaviour of glass structures. Thank you
for always having the door open and taking your time helping us.
During our time at the Department of Construction Sciences there has never been a
problem for us to ask questions and get advice from anyone of the staff. We are
sincerely grateful to have had this opportunity and to be a part of the interesting
coffee breaks and meetings at the institution.
Our time at the university has been an interesting journey and the years passed in
Lund are never to be forgotten. We would like to thank friends we gained during our
time here, without you this journey would never have been the same. Special thanks
to Martin Andersson and Mark Bellingham for proofreading this report.
Finally we would like to thank our families for all your support throughout our
education.
Lund, March 2014
Pontus Dufvenberg and Fredrik Jnsson
I
II
Abstract
Glass is by procurers and architects regarded as a material with desirable aesthetic
properties and is therefore more frequently utilized as a building material. A problem
though, is that glass is a brittle material sensitive to stress concentrations and
imperfections. Knowledge about glass as a bearing structural element is limited, but
is steadily improving. The aim of this report was to design a load bearing structure
consisting of a glass floor supported by glass beams. The analyses were carried out
using heat strengthened glass layers with SentryGlasPlus as laminating interlayers.
Analyses of the system were mainly carried out using the finite element software
Abaqus CAE. Different cross sections of glass plates were analysed with the purpose
to determine stresses and deflections in the profiles. Cracks were introduced to the
plates and the influence of these was investigated with approximate analytical
calculations and reference work. A laminated glass plate consisting of two 12 mm
glass layers in the centre and two 8 mm glass layers outermost was considered
acceptable when carrying a uniformly distributed load on a simply supported glass
plate of length 1.5 m. The glass profile was considered adequate both for a cracked
and an uncracked profile.
The beams were analysed using static- and buckling analyses in Abaqus. When the
static analyses were performed, both a cracked and an uncracked profile were tested.
Distributions of the cracks were determined with a previously performed test study
and calculations performed in Abaqus. Stresses, strains and deflections were
determined in the cross section to validate the chosen profile. A reinforcing steel bar
was decided to act in the bottom of the beam to prevent a hasty breakage if the glass
would start to crack. A beam consisting of three laminated glass layers with a
thickness of 15 mm each was decided as the cross section. A quadratic bar of 15x15
mm2 steel reinforcement was decided to act in the bottom of the centric glass layer.
The total height of the beam was chosen to be 250 mm and the total length was 4 m.
Analyses were carried out concerning vibrations using a combined structure of beams
and plates. The response of both vertical and lateral vibrations was investigated
concerning the system. The calculated vibrations were below the allowed limits.
A simplified calculation of the systems resistance against fire was performed and a
few suggestions concerning actions to construct a resistant glass system is presented.
Finally a discussion concerning the entire report and suggestions for further work are
presented.
III
IV
Sammanfattning
Glas r av arkitekter och bestllare ansett som ett material med tillfredstllande
estetiska egenskaper och anvnds drfr allt mer frekvent som byggnadsmaterial. Ett
problem r dock det faktum att glas r ett sprtt material, knsligt fr
spnningskoncentrationer och imperfektioner. Kunskapen om glas som brande
element r begrnsad, men r under stndig utveckling. Syftet med denna rapport var
att dimensionera ett glasgolv uppburet av glasbalkar. Vrmefrstrkta glasskikt
anvndes med SentryGlasPlus som laminat mellan glasskivorna.
Analyserna utfrdes frmst med hjlp av programvaran Abaqus CAE. Olika tvrsnitt
av glasplattor analyserades med syfte att bestmma spnningar och frskjutningar i
profilerna. Sprickor introducerades ven i plattorna och dess pverkan utvrderades
genom approximativa analytiska berkningar och referensarbeten. En laminerad
glasplatta bestende av tv 12 mm glasskikt centralt och tv 8 mm glasskikt ytterst
ansgs tillrckligt gllande brfrmga av en jmt utbredd last p en 1.5 m lng fritt
upplagd platta. Profilen ansgs tillrcklig gllande bde ett sprucket och ett intakt
tvrsnitt.
Balkarna analyserades med en statisk analys, samt med en instabilitetsanalys i

Abaqus. Nr den statiska analysen genomfrdes studerades bde en sprucken och en
intakt profil. Sprickornas utbredning bestmdes genom en jmfrelse med tidigare
genomfrd studie, samt berkningar i Abaqus. Spnningar, tjningar och nedbjning
bestmdes i tvrsnittet fr att tillgodose en tillrcklig profil fr syftet. Ett
armeringsband tillverkat av stl bestmdes verka i botten av balken fr att frebygga
ett sprtt brott om sprickbildning i glaset skulle uppst. En balk innehllande tre
laminerade glasskikt med en tjocklek p 15 mm vardera bestmdes verka tillsammans
i tvrsnittet. Ett kvadratiskt 15x15 mm2 armeringsband av stl valdes verka i botten
av det centriska glasskiktet. Balkens totala hjd sattes till 250 mm och den totala
lngden till 4 m.
En analys gllande vibrationer i ett kombinerat system av balkar och plattor

genomfrdes och responsen av bde vertikala och horisontala vibrationer i systemet
utvrderades. Berknade vibrationer visade sig vara under givna riktvrden.
En frenklad berkning genomfrdes gllande systemets motstndskraft mot brand

och ngra tgrder betrffande uppfrandet av ett brandmotstndskraftigt system
presenterades.
Slutligen frdes en diskussion gllande hela rapporten dr rekommendationer fr

fortsatta studier presenterades.
V
VI
Contents
1 Introduction 1
1.1 Background ......................................................................................... 1
1.2 Objective and method ......................................................................... 1
1.3 Disposition .......................................................................................... 2
2 Description of the glass system 3
2.1 Intended system .................................................................................. 3
2.2 Reference work ................................................................................... 3
3 Materials 7
3.1 Glass ................................................................................................... 7
3.1.1 Annealed glass ........................................................................ 8
3.1.2 Heat strengthened glass ........................................................... 8
3.1.3 Tempered glass........................................................................ 8
3.1.4 Comparison and choice of glass material................................ 8
3.2 Polymer interlayer .............................................................................. 8
3.3 Rubber................................................................................................. 9
3.4 Adhesive ........................................................................................... 10
3.5 Steel .................................................................................................. 10
4 Theory 11
4.1 The Finite element method ............................................................... 11
4.1.1 Introduction ........................................................................... 11
4.1.2 Modelling of linear-elastic materials .................................... 11
4.1.3 Equation of motion................................................................ 12
4.1.4 Finite elements ...................................................................... 12
4.1.5 Isoparametric finite elements ................................................ 12
4.2 Structural dynamics .......................................................................... 13
4.2.1 Springs .................................................................................. 13
4.2.2 Modelling rubber boundaries ................................................ 13
4.2.3 Steady state............................................................................ 13
4.2.4 Damping ................................................................................ 14
4.2.5 Rayleigh damping ................................................................. 15
4.3 Buckling analysis .............................................................................. 16
4.4 Abaqus modelling ............................................................................. 16
5 Eurocode and standards 17
5.1 Design value of strength for heat strengthened glass ....................... 17
5.2 Design of the glass structure ............................................................. 18
5.2.1 Design value for loading in ultimate limit state .................... 18
5.2.2 Serviceability limit state ........................................................ 18
5.3 Vibration analysis ............................................................................. 19
6 Design of glass plates 21
6.1 Estimation of a glass plate ................................................................ 21
6.1.1 Conclusion considering shear force ...................................... 22
6.2 Analysis of stresses and deflections ................................................. 22
6.2.1 Abaqus modelling ................................................................. 22
6.2.2 Meshing of the glass plates ................................................... 23
6.2.3 Description of the analysis .................................................... 24
6.3 Results from the static analyses ........................................................ 26
6.3.1 Conclusion static analysis ..................................................... 27
VII
6.4 Cracked glass plates .......................................................................... 28
6.5 Analytically calculated strength of cracked glass plates .................. 29
6.6 Conclusions and choice of glass plates ............................................. 32
7 Design of glass beams 33
7.1 Estimation of a glass beam ............................................................... 33
7.1.1 Conclusion considering shear force ...................................... 33
7.2 Analysis of stresses, strains and deflections ..................................... 33
7.2.3 Modelling of the cracks in the beam ..................................... 35
7.2.4 Modelling of multiple cracks in the beam............................. 37
7.3 Analysis of buckling ......................................................................... 38
7.3.2 Description of the buckling analysis ..................................... 39
7.4 Results............................................................................................... 42
7.4.1 Static analysis ........................................................................ 42
7.4.2 Buckling analysis .................................................................. 43
7.5 Modelling of beams presented from previous study ......................... 43
7.5.2 Results ................................................................................... 44
7.6 Conclusions....................................................................................... 45
8 Vibration analysis 47
8.1 Analysis of the system ...................................................................... 47
8.1.2 Evaluation of vibrations ........................................................ 48
8.2 Results............................................................................................... 50
8.2.1 Damping coefficients ............................................................ 50
8.2.2 Vibrations .............................................................................. 50
8.3 Conclusions of the vibration analysis ............................................... 52
9 Design of the whole system involving boundaries 53
9.1 Description of the system with dimensions ...................................... 53
9.2 Wear layer ......................................................................................... 54
9.3 Attachments ...................................................................................... 54
9.4 Erection of the system ...................................................................... 54
10 Design concerning resistance against fire 57
10.1 Fire progression ............................................................................... 57
10.2 Fire safety requirements .................................................................. 57
10.3 Fire resistance of glass ..................................................................... 58
10.4 Fire resistance of polymers .............................................................. 58
10.5 Simulation of fire ............................................................................. 58
10.6 Conclusion ....................................................................................... 60
11 Final remarks 61
11.1 Conclusions ..................................................................................... 61
11.2 Further studies ................................................................................. 61
12 Bibliography 63
VIII
1 Introduction
1.1 Background
The usage of glass as a structural element is common around the globe today; it is
regarded as a material with desirable aesthetical properties by procurers and architects.
Technological developments have made it possible to have glass elements with
relatively slender profiles as the main bearing system. The problem though is that
glass is a brittle material sensitive to stress concentrations at supports and to
imperfections, such as micro-cracks. This makes glass a quite unreliable material
concerning safety and breakage. The usage of polymer interlayers makes it possible to
hold several glass layers together even if cracks would occur and it reduces the risk
for cracking to spread between the laminated sections.
To imagine a bearing structure containing a glass floor carried by slender glass beams
is a fascinating idea, which as far as the authors are aware, has never been carried out.
Several similar solutions have been managed on the other hand, such as the Apple
glass cube in New York. In this structure, a glass beam frame carries a box of glass
which is the entrance to one of the Apple stores in the city. In other examples glass is
used in stairways, or as a floor which is the case of the Grand Canyon Skywalk.
1.2 Objective and method

The aim of this masters dissertation is to design a load bearing structure consisting of
a glass floor supported by glass beams. Supports will also be considered. The
dimensions of the glass structure members with attachments will be determined with
calculations performed with the software Abaqus.
Analysis will be carried out concerning static loading and buckling of the beams.
Evaluations will be made for dynamic loads acting on the system.
The calculations concerning the beams in this report will be confirmed by an analysis
in Abaqus of reinforced glass beams that have previously been tested in a laboratory
study carried out by [5].
1
1.3 Disposition
The report includes the following chapters:
In Chapter 2 the intended glass system is described.
In Chapter 3 the materials glass, polymer interlayer, rubber, adhesive and steel
are generally described.
In Chapter 4 the finite element method is generally described and structural
dynamics theory is introduced as well as vibration and buckling theory.
In Chapter 5 Eurocode and standards are presented.
In Chapter 6 the design of the glass plates with results is presented.
In Chapter 7 the design of the glass beams with results is presented. An analysis
regarding the tests carried out by [5] is also presented as verification.
In Chapter 8 a study of the glass system concerning vibrations is carried out.
In Chapter 9 the whole system is presented with connections.
In Chapter 10 the systems resistance to fire is discussed.
In Chapter 11 final remarks and suggestions for further work are presented.
2
2 Description of the glass system
In this chapter a brief description of the indented glass system is given with some
references to a previously performed study.
2.1 Intended system

The system contains a glass floor supported by glass beams, in this case carried by
steel columns. All beams and plates are simply supported. The intended structure can
be seen in Figure 2.1.
Figure 2.1: Sketch of the intended structure.
The beams that were investigated in the work presented in this report were decided to
have a span of 4 m and a spacing of 1.5 m between each other. The beams are simply
supported with boundaries based on steel columns. Each meter of a beam carries a
load of 1.5 m glass plate. The glass plates have a dimension of 0.5x1.5 m each and
are connected to the beams with a silicone adhesive and spacers made of EPDM-
rubber. Silicone is also used in the connection between all plates. The beams are
attached to the steel columns with U-formed boundaries made out of steel with the
inside covered with rubber. A rubber cover is also placed in the connection between
every simply supported beam at the boundary on the columns.
2.2 Reference work

The report described in [5] presents tests concerning three different beams, all built
up from three glass layers laminated together with a steel reinforcement in the bottom
of the mid-layer. The dimensions of a single beam are presented in Figure 2.2.
3
Figure 2.2: Dimensions of the reference work beam [5].
The beam shown in Figure 2.2 has several benefits which are stated below:
Glass is a brittle material and therefore sensitive to cracking. By

using a plastic material in between and laminate several layers of
glass together, it enables a profile that is still capable to carry load
even if cracking has occurred.
Flat thin profiles are desirable in the manufacturing process. It is
easier to make them affordable and to control the quality of the
material.
This kind of profile enables a bar of steel reinforcement to be easily
installed in the centre at the bottom of the cross section. The
purpose of the reinforcement is to take the tensile stresses if
cracking occurs.
4
The strength of the beams in [5] was investigated for three choices of glass types. The
beams tested consisted of annealed glass, heat strengthened glass and fully tempered
glass. Each beam had a support span of 1400 mm and was subjected to a four point
bending test as can be seen in Figure 2.3. The study showed that fully tempered glass
gives the best results concerning the initial breakage load, since it was capable of
taking the highest load. Heat strengthened glass on the other hand showed a better
result concerning the maximum post breakage load.
Figure 2.3: Reference work four-point bending test setup [5].
The results concerning the beams are interesting in the verification of the theoretical
calculations carried out in this report. In Section 7.6 a comparison with the result
from [5] and a model made in Abaqus will be performed.
5
6
3 Materials
When glass is used as a material in plates, different options are possible. One
possibility is to use a single solid piece of glass. Another option is to put several
layers of smaller glass plates together with plastic layers in between. This procedure
is called lamination and the product is known as a laminated plate.
Glass is a material with a brittle behaviour. When glass is critically loaded, micro
cracks which exist in the material will instantly grow resulting in total breakage of the
glass profile. This kind of failure happens instantly as the required amount of fracture
energy is low. Typical for the failure surfaces is that they will not deform during the
process [31].
Concerning a beam element, glass can be used as a solid. However, the development
of cracks and the difficulties for the manufacturer to make a profile big enough are
problems to be solved. If the glass is laminated in a few layers it will result in a more
ductile and reliable cross section, which also is easier to fabricate. Glass is a high
strength material for compression loading but not as good when considering tensile
loading. This is due to the micro cracks in the surface which will weaken the material
considerably [1].
When the strength of a glass material is exceeded, continuous cracks will develop fast
in the material if a tensile state is present. Therefore it is necessary to have some kind
of safety built in to prevent fast brittle breakage. In this report, the safety added is a
steel bar of reinforcement in the bottom of the beams where tensile stresses act. If a
crack occurs, the reinforcement will take the tensile stresses and prevent a sudden
failure of the structure.
3.1 Glass
Glass is a non-crystalline product produced by sand and alkalis fused together [1].
Glass has a plastic behaviour in the molten state, soft and malleable when hot and
brittle when it is cold. Normal room temperature is considered cold; hence glass has a
brittle behaviour.
Fracturing in a glass section occurs at much lower stresses when the specimen is
loaded in tension than when it is loaded in compression. The theoretical compressive
strength of glass can be as high as 16 GPa [23], however this value is well above
experimental values.
Melting is the central phase in glass manufacturing [1]. The individual raw materials
react and combine in high temperatures around 1400C. The glass is then cooled
down to a lower temperature where it is shaped to the desirable form. After shaping,
the material must be cooled, initially at a temperature just below where the glass
begins to soften (450-550 C). The temperature is then slowly lowered until room
temperature is reached to remove residual stresses inside the glass. If the cooling is
carried out too quickly, tension will remain inside the glass cross section. This will
result in stresses built into the section, which may cause cracking. However if the
7
temperature is slowly lowered to the cold state in a correct way, no stresses will
remain inside the cross section.
Sometimes stresses are desirable in a glass plate. For structural design purposes,
tension inside a glass plate and compression at the surface means that the plate can
take much higher loads [2]; this is called a heat-treated glass. The mechanical strength
of heat-treated glass varies significantly depending on the glass surface condition.
This is also the case concerning cracking behaviour. Glass can be divided into several
groups depending on the fabrication process. In this report three groups are of interest:
Annealed glass, Heat strengthened glass and Tempered glass.
3.1.1 Annealed glass

Annealed glass is raw glass with low residual stresses [2]. The fracture behaviour of
annealed glass profiles is a few long continuous cracks which will not expand with a
chain reaction over the surface. This enables cutting of a profile during production.
3.1.2 Heat strengthened glass

Glass that has been heat-treated to have a surface compression of 70 MPa is called
heat strengthened glass [20]. It has a fracture behaviour similar to annealed glass.
3.1.3 Tempered glass

Glass that has been heat-treated to have a surface compression of 120 MPa is called a
tempered glass [20]. This type of glass is about three times stronger than annealed
glass and breaks into small pieces at failure. This means that the entire profile most
likely shatters when a single crack occurs [2].
3.1.4 Comparison and choice of glass material

Normally annealed glass is used in laminated plates, as the breakage of the glass in
failure into big sharp pieces is ideal for the laminate. In comparison tempered glass
breaks into small pieces the size of gravel, which is harder for the laminate interlayer
to hold together [4]. The choice of glass type in laminated glass is, however,
dependant of the application. Heat strengthened glass works in the span between the
two other mentioned types and is possible to fabricate with a behaviour near annealed
glass considering cracking. A test carried out by [3] shows that heat strengthened
glass is favourable in lamination of plates, compared with tempered glass.
The fact that heat strengthened glass is about two times stronger than annealed glass
and has fracture behaviour similar to annealed glass [4], leads to the conclusion that
this will be the glass to be used for the design in this report. Glass normally has a
density of 2500 kg/m3, a Poissons ratio of 0.22 and a Youngs modulus of 70 GPa
[5].
3.2 Polymer interlayer

A polymer can be synthetic or natural, and consists of chain-shaped molecules [6].
All the parts in the chain-shaped molecules are bound with covalent bindings.
Polymers are usually viscoelastic and will exhibit creep strains when loaded.
The glass layers considered in this report are laminated together using polymer
interlayers. If the glass cracks it can still carry compressive forces; the interlayers will
8
help to keep the glass in its place and still allow it to be a bearing element in the
structure. The glass sheets will be laminated together using SentryGlasPlus (SGP)
interlayers from DuPont [7]. These interlayers are considered 5 times tougher and up
to 100 times stiffer than conventional interlayer materials like PVB. The interlayers
can thus carry more load and contribute more as a bearing element than other
conventional materials.
SGP has a mass density of 950 kg/m3 [7]. The stiffness and Poissons ratio of the
polymers varies with temperature and duration of the loading. For a long lasting load
of 10 years with a temperature of 24C, SGP has a stiffness of 129 MPa and a
Poissons ratio of 0.489. These material parameters will be used in the calculations of
an uncracked beam with long term loading scenarios. For a short lasting load of 1
minute and a temperature of 24C, SGP has a stiffness of 505 MPa and a Poissons
ratio of 0.458. These material parameters will be used both in the calculations of a
cracked beam and in the dynamic analysis where short term loads are acting. The
plastic yield stress of SGP is 23 MPa and the breaking strength is about 34.5 MPa
[22].
The stress strain relation concerning SGP can be seen in Figure 3.1 [17].
Figure 3.1: Stress-strain relation curve for SGP [17].
3.3 Rubber
Rubber is a special group of polymeric materials [8]. There are natural rubbers that
are created by nature and synthetic rubbers that are manmade. Rubber is characterized
by a process called vulcanization. When it undergoes vulcanization it switches to an
elastic state. During the vulcanization sulphur is added and cross links are created
between the molecule-chains so that a network is formed. This network gives rubber
its very high elastic characteristics; the sparse network structure can be deformed
when loaded and regain its original shape when unloaded [8]. An important property
of a rubber component is the possibility to modify its stiffness. The stiffness of a
component can be modified when the rubber is created by adding fillers or afterwards
by changing the thickness of the rubber.
The boundaries of the beams will be covered with EPDM-rubber, and the spacers
between the glass beams and the glass plates will be made of EPDM-rubber. This
type of rubber is very resistant to aging and external aggressive conditions including
9
severe temperature changes [9]. Hence, the material is widely used in the construction
industry and therefore assumed to be reliable.
The rubber that will be used in this report has a density of 1300 kg/m3, a Poissons
ratio of 0.49 and a Youngs modulus of 70 GPa according to [9]. The material can
take 8 MPa in tension and 400 % in elongation.
3.4 Adhesive
An adhesive is a substance that binds two objects together. The connection is
accomplished by adhesion between the adhesive and the objects boundary surfaces
and through cohesion in the glue joint [10]. It is required that the adhesive has low
viscosity when applied and that the surface of the object has good wetting against the
adhesive so that it can spread across the surface.
Glass is a brittle material which makes it sensitive to stress concentrations. Therefore

adhesives are good alternatives to mechanical glass joints since it spreads the stresses
over the surface of the joint. In the work presented in this report a silicone adhesive is
used to join the glass beams and the glass plates. Silicone has good durability and
resistance to weather, temperature, age and ultraviolet light [11]. As silicone is a soft
material, spacers made out of EPDM-rubber are placed in the boundary between the
glass beams and the glass plates to stiffen the connection.
3.5 Steel
Steel is the term for materials with elemental iron as the main constituent. Steel as a
metal is composed of crystals having a regular array of nuclei [8]. Various types of
processed steel with different kinds of properties are possible to manufacture from the
molten state, most of them are isotropic materials which means that they have the
same behaviour in all directions. At normal room temperature low-grade steel has
ductile failure behaviour. The material will behave elastically until the upper yield
stress point and will thereafter act plastically until breakage. Steel is said to have the
same strength both in tensile loading as in compressive loading.
In this case austenitic stainless steel will be used as reinforcement inside the glass
beams, as the corrosion risk might be high. This kind of steel has a density of 7950
kg/m3, a Poissons ratio of 0.2 and a Youngs modulus of 203 GPa according to [12].
Steel can be produced with a strength reaching over 1000 MPa depending on the
hardening process and the choice of alloy.
10
4 Theory
In this chapter theoretical backgrounds to the calculations which are performed in
later chapters are given. References to further reading are also suggested. An
introduction to the Finite element method, structural dynamics- and vibrations theory,
as well as the theory behind eigenfrequency analysis and buckling analysis is given.
4.1 The Finite element method

In this section an introduction to the theory of the finite element analysis is given. For
further reading about the method see [14].
4.1.1 Introduction
The Finite element method solves differential equations in an approximate manner
using a numerical approach. It is used for solving engineering problems that are too
complicated to solve analytically.
Consider a variable that has an arbitrary variation over a region, then it is a good
approximation to assume that it varies in a linear manner for small elements in this
region. This is the basis of the Finite element method, to divide regions into smaller
elements and then solve the problem approximately for each element. A region with
elements is called a finite element mesh. With a finer element mesh the solution
converges towards the exact solution [14].
Calculating deflections and forces with the finite element analysis is done by solving
the following equation system
(4.1)
where is the stiffness matrix, is the displacements and is the forces.
4.1.2 Modelling of linear-elastic materials

The following text is a summary of the finite element formulation of a linear-elastic
material, for a more extensive derivation see pages 235-260 in [14]. Theory for large
deformations can be found in [25].
The relation between stresses and strains is called a constitutive relation. The simplest
constitutive relation is linear elasticity expressed by Hookes law in one dimension by
eq. (4.2)
(4.2)
where is the stress, E is Youngs modulus and is the strain [14]. This relation also
holds when there are several stress and strain components. In three dimensions the
generalized Hookes law describes the stresses and strains given by [14] as
(4.3)
where
11
(4.4)
and
(4.5)
In modelling of plasticity concerning the material, theory is to be found in [26].
4.1.3 Equation of motion

The linear dynamic FE-formulation of an undamped MDOF-system is given by [16]
as
(4.6)
where M is mass matrix, is the accelerations, K is the stiffness matrix, is the

displacements and f is the load vector.
4.1.4 Finite elements

The elements used in the FE-calculations are three dimensional deformable solids.
Assuming that the displacements vary in a quadratic manner in the elements the
geometric order is chosen to be quadratic. The solid elements are thus 20 node brick
elements.
4.1.5 Isoparametric finite elements

For a brick element to behave in a compatible manner the sides of the element must
be parallel to the coordinate axes. This leads to restrictions concerning modelling of
bodies with arbitrary geometry. However, with the use of isoparametric finite
elements it is possible to establish compatible finite elements that have curved
boundaries [14].
This is achieved by mapping of one region into another region. The region that
contains the arbitrary geometry is called the global domain and the region containing
a square geometry is called the parent domain. For three dimensional finite elements
the parent domain is a cube bound by the lines =1, =1 and 1 according to
Figure 4.1. This mapping is given by [14] as
(4.7)
12
The mapping results in that every point in the parent domain has a corresponding
point in the global domain which can be seen in Figure 4.1.
Figure 4.1: Twenty-node three dimensional isoparametric brick element [14].
4.2 Structural dynamics

An introduction to structural dynamics- and vibrations theory, as well as the theory
behind buckling analysis is given in this section. For further reading about structural
dynamics see [15].
4.2.1 Springs
The force needed to extend or compress a spring a distance, is linearly proportional to
that distance. This relation is described by Hookes law according to eq. (4.8)
(4.8)
where k is the stiffness, is the force with and is the displacement [14].
4.2.2 Modelling rubber boundaries

When modelling rubber boundaries the rubber can be replaced by springs with equal
stiffness. By doing this, the stiffness of the rubber will be approximated to a few
springs. This prevents the FE-analysis from calculating the modeshapes of the rubber
and instead calculating the modeshapes of the entire system. The stiffness of a spring
element is calculated according to eq. (4.9)
(4.9)
where E is the elasticity of the rubber, A is the area of the rubber and L is
the thickness of the rubber [14].
4.2.3 Steady state

In a steady state solution the dynamic system is transformed into a time independent
system of equations for the amplitudes. Consider the undamped system in eq. (4.10).
(4.10)
13
This system can be solved using a trial solution according to eq. (4.11), [16].
(4.11)
The second derivative of the displacement gives the acceleration according to eq.
(4.12)
(4.12)
Inserting eq. (4.11) and eq. (4.12) in eq. (4.10) gives the steady state solution for an
undamped system according to eq. (4.13), [16].
(4.13)
4.2.4 Damping
The process which makes the free vibrations diminish is called damping. The
damping dissipates the vibration energy from the system due to different mechanisms.
It is impossible to establish all the mechanisms in a structure that contribute to the
damping. Some examples of damping are energy dissipation from repeated elastic
straining, internal friction when a solid is deformed, and opening and closing of micro
cracks in concrete when the structure is subjected to a vibration load [15].
Since it is difficult to determine all the mechanisms that contribute to damping, the
damping in structures is idealized. In a MDOF-system the damping can be described
with the equation
(4.14)
where is the damping coefficient with the unit of Ns/m, is the forces with the
unit N and is the velocity with the unit m/s [15].
The equation of motion for a damped system can be seen in eq. (4.15), [16].
(4.15)
In a damped system the displacements will also be harmonic but with a phase lag
relative to the force. A damped system can be described with complex notation which
takes the phase lag into account. Using complex representation the displacements can
be rewritten according to eq. (4.16), [16].
(4.16)
The first- and second derivative of the displacement gives the velocity and
acceleration according to eq. (4.17) and (4.18).
(4.17)
(4.18)
14
Inserting eq. (4.17) and eq. (4.18) in eq. (4.15) gives the steady state solution for a
damped system according to eq. (4.19).
(4.19)
4.2.5 Rayleigh damping

The basis for Rayleigh damping is that for low eigenfrequencies the damping
primarily depends on the mass and for high eigenfrequencies damping primarily
depends on the stiffness [15], as can be seen in Figure 4.2.
Figure 4.2: Rayleigh damping [15].
The damping ratio for the nth mode of a system is calculated with eq. (4.20)
(4.20)
where and are Rayleigh coefficients, is the nth damping ratio and is the
th
n eigenfrequency [15].
The coefficients and can be determined from two angular frequencies and
as shown in eq. (4.21).
(4.21)
For Rayleigh damping the C matrix consists of the Rayleigh coefficients according to
eq. (4.22), [16].
(4.22)
15
4.3 Buckling analysis
In the eigenvalue buckling problem, the loads sought after are the ones for which the
stiffness matrix of the structure becomes singular, so that the problem
(4.23)
has nontrivial solutions [19]. is the tangent stiffness matrix when the loads are
applied and are the nontrivial displacement solutions. The magnitude of the loading
required to achieve buckling is scaled by the load multipliers found in the
eigenvalue problem
(4.24)
where is the stiffness matrix corresponding to the base state, is the differential
initial stress and load stiffness matrix due to the incremental loading pattern, are the
eigenvalues and are the buckling mode shapes [19].
4.4 Abaqus modelling

To model the structural components regarding its strength and dynamic response, the
software Abaqus CAE is used in this work. Abaqus is a software suite for finite
element analysis and computer aided engineering, based on the scripting language
Python. Abaqus CAE is a user interface that can be used both for modelling of
mechanical components and assemblies as well as visualizing the finite element
analysis results.
All elements that were used for modelling of the glass structure in this report in
Abaqus were deformable 20 node quadratic 3D solids. The material properties were
chosen as the material properties given in Chapter 3. The interlayers were considered
to be ideally plastic materials and were allowed to deform plastically.
16
5 Eurocode and standards
The design values considering the maximum strength of glass, the ultimate limit state
(ULS) and the serviceability state (SLS) are calculated in this chapter. The technical
rules used are all developed by the European Committee for Standardisation, so
called Eurocode. Vibration analysis is also discussed in this chapter.
5.1 Design value of strength for heat strengthened glass

The design value of strength for a heat strengthened glass material is calculated
according to the equation for pre stressed glass in Eurocode [20]
(5.1)
where is a factor concerning the load duration, is a factor for the glass
surface profile and is the characteristic value of the bending strength. The
material partial factor is called for annealed glass. The factor does consider
strengthening of pre-stressed glass, is the characteristic value of pre-stressed
glass and is a material partial factor for surface pre stressed glass.
Values for the calculation of the design value for heat strengthened glass are seen in
Table 5.1. All values are found in [20]. The variable is chosen considering the
worst case scenario to be a personnel load with 30 seconds of duration. This value
was chosen out of the condition that it is not likely that people in a worst case
scenario will stay longer than a few seconds on the same part of the floor.
Parameter Value Source

0.89 Table 6
1.0 Table 5
45 MPa Section 8.1.1
1.8 Table 2
1.0 Table 8
70 MPa Table 7
1.2 Table 2
Table 5.1: Values for calculation of the design value for heat strengthened glass [20].
Insertion of these values into eq. (5.1) gives the design value of strength for heat
strengthened glass.
17
5.2 Design of the glass structure
5.2.1 Design value for loading in ultimate limit state
According to Eurocode the load to be applied on a structure in the ultimate limit state
shall be calculated with eq. 6.10b in [13], as can be seen in eq. (5.2)
(5.2)
where is a factor concerning the safety class, is the permanent loading and
is the variable concentrated loading. Depending on the worst case scenario the
variable load can be placed either as a concentrated load or as a distributed load. If the
variable load is introduced as a distributed load, the parameter is replaced with
in equation (5.2).
The factor is in this case 1.0, as the risk for personal injury is high. The factor
varies depending on the weight of the structure. The characteristic value
is for a congregation area 3 kPa. These values are given in [13], Table 1.7.
5.2.2 Serviceability limit state

The load combination used for serviceability limit state is quasi-permanent loading.
According to Eurocode, the load on a structure when exposed to long time loading is
calculated with equation 6.16b [13], as can be seen in eq. (5.3).
(5.3)
where is the permanent loading, is a factor concerning variable loads, and

is the variable loading.
The parameter varies depending on the weight of the structure. For a

congregation area is 0.6, which is given in [13] at Table 1.6. The characteristic
for a congregation area is 3 kPa. The values are to be found in [13], Table 1.7.
18
5.3 Vibration analysis
The vibration analysis will be performed as described in chapter 2 in Acceptance
criteria for human comfort, by [18]. The acceleration at an arbitrary node in the
structure is multiplied with a reduction factor according to Table 5.2. Which one of
the factors the acceleration is multiplied with is dependent on what frequency that is
exciting the structure. In [18] the first frequency interval does not stop where the
second frequency interval starts and the second frequency interval does not stop
where the third frequency interval starts and so on. Therefore to cover the entire
frequency span the frequency intervals are slightly modified to start where the
previous interval stops as can be seen in Table 5.2.
Frequency interval (Hz) 1.6-3.2 3.2-4.8 4.8-6.4 6.4-8.8

0.5 0.2 0.1 0.05
Table 5.2: Vibration reduction factors [18].
Apart from multiplying the acceleration with the stated calculations also have to
take the persons moving around on the floor in consideration. This is taken into
account by multiplying the weight of a person, typically 700 N with the square root of
the number of persons moving around on the floor [18]. The square root takes the
uncorrelated movement of the persons into consideration. For example if nine persons
are walking uncorrelated it gives the same effect on the vibrations as three persons
walking correlated.
When calculating the accelerations on the structural element the load is to be put as a
total of 1 N. The accelerations of the structural element is linearly proportional to the
load applied which leads to simple calculations; in this case the loads from persons
moving on the floor will hence be multiplied by 1.
After multiplying the acceleration in an arbitrary node of the structure with , the
weight of a person and the square root of the number of persons moving on the
structure, the value gained is divided with the acceleration of gravity to get the node-
acceleration in relation to the acceleration of gravity. This can be seen in eq. (5.4).
All the calculated values of the node-accelerations related to the acceleration of

gravity are then compared with approved values of vibrations for different activities.
The complete equation used for calculating the vibrations is
(5.4)
where is the weight of an average person estimated to 700 N, is the

acceleration at a specific frequency, is a reduction factor within a frequency
interval, is the acceleration of gravity assumed to be 9.81 m/s2 and is the
number of persons moving on the floor.
19
20
6 Design of glass plates
In this chapter the design of the glass plates that are placed on the glass beams in the
floor-structure is performed. The analysis was carried out when a uniformly
distributed load or a concentrated load were acting on the surface. Requirements
concerning deflection were also considered in the analysis. The behaviour of a plate
when cracking occurs through the section was considered as well.
6.1 Estimation of a glass plate

The maximum moment capacity of a rectangular beam is given by [21], as eq. (6.1)
(6.1)
and the maximum moment at the midpoint of a simply supported beam subjected to a
uniformly distributed load is given by [13], as eq. (6.2).
(6.2)
where M is the maximum moment, is the maximum normal stress, b is the width, h
is the height, L is the length and q is the line load.
A rough estimation of a glass plates dimensions was carried out with a calculation,
where one meter of a glass plate in width where calculated as a beam. A roughly
estimated load of 6 kPa was assumed to act along a plate. Combining eq. (6.1) and eq.
(6.2) gives
15 mm is a rather thin glass layer and the dimension may be higher when more
accurate calculations are carried out. To make the profile efficient and resistant if
cracking would occur, it was desirable to have the laminate a bit out from the centre
of the section. A decision was made to have three or four layers of glass laminated
together with 1.52 mm SGP.
The shear capacity of a cross section was investigated and is given by [13], as eq. (6.3)
(6.3)
where is the first moment of area, is the shear force, is the second moment of
area and is the width. To calculate the maximum shear capacity of a rectangular
beam, eq. (6.3) can be written as eq. (6.4).
21
(6.4)
The maximum shear force of a simply supported beam subjected to a uniformly

distributed load is given by [13] as eq. (6.5)
(6.5)
where is the uniformly distributed load and is the length of the beam. Combining
eq. (6.4) and eq. (6.5) gives the required height of the plate when considering shear
capacity.
6.1.1 Conclusion considering shear force

The height required of the section to take the shear stresses was just a fraction of the
height required to take the moment. An assumption was made that the moment
capacity of the cross section would be the design basis of the plates.
6.2 Analysis of stresses and deflections

6.2.1 Abaqus modelling
The elements which were chosen for modelling the glass plates were 20 node
quadratic brick elements with reduced integration. The material properties were
decided to be as stated in Chapter 3. The SGPs were assumed to deform as an ideally
plastic material after reaching the stress 23 MPa where plastic deformation begun.
The glass plates were analysed using a static load step. In this analysis the plates were
modelled as simply supported, see Figure 6.1. At the bottom of both ends, the edge
was prevented from moving in the z-direction. The plates were prevented to move in
the x-direction at the bottom of one end and prevented from moving in the y-direction
in one node at the bottom of each end. These boundary conditions allowed the plates
to expand in the longitudinal and lateral directions but prevented rigid body
displacements.
22
Figure 6.1: Boundary conditions for the glass plates.
6.2.2 Meshing of the glass plates

Calculations to verify the needed mesh size for a given glass plate were carried out to
decide the meshing size needed for a good approximation of the stresses. The
analyses were performed on a plate exposed to a distributed load of 0.9 kPa or a
concentrated force of 4.5 kN on an area of 50x50 mm2 in the middle of the plate at
one edge. The meshed plate is visualized in Figure 6.2.
Figure 6.2: Meshing of a plate.
First the number of elements required for the xy-plane was determined by varying the
size in the xy-plane and keeping the mesh in the z-direction to be one element in all
layers through the entire plate. In Table 6.1 an evaluation where the global mesh size
in the xy-plane is varied, is compared with a mesh where the y-direction is decided to
be 50 mm and the x-direction varies.
23
Mesh width (mm) 50 25 12.5
Stress (MPa), global mesh 28 28.9 29.2
Stress (MPa), mesh constant 50 in y-direction, varies 28 28.9 29.1
in x-direction
Table 6.1: Comparison horizontal meshing.
In Table 6.2 the results from an evaluation is shown where a 50x12.5 mesh and a
50x25 mesh like the ones described above is compared for various mesh sizes in the
z-direction. Three different cases were tested where each glass plate was divided into
1, 2 and 3 elements in the thickness direction for each material layer.
Number of elements/layer 1 2 3
Stress (MPa), mesh: 50x12.5, vertical varies 29.1 34.2 36
Stress (MPa),mesh: 50x25, vertical varies 28.9 34 35.7
Table 6.2: Comparison of 50x12.5 and 50x25 in vertical meshing.
A conclusion was reached that a meshing pattern of 50x25 is enough in the xy-plane.
The required meshing in the z-direction was also decided, in Figure 6.3 the results
from an evaluation is shown. The horizontal axis shows the number of elements that
each glass layer was divided into. When divided into 8 numbers of elements per layer,
the SGP-layers were divided as well, into 2 layers per element.
Comparison in vertical meshing

45
40
35
Stress (MPa)
30
25
20
15 50x25
10
5
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Number of elements/layer
Figure 6.3: Comparison in vertical meshing.
The meshing in the vertical direction was decided so that each glass part had 3
elements. This mesh size was considered as a meshing accurate enough to be used
when calculating the stresses in the plates. The meshing does not provide an exact
result, but a good estimation to fit the purpose.
6.2.3 Description of the analysis

The stresses and the deflections in the glass plates were to be decided. Different plate
dimensions were modelled and tested. One single plate had the dimension of 1.5x0.5
24
m in size. Firstly a laminated plate with three equally thick glass layers with
interlayers of SGP were modelled and secondly a laminated plate with four glass
layers consisting of two differing thicknesses. The laminated plates were modelled as
simply supported.
Glass laminates with five types of sections were analysed, see Figure 6.4 . The first
consisting of three 8 mm glass layers, the second of three 10 mm glass layers and the
third of three 12 mm glass layers. The fourth section tested consisted of two 10 mm
glass layers closest to the centre of the laminate and two 8 mm glass layers outermost.
The fifth section tested consisted of two 12 mm glass layers closest to the centre of
the laminate and two 8 mm layers outermost.
Figure 6.4: Glass plates to be analysed.
Stresses and deflections in the different sections were analysed by means of the FE-
method for two types of loading. The first was a distributed load of 4.5 kPa, the
second a concentrated load of 4.5 kN. The concentrated load was placed on a surface
of a 50x50 mm square area at the centre of the plate at an edge, which was the worst
location for a concentrated load on the plate. Concerning loads in the serviceability
state a distributed load of 1.8 kPa was employed. Self weight was added to all load
cases.
25
Table 6.3 shows all data with forces acting on the plates. The loads that were imposed
to the plates are given and calculated according to sections 5.2.1, 5.2.2 and [13].
Dimension ULS SLS

Plate type Imp. Imp. Self Imp. Imp. Self
(mm) load load weight load load weight
kPa kN kPa kPa kN kPa
3x8 4.5 4.5 0.7 1.8 1.8 0.6
3x10 4.5 4.5 0.9 1.8 1.8 0.8
3x12 4.5 4.5 1.1 1.8 1.8 0.9
2x8+2x10 4.5 4.5 1.1 1.8 1.8 0.9
2x8+2x12 4.5 4.5 1.2 1.8 1.8 1.0
Table 6.3: Loads acting on the plates.
Analyses concerning loading of the five plates in ULS were performed for both a
distributed load and a concentrated load. The maximum principal stress in the glass
was determined in order to verify that it did not exceed the allowed stress of 43.1
MPa.
Analyses concerning loading of the five plates in the SLS was also made for a
distributed load and a concentrated load. Maximum deflection was determined to
verify that did not exceed a deflection of L/300, which is a commonly used, rather
high requirement [13].
An analysis was also carried out to verify the durability of the section when cracking
has occurred. This analysis was only performed on one of the plates.
6.3 Results from the static analyses

The results from the static analyses for the five different plates in ULS, when exposed
to a distributed load or a concentrated load, are shown in Figure 6.5.
26
Maximum stresses in the plates, when exposed to distributed or
concentrated load.
60
52.8
50
40 Strength limit
Stress (MPa)
35.7 43.1 MPa

30
25.8 25.2 Distributed load
20 19.9 Concentrated load
12.2
10 8.4
6.2 6.1 5.2
0
8x3 10x3 12x3 10x2+8x2 12x2+8x2
Dimensions (mm)
Figure 6.5: Maximum stresses in the plates, when exposed to distributed- and concentrated load.
The results of the analyses considering the five different plates concerning the
maximum deflection when exposed to a distributed load or a concentrated load in
SLS are shown in Figure 6.6.
Maximum deflection in the plates, when exposed to

distributed and concentrated load.
5
4.5
4
3.7
Deflection (mm)
3.5 Maximum
3 deflection, 5
2.5 mm
2.1 Distributed load
2
1.5 1.6 Concentrated load
1.3 1.3
1 1 1
0.5 0.6 0.6 0.5
0
8x3 10x3 12x3 10x2+8x2 12x2+8x2
Dimensions (mm)
Figure 6.6: Maximum deflection in the plates, when exposed to distributed- and concentrated load.
6.3.1 Conclusion static analysis

The concentrated loads gave higher stresses and larger deflections in the glass than
the distributed loads. All of the analysed glass plates, except the 3x8 mm plate, could
carry the loads without reaching critical stresses in the glass of 43.1 MPa and without
exceeding the maximum deflection of 5 mm.
27
6.4 Cracked glass plates
It is important for safety reasons that the glass floor does not collapse in case cracking
occurs. To verify that the glass floor could carry the loads with cracks present, a
comparison with a laboratory testing and an approximate calculation was performed.
Calculations concerning cracked glass plates are rather complicated. If cracking

occurs in a single glass layer it is difficult to predict how a crack will develop through
the material and if the crack will break through all layers or stay in one layer. If the
cracking starts in one single layer and breaks through it, fracture mechanics is
necessary to validate if the crack will continue through the next layer or stop at the
interlayer between the glass layers. The exact pattern of the cracking is needed for the
calculations to be correct and the pattern concerning glass is hard to decide. The
residual strength of the glass plate may, however, be estimated for a worst case
scenario.
Several test studies have been carried out concerning broken laminated glass plates.
In [3] a canopy of heat strengthened glass is subjected to a uniformly distributed static
load of 1.5 kPa, see Figure 6.7. The dimension of the canopy was 1x1 m2 and it
consisted of two 8 mm thick layers of heat strengthened glass laminated together with
1.52 mm SGP. The plates were completely cracked before the load was applied on the
section and were acting for 24 hours resulting in that the plate was still able to carry
the load.
Figure 6.7: Loading of cracked glass plate. Testing carried out by [3].
28
6.5 Analytically calculated strength of cracked glass plates
The study carried out by [3] shows that two cracked glass layers, laminated together
with 1.52 mm SGP, can carry a weight of 1.5 kPa for 24 hours. Therefore it should be
realistic to consider four layers of glass, laminated together with 3 layers of SGP
enough to carry at least three times the load. The profile consisting of 8-12-12-8 mm
laminate will probably be enough to carry a distributed load of 4.5 kPa on a span of
1.5 m, when all the plates are completely cracked.
A calculation of a cracked glass plate was carried out on an 8-12-12-8 mm laminated

plate, where the section was considered to be cracked. In order to calculate the
strength of the cracked glass plates, an analytical calculation on a simplified model
was carried out. To determine that the residual strength of the glass plate had
sufficient moment capacity, the cracked section was compared with the moment
generated from the loading. The cracked section can be seen in Figure 6.8.
Figure 6.8: Cracked section of a glass plate.
In Figure 6.9 the stresses in the cross section can be seen before cracks have occurred.
The glass will in this state take compressive- and tensile stresses. The interlayers will
have a negligible impact on the stress distribution and are therefore not illustrated in
this figure.
Figure 6.9: Stress distribution in uncracked section.
If the glass cracks it can still carry compressive stresses. The worst scenario would be
that all the glass layers crack, leading to a redistribution of the stresses in the glass
layers and in the interlayers. An assumption regarding this scenario is that the upper
part of the top plate takes all the compressive stresses while the three interlayers take
29
the tensile stresses, as shown in Figure 6.10. The failure strength of the interlayer
material is around 34.5 MPa according to [22]. The interlayers were assumed to be of
an ideally plastic material. All of the interlayers in the model were assumed to have
reached the failure strength of the material.
Figure 6.10: Stress distribution in cracked section.
A section of the cracked glass plate can be seen in Figure 6.11. All of the glass layers
were cracked but the interlayers was assumed to remain intact. When the glass layers
are cracked, the plate still has to carry the moment and shear forces from the loading.
Considering the high compressive stresses that occurred in the top of the glass plate,
an assumption was made that this would be enough to carry the shear force in the
section, and no verification of the shear force capacity was performed.
Figure 6.11: Cracked section of a glass plate.
The forces in the interlayers in Figure 6.12 were calculated per meter in the lateral
direction of the plates. The stress in the top glass plate equals the stresses in the
interlayers in order to achieve lateral equilibrium. The compressive stresses were
assumed to be redirected around a joint in the upper part of the top glass layer.
Moment equilibrium was implemented from the joint according to eq. (6.6)
(6.6)
where the failure strength limit of the interlayers was 34.5 MPa, the thickness of
the interlayers was 1.52 mm and are levers to the interlayers, where 4.76
mm, 18.28 mm and 31.8 mm.
30
Figure 6.12: Stress distribution in the cracked section.
The values were inserted into eq. (6.6) and the moment capacity of the glass plates
per meter was calculated according to eq. (6.7).
(6.7)
This moment capacity was compared with the moment generated from the loading.
Assuming a concentrated force in the middle of the section the moment became
(6.8)
where F was the line load acting on the glass plate, and was the length of the plate.
The maximum loading per meter, was calculated as shown in eq. (6.9).
(6.9)
This corresponded to a concentrated force of 4.5 kN acting on half a meter of glass

plate, which was about the same load as the imposed load in Table 6.3.
31
6.6 Conclusions and choice of glass plates
All plates modelled in Abaqus, except the one with 3x8 mm glass, had the required
load bearing capacity uncracked. Considering cracked glass layers, tensile stresses
cannot be taken, and thus knowing which layers that are cracked was important in
order to perform an accurate analysis. However knowing how cracks are spreading
through the laminated glass section is difficult and assumptions had to be made in
order to make calculations. The assumption made was that all the glass layers were
cracked which was considered the worst case scenario. The very approximate
analytical calculation that was conducted in Section 6.5 showed that the glass floor
will maintain its required load bearing capacity even when all the glass plates were
cracked. The results in [3] and the results found in the analytical solution made a
strong argument that the glass floor built up by 8-12-12-8 mm glass plates and 3x1.52
mm SGP interlayers would be able to carry the required loads even when cracking
occurred in the glass plates.
If cracking occurs, the tensile forces acting on that plate have to be redistributed to
the other glass plates and to the interlayers. The interlayers would deform plastically
proportionally to time, as a consequence of the heavy local loads acting on the section.
Testing on laminated plates however, as can be seen in [3], showed that no brittle
breakage will develop in the section. People standing on one of the 1.5x0.5 plates will
have plenty of time to move away from the cracked plate so that the glass can be
replaced.
32
7 Design of glass beams
In this chapter the design of the beams in the system is performed. A static analysis of
a beam was carried out to determine its load-bearing capacity. An evaluation
concerning buckling of the beam was also performed.
7.1 Estimation of a glass beam

The maximum moment capacity of a rectangular beam is given by [21], as eq. (6.1)
(7.1)
and the maximum moment at the midpoint of a simply supported beam subjected to a
distributed load is given by [13], as eq. (6.2).
(7.2)
where M is the maximum moment, is the maximum normal stress, b is the width, h
is the height, L is the length and q is the line load.
A rough estimation of the glass beams dimensions was carried out with a calculation
where the load acting was assumed to be 6 kPa acting on a single beam. This gives a
line load of 6 1.5= 9 kN/m. Combining eq. (7.1) and eq. (7.2) gives
To be on the safe side a section height of the beam was chosen to be 250 mm. The
shape of the reinforcement was decided to be quadratic with a dimension of 15x15
mm2 in stainless steel.
7.1.1 Conclusion considering shear force

As stated in Section 6.1 the shear force would not be a design basis of the plates. The
same assumption was made considering the beams, that the moment capacity of the
cross section would be a design basis of the beams.
7.2 Analysis of stresses, strains and deflections

The elements that were chosen for modelling the glass plates were 20 node quadratic
brick elements with reduced integration. The material properties were decided to be
as stated in Chapter 3. The SGPs were assumed to deform as an ideally plastic
material after reaching the stress 23 MPa where plastic deformation begun.
33
The glass beams were analysed using a static load step. The load was applied as
distributed on top of a beam. In this analysis the beams were modelled as simply
supported, see Figure 7.1. At the bottom of both ends the edge of a beam was
prevented from moving in the y-direction. At the bottom of one end, the edge was
prevented from moving in the z-direction. The steel is thus not prevented from
moving in the z-direction, to get a more symmetrical deformation and to avoid stress
concentrations. When modelled, a beam was prevented from moving in the x-
direction at one node at the bottom of each end and along the entire longitudinal edge
at the top. This allowed expansion in the x-direction. The boundary conditions can be
seen in Figure 7.1.
Figure 7.1: Boundary conditions for the glass beam.

The beam was 4 meters long and consisted of three 15 mm thick glass layers,
laminated together with two 1.52 mm thick layers of SGP in between. The height of a
beam was 250 mm. A quadratic 15x15 mm2 reinforcement made out of steel acted in
the bottom of the mid-section. The beam can be seen in Figure 7.2.
Figure 7.2: Glass beam.
34
One type of loading was performed in ULS. It was the already calculated distributed
load of a glass plate acting on a beam, combined with the self weight of the beam.
The beam was also tested concerning deflection in SLS. The load acting on the glass
plates was multiplied with the length of a plate (1.5 m) to get total load acting on a
beam.
The loads acting on a single beam are presented in Table 7.1 and calculated according
to [13].
Dimension ULS (kPa) SLS (kPa)

(mm)
Variable Self Variable Self weight
load weight load
3x15 140.5 40.0 56.2 33.3
Table 7.1: Loads acting on a single beam.
An analysis concerning loading in ULS was performed when a distributed load was
applied. The maximum stress in the beam was decided to verify that it would not
exceed the design strength value of heat strengthened glass, which was 43.1 MPa.
The maximum stress in the steel as well as the maximum strain in the laminate was
also to be decided.
An analysis concerning loading in SLS took part when a distributed load was applied.
The maximum deflection in the beam was determined, and verification was carried
out to confirm that it did not exceed a deflection of L/300 [13].
An analysis was carried out concerning a scenario in ULS when cracks had occurred
throughout the midspan of a beam, causing a worst case scenario. The cracks did go
through the cross-section from the bottom of the beam where tension did act until the
top of the beam where compression started to act. The cracks developed in a
triangular pattern through the section as can be seen in Figure 7.3. The modelling of
the cracks is closely described in Section 7.2.3 and Section 7.2.4.
When the beam contained cracks, the maximum stress of the glass was decided in the
section. The maximum strain in the interlayers was also calculated to verify that an
extensive displacement did not occur. The maximum stress in the steel was finally
decided.
7.2.3 Modelling of the cracks in the beam

When glass exceeds its tensile strength cracks occur. When cracks occur the glass can
no longer carry tensile stresses, but the interlayers makes it possible for the glass to
still take compressive stresses. The steel reinforcement has the purpose to carry the
tensile stresses when cracks occur.
The cracks where shaped as thin lines that went through the vertical direction of the
beam as can be seen in Figure 7.3. Looking at the shape of these cracks in Figure 7.3,
the cracks were approximated by upside down triangles when they were modelled. An
assumption was made that the glass within these triangles did not have any
contributions to the load bearing capacity of the beam. This assumption was made as
35
the glass in the triangles was located in the tension member of the beam and hence
could not carry any tensile stresses while surrounded by cracks.
Figure 7.3: Cracks in a heat strengthened beam, [5].
When modelling the cracks in Abaqus an assumption was made that cracks only
occur in the tension part of the beam, therefore the cracks were stopped when they
reached the compression part of the beam. The exact height position in the beam
where the cracking stopped was located through trial and error. By completing the
analysis of the beam and hence visualizing the stresses in the longitudinal direction at
the top of the cracks, one can see if there are tensile or compressive stresses. If there
were tensile stresses, the cracks were raised, and if there were compressive stresses,
the cracks were lowered. The modelling was considered completed when the top of
the cracks was located at the neutral layer of the cross section, where the compressive
and tensile stresses met.
The process of finding the neutral layer, where the cracks will stop developing, was
carried out with trial and error and can be seen in Figure 7.4 - Figure 7.7. In these
figures the steel reinforcement and the interlayers are hidden and only the glass is
visible. The compressive stresses are shown as black and the tensile stresses are
coloured.
Figure 7.4: Cracks 10 cm from the top. Figure 7.5: Cracks 8 cm from the top.
Figure 7.6: Cracks 7.5 cm from the top. Figure 7.7: Cracks 7.0 cm from the top.
36
In Figure 7.4 the height of the cracks was decided to be 10 cm from the top. This
crack height was too low as the top of the cracks contained tensile stresses, thus the
cracks needed to be raised.
In Figure 7.5 the cracks were raised to be 8 cm from the top of the beam. As tensile
stresses still existed in the top of the cracked section, the cracks needed to be raised
further.
In Figure 7.6 the cracks were raised to be 7.5 cm from the top of the beam. As tensile
stresses still existed in the top of the cracked section, but was about to shift to
compressive stresses, the top of the cracks were almost placed at the neutral layer.
However since the top of the cracks still contained tensile stresses it was possible for
the cracks to go further up through the section. Therefore the cracks were raised a bit
further.
The cracks were raised another 0.5 cm which can be seen in Figure 7.7. As can be
seen, the top of the cracks only contained compressive stresses, which meant that the
cracks would not go further up, and thus the correct height of the cracks was found.
This was an example of how the cracks were modelled in Abaqus. The cracks had a
different height, depending on the load acting, the span of the beam, and the
dimension of the beam. Figure 7.4 - Figure 7.7 was made in ULS on the 4 m long
beam containing three 15 mm glass plates with two SGP interlayers in between. The
crack height was different concerning modelling of the beams from the test study [5]
as can be seen in Section 7.5, since they differed in applied load, dimension and span.
As can be seen in Figure 7.4 - Figure 7.7 the top of the beam takes compressive
stresses. An assumption was made that the compressive part of the beam takes the
shear stresses when the cross section is cracked. No further analysis concerning the
shear stresses in the beam was carried out.
7.2.4 Modelling of multiple cracks in the beam

More than one set of cracks can occur, therefore an investigation was carried out
concerning three sets of cracks. Figure 7.8 shows one beam with one set of cracks and
one beam with three sets of cracks.
Figure 7.8: Beams with one and three sets of cracks.
To decide whether multiple cracks present a more unfavourable scenario some

comparisons were carried out between the beams. The stress in the glass, the stress in
37
the steel, the strain in the SGP and the total deflection of the beams was compared.
All these comparisons considered the maximum values and the results can be seen in
Table 7.2.
Stress glass Stress steel Strain Deflection

(MPa) (MPa) interlayer (%) (mm)
1 crack 39 364 14 8
3 cracks 39 363 14 10
7.2: Comparison between one- and three sets of cracks.
As shown in Table 7.2 it did not make much of a difference if there were one or
multiple cracks in a beam. The deflection did vary a bit, but when a beam is cracked,
people will be evacuated from the glass floor and the deflection that has appeared will
not to a great extent impact the ability to perform the evacuation.
Since multiple cracks in the beam did not matter that much, the further analyses of the
beam were only carried with one set of cracks.
7.3 Analysis of buckling

The buckling analysis of a beam was carried out with use of two steps in Abaqus. The
first step was a static load step, which introduced a deformation in the beam. The
second step was a buckling step which performed a buckling analysis of the beam.
In the static load step a distributed load according to Table 7.1 was applied on top of
the beam. Another load of 1000 N was applied in the lateral direction in the midspan
of the beam. This force introduced a deformation to the beam before the buckling
analysis was carried out. The option to perform the analysis using large deformations
was also chosen in Abaqus.
In the buckling step a distributed load of 1 Pa was applied on top of the beam. In the
buckling analysis, a requested number of eigenfrequencies and eigenmodes were
calculated. In order to get the load at which the mode shapes occurred, the applied
load was multiplied with the eigenvalue of the mode shape. Having the load as 1 Pa
was convenient since the load at which the mode shapes occurred at could be
determined directly from the eigenvalue, since it was multiplied with the load 1 Pa.
To better simulate the real boundaries of the beam, boundary conditions were
changed to consist of rubber which had the form of a U with a plate on the back.
Springs were introduced to simulate the rubber at these boundaries as problems with
eigenvalues and eigenfrequencies occurred in the rubber layer when rubber as
material was modelled as 3d solids. The material parameters of rubber were
recalculated to be valid for the springs. These springs were the only boundary
conditions on the beam which meant that the beam was not stabilized in the lateral
direction on the top. In reality the beam will to some extent be stabilized in this
direction due to the glass floor. However by ignoring the stabilizing addition from the
glass floor, the analysis is performed using a more unfavourable case scenario.
38
7.3.2 Description of the buckling analysis
The stiffness of rubber is dependent on the dimensions of the rubber profiles. Since
the rubber placed at the ends of the beam consisted of different sized rubber plates,
they had different Youngs moduli. A rubber boundary can be seen in Figure 7.9.
Figure 7.9: Rubber boundary.
The boundaries were assumed to consist of four rubber plates, one bottom plate, two
side plates and an end plate. The stiffness was calculated according to [24]. A shape
factor was calculated according to eq. (7.3) and the location of different areas can be
seen in Figure 7.9.
(7.3)
The shape factor S was calculated for the rubber at the location of the end, side and
bottom rubber boundary.
39
With the shape factors and the hardness of the rubber used, Youngs moduli could be
calculated. The EPDM rubber used in this report was assumed to have a hardness of 70
IRHD. Looking at the diagram in Figure 7.10 the following Youngs moduli were found.
Figure 7.10: Definition of areas on loaded rubber [24].
40
The thickness of the rubber was decided to be 3 mm. The calculation concerning the
stiffness was calculated as shown below.
The springs were placed according to Figure 7.11, with 6 springs on each side of the
beam, 6 springs on the end and 6 springs on the bottom. The stiffnesses were divided
so that the outer springs each contained 1/8 of the stiffness and the springs in the
middle each contained 2/8 of the stiffness in the respective directions. This refers to
the springs located on the sides directed in the x-direction, springs on the bottom in
the y-direction and springs on the ends in the z-direction.
Some nodes had spring stiffnesses in more than one direction, for example both nodes
in the bottom corners of the beam had spring stiffnesses in all three directions. A
rough explanation of the placement of the springs can be seen in Figure 7.11. The
purpose of this figure is to explain where on the profile kEnd rubber, kSide rubber and kBottom
rubber are defined, therefore all positions were not written in the figure. The stiffness
values acting on the beam can be seen in Table 7.3.
Position End, End, Side, Side, Bottom, Bottom,

mid outer mid outer mid outer
Stiffness 150.1 75.1 171.9 85.9 25.0 12.5
(MN/m)
Table 7.3: Spring stiffness values on the beam.
41
Figure 7.11: Boundary conditions for buckling analysis.
7.4 Results
7.4.1 Static analysis
Results from the static analysis of the beam are presented in Table 7.4.
Stress glass Stress steel Strain Deflection

(MPa) (MPa) interlayer (mm)
Uncracked 32 90 3 3
beam
Cracked 38 380 14 -
beam
Table 7.4: Results from the static analysis of the beam.
42
7.4.2 Buckling analysis
The first eigenvalue was 1.52106, thus the first buckling mode occurs at a distributed
load of 1.52 MPa. The buckling mode shape can be seen in Figure 7.12.
Figure 7.12: Buckling mode shape.
7.5 Modelling of beams presented from previous study

The beams tested by [5] were analysed to verify the results gained in this chapter
considering the glass beams. Modelling of the beams was carried out in the same way
as described in Section 7.2.

The beams were 1.6 meters long and consisted of two 6 mm thick glass layers on the
sides and a 10 mm thick layer in the centre of the section, laminated together with
two 1.52 mm thick layers of SGP. The beams were 125 mm high. A quadratic
hollowed reinforcement made out of steel with dimension 10x10x1 mm acted in the
bottom of the section. Dimensions of the beams can be seen in Figure 2.2.
The three different beams were of sort annealed glass, heat strengthened glass and
fully tempered glass. The material data of the beams are shown in Table 7.5.
43
Property Annealed Heat Fully tempered Steel SGP
glass strengthened glass
glass
Tensile 45 70 120 520-750 34.5
strength
[Pa]
Elastic 70x103 70x103 70x103 200x103 300
modulus
[N/mm2]
Elongations - - - 45 400
at tear
[%]
Density 2500 2500 2500 7900 950
[kg/m3]
Table 7.5: Material data of beams in the test study, [5].
The loading of the beams were carried out as four point loading, as can be seen in
Figure 2.3. The loading leading to breakage carried out by [5] was 10.9 kN for the
annealed glass beam, 33.1 kN for the heat strengthened glass beam, and 42.1 kN for
the fully tempered glass beam. These were the loads that were applied during the
analysis in Abaqus, along with a self weight of 0.3 MPa.
7.5.2 Results
The results from the analysis together with the characteristic values are presented in
Table 7.6. The first column describes the characteristic tensile strength (MPa), the
stress values in the glass from the Abaqus modelling without cracks (MPa) and
the stress values in the steel reinforcement from the Abaqus modelling with cracks
(MPa).
Annealed glass Heat strengthened glass Tempered glass

(MPa) 45 70 120
(MPa) 42.4 129 164
(MPa) 460 458 400
Table 7.6: Results from analyses in Abaqus with input from previously performed test study, [5].
44
7.6 Conclusions
The intended beam met the requirements caused by the different types of loading.
This as the stresses found in the glass were less than the design value of 43.1 MPa,
and the deflection were less than 4000/300=13.3 mm.
When cracking occurs, the steel must be able to carry a load of 380 MPa or more. The
ferrit-austenitic steel: S32205 (EN 14462) [12] with a proof strength of 450 MPa and
a tensile strength of 650-880 MPa, is chosen. The design strength value according to
[13] is the same value as the characteristic value. It was also concluded that the
number of cracks which occurs has negligible affects on the stresses in the beam and
the difference in deflections between one and three sets of cracks were also negligible.
The buckling analysis gave the first instability at the load 1.52 MPa which was
substantially higher than the applied load of 180.5 kPa. The beam will thus not break
as a consequence of instability.
Concerning the previous carried out study it can be noticed that the stresses calculated
in the Abaqus model were higher than the characteristic stresses considering heat-
strengthened glass. This was expected since characteristic values are taken from the
lower five percent fractile. Deviations from the characteristic values also depend on
the glass used at the laboratory testing and on the strength of the glass. The actual
bearing capacity of heat-strengthened glass is thus generally higher than 70 MPa. The
stress value obtained in Abaqus concerning heat strengthened glass was 129 MPa
when the failure load from the study was applied. Since the stress value obtained in
the Abaqus model for heat-strengthened glass was a bit higher than the characteristic
value, it is reasonable to conclude that the Abaqus model gives a good approximation
of the actual glass beam.
The stress values in the annealed and tempered glass were also reasonably consistent
with the characteristic values, with some deviations for the annealed glass which
almost had the same stress in between the Abaqus model and the characteristic value.
The stresses in the steel reinforcement were below the characteristic values of the
material used. This is consistent with the laboratory testing since failure in the steel
did not occur. These similarities between the Abaqus models and the laboratory
testings concerning stress values in tempered glass, annealed glass and steel also
implicates that the Abaqus model used is consistent.
45
46
8 Vibration analysis
Calculations concerning vibrations in the floor structure from dynamic loads acting
on the system are performed in this chapter. The calculated vibrations were compared
with guidelines given by [18].
8.1 Analysis of the system

A combined structure consisting of beams and plates formed the model for the
vibration analysis. It consisted of three beams and two rows of glass plates in between,
as shown in Figure 8.1.
Figure 8.1: Model of the floor system used in the vibration analysis.
The boundaries between the plates and the beams consisted of silicone with rubber
spacers. Since rubber and silicone have similar mechanical properties, all boundaries
were assumed to be made of 3 mm thick rubber. At the edges of the beams layers of
rubber material were assumed. These rubber layers had the purpose to form a soft
boundary and to act as dampers. In Figure 8.2 such a boundary layer, mounted on a
beam, is shown.
Figure 8.2: Rubber boundaries.
47
Damping parameters were introduced as material damping for the rubber and the SGP
layers. Rayleigh damping was employed in this modelling and the two damping
coefficients a0 and a1 were calculated with eq. (4.21) using the frequencies 1 Hz and
10 Hz. Both frequencies were assumed to have the same damping ratio, which was
7 %. A commonly used standard value for damping in building codes is 5 %.
Recommended damping values can vary between about 2-20 % though, where steel
normally has a damping ratio of 5 % and wood 15 %, just below the yield point [15].
These damping ratios can be used directly for the linearly elastic analysis of structures
with classical damping [15]. The rubber was assumed to have a damping ratio of 7 %,
which should be on the safe side concerning propagation of vibrations.
Two different steady state vibration analyses were performed. The first analysis was
to verify the response of the structure concerning vibrations acting vertically on the
floor, and the second analysis was to verify the response of the structure concerning
vibrations acting laterally on the floor.
8.1.2 Evaluation of vibrations

The allowed maximum values concerning vibrations at different sites are shown in
Figure 8.3 [18].
Figure 8.3: Maximum values concerning vibrations at different sites, [18].
The load acting on the structure will mainly come from people walking on the floor
and the vibrations caused by their footsteps. The load acting in the vertical direction
was decided to be 1 N in total, which was spread out over the total floor area of 12 m2.
A load of 1 N was applied since the relation between the load and the accelerations is
linear, which allows loading from the people to be varied after the accelerations from
the steady state analysis were calculated. The total load of 1 N resulted in a uniformly
distributed load of 0.083 N/m2. Concerning vibrations in the vertical direction, 12
48
persons were assumed to be walking on the floor at the same time. The position of the
largest vertical accelerations is shown with a red circle in Figure 8.4. These
accelerations were compared to the allowed maximum acceleration values.
Figure 8.4: Largest vertical accelerations.
In the analysis concerning the lateral vibrations 3 people was assumed to act as
concentrated loads, as shown in Figure 8.5. The total load acting in the lateral
direction was decided to be 1 N for the same reasons as described for the vertical
vibrations. This resulted in three forces of 0.33 N each. The largest accelerations were
those at the tip of the mid arrow in Figure 8.5. These accelerations were compared to
the allowed maximum acceleration values.
Figure 8.5: 3 Lateral point loads.
49
The mass of the people was added to the mass density of the beams during the
analysis. The accelerations of the structure when excited by dynamic forces with
frequencies between 1-10 Hz were inserted into a MATLAB program. In this
program the amount of people moving on the floor could be decided and the
vibrations could be calculated and plotted.
8.2 Results
8.2.1 Damping coefficients
By solving eq. (4.21) with the frequencies 1 Hz and 10 Hz combined with a damping
ratio of 7 %, the Rayleigh damping coefficients were calculated to a0=0.8 and
a1=0.002.
8.2.2 Vibrations
The accelerations of the system related to the acceleration of gravity were obtained
with eq. (5.4). The results concerning the vertical vibrations of the structure can be
seen in Figure 8.6 and the results concerning lateral vibrations of the structure can be
seen in Figure 8.7. The accelerations were compared with two curves. The pink line
in Figure 8.6 is the baseline curve and the black line is the limit for offices and
residences [18]. The pink line in Figure 8.7 is the baseline curve [18]. Vibrations
under the baseline curve represent vibrations that cannot be felt by a human being.
The red lines in Figure 8.6 and Figure 8.7 represent the systems actual accelerations,
without the reduction coefficients found in Table 5.2.
50
Vertical vibrations
10
reduced vibrations
vibrations without reduction
baseline curve
offices, residences
1
a/g (%)
0,1
0,01
1 2 3 4 5 6 7 8 9 10
Frequency (Hz)
Figure 8.6: Vertical vibrations of the glass system.
Lateral vibrations
0,1
reduced vibrations
vibrations without reduction
baseline curve
a/g (%)
0,01
0,001
1 2 3 4 5 6 7 8 9 10
Frequency (Hz)
Figure 8.7: Lateral vibrations of the glass system.
51
8.3 Conclusions of the vibration analysis
For a glass floor in a public place the vibrations should not exceed the line which
corresponds to Indoor Footbridges, Shopping Malls and Dining and Dancing; as
shown in, Figure 8.3. The vertical vibrations obtained from the analysis were not only
below this limit but also below the limit corresponding to Offices and Residences in
Figure 8.3. The lateral vibrations had values that were even lower than the vertical
vibrations. A conclusion that can be made from the vibration analyses is that the glass
system will be well under the limit for both vertical and lateral vibrations given by
[18]. However the analysis was performed with some assumptions that may affect the
results.
One assumption was the damping ratios used for the rubber and interlayers. Damping
ratios vary between different structures and it is hard to obtain the correct value
without performing tests on the actual structure. Since the structure discussed in this
report has not been erected, experimental testing was not possible. The damping
ratios affect the vibrations and with different assumptions of damping ratios, different
results will be obtained. Also the damping ratio of the glass was neglected which
could have a minor affect on the results. Another assumption when performing the
analysis was the amount of people walking on the floor. One person per square metre
was assumed for the vertical vibrations and three people acting as concentrated loads
were assumed for the lateral vibrations. These values could be both higher and lower
than the assumed values and thus have an impact on the results.
At the ends of the beams, rubber boundaries were placed which can be seen in Figure
8.5. The glass is not fastened in these rubber boundaries but is held in place by
friction. In the case of dynamic loading this friction connection will allow the glass to
slide small distances relative to the rubber. However, in the vibration analyses the
glass was fastened all along the rubber boundaries which created a stiffer connection
than intended in reality. If the glass would be allowed to move small distances in the
rubber boundaries, it could result in slightly larger vibrations in the model.
It can be concluded from the red lines in Figure 8.6 and Figure 8.7 that the system did
not have any natural frequencies between 1-10 Hz. If a natural frequency would exist
in this frequency interval the system would exhibit resonance and the red line would
have had a peak. The lack of a natural frequency in or close to the interval 1-10 Hz is
another reason to that the accelerations were small.
52
9 Design of the whole system involving
boundaries
This chapter describes the system with dimensions for the complete floor structure.
Parts that have been described in earlier chapters are here shown added together with
figures. The purpose of this chapter is to give an understanding of the system and the
connections. The conclusions made concerning sizes of different sections in earlier
chapters are put together and the whole system is shown. Practical issues are
discussed and suggestions concerning a wear layer and erection method are presented.
9.1 Description of the system with dimensions

As verified in previous chapters, a single glass plate has a dimension of 1.5x0.5 m,
with a thickness of 8-12-12-8 mm glass, with 3x1.52 mm polymer interlayer in
between. The beams are 4 m long and 250 mm high. The cross section of the beams
consists of 3x15 mm heat strengthened glass plates with 2x1.52 mm SGP interlayer in
between. A quadratic profile of steel with dimension 15x15 mm2 runs along the
bottom of the centric glass plate, laminated together with the rest of the glass beam.
The steel is of type: S32205 (EN 14462).
The beams are simply supported on U-shaped boundaries at the ends. The boundaries
are connected to steel columns, for example VKR-40x40-2.5. However a closer
calculation of the columns has to be performed in order to verify the exact dimension.
At the location where two beams meet each other on a column there is a layer of
rubber to prevent the beams to interfere with each other. A basic detail of the section
can be seen in Figure 9.1.
The glass plates are glued to the beams using a silicon adhesive with rubber spacers
in between. In the connection between all the glass plates will be a soft silicon
adhesive.
Figure 9.1: Basic detail of connection between two beams.
A glass with a low content of iron will provide the best results in the intended glass
structure, since iron colours glass and give it a tone of green that can be undesirable
53
especially for thicker profiles. The green tone in normal glass gets stronger with an
increased thickness and will therefore probably dominate the appearance of the
structure.
9.2 Wear layer

On top of the glass plates a surface wear layer must be employed. This is to prevent
scratching of the glass plates and to allow impacts to some extent. Scratching is not
desirable as it can affect the durability of the structural strength of the glass plates. In
the case that the surface is exposed to a severe impact, a wear layer can be useful as it
can take the main concentrated part of the impact and smear it out to the other plates.
If a wear layer cracks it is rather easy and cheap to replace, whereas a replacement of
an entire glass plate is much more expensive. The surface wear layer will be a single
glass layer attached on top of the glass plates. It can be desirable to make the layer
thin since a thinner layer is easier to replace.
9.3 Attachments
The beams will be simply supported on U-shaped boundaries at the ends covered with
rubber. The boundary between the glass beams and the glass plates will consist of
silicone with rubber spacers. Between each glass plate, silicon is used as the adhesive.
No screws or bolts will be necessary at the ends of the beams, friction between the
rubber and the glass will keep the structure in place.
9.4 Erection of the system

When the system is to be erected, the columns are first put in place. The next step is
to install joist hangers at top of the columns. These will be clad in rubber as shown in
Figure 7.9. The beams will then be lifted into place on the rubber boundaries and will
hence be simply supported. After the beams are put in place the glass plates are to be
installed. The glass plates will be put on the rubber spacers on top of the beams and
glued to the beams using silicone. Silicone will also be used between each plate along
the beams. All the plates are to be simply supported just as the beams. Finally when
all the structural elements are in place the wear layer can be installed on top of the
system. Actions concerning safety against fire are discussed in Chapter 10. The glass
system containing glass plates, beams and rubber boundaries is shown in Figure 9.2
and Figure 9.3.
54
Figure 9.2: Section of the glass floor, seen from above.
Figure 9.3: Section of the glass floor, seen from below.
55
56
10 Design concerning resistance against fire
This chapter is dealing with fire and what a fire development looks like. Fire
resistance of the system and its ability to withstand a fully developed fire is here
taken into account. Actions to reduce a fire development are hence introduced.
10.1 Fire progression

A fire normally includes two important phases; initial fire and fully developed fire.
After this phases, a cooling phase follows. The temperature increase of a fire is rather
rapid and a few minutes are all that is needed for it to fully develop.
A fire in a building usually starts in the decor, for example in furniture or curtains.
This is often a result of carelessness with candles, cigarettes and the like [28]. During
a fire a smoke cloud rises towards the ceiling and a hot layer of fire gases forms. This
is followed by a quickly raising temperature which can lead to a flashover. A
flashover is when all the combustible materials and surfaces in the room catch fire,
the cloud of fire gases will drop towards the floor and flames will appear in the smoke
cloud. The temperature when this happens is about 500-600C and swiftly increasing
during this stage [28].
A flashover is followed by a fully developed fire, which can last for a few minutes up
to a few hours, depending on the supply of fuel and oxygen. The radiation levels
towards the floor will be about 15-20 kW/m2 during this stage, which is more than a
human being can endure [28]. How the fire develops in case of temperature and
intensity is decided by how much oxygen the fire has access to, how much material
that is available as fuel, and the ignition temperature of these. The cooling phase
begins when the fuel in the room runs out. The temperature and intensity of the fire
will then decrease over a period of one to a few hours [28].
10.2 Fire safety requirements

The Construction Products Regulation (CPR) contains seven essential requirements
for safety in buildings. One of these requirements concerns fire safety [27].
According to this requirement a building shall be designed and built so that the
following is achieved in the case of a fire [27]:
The constructions bearing capacity is assumed to be intact during a determined

time.
Development and spread of fire and smoke within the construction are limited.
Spreading of fire to neighbouring constructions is limited.
People who are in the construction can leave it or be saved in some other way.
Safety of rescue teams is considered.
What the above mentioned statements mean for the glass structure discussed in this
report is that the glass floor must be able to still carry load for a determined time
during a fire. People that are on the floor must at the same time be able to evacuate,
and the safety of the rescue team which will put out the fire, must be considered.
57
10.3 Fire resistance of glass
In order to construct a glass system that meets the requirements of fire safety it is
natural to look at other materials and how to make them fire resistant. Glass and steel
are normally considered bad materials considering fire safety, but fire resistant
buildings are constructed using steel as the load bearing material. To create fire safe
buildings in steel, the material can be cast into concrete or painted with a fire resistant
coating. These solutions protect the steel against too rapid a rise in temperature [8],
which gives the people in the building more time to evacuate. However these
solutions cannot be applied on glass since it is a transparent material and insulation or
paint would ruin the appearance and hence disregard the main reason to use glass as
construction material.
The National Research Council of Canada (NRCC) have done tests concerning the
effect of fire on annealed glass, heat strengthened glass and tempered glass [29].
When the annealed glass was exposed to high radiation it broke within minutes with a
temperature on the exposed side of 150-175C.
The heat strengthened glass and tempered glass did not crack until the temperature
reached 350C [29]. The radiation level used was 10 and 40 kW/m2 for plain glass
and 43 kW/m2 for heat strengthened glass and tempered glass.
The test carried out by NRCC also included a study where sprinklers were used to
verify which effect the thermal shock of water spray had on the glass. The results
from these tests were that the annealed glass could withstand the impact of water
spray at 80-90C, the heat strengthened glass at 150-165C and tempered glass at
200C [29]. If the glass specimens had a higher temperature than this, the glass would
crack when water spray reached the surface.
The tests carried out by the NRCC showed that the worst case scenario when using
sprinklers to put out a fire was when a small fire developed close to the glass. In this
scenario standard sprinklers were not efficient in protecting the glass, only fast
response sprinklers located close to the glass would activate early enough to protect
the glass from cracking [29].
10.4 Fire resistance of polymers

Normally polymers are not designed to work in extreme temperatures and will melt or
burn when a high temperature is reached in the material. The SGP interlayer has a
melting point at 94C [30]. When this temperature is reached its ability to act as an
interlayer is considerably decreased. EPDM rubber will lose its properties after 100C
[9] and will start to burn at about 300C. Silicone is normally heat resistant until
about 200C is reached [11].
10.5 Simulation of fire

A simplified simulation of a fire with an ISO 834 standard heating curve in the
software Argos was carried out. The surface temperature development in the glass
profile was calculated and plotted by the program with the assumption that the worst
case scenario is when a fire affects the glass beam on both sides. The analysis was
58
performed and plotted until the transition point of glass was reached, which is about
600C [1]. Figure 10.1 shows the temperature distribution through the glass beam
during a fire without any measures on site to decrease the influence of the fire. Figure
10.2 shows the surface temperature only, of the same beam.
As can be seen in Figure 10.1 and Figure 10.2 a temperature of 150C is reached after
about 90-120 seconds at the surface of the beam. The temperature 15 mm into the
beam at the laminate layer between the outer and the inner glass plates, has reached a
temperature of 100C after about 400 seconds. The temperature at the surface will at
the same time be about 400C.
Figure 10.1: Temperature in a cross section of a glass beam exposed to fire, plotted against time.
Figure 10.2: Surface temperature off a glass beam exposed to fire, plotted against time.
59
At the beginning of a fire the floor consisting of glass plates will most likely only be
affected by the fire on one side of the floor. As shown in Chapter 6, three glass plates
will be enough to carry the load acting on the plates in a worst case scenario. This
means that the durability of the glass plates probably will be longer than the durability
of the beams. On the other hand, if fire affects the plates on both sides the durability
time will reduce and hence reach a value below the one calculated for a beam.
10.6 Conclusion
Heat strengthened glass can withstand temperatures of 150-165C without cracking
when sprayed with water from a sprinkler system. However the interlayers can only
withstand temperatures of around 100C. To reach a temperature of 150C on the
surface, will take about 90-120 seconds in a normal fire development, and to reach a
temperature of 100C in the outermost interlayer will take about 400 seconds. The
temperature will thereby reach a critical level in the glass at the surface before it
reaches a critical level in the interlayer.
In order to prevent the structure from collapsing, the sprinkler system must be
activated 90 seconds after the fire starts and there must be sprinklers to spread water
above and underneath the floor. If a fire concentrated to a small area close to the glass
structure were to occur, only fast response sprinklers close to the glass structure
would be able to activate early enough to protect the glass [29]; however placing
sprinklers close to the glass floor can be problematic from an aesthetical and practical
point of view.
Another solution is to put the sprinklers underneath the glass floor and further away
on top of the floor. This would ensure an efficient sprinkler system close to the glass
at least considering the beams. This solution would protect the glass beams fully and
the plates from underneath. On the other hand, if the upper plate starts to crack due to
a slow acting sprinkler system, this would not be catastrophic, as the floor can carry
the load with less than four plates.
Ensuring that a bearing glass structure is fully resistant against fire is a hard task and
is not handled in this report. In order to ensure a fireproof glass structure the
temperature of the glass must be limited in some way, one possible solution is the
previously discussed one using sprinklers, but for this kind of structure the sprinkler
system must be very reliable and well built.
60
11 Final remarks
In this chapter the authors give a final conclusion considering the results in the report
and give some suggestions for further work.
11.1 Conclusions
Out of the results gained in this report the authors conclude that the glass system
presented will be able to carry the required loads concerning a congregation area
given by Eurocode. The system can handle the loads given without reaching critical
stress levels or too large deflections in the profiles, even when cracks are introduced.
The results also showed that no instability considering buckling will occur and that
the vibration levels in the system will be well below the limit presented in guidelines.
Considering fire resistance of the structure it is clear that further investigations must
be carried out. The investigations presented in this report however, showed that fire is
a critical factor that will be a hard task to solve, bearing in mind that the transparent
properties of the system must be preserved.
11.2 Further studies

The results presented in this report came from Abaqus models and no laboratory
testing has been done. To establish that the system works as intended by the models,
laboratory testing has to be conducted. A suggestion for further work is therefore to
construct the glass system presented in this report and to measure stresses,
deformations, instability and vibrations at site. In laboratory testing of the glass
system more accurate damping values can be found, which would enable a computer
model to present more accurate results.
Another suggestion for further work is to perform laboratory testing on the rubber
boundaries. These tests should be focused on the adhesion between the glass plates
and the glass beams, and the friction connection between the end of a beam and its
rubber boundary. This study would conclude if these boundaries would behave as
intended by the computer models.
The size and placement of the cracks in the system were based on a previous study [5]
and approximations made by the authors. The stresses in the system were critical
when cracks were present and since the size and placement of the cracks were
somewhat approximated, the dimensions of the glass plates and glass beams was also
approximated. In order to better optimize the use of material in the structure a more
extensive analysis would have to be carried out on the cracks in the system.
One big issue when constructing glass systems is to make them able to manage safety
requirements during a fire. Only a brief introduction to the effects of fire on glass and
a few suggestions on how to make a glass system safe against fire were given in this
report. To be able to build glass systems, solutions that ensure a glass systems safety
during a fire have to be developed and tested. Another suggestion for further work is
thus to find out how to improve the fire resistance of glass and how to construct a
system that can manage the various requirements given for a fire-proof structure.
61
62
12 Bibliography
[1] Flygt, Elisabeth et al. (2004), Boken om glas, Glafo
[2] Capital Glass. What is the difference between Annealed, Heat Strengthened, and
Tempered Glass?, Capital Glass, Inc.
http://www.capitalglassonline.com/FAQRetrieve.aspx?ID=39330, visited 2014-02-25
[3] Belis, Jan et al. (2012). Experimental Investigation of Unconventional Canopy

Prototypes, Suspended by Adhesive Bonds, Challenging Glass 3 Conference on
Architectual and Structural Applications of Glass
[4] Forserum Safety Glass AB. Vrmefrstrkt glas, Forserum Safety Glass AB.
http://www.fsglass.se/Svenska/Produkter/V%C3%A4rmef%C3%B6rst%C3%A4rktgl
as.aspx, visited 2014-02-25
[5] Louter, Christian et al. (2012). Reinforced Glass Beams Composed of Annealed,
Heat- Strengthened and Fully Tempered Glass, Challenging Glass 3 Conference on
Architectual and Structural Applications of Glass
[6] Terselius, Bjrn (2013). Polymer, Nationalencyklopedin,

http://www.ne.se/lang/polymer, visited 2014-02-25
[7] SGP (2008), SGP Plus ElasticProperties (SGP 5000), DuPont
[8] Burstrm, Per Gunnar (2003). Byggnadsmaterial, Studentlitteratur
[9] Trelleborg (2005). TEKNISK SPECIFIKATION MATERIAL: 7001 EPDM,

Trelleborg AB,
http://www.trelleborg.com/upload/Elastomer%20Laminates/Swedish%20version/pdf/
Teknspec_7001.pdf, visited 2014-02-25
[10] Terselius, Bjrn (2013). Lim, Nationalencyklopedin,

http://www.ne.se/lang/lim/241669, visited 2014-02-25
[11] MASTERBOND (2013). MasterSil 702 Product Description, Master Bond inc.
http://www.masterbond.com/tds/mastersil-702, visited 2014-02-25
[12] Brderna Edstrand [1999]. Rda katalogen: Teknisk handbook, Brderna

Edstrand AB
[13] Isaksson, Mrtensson (2010). Byggkonstruktion, Studentlitteratur
[14] Ottosen, Petersson (1992). Introduction to the Finite Element Method, Prentice
Hall
[15] Chopra, Anill K. (1995). DYNAMICS OF STRUCTURES, Prentice Hall
63
[16] Kursmaterial. (2013). Strukturdynamiska berkningar, VSMN10, Faculty of
Engineering at Lund University
[17] Delinc, Didier et al. (2008). Post-breakage behaviour of laminated glass in

structural applications, Challenging Glass Conference on Architectual and
Structural Applications of Glass
[18] Murray, Allen, Ungar (2003). Floor Vibrations Due to Human Activity,
American institute of steel Construction Inc.
[19] Dassault Systmes SIMULIA (2013). Abaqus 6.11 Documentation, Daussault

Systmes Inc S.A, http://abaqus.ethz.ch:2080/v6.11/, visited 2014-02-25
[20] European Standard (2013). Glass in building- Determination of the load

resistance of glass panes by calculation and testing, Technical Committee CEN/TC
129
[21] Kursmaterial. (2012). BALKTEORI, KFS AB
[22] SGP, SGP Plus vs. PVB for laminated Glass, DuPont.
http://www2.dupont.com/Building_Innovations/zh_CN/assets/downloads/SGPintro_E
.pdf, visited 2014-02-25
[23] Le Bourhis, Eric (2007). Glass, WILEY-VCH
[24] Lindley, P. B. (1978). Engineering design with natural rubber, Hertford Offset
Ltd
[25] Krenk, Steen [2009]. Non-linear Modelling and Analysis of Solids and
Structures,, Cambridge University Press
[26] Ottosen, Ristinmaa (2005). The Mechanics of Constitutive Modelling, Elsevier

Ltd
[27] Europarlamentetets och Rdets Frordningar [2011]. Byggproduktfrordningen

(CPR) 305/2011, Official Journal of the European Union
[28] Svensk Tr [2014]. TrgGuiden, Svenskt Tr.

http://www.traguiden.se/TGtemplates/GeneralPage.aspx?id=8046, visited 2014-02-25
[29] Kim, Lougheed [1990]. Journal of Fire Protection Engineering, SAGE
[30] SGP, SGP Brochure, DuPont.

http://www2.dupont.com/SafetyGlass/en_US/assets/pdfs/SGP_brochure.pdf, visited
2014-02-25
[31] Carlsson, Janne (2014). Brottmekanik, Nationalencyklopedin,

http://www.ne.se/lang/brottmekanik, visited 2014-02-19
64

Web 5192

Uploaded by

Copyright:

Available Formats

Web 5192

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Web 5192

Uploaded by

Copyright:

Available Formats

DESIGN OF A

GLASS FLOOR STRUCTURE

PONTUS DUFVENBERG and FREDRIK JNSSON

DIVISION OF STRUCTURAL MECHANICS

PONTUS DUFVENBERG and FREDRIK JNSSON

Copyright 2014 Division of Structural Mechanics

For information, address:

Lund, March 2014

Pontus Dufvenberg and Fredrik Jnsson

Balkarna analyserades med en statisk analys, samt med en instabilitetsanalys i

En analys gllande vibrationer i ett kombinerat system av balkar och plattor

En frenklad berkning genomfrdes gllande systemets motstndskraft mot brand

Slutligen frdes en diskussion gllande hela rapporten dr rekommendationer fr

1.2 Objective and method

2.1 Intended system

Figure 2.1: Sketch of the intended structure.

2.2 Reference work

Glass is a brittle material and therefore sensitive to cracking. By

Figure 2.3: Reference work four-point bending test setup [5].

3.1.1 Annealed glass

3.1.2 Heat strengthened glass

3.1.3 Tempered glass

3.1.4 Comparison and choice of glass material

3.2 Polymer interlayer

Figure 3.1: Stress-strain relation curve for SGP [17].

Glass is a brittle material which makes it sensitive to stress concentrations. Therefore

4.1 The Finite element method

where is the stiffness matrix, is the displacements and is the forces.

4.1.2 Modelling of linear-elastic materials

In modelling of plasticity concerning the material, theory is to be found in [26].

4.1.3 Equation of motion

where M is mass matrix, is the accelerations, K is the stiffness matrix, is the

4.1.4 Finite elements

4.1.5 Isoparametric finite elements

Figure 4.1: Twenty-node three dimensional isoparametric brick element [14].

4.2 Structural dynamics

4.2.2 Modelling rubber boundaries

4.2.3 Steady state

4.2.5 Rayleigh damping

Figure 4.2: Rayleigh damping [15].

4.4 Abaqus modelling

5.1 Design value of strength for heat strengthened glass

Parameter Value Source

5.2.2 Serviceability limit state

where is the permanent loading, is a factor concerning variable loads, and

The parameter varies depending on the weight of the structure. For a

Frequency interval (Hz) 1.6-3.2 3.2-4.8 4.8-6.4 6.4-8.8

All the calculated values of the node-accelerations related to the acceleration of

where is the weight of an average person estimated to 700 N, is the

6.1 Estimation of a glass plate

The maximum shear force of a simply supported beam subjected to a uniformly

6.1.1 Conclusion considering shear force

6.2 Analysis of stresses and deflections

6.2.2 Meshing of the glass plates

Figure 6.2: Meshing of a plate.

Comparison in vertical meshing

Figure 6.3: Comparison in vertical meshing.

6.2.3 Description of the analysis

Figure 6.4: Glass plates to be analysed.

Dimension ULS SLS