Académique Documents
Professionnel Documents
Culture Documents
shell structures
submitted by
Sushant Goel
in
November, 2014
Master
thesis
by
Sushant Goel
Declaration
This thesis is a presentation of my original work.
Wherever contributions of others are involved, every effort has been made to indicate
this clearly, with due reference to the literature, and acknowledgement of collaborative
discussions and research.
The submitted thesis was and is not part of any other examination procedure neither
completely nor in parts.
Thesis was published neither in whole nor in parts before.
The electronic copy is identical to the written, bound ones.
Institut fr Baustatik
und Baudynamik
Baustatik und Baudynamik
Institut fr Tragkonstruktionen
und Konstruktives Entwerfen
Prof. Dr.-Ing. J. Knippers
Master Thesis
Analysis of segmented shell structures
Monolithic shells made from concrete or other, appropriate materials have known many innovations
and widespread use particularly in between 1920 and 1970. Since then, rising costs for labour and
formwork as well as the shape and feasibility limits imposed by concrete material lead to a significant
decline of realizations in architecture. Fluent, curved glass facade or roof shapes are increasingly
demanded in recent free-form architecture. However, typical envelope materials such as glass, fiberreinforced polymers or metals can only be produced and shaped in limited size. Thus, assembly of
individual curved segments is necessary to obtain a larger, modular shell. Segmentation offers high
potential not only for application of new materials, but could also allow high-quality pre-fabrication of
shell segments and fast assembly on site without formwork.
The topic of modular or segmented shells is not yet well covered by research and necessitates
principal considerations concerning the interaction of joints, geometrical shape and load-bearing
behaviour. Based on the state of research, studies of these complex interdependencies focusing on
three basic shell shapes of a spherical dome, a cylindrical roof and a hyperbolic paraboloid (hypar) is
the main aim of the master thesis.
Abstract
Abstract
Monolithic glass shells can be constructed in limited sizes. Segmented shells allow coverage of larger spans. Three shell systems - spherical dome, cylindrical roof and hypar were
constructed. Four different segmentation patterns were made on these shell systems using
Rhinoceros, Grasshopper and MAPDL. Three joint materials were analysed- glass, silicone
(soft adhesive) and epoxy (hard adhesive). A Reissner-Mindlin finite element has been used
in ANSYS to discretize the shell geometry. The boundary conditions were setup keeping in
mind the favourable membrane behaviour of shell structures. It was found out that segmentation significantly influences the shell behaviour. Optimal patterns show similar behaviour
in comparison to monolithic glass shell. Others show a significant increase in the deflection or
stress resultant values and fall in critical buckling load values which is unwanted. The in-plane
membrane forces remain mostly unaffected. The bending moment and shear force values show
jumps at the joints due to a drop in material stiffness.
Keywords: Segmented shell, glass, MAPDL, Reissner-Mindlin, silicone, epoxy
Acknowledgement
Acknowledgement
I would like to thank Prof. Bischoff and Prof. Knippers for providing me an opportunity to get
associated with their institutes for this thesis project. I would like to thank my thesis advisors
Anne-Kathrin Schuble and Thiemo Fildhuth for their continuous support and feedback. The
two group meetings held with Prof. Bischoff, Anne-Kathrin Schuble and Thiemo Fildhuth
were very crucial for giving the right direction to this research work. I would also like to thank
Anne Bagger for her responses to my email queries about glass shells and her insistence on
using Rhinoceros-Grasshopper tools to construct segmented shell geometry.
Sushant Goel
ii
Contents
1. Introduction
1.1. Form . . . . . . . . . . . . . . . . . .
1.2. Material . . . . . . . . . . . . . . . .
1.3. Form and material: Glass shell . . .
1.3.1. Glass on steel grid . . . . . .
1.3.2. Glass bonding . . . . . . . . .
1.4. Motivation . . . . . . . . . . . . . .
1.4.1. Patterns in nature . . . . . .
1.4.2. Patterns in built environment
1.5. Thesis objectives . . . . . . . . . . .
1.6. Thesis outline . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
1
1
2
3
3
3
5
5
6
7
8
.
.
.
.
.
.
.
.
.
9
9
10
10
11
11
12
12
13
14
3. Geometry modelling
3.1. Shell dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15
15
2. Shell structures
2.1. Some definitions . . . . . . . .
2.2. Shell classification . . . . . . .
2.3. Evolution of shell formulations
2.4. Membrane momentless theory
2.5. Structural behaviour . . . . . .
2.5.1. Spherical dome . . . . .
2.5.2. Cylindrical roof . . . . .
2.5.3. Hypar . . . . . . . . . .
2.6. Linear stability analysis . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
iii
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Contents
3.2. Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.1. Method 1: Only MAPDL . . . . . . . . . . . . . .
3.2.2. Method 2: Rhinoceros and Grasshopper . . . . . .
3.2.3. Method 3: Only Rhinoceros . . . . . . . . . . . . .
3.2.4. Method 4: Rhinoceros and ANSYS DesignModeler
3.3. Discussion on modelling methods . . . . . . . . . . . . . .
3.4. Monolithic shell construction . . . . . . . . . . . . . . . .
3.4.1. Spherical dome . . . . . . . . . . . . . . . . . . . .
3.4.2. Barrel . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.3. Hypar . . . . . . . . . . . . . . . . . . . . . . . . .
3.5. Segmentation of monolithic shell . . . . . . . . . . . . . .
3.5.1. Patterns . . . . . . . . . . . . . . . . . . . . . . . .
3.5.2. Model accuracy . . . . . . . . . . . . . . . . . . . .
4. FE model
4.1. Material properties and loads . . . . .
4.2. Shell finite elements . . . . . . . . . .
4.2.1. Membrane shell element . . . .
4.2.2. Kirchhoff-Love shell element . .
4.2.3. Reissner-Mindlin shell element
4.3. Element properties . . . . . . . . . . .
4.3.1. Element size . . . . . . . . . .
4.3.2. Element coordinate systems . .
4.4. MAPDL script . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
16
16
17
17
17
17
18
18
19
19
20
20
21
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
24
24
25
25
25
25
26
26
27
28
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
29
29
30
30
30
31
32
34
34
36
38
.
.
.
.
40
41
41
41
41
. .
. .
N2
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
iv
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Contents
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
42
42
42
43
44
44
51
.
.
.
.
.
.
.
.
.
.
.
.
.
53
54
54
54
54
54
55
55
55
56
57
57
58
66
.
.
.
.
.
68
69
69
70
70
75
.
.
.
.
77
77
78
79
80
81
83
84
Bibliography
94
1
Introduction
Columns, beams, trusses, cables, arches and shells are the possible structural forms. Principally speaking, a structure channels loads to the ground (Adriaenssens u. a., 2014). A
smooth channelling of loads is shown by cables, shells and arches. This report focusses on
shell structures.
1.1. Form
Shell structures are intriguing, efficient and moody.
A shell personifies the dictum - Less is more (Robbin, 1996). They exist all around us- eggs,
car bodies, aircraft fuselage, beer cans, blood vessels, bones, human skull, roof coverings, etc.
Shell roofs exude lightness due to their slender form and can spread over large areas. A thin
curved roof providing shelter without any obstructing verticals or sharp corners gives a sense
of freedom to the inhabitant.
Shell structures are inherently optimised due to their shape. Each material fiber has equal
contribution in the transfer of transversal loads by in-plane (membrane) action. Shells are
1.2. Material
bendophobic structures and restrain from the disadvantageous state of bending (Ramm und
Wall, 2004).
Well-designed shells are mostly loaded in compression. The combination of compressive forces
and slender form is bad for shells as this can lead to stability failure without any warning.
Figure 1.3.: Other forms of shell roof - cylindrical and hypar (Paaskerk, 1963), (Structurae,
1935)
1.2. Material
Glass is fascinating, underused structurally and brittle.
Structural materials like steel, aluminium, wood make their presence felt unlike glass which
diffuses into the surrounding environment. It allows complete transfer of light and gives a
sense of lightness even though its density is same as concrete.
Glass has double the strength of mild steel in compression and five times more rigid than wood
but its not widely used as a structural material because of brittleness. One solution for this
is laminating thin glass sheets together (Wester, 1997).
Figure 1.4.: Apples iconic fifteen glass panel cube store in Manhattan, USA (Apple, 2012)
Line joint
Point joint
Fixed joint
Hinged joint
Joint in-between the plates
Joint over the plates
Joint under the plates
Joint on both sides of the plates
Figure 1.6.: Dome of the Reichstag building, Berlin, Germany (Nytimes, 2014)
Another option is gluing of glass segments with adhesives. Adhesives are monomer compositions which can join material surfaces together by forming polymers (Blandini, 2005). In
2002, designers at Delft University of Technology built a structural glass dome using flat glass
segments. A linear joint system was used with free edges to allow for tolerances. The joint
system is composed of aluminium strips glued onto the glass edges using a thick flexible adhesive. The aluminium strips are clamped together using two more strips and very small bolts
(Veer u. a., 2003), see Figure 1.7.
Lucio Blandini had the idea of constructing a frameless glass dome using only adhesive joints
without any discrete metallic clamping systems. This research project was undertaken at the
Institute for Lightweight Structures and Conceptual Design (ILEK), University of Stuttgart,
Germany. Numerous tests were conducted under tensile, shear and bending loads to choose an
optimal adhesive for a butt joint (Blandini, 2005). The end result of this research was a built
prototype in the year 2004, see Figure 1.8. It is a spherical calotte with a span of 8.5 meter,
curvature radius of 6 m and 1 cm thick laminated glass. The width and thickness of adhesive
butt joint is 1 cm. The prototype, though planned as a temporary structure, has stood for ten
1.4. Motivation
Overhead view
Figure 1.7.: Structural glass dome at TU Delft, Netherlands (Veer u. a., 2003)
years under the action of wind, symmetrical and asymmetrical snow loadings (Weller und
Tasche, 2014).
1.4. Motivation
1.4.1. Patterns in nature
Every natural structure has a pattern on a macro or micro level, see Figure 1.9. The pattern
on a giraffes skin or turtle shell have the same principle. Natural structures continuously
evolve (optimisation process) and adapt to the surroundings to reach an ideal performance
state (Dimcic, 2011).
Sea urchin shell is an example of a natural plate structure. The shell is composed of calcite
plates and three plates meet at a vertex. The plates transfer shear forces using the hinged
teeth along plate edges, see Figure 1.10 (Robbin, 1996).
The shape of a spiders web is governed by its functioning. The web absorbs the kinetic
energy of a flying insect by allowing large elongation of threads (strain energy). The multiple
1.4. Motivation
Leaf
Giraffe
Turtle
Spider web
redundancy of radial threads ensures that the web will function even if many radials break.
The load travels from the spirals to the supports via the radials (Rice, 1998).
The Eden project was built in 2001 in Cornwall, United Kigdom. The two greenhouses are
giant domes built up from steel structure which is filled in with EFTE foil air cushions. The
purpose of using hexagonal segments was to allow more light inside the dome. The domes
showed large deformations and a second structural layer was added to avoid this problem
(Aanhaanen, 2008).
The choice of segment shapes (square, rhombus, hexagon, etc.) usually depends on the aesthetics, feasibility of construction and cost effectiveness. Structurally speaking, which shape
is suitable for optimal shell behaviour?
2
Shell structures
2.1. Some definitions
Shell
Shell structures are form-passive 3-D curved surfaces which have one dimension significantly
smaller compared to the other two.
Form-passive
A form-passive structural system does not actively change its shape under varying load conditions. Cables and membrane structures are form-active systems (Adriaenssens u. a., 2014).
Thin shells
A shell is regarded as thin if the ratio of thickness to radius of curvature at any point is less
1
than 20
(Novozhilov, 1959).
Slenderness ratio
The ratio of shell thickness to a typical length measure (edge length or radius of curvature) is
called slenderness ratio.
Gaussian curvature
The Gaussian curvature is defined by the fraction R11R2 where R1 and R2 are the radii of
curvature in shell principal directions (Bischoff u. a.).
Stress resultants
There are total 8 stress resultants in shell theory. The total forces acting per unit length of
shell middle surface, N1 , N2 , N12 , Q1 , Q2 and the total moments acting per unit length of shell
middle surface, M1 , M2 , M12 . These are all integrals of stress over the thickness (Billington,
1965).
Negative
Zero
Surface
Doubly curved
Doubly curved
Singly curved
Name
Synclastic
Anticlastic
Developable
Example
Sphere
Hyperbolic paraboloid
Cylinder
Developability
Non-developable
Non-developable
Developable
10
multi-layer (for laminates) and multi-director were also presented. The multi-director model
takes into consideration the fact that shell normal doesnt remain straight during deformation
(Zingoni, 1997)(Bischoff u. a.).
Kirchhoff-Love
Reissner-Mindlin
7-parameter
Multi-director
11
In 1705, Richard Waller gave the solution to this anagram - Ut pendet continuum flexile, sic
stabit contiguum rigidum inversum (As hangs the flexible line, so but inverted will stand the
rigid arch). A hanging chain is in pure tension and without any bending moment. If it is
inverted, we get an arch shape which is under pure compression. This is a case of chain under
self-weight for which the ideal arch shape is a hyperbolic cosine (catenary). For a constant
load, the arch shape is a quadratic parabola. Thus, in case of arches, it is possible to attain
equilibrium without bending. But this works only for specific combinations of applied load
and arch geometry.
A dome can be thought of as a combination of arches along the meridian and hoop lines.
The meridian arches can carry any arbitrary (sufficiently smooth) load without bending. This
is made possible by the other set of hoop arches which act as stiff rings and dampen the
meridional bending (Adriaenssens u. a., 2014)(Bischoff u. a.).
Meridian arches
Hoop arches
Figure 2.3.: Spherical dome
12
Longitudinal beams
Transverse arches
Name
R/L
1
2
3
Long barrel
Intermediate barrel
Short barrel
< 0.4
Between 0.4 and 2.0
> 2.0
2.5.3. Hypar
Cable
Arch
Cable tension
Shear
Arch thrust
The hyperbolic paraboloid or hypar can be thought of as a combination of arches and cables
which are placed at right angles to each other. Arches carry loads in pure compression and
cables carry loads in pure tension. Along the entire edge of hypar, the normal components of
tension in the cable and compression in the arch are equal and opposite. Hence, these cancel
each other and the tangential components add up to give shear along edges. This shear force
needs edge members which can be loaded axially (Ramaswamy, 1984).
13
14
3
Geometry modelling
Geometry represents the design and is the starting point of all engineering simulation. The
goal is to construct three segmented shell systems- dome, barrel and hypar.
Number
Year
Location
1998
Duesseldorf
2002
Delft
2003
Stuttgart
2006
Barcelona
15
3.2. Modelling
3.2. Modelling
Four different modelling approaches were tried to get the required shell geometry.
P
Y
X
Z = Z (X ,Y )
1
1
Z = k11 X 2 + k12 XY + k22 Y 2 + Higher order terms
2
2
1
k11 =
R1
1
k22 =
R2
(3.1)
This method was implemented in MAPDL using a script file which is given in Appendix
A(Ramos und Hoogenboom, 2013). Amongst the various input parameters, k11 and k22
are the most important ones. As shown in Figure 3.3, same sign numerical values for these
parameters yield a synclastic surface, opposite sign values yield an anticlastic surface and if
either of the two values is zero, then the resulting surface is developable. The joint patterns
can be constructed on the shell surface by connecting the nodes in a sequence relevant to the
particular segmentation pattern.
16
1:
2:
3:
4:
5:
6:
7:
17
equation needs to be modified to have bottom of shell structure flat. Method 2 involves
creation of nodes and elements (meshing operation) in Rhinoceros. Thus, the user cannot
explore flexible meshing options provided by MAPDL and any change in elements or nodes at
a later stage would mean reworking the whole CAD file. Method 3 leads to significant data
loss due to incompatibility issues between IGES file format and MAPDL.
Method 4 is the suitable method to create segmented shell or other complicated geometries.
There is no data loss as MAPDL is highly compatible with ANF format.
18
2m
7.25 m
10 m
43.6
3.4.2. Barrel
Barrel was created by moving a straight line (generatrix) over an arc of a circle (directrix).
The curvature radius of this arc is 7.25 m and height is 2 m. Longitudinal span is 10 m, chord
length (projection of circular arc) is 10 m and cutting angle is 43.6 . The slenderness ratio of
barrel is 1:725.
2m
Traverses
7.25 m
Span=10 m
43.6
Chord length= 10 m
Figure 3.6.: Barrel geometry
3.4.3. Hypar
The hypar is formed by running a set of lines, which are parallel on plan, across two edge
lines which are also parallel on plan but not coplanar. Such a surface contains another set of
lines parallel on plan to the two edge lines. These two set of lines are known as straight line
generators (Orton, 2013). In Rhinoceros, a 10 m x 10 m surface was constructed on a plane.
Two opposite corners of this surface were raised by 2 m while the other two corners remained
at the original level. The ratio of thickness to edge length is 1:1000.
19
2m
10 m
10 m
Figure 3.7.: Hypar geometry
1.5 m
0.23 m
0.2 m
0.93 m
SQUARE
DIAMOND
HEXAGON
0.93 m
HEXALOCK
3.5.1. Patterns
Four symmetric patterns were chosen for segmenting the shells and each pattern comprises of
repeating glass units, see Figure 3.8. These have been named as square, diamond, hexagon
and hexalock patterns. The hexalock pattern was constructed using a Grasshopper visual
script given in Appendix B. The projection of square and diamond patterns on dome resulted
in similar segmentations. Therefore, the pie pattern was additionally constructed for the
dome. Each repeating unit (glass pane) has 20 mm wide rectangular strips (adhesives) along
the edges. The joints between glass panes should not be continued around the pane vertices
(Aanhaanen, 2008). Local damage and subsequent repair or replacement of any joint should
not affect the neighbouring joints. The triangular areas (in red) at pane vertices have been
removed from the geometry model, see Figure 3.9.
The size of glass pane can influence the membrane behaviour of shell as a very big pane can
develop high local bending stresses. This was taken into consideration while choosing the pane
size. The planar surface area of each glass unit is 2.25 m 2 .
20
SQUARE
DIAMOND
HEXAGON
HEXALOCK
20 mm
20 mm
PIE
Figure 3.9.: Patterns used for segmentation
21
draw all patterns directly on curved shell surface using OffsetCrvOnSrf command. This process is very complicated for hexagon and hexalock patterns. Therefore, the vertical projection
method was chosen. Figure 3.11 shows the size deviation in glass panes and adhesive strips.
These size deviations are spread symmetrically on the shell surface.
20.11 mm
1502 mm
1541 mm
1679 mm
21.06 mm
The twelve segmented shells with symmetric patterns are shown in Figure 3.12 - 3.14.
22
23
4
FE model
Over the past sixty years, numerical solution methods (notably the finite element method) have
become more popular as compared to finding analytical solutions of rigorous shell formulations.
It is of utmost importance to understand the properties and behaviour of a finite element
making up the whole structure (Zingoni, 1997). The goal is to set up a FE model of segmented
shell geometry constructed in the previous chapter. FE package ANSYS Mechanical 13.0 has
been used for the analysis.
G=
E
2(1 + )
(4.1)
Material
Density [kg/m 3 ]
Poissons ratio
Glass
Epoxy
Silicone
2500
1100
1100
70000
1500
10
0.23
0.35
0.46
28455
556
3.4
24
25
SY
SX
SX(TOP)
SX(MID)
SX(BOT)
2-direction
Q13
M11
3-direction
M12
N11
M11-axis
Q23
M22
N12
M21
M22-axis
1-direction
N12
N22
STRESS RESULTANTS
26
This could result in bad element shapes and there is only one element in the joint width
direction. Therefore, the joints were meshed first followed by glass panes.
0.0942
0.094
Max. von Mises stress (MPa)
0.0938
0.094
0.0936
0.0937
0.0934
0.0932
0.093
0.0931
0.0928
0.0929
0.0929
0.0929
50
40
0.0926
0.0924
0.0922
200
150
100
75
Element size (mm)
Figure 4.2.: Maximum von Mises stress convergence for spherical dome
-0.017
-0.0178
-0.018
-0.019
-0.02
-0.0214
-0.021
-0.022
-0.023
-0.024
-0.024
-0.0249
-0.025
-0.025
-0.0251
50
40
-0.026
200
150
100
75
Element size (mm)
27
70000
60795
Number of elements
60000
50000
39646
40000
30000
18871
20000
10926
10000
3205
5347
0
200
150
100
75
50
40
used for the element x-axis direction (ANSYS, 2009). This error was corrected by placing the
topmost centre point of shell surface at coordinates 0,0,7250 (in mm). This is in accordance
with the curvature radius of dome or barrel and facilitates application of boundary conditions
that are tangential to shell mid-surface.
Cylindrical
Cartesian
28
5
Boundary conditions and model
verification
Thin shells should be designed to carry loads primarily through membrane forces. Bending
due to boundary conditions should be avoided by selecting an optimal support system for the
shell, although in practice it is not possible to completely ignore bending (Bischoff u. a.).
The goal is to set up such boundary conditions for dome, barrel and hypar.
Extensional deformation
Inextensional deformation
In Figure 5.1, a small plastic spherical cap is loaded with a force. In case of extensional
deformation, the shell mid-surface is extending and load is carried mostly by membrane forces.
In case of inextensional deformation, the load is mostly carried by bending. It is important
to avoid inextensional deformations in thin shells as large bending deformation can lead to
damage (Hoogenboom). The flat surface on which the cap is placed in first case is equivalent
29
to supporting the cap edge with vertical supports. This clearly shows that the boundary
conditions influence the deformation behaviour of shell.
F
Nm
Ring beam
R
Figure 5.2.: Usual practice of supporting domes
Another option is to provide tangential hinge support to shell mid-surface, see Figure 5.3.
In this case, the inclined rollers permit free movement of shell edge in radial direction, the
hinge allows free rotation of edge in the meridional plane. The shear forces do not eminate
and bending is not needed for equilibrium (Zingoni, 1997). Ring beam is not required and
the dome is in an ideal membrane state. This type of boundary condition has been used for
the dome FE models. Three additional hinge supports are provided at shell edge (parallel to
hoop direction) to avoid rigid body modes.
30
5.3.2. Barrel
Along the longitudinal edge of barrel, tangential hinge support is provided to shell mid-surface.
Such supports are not effective along the traverses. It is usually assumed that thick barrels are
simply supported (vertical hinge) at the traverses (Ramaswamy, 1984). For monolithic glass
barrel, two types of boundary conditions for traverses were implemented in the FE model for
comparison- simple support and circumferential hinge support, see Figure 5.4.
Simple support
Circumferential support
The FE plots show top view of barrel with traverses on left and right side.
Deflections (Figure 5.5)
The maximum deflection value of -0.26 mm is spread over a large portion of simply supported
traverses. The same value is localized in case of circumferentially supported traverses and
major portion shows a deflection value of -.079 mm. This observation points towards making
use of circumferential supports at the traverses.
Bending moment in transverse direction, M1 (Figure 5.6)
In simply supported traverses, significantly higher moment values ranging from -8.5 Nmm/mm
to 4.6 Nmm/mm are observed. The four corners of barrel show high bending moments. Using
the circumferential supports, the four corners show moment values of -1.17 Nmm/mm which
is much lower. This observation shows the compatibility of circumferential supports with
tangential supports at corners of barrel.
Bending moment in longitudinal direction, M2 (Figure 5.7)
In simply supported traverses, the moment value at the traverses vary from -17.3 Nmm/mm
(hogging) at centre to 23.2 Nmm/mm (sagging) at corners. This change of moment sign is
31
Simple support
Circumferential support
Figure 5.5.: Deflection values for different traverse supports [in mm]
Simple support
Circumferential support
not observed in the other barrel and it shows a much lower value of -0.03 Nmm/mm at the
traverses.
From these observations, it is evident that barrel with simply supported traverses shows higher
bending moments. This stress state is undesirable and thus circumferentially supported traverses are chosen for barrel FE models. Two additional hinge supports are provided at the
centre of each longitudinal edge (parallel to longitudinal direction) to avoid rigid body modes,
see Figure 5.8.
5.3.3. Hypar
The choice of most suitable boundary conditions for hypar is based on its loading behaviour,
see Section 2.5.3. Shear supports along all the four edges were set up in FE model. The
analysis did not converge due to insufficiently constrained model. Vertical hinge supports were
added at the two upper corners. This made the solution converge but hypar showed large
32
Simple support
Circumferential support
deformations and unsymmetrical behaviour. Therefore, two more vertical hinge supports were
added at the lower corners, see Figure 5.9.
33
Figure 5.10.: Deflection values for segmented glass dome with glass joints [in mm]
34
Figure 5.11.: Deflection values for segmented glass shell with glass joints [in mm]
35
11t = R(
1
cos )
1 + cos
1
cos 0)
1 + cos 0
(5.1)
1
cos 43.6)
1 + cos 43.6
(5.2)
1
cos )
1 + cos
36
Barrel
Using equilibrium equations of barrel, the membrane stress in transverse direction is calculated,
see Equation 5.3 and 5.4 (Ramaswamy, 1984). The values from FE model are matching
and shown in Figure 5.13. 11t is stress at top of barrel (dark blue in plot) and 11b is at
longitudinal edge (light blue).
11t = R cos
11t = (25000 109 ) 7250 cos 0
(5.3)
11t = 0.1812MPa
11b = R cos
11t = (25000 109 ) 7250 cos 43.6
(5.4)
11b = 0.1312MPa
Figure 5.13.: Transverse direction membrane stress (11 ) in segmented barrel with glass joints
Hypar
The in-plane shear stress (12 ) in hypar is calculated using Equation 5.5 (Ramaswamy, 1984).
This formula is valid when one corner of hypar is pulled down and other three corners remain
at original level.
a - Projected length of hypar
b - Projected width of hypar
f - Distance between upper and lower corners
37
ab
2f
25000 109 10000 10000
=
2 2000
= 0.625MPa
12 =
12
12
(5.5)
When two corners are pulled down, the shear stress is halved. This is in agreement with shear
stress plot shown in Figure 5.14.
12 =
0.625
= 0.312MPa
2
(5.6)
Figure 5.14.: In-plane shear stress (12 ) in segmented hypar with glass joints
38
2
t2
p
2
2
R
3(1 2 )
102
2
= 0.4 70000
p
2
2
7250
3(1 0.222 )
cr = Ck E
cr
cr = 0.06306MPa
g = 25000 108 MPa
cr
BLF =
g
63060 106
BLF =
25000 108
BLF = 252.2
39
(5.7)
6
Results for segmented dome
The FE results for segmented domes are presented in this chapter. Different patterns and
joints have been compared using graphs for paths along meridian and hoop directions. Four
paths were constructed out of which results from two have been presented, see Figure 6.1.
Square
Pie
Hexagon
Hexalock
Figure 6.1.: Meridian (marked in green) and hoop (marked in black) paths of segmented dome
40
In some graphs, the plotted curves are discontinuous. These discontinuities occur when the
constructed path passes over the glass pane vertices where no material exists. The curve in
black colour represents the monolithic glass dome.
41
in the joints as shown in Figure 6.2. The hexagon pattern develops least compressive force of
-0.65 N/mm in glass panes.
Hexagon
Hexalock
Figure 6.2.: Segmented dome with silicone joints: Contour plot of hoop in-plane force, N1
42
Hexagon
Pie
Figure 6.3.: Segmented dome with silicone joints: Contour plot of meridional bending moment,
M2
Hexagon
Pie
Figure 6.4.: Segmented dome with silicone joints: Bending moment in hoop direction, M1
43
Square
Pie
Figure 6.5.: Segmented dome with epoxy joints: Meridional transverse shear force, Q23
Square
Pie
Figure 6.6.: Segmented dome with epoxy joints: Hoop transverse shear force, Q13
44
Square
Glass
Hexagon
Pie
Hexalock
0.2
Deflection (mm)
-0.2
-0.4
-0.6
-0.8
-1
-1.2
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
0.01
0.005
Deflection (mm)
-0.005
-0.01
-0.015
-0.02
-0.025
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
45
10000
11000
Square
Glass
Hexagon
Pie
Hexalock
3
2.5
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
-3.5
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
Figure 6.9.: Segmented dome with silicone joints: Meridional in-plane force, N2
-0.2
-0.4
-0.6
-0.8
-1
-1.2
-1.4
-1.6
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Figure 6.10.: Segmented dome with epoxy joints: Meridional in-plane force, N2
46
11000
Square
Glass
Hexagon
Pie
Hexalock
-0.3
-0.4
-0.5
-0.6
-0.7
-0.8
-0.9
-1
-1.1
-1.2
-1.3
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
Figure 6.11.: Segmented dome with silicone joints: Hoop in-plane force, N1
-0.45
-0.5
-0.55
-0.6
-0.65
-0.7
-0.75
-0.8
-0.85
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
Figure 6.12.: Segmented dome with epoxy joints: Hoop in-plane force, N1
47
20000
Square
Glass
Hexagon
Pie
Hexalock
-1
-2
-3
-4
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
Figure 6.13.: Segmented dome with silicone joints: Meridional bending moment, M2
0.3
0.2
0.1
-0.1
-0.2
-0.3
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Figure 6.14.: Segmented dome with epoxy joints: Meridional bending moment, M2
48
11000
Square
Glass
Hexagon
Pie
Hexalock
2.5
1.5
0.5
-0.5
-1
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
Figure 6.15.: Segmented dome with silicone joints: Hoop bending moment, M1
0.3
0.2
0.1
-0.1
-0.2
-0.3
2000
4000
6000
8000
10000
12000
14000
16000
18000
Figure 6.16.: Segmented dome with epoxy joints: Hoop bending moment, M1
49
20000
Square
Glass
Pie
0.0025
0.002
0.0015
0.001
0.0005
0
-0.0005
-0.001
-0.0015
-0.002
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
Figure 6.17.: Segmented dome with epoxy joints: Meridional transverse shear force, Q23
0.0004
0.0003
0.0002
0.0001
-0.0001
-0.0002
-0.0003
-0.0004
2000
4000
6000
8000
10000
12000
14000
16000
18000
Figure 6.18.: Segmented dome with epoxy joints: Hoop transverse shear force, Q13
50
20000
Monolithic glass dome has a BLF of 277.2, see figure 6.19. The BLF for segmented dome with
silicone and epoxy joints is shown in Figure 6.20.
For silicone joints, the square pattern shows global buckling with highest BLF of 29.8. The
glass panes govern buckling in square pattern and hence give maximum BLF. All other patterns
show local buckling of silicone joints. The hexalock pattern shows the least BLF of 5.9.
For epoxy joints, the pie pattern undergoes global buckling with highest BLF of 240.6. The
hexagon pattern shows the least BLF of 172. Most patterns show buckling near dome edge and
the epoxy joints do not govern buckling. This behaviour is comparable to the edge buckling
shown by monolithic glass dome.
51
Square, 29.8
Square, 206.3
Pie, 19.6
Pie, 240.6
Hexagon, 20.0
Hexagon, 172.0
Hexalock, 5.9
Hexalock, 186.0
Figure 6.20.: BLF for segmented domes with silicone (left) and epoxy (right) joints
52
7
Results for segmented barrel
The FE results for segmented barrels are presented in this chapter. Different patterns and
joints have been compared using graphs for paths along longitudinal and transverse directions.
In all graphs, the black colour curve represents monolithic glass barrel. Three paths were
constructed out of which results from two have been presented, see Figure 7.1.
Square
Diamond
Hexagon
Hexalock
Figure 7.1.: Longitudinal (marked in green) and transverse (marked in black) paths of barrel
53
54
Silicone joints
Epoxy joints
55
Square
Diamond
Figure 7.3.: Segmented barrel with silicone joints: Longitudinal bending moment, M2
Hexagon
Hexalock
Figure 7.4.: Segmented barrel with epoxy joints: Longitudinal bending moment, M2
56
Diamond
Hexalock
Figure 7.5.: Segmented barrel with silicone joints: Transverse bending moment, M1
Square
Diamond
Figure 7.6.: Segmented barrel with epoxy joints: Transverse shear force in longitudinal direction,
Q23
57
Hexagon
Square
Figure 7.7.: Segmented barrel with epoxy joints: Transverse shear force in transverse direction,
Q13
58
Square
Monolithic
Hexagon
Diamond
Hexalock
1
0
-1
Deflection (mm)
-2
-3
-4
-5
-6
-7
-8
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Figure 7.8.: Segmented barrel with silicone joints: Deflection values along longitudinal path
0.06
0.05
0.04
Deflection (mm)
0.03
0.02
0.01
0
-0.01
-0.02
-0.03
-0.04
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Figure 7.9.: Segmented barrel with epoxy joints: Deflection values along longitudinal path
59
Square
Monolithic
Diamond
Hexagon
Hexalock
20
15
Deflection (mm)
10
-5
-10
-15
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
Figure 7.10.: Segmented barrel with silicone joints: Deflection values along transverse path
0.8
0.6
Deflection (mm)
0.4
0.2
-0.2
-0.4
-0.6
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
Figure 7.11.: Segmented barrel with epoxy joints: Deflection values along transverse path
60
Monolithic
Square
Hexagon
Diamond
Hexalock
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
-3.5
-4
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Figure 7.12.: Segmented barrel with silicone joints: Longitudinal in-plane force, N2
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.7
-0.8
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
Figure 7.13.: Segmented barrel with epoxy joints: Longitudinal in-plane force, N2
61
10000
Square
Monolithic
Diamond
Hexagon
Hexalock
0.5
-0.5
-1
-1.5
-2
-2.5
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
Figure 7.14.: Segmented barrel with silicone joints: Transverse in-plane force, N1
-1
-1.2
-1.4
-1.6
-1.8
-2
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Figure 7.15.: Segmented barrel with epoxy joints: Transverse in-plane force, N1
62
11000
Square
Monolithic
Hexagon
Diamond
Hexalock
-2
-4
-6
-8
-10
-12
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Figure 7.16.: Segmented barrel with silicone joints: Longitudinal bending moment, M2
0.6
0.4
0.2
-0.2
-0.4
-0.6
-0.8
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Figure 7.17.: Segmented barrel with epoxy joints: Longitudinal bending moment, M2
63
Square
Monolithic
Hexagon
Diamond
Hexalock
10
-5
-10
-15
-20
-25
-30
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
Figure 7.18.: Segmented barrel with silicone joints: Transverse bending moment, M1
-1
-2
-3
-4
-5
-6
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Figure 7.19.: Segmented barrel with epoxy joints: Transverse bending moment, M1
64
11000
Monolithic
Square
Hexagon
Diamond
0.006
0.004
0.002
-0.002
-0.004
-0.006
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Figure 7.20.: Segmented barrel with epoxy joints: Transverse shear force along longitudinal path,
Q23
0.009
0.006
0.003
-0.003
-0.006
-0.009
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
Figure 7.21.: Segmented barrel with epoxy joints: Transverse shear force along transverse path,
Q13
65
Monolithic glass barrel has a BLF of 8.62, see Figure 7.22. The BLF for segmented barrel
with silicone and epoxy joints is shown in Figure 7.23.
For silicone joints, the square pattern shows instability with the lowest BLF of 0.59. This
means that the applied load already exceeds the estimated critical buckling load. The hexalock
pattern shows the highest BLF of 2.22 and hence the most stable of all patterns.
For epoxy joints, the square pattern shows the lowest BLF of 5.48. Unlike with silicone joints,
the square pattern with epoxy joints show stability. The hexalock pattern shows the highest
BLF of 6.88.
66
Square, 0.59
Square, 5.48
Diamond, 1.84
Diamond, 6.66
Hexagon, 1.86
Hexagon, 6.67
Hexalock, 2.22
Hexalock, 6.88
Figure 7.23.: BLF for segmented barrels with silicone (left) and epoxy (right) joints
67
8
Results for segmented hypar
The FE results for segmented hypar are presented in this chapter. Different patterns and
joints have been compared using graphs for paths along principal hypar directions. Two paths
were constructed - one connecting the upper corners and other connecting the lower corners
of hypar, see Figure 8.1.
Square
Diamond
Hexagon
Hexalock
Figure 8.1.: Path between upper corners of hypar (marked with black arrow)
68
The two paths show same but opposite behaviour - one in tension and other in compression.
Results from the path connecting upper corners (in tension) have been presented. In all graphs,
the black colour curve represents monolithic glass hypar.
Square
Hexalock
69
have a better comparison of other patterns. The hexagon pattern shows high tensile force
value of 3.98 N/mm in the glass panes. The square pattern show the least tensile force of 3.63
N/mm.
Hexalock
Diamond
Figure 8.3.: Segmented hypar with silicone joints: Contour plot of bending moment
70
Square
Monolithic
Diamond
Hexagon
Hexalock
15
-5
Deflection (mm)
-15
-25
-35
-45
-55
-65
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
12000
13000
14000
13000
14000
1
0.8
0.6
0.4
Deflection (mm)
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
-1.2
-1.4
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
12000
71
Square
Monolithic
Diamond
Hexagon
Hexalock
10
-2
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
12000
13000
14000
Figure 8.6.: Segmented hypar with silicone joints: In-plane shear force, N12
5.5
4.5
3.5
2.5
1.5
0.5
-0.5
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
12000
13000
Figure 8.7.: Segmented hypar with epoxy joints: In-plane shear force, N12
72
14000
Square
Monolithic
Diamond
Hexagon
Hexalock
10
7.5
2.5
-2.5
-5
-7.5
-10
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
12000
13000
14000
13000
14000
-2
-4
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
12000
73
Square
Glass
Diamond
0.0006
0.0004
0.0002
-0.0002
-0.0004
-0.0006
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
Figure 8.10.: Segmented hypar with silicone joints: Transverse shear forces
0.0006
0.0004
0.0002
-0.0002
-0.0004
-0.0006
-0.0008
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
Figure 8.11.: Segmented hypar with epoxy joints: Transverse shear forces
74
Initial BLFs for monolithic and segmented hypar were less than -1 which implies buckling is not
expected even if the loads are reversed (SolidWorksHelp). This stability against buckling
is due to the tensile forces carried by hypar in one of the principal directions.
Monolithic glass hypar has the first positive BLF of 3.33, see Figure 8.12. The BLF for
segmented hypar with silicone and epoxy joints is shown in Figure 8.13. For silicone joints,
the hexalock pattern shows the highest first positive BLF. For epoxy joints, the hexagon
pattern shows the highest first positive BLF of 2.95.
75
Square, 0.108
Square, 2.33
Diamond, 0.37
Diamond, 2.66
Hexagon, 0.28
Hexagon, 2.95
Hexalock, 0.84
Hexalock, 2.57
Figure 8.13.: BLF for segmented hypar with silicone (left) and epoxy (right) joints
76
9
Conclusions and recommendations
Segmented shells with various patterns (square, diamond/pie, hexagon and hexalock) were
analysed in this report. CAD tools like Rhinoceros and Grasshopper were used to model
the geometry of segmented shells. ANSYS package was used for FE analysis and a ReissnerMindlin finite element was chosen to discretize the geometry model.
Segmentation does have an effect on the shell behaviour. Some patterns show similar behaviour
in comparison to monolithic glass shell. Others show a significant increase in the deflection
or stress resultant values which is unwanted. The in-plane membrane forces mostly remain
unaffected. The bending moment and transverse shear force values show jumps at the joints
due to a drop in material stiffness.
Conclusions have been drawn for each shell system separately. The tables given below show
Force results for segmented shell
Forcewith silicone joints
the results for monolithic glass shell and optimal
In-plane force (N/mm)
Monolithic
Silicone
increase
Pattern
Epoxy
increase
Pattern
and epoxy joints. The increase factor is thefactor
ratio of monolithic glass shellfactor
results to segmented
shell Meridian
results.direction
The recommendations
for continuation
of research -1.02
work on segmented
shells have
-1.03
-1.09
1.1
Hexagon
1.0
Square/Hexagon
Hoop direction
-0.65
1.0
Hexagon
-0.65
1.0
Hexagon
been made
at the end. -0.65
Monolithic
Silicone
Moment
increase
factor
Pattern
Epoxy
Moment
increase
factor
Pattern
Meridian direction
-0.012
-1.05
87.5
Hexagon
0.109
9.1
Hexagon
Deflection (mm)
Monolithic
Silicone
Deflection
increase
factor
Pattern
Epoxy
Deflection
increase
factor
Pattern
Meridian direction
-0.012
-0.73
60.8
Hexalock
-0.019
1.6
Hexagon
moment
9.1.Bending
Dome
(Nmm/mm)
The epoxy joints show optimal deflection behaviour with hexagon pattern. The silicone joints
Hoop direction
-0.005
-0.2
40.0
Pie
-0.018
3.6
Pie
show much higher deflection, see Table 9.1.
77
9.2. Barrel
Monolithic
Silicone
Force
increase
factor
Pattern
Epoxy
Force
increase
factor
Pattern
Meridian direction
-1.03
-0.65
-1.09
-0.65
1.1
1.0
Hexagon
Hexagon
-1.02
-0.65
1.0
1.0
Square/Hexagon
Hexagon
Hoop direction
Monolithic
Silicone
Moment
increase
Pattern
factor
are independent
Moment
increase
factorwhen
stiffness
Epoxy
-0.005
-0.2
40.0
Force
Pie
9.1
3.6
Force
-0.018
Pattern
the hexagon
Hexagon
Pie
TheIn-plane
low force
bending
values Silicone
of monolithic
dome show
it is in a Pattern
membrane
(N/mm) moment
Monolithic
increase glass
Pattern
Epoxy that
increase
factor
factor
state. The hexagon pattern is optimal for bending in meridian direction and the pie pattern
Meridian direction
-1.03
-1.09
1.1
Hexagon
-1.02
1.0
Square/Hexagon
is optimal for bending in hoop direction. Silicone joints are not suitable for maintaining the
Hoop direction
-0.65
-0.65
1.0
Hexagon
-0.65
1.0
Hexagon
membrane state of a segmented dome , see Table 9.3.
Bending moment
(Nmm/mm)
Meridian direction
In-plane
(N/mm)
Hoop force
direction
Meridian direction
Hoop direction
Monolithic
Silicone
-0.012
-1.05
Monolithic
-0.005
Silicone
-0.2
Moment
increase
factor
87.5
Force
increase
40.0
factor
Pattern
Epoxy
Hexagon
Pattern
Pie
0.109
Epoxy
-0.018
Moment
increase
factor
9.1
Force
increase
3.6
factor
-1.09dome 1.1
Hexagon Bending
-1.02
1.0
Table-1.03
9.3.: Optimal
segmentation:
moments
-0.65
-0.65
1.0
Hexagon
-0.65
1.0
Pattern
Hexagon
Pattern
Pie
Square/Hexagon
Hexagon
The square pattern with epoxy joints shows the least transverse shear force value.
Moment
Moment
The pie
pattern
Bending
momentwith epoxy joints shows the highest BLF of 240.6. The buckling stability is
Monolithic
Silicone
increase
Pattern
Epoxy
increase
Pattern
(Nmm/mm)
highly dependent
on joint stiffness as the value
joints.
factor drops to 19.6 with silicone
factor
Meridian direction
Hoop direction
-0.012
-0.005
-1.05
-0.2
87.5
40.0
Hexagon
Pie
0.109
-0.018
9.1
3.6
Hexagon
Pie
9.2. Barrel
Deflection
Deflection
Deflection (mm)
Monolithic
Silicone deflection
increase
Pattern the transverse
Epoxy
increase
Pattern
The monolithic
glass barrel
shows higher
along
direction compared
to
factor
factor
the longitudinal
direction. The hexalock pattern is optimal for silicone joints and the square
Meridian direction
-0.012
-0.73
60.8
Hexalock
-0.019
1.6
Hexagon
pattern is optimal for epoxy joints, see Table 9.4.
Deflection (mm)
Monolithic
Silicone
Deflection
increase
factor
Pattern
Epoxy
Deflection
increase
factor
Pattern
Longitudinal direction
-0.018
-0.27
-4.3
-5.5
238.9
20.4
Hexalock
Hexalock
-0.024
-0.39
1.3
1.4
Square
Square
Transverse direction
The longitudinal and transversal in-plane forces remain unaffected with epoxy joints when the
`
square pattern is used. The forces increase in the longitudinal direction with silicone joints,
see Table 9.5.
78
Hoop direction
-0.012
-0.012
-0.005
-0.73
-1.05
-0.2
Deflection (mm)
Monolithic
Silicone
Deflection direction
(mm)
Longitudinal
Monolithic
-0.018
Silicone
-4.3
Transverse direction
Meridian direction
-0.27
-0.012
-5.5
-0.73
Monolithic
Silicone
Deflection direction
(mm)
Longitudinal
Monolithic
-0.56
Silicone
-1.18
Meridian direction
Meridian direction
-1.76
-1.82
-0.018
-4.3
-0.27
-5.5
Table 9.5.: Optimal
Transverse direction
Longitudinal direction
Transverse direction
Bending moment
(Nmm/mm)
Monolithic
factor
60.8
87.5
40.0
Hexalock
Hexagon
Pie
Deflection
increase
Pattern
factor
Deflection
9.3.
Hypar
increase
Pattern
238.9
Hexalock
factor
20.4
Hexalock
60.8
Force
increase
factor
Deflection
increase
2.1
factor
-0.019
0.109
-0.018
Epoxy
Hexalock
Epoxy
-0.024
-0.39
-0.019
Pattern
Epoxy
factor
1.6
Hexagon
Hexagon
Pie
9.1
3.6
Deflection
increase
factor
Deflection
increase
1.3
factor
1.4
Pattern
Pattern
Square
Square
Hexagon
1.6
Force
increase
factor
Deflection
increase
1.0
factor
Pattern
Pattern
Epoxy
Square
-0.57
1.0
Square
-1.75
1.0
Hexalock
-0.024
1.3
` 238.9
20.4segmentation:
Hexalock In-plane
-0.39 forces
1.4
Moment
Moment
barrel
Silicone
increase
Pattern
factor
Force
Silicone
increase
Pattern
1.98
79.2
Square
direction
bending
moment
factor
Epoxy
Pattern
Square
Square
Square
Square
increase
factor
Force
increase
8.3
monolithic
factor
Pattern
forcedirection
(N/mm)the Monolithic
Epoxy of
Pattern
Longitudinal
-0.025
-0.207
Square
It isIn-plane
evident
from
transverse
values
glass
barrel
Transverse
direction
-3.85
-3.68
1.0
Square
-3.14
0.8
Square
thatLongitudinal
the barrel
is not in -0.56
a pure membrane
state.
TheSquare
bending moments
in1.0
transverseSquare
direction
direction
-1.18
2.1
-0.57
are not
affected
by joint-1.76
stiffness when
pattern
Transverse
direction
-1.82 the square
1.0
Square is used,
-1.75see Table
1.0 9.6. Square
`
Bending moment
(Nmm/mm)
Longitudinal direction
Transverse direction
Monolithic
Silicone
Moment
increase
factor
Pattern
Epoxy
Moment
increase
factor
Pattern
-0.025
-3.85
1.98
-3.68
79.2
1.0
Square
Square
-0.207
-3.14
8.3
0.8
Square
Square
The square pattern with epoxy joints shows the least transverse shear force value.
The BLF of monolithic glass barrel is very low as compared to the glass dome. The hexalock
pattern shows maximum buckling stability for both types of joints. In general, the hexalock
pattern is not suitable for segmented shells which show bending. The glass portion projecting
out from the pane sides attracts high bending moments.
9.3. Hypar
The diamond pattern with epoxy joints shows similar deflection values as the monolithic glass
hypar, see Table 9.7.
Deflection (mm)
Monolithic
Silicone
Deflection
increase
factor
Pattern
Epoxy
Deflection
increase
factor
Pattern
-0.725
-9.83
13.6
Hexalock
-0.97
1.3
Diamond
Monolithic
increase
factor
Pattern
Epoxy
increase
factor
Pattern
ThePath
in-plane
of the joint
when square
see Table
b/w upperforces
corners are independent
3.55
4.04
1.1 stiffness
Square
3.63 pattern
1.0 is used,Square
9.8.
Moment
Moment
Bending is
moment
The hypar
not in a membrane
as it develops
moments.
The bendingPattern
moments
Monolithic state
Silicone
increase significant
Pattern
Epoxy
increase
(Nmm/mm)
factor is used with silicone joints
factor and the hexagon
remain unaffected when the diamond pattern
Path b/w upper corners
5.94joints, see
6.58Table 1.1
5.17
0.9
Hexagon
pattern
is used with epoxy
9.9. Diamond
The square pattern with epoxy joints shows the least transverse shear force value.
79
Deflection (mm)
Monolithic
Monolithic
Silicone
Path
b/w force
upper(N/mm)
corners
In-plane
-0.725
Monolithic
-9.83
Silicone
Monolithic
Bending
moment
Path
b/w upper
corners
(Nmm/mm)
-0.725
3.55
Silicone
9.4.
-9.83
4.04
Silicone
6.58
Path
b/w upper
corners
Bending
moment
(Nmm/mm)
Monolithic
Silicone
5.94
6.58
Deflection
increase
Pattern
Recommendations
factor
Deflection
13.6
Hexalock
increase
Pattern
factor
Force
13.6
Hexalock
increase
Pattern
factor
Force
1.1
Square
increase
Pattern
factor
Moment
hypar
segmentation:
1.1
Square
increase
Pattern
factor
Moment
1.1
Diamond
increase
Pattern
factor
1.1
Diamond
Epoxy
-0.97
Epoxy
-0.97
Epoxy
3.63
Epoxy
In-plane
3.63
Epoxy
5.17
Epoxy
5.17
Deflection
increase
factor
Deflection
1.3
increase
factor
Force
1.3
increase
factor
Force
1.0
increase
factor
Moment
forces
1.0
increase
factor
Moment
0.9
increase
factor
0.9
Pattern
Diamond
Pattern
Diamond
Pattern
Square
Pattern
Square
Pattern
Hexagon
Pattern
Hexagon
The hexagon pattern with epoxy joints shows maximum buckling stability.
9.4. Recommendations
The following recommendations for further analysis are suggested.
The glass pane and adhesive dimensions are not exactly same throughout the model
because the geometry has been constructed using vertical projection. Other construction
methods were tried but didnt result in same size panes and adhesives. This may have
an influence on the segmented shell results and should be checked.
Imperfections can have a significant effect on structural stability of shells. It is important
to identify the correct imperfection shape and do a geometrically non-linear (buckling)
analysis for segmented shells.
In this study, the only applied load is shell self-weight without any safety factor. Appropriate safety factors should be used and other loads like snow, wind suction and pressure,
maintenance need to be applied.
The effect of temperature on adhesives should be analysed as it could be a critical factor.
Ideal support conditions have been implemented in the shell FE models. In practice, such
conditions would not be achievable in built shells and additional stresses will develop near
the shell edges. Also, support settlement is another issue which can cause shell instability.
All materials are assumed to be linearly elastic and isotropic. Adhesives with non-linear
elasticity and anisotropic properties should be implemented in the FE model.
Stress resultants and deflections have been compared only for certain paths. More path
and global shell data should be analysed to make general conclusions about segmented
shells.
80
A
MAPDL script for method 1
!define variables
MPTEMP,1,0
z1=0.5*k11*x*x
/PREP7
z2=k12*x*y
z3=0.5*k22*y*y
81
z=z1+z2+z3
z=z1+z2+z3
Nx,y,z,
Nx,y,z,
*ENDDO
*DO,i,0,nx
k3=1+(i-1)*2+j*(3*nx+2)
z2=k12*x*y
z3=0.5*k22*y*y
*ENDDO
z=z1+z2+z3
*ENDDO
Nx,y,z,
*ENDDO
D,(2*nx+1)*ny+(nx+1)*ny+i 0UX,UY,UZ
*ENDDO
D,i 0UX,UY,UZ,ROTX,ROTY,ROTZ
*DO,i,0,2*nx
*ENDDO
x=(i-nx)*((nx*h-lx)*i*i+qx*i+mx)/gx
*DO,j,2,ny
y=ly/2
D,1+(j-1)*(3*nx+2) 0UX,UY,UZ
z1=0.5*k11*x*x
D,1+(j-1)*(3*nx+2)+2*nx 0UX,UY,UZ
z2=k12*x*y
*ENDDO
z3=0.5*k22*y*y
FINISH
82
B
Grasshopper script for hexalock
pattern
83
C
MAPDL script used for FE analyses
!NOTE:ANF INPUT REQUIRED BE- MPDATA,DENS,22.5e-9
FORE RUNNING SCRIPT
SECTYPE,1,SHELLHardAdh
!NOTE:GRAPHICAL PICKING OF ARSECDATA,10,1,0,3
EAS AND NODES ACTIVE
SECOFFSET,MID
/PREP7
SECTYPE,2,SHELLGlass
SHPP,ON
SECDATA,10,2,0,3
LOCAL,12,2
SECOFFSET,MID
ET,1,SHELL181
MSHKEY,0
KEYOPT,1,1,0
AESIZE,ALL,50
KEYOPT,1,3,0
ASEL,S,P !select areas
KEYOPT,1,8,2
AATT,1, ,1,12,1
KEYOPT,1,9,0
ASEL,S,MAT1
MPTEMP,1,0
AMESH,ALL
MPDATA,EX,11500!HARD Adhesive
ALLSEL
MPDATA,PRXY,10.35
MPDATA,DENS,11.1e-9
AATT,2, ,1,12,2
MPTEMP,1,0
ASEL,S,MAT2
MPDATA,EX,270000!Glass
AMESH,ALL
MPDATA,PRXY,20.22
ALLSEL
84
NROTAT,ALL
FINISH
FINISH
/POST1
/SOLU
RSYS,12
ALLSEL
ANTYPE,STATIC
ETABLE,N11,SMISC,1
(per unit length)
PSTRES,ON
ETABLE,N22,SMISC,2
ETABLE,N12,SMISC,3
!In-plane
forces
ETABLE,M22,SMISC,5
ETABLE,M12,SMISC,6
D,ALL,UY,0
ALLSEL
/PSYMB,CS,0
ETABLE,Q23,SMISC,8
/PLOPTS,INFO,3
ETABLE,Sm11,SMISC,34
stresses
!Membrane
/PLOPTS,LEG1,1
ETABLE,Sm22,SMISC,35
/PLOPTS,LEG2,1
ETABLE,Sm12,SMISC,36
/PLOPTS,LEG3,1
/PLOPTS,FRAME,1
ETABLE,Sb22,SMISC,38
/PLOPTS,TITLE,0
ETABLE,Sb12,SMISC,39
/PLOPTS,MINM,1
/PLOPTS,FILE,1
/PLOPTS,LOGO,1
!ETABLE,Sp22b,SMISC,41
/PLOPTS,WINS,1
!ETABLE,Sp12b,SMISC,42
/PLOPTS,WP,0
/PLOPTS,DATE,0
/TRIAD,OFF
!ETABLE,Sp22t,SMISC,44
ALLSEL
!ETABLE,Sp12t,SMISC,45
ACEL,9810
SOLVE
85
ETABLE,St23,SMISC,47
CSWPLA,17
/UIS, MSGPOP,3
WPOFF,0,0
/REPLOT,RESIZE
CSWPLA,18,2
WPSTYLE1
CSCIR,18,2
wprot,0,0,-90
PPATH,1,0,3535.534,3535.534,5250
/POST1
PPATH,2,0,-3535.534,-3535.534,5250
PMAP,UNI
CSWPLA,13
WPCSYS,1,17
WPOFF,0,0
CSWPLA,17
CSWPLA,14,2
WPSTYL,DEFA
CSCIR,14,2
WPOFF,0,0,6250
PPATH,1,0,-5000,0,5250
PATH,Syn4,2,30,4000,!hoop path
PPATH,2,0,5000,0,5250
CSWPLA,19
PMAP,UNI
WPOFF,0,0
WPCSYS,1,13
CSWPLA,20,1
CSWPLA,13
CSCIR,20,1
wprot,0,90,0
PPATH,1,0,3211.000001,0.1E-05,6500
PPATH,2,0,3210.999999,-0.1E-05,6500
CSWPLA,15
PMAP,UNI
WPOFF,0,0
WPCSYS,1,19
CSWPLA,16,2
CSWPLA,19
CSCIR,16,2
PPATH,1,0,0,5000,5250
PPATH,2,0,0,-5000,5250
PMAP,UNI
PATH,Syn1
WPCSYS,1,15
AVPRIN,0, ,
CSWPLA,15
RSYS,12
wprot,0,-45,0
PDEF, ,ETAB,N11,NOAVG
PATH,Syn3,2,30,2000,!inclined path
PDEF, ,ETAB,Sm11,NOAVG
86
PDEF, ,ETAB,N22,NOAVG
PDEF, ,ETAB,Sm11,NOAVG
PDEF, ,ETAB,Sm22,NOAVG
PDEF, ,ETAB,N22,NOAVG
PDEF, ,ETAB,N12,NOAVG
PDEF, ,ETAB,Sm22,NOAVG
PDEF, ,ETAB,Sm12,NOAVG
PDEF, ,ETAB,N12,NOAVG
PDEF, ,ETAB,M11,NOAVG
PDEF, ,ETAB,Sm12,NOAVG
PDEF, ,ETAB,Sb11,NOAVG
PDEF, ,ETAB,M11,NOAVG
PDEF, ,ETAB,M22,NOAVG
PDEF, ,ETAB,Sb11,NOAVG
PDEF, ,ETAB,Sb22,NOAVG
PDEF, ,ETAB,M22,NOAVG
PDEF, ,ETAB,M12,NOAVG
PDEF, ,ETAB,Sb22,NOAVG
PDEF, ,ETAB,Sb12,NOAVG
PDEF, ,ETAB,M12,NOAVG
PDEF, ,ETAB,Q13,NOAVG
PDEF, ,ETAB,Sb12,NOAVG
PDEF, ,ETAB,St13,NOAVG
PDEF, ,ETAB,Q13,NOAVG
PDEF, ,ETAB,Q23,NOAVG
PDEF, ,ETAB,St13,NOAVG
PDEF, ,ETAB,St23,NOAVG
PDEF, ,ETAB,Q23,NOAVG
PDEF, ,U,X,NOAVG
PDEF, ,ETAB,St23,NOAVG
PDEF, ,U,Y,NOAVG
PDEF, ,U,X,NOAVG
PDEF, ,U,Z,NOAVG
PDEF, ,U,Y,NOAVG
RSYS,0
PDEF, ,U,Z,NOAVG
PDEF,UZCartesian,U,Z,NOAVG
RSYS,0
RSYS,12
PDEF,UZCartesian,U,Z,NOAVG
/STITLESyn1
RSYS,12
/HEADER,OFF,ON,OFF,OFFOFF
/STITLESyn2
/PAGE,200000200000
/HEADER,OFF,ON,OFF,OFFOFF
/FORMAT,17,9
/PAGE,200000200000
/FORMAT,17,9
PATH,Syn2
AVPRIN,0, ,
PATH,Syn3
RSYS,12
AVPRIN,0, ,
PDEF, ,ETAB,N11,NOAVG
RSYS,12
87
PDEF, ,ETAB,N11,NOAVG
RSYS,12
PDEF, ,ETAB,Sm11,NOAVG
PDEF, ,ETAB,N11,NOAVG
PDEF, ,ETAB,N22,NOAVG
PDEF, ,ETAB,Sm11,NOAVG
PDEF, ,ETAB,Sm22,NOAVG
PDEF, ,ETAB,N22,NOAVG
PDEF, ,ETAB,N12,NOAVG
PDEF, ,ETAB,Sm22,NOAVG
PDEF, ,ETAB,Sm12,NOAVG
PDEF, ,ETAB,N12,NOAVG
PDEF, ,ETAB,M11,NOAVG
PDEF, ,ETAB,Sm12,NOAVG
PDEF, ,ETAB,Sb11,NOAVG
PDEF, ,ETAB,M11,NOAVG
PDEF, ,ETAB,M22,NOAVG
PDEF, ,ETAB,Sb11,NOAVG
PDEF, ,ETAB,Sb22,NOAVG
PDEF, ,ETAB,M22,NOAVG
PDEF, ,ETAB,M12,NOAVG
PDEF, ,ETAB,Sb22,NOAVG
PDEF, ,ETAB,Sb12,NOAVG
PDEF, ,ETAB,M12,NOAVG
PDEF, ,ETAB,Q13,NOAVG
PDEF, ,ETAB,Sb12,NOAVG
PDEF, ,ETAB,St13,NOAVG
PDEF, ,ETAB,Q13,NOAVG
PDEF, ,ETAB,Q23,NOAVG
PDEF, ,ETAB,St13,NOAVG
PDEF, ,ETAB,St23,NOAVG
PDEF, ,ETAB,Q23,NOAVG
PDEF, ,U,X,NOAVG
PDEF, ,ETAB,St23,NOAVG
PDEF, ,U,Y,NOAVG
PDEF, ,U,X,NOAVG
PDEF, ,U,Z,NOAVG
PDEF, ,U,Y,NOAVG
RSYS,0
PDEF, ,U,Z,NOAVG
PDEF,UZCartesian,U,Z,NOAVG
RSYS,0
RSYS,12
PDEF,UZCartesian,U,Z,NOAVG
/STITLESyn3
RSYS,12
/HEADER,OFF,ON,OFF,OFFOFF
/STITLESyn4
/PAGE,200000200000
/HEADER,OFF,ON,OFF,OFFOFF
/FORMAT,17,9
/PAGE,200000200000
/FORMAT,17,9
PATH,Syn4
!PRANGE0,4000
AVPRIN,0, ,
88
PASAVE,ALL,SynpathsPASAVE
/POST1
/UIS, MSGPOP,3
/VIEW,1,1
/AUTO,1
/RGB,INDEX,100,100,100, 0
RSYS,12
PLETAB,N11,NOAV
PLETAB,Sm11,NOAV
/RGB,INDEX, 0, 0, 0,15
PLETAB,N22,NOAV
/GRAPHICS,POWER
PLETAB,Sm22,NOAV
/SHADE,ALL,2
PLETAB,N12,NOAV
PLETAB,Sm12,NOAV
/REPLO
/SHOW,PNG0
PNGR,COMP,1,-1
PNGR,ORIENT,HORIZ
PNGR,COLOR,2
PLETAB,M11,NOAV
PNGR,TMOD,1
PLETAB,Sb11,NOAV
/GFILE,1200,
PLETAB,M22,NOAV
/EDGE,1,0,45
PLETAB,Sb22,NOAV
/GLINE,1,-1
PLETAB,M12,NOAV
SHELL,TOP
PLETAB,Sb12,NOAV
AVRES,4
PLETAB,Q13,NOAV
RSYS,0
PLETAB,St13,NOAV
PLNSOL,U,X
PLETAB,Q23,NOAV
PLNSOL,U,Y
PLETAB,St23,NOAV
PLNSOL,U,Z
ESEL,S,MAT2
PLNSOL,U,SUM
RSYS,0
PLNSOL,U,X
PLNSOL,U,Y
89
PLNSOL,U,Z
/PLOPTS,MINM,0
/GFORMAT,F,12,4,
EPLOT
PLNSOL,U,Z
ESEL,S,MAT2
/GFORMAT,F,12,3,
/PBC,PATH,1
PLNSOL,U,Z
/REPLOT
/GFORMAT,DEFA,12,4,
/PLOPTS,MINM,1
PLNSOL,U,SUM
/PBC,PATH,0
ALLSEL
/TRIAD,ORIG
/REPLOT
/VIEW,1-1
RSYS,12
/ANG,1
PLETAB,N11,NOAV
/USER, 1
PLETAB,Sm11,NOAV
PLETAB,N22,NOAV
PLETAB,Sm22,NOAV
PLETAB,N12,NOAV
PLETAB,Sm12,NOAV
PLETAB,M11,NOAV
PLETAB,Sb11,NOAV
PLETAB,M22,NOAV
/REPLOT
PLETAB,Sb22,NOAV
/TRIAD,OFF
PLETAB,M12,NOAV
/VIEW,1,1
PLETAB,Sb12,NOAV
/AUTO,1
PLETAB,Q13,NOAV
PATH,Syn1
PLETAB,St13,NOAV
PLPAGM,UX,20000,NODE
PLETAB,Q23,NOAV
PLPAGM,UY,20000,NODE
PLETAB,St23,NOAV
PLPAGM,UZ,20000,NODE
ALLSEL
PLPAGM,UZCartesian,20000,NODE
90
PLPAGM,ETABN11,20000,NODE
PLPAGM,ETABM22,30000,NODE
PLPAGM,ETABSm11,20000,NODE
PLPAGM,ETABSb22,30000,NODE
PLPAGM,ETABN22,20000,NODE
PLPAGM,ETABM12,30000,NODE
PLPAGM,ETABSm22,20000,NODE
PLPAGM,ETABSb12,30000,NODE
PLPAGM,ETABN12,20000,NODE
PLPAGM,ETABQ13,30000,NODE
PLPAGM,ETABSm12,20000,NODE
PLPAGM,ETABSt13,30000,NODE
PLPAGM,ETABM11,30000,NODE
PLPAGM,ETABQ23,30000,NODE
PLPAGM,ETABSb11,30000,NODE
PLPAGM,ETABSt23,30000,NODE
PLPAGM,ETABM22,30000,NODE
PATH,Syn3
PLPAGM,ETABSb22,30000,NODE
PLPAGM,UX,20000,NODE
PLPAGM,ETABM12,30000,NODE
PLPAGM,UY,20000,NODE
PLPAGM,ETABSb12,30000,NODE
PLPAGM,UZ,20000,NODE
PLPAGM,ETABQ13,30000,NODE
PLPAGM,UZCartesian,20000,NODE
PLPAGM,ETABSt13,30000,NODE
PLPAGM,ETABN11,20000,NODE
PLPAGM,ETABQ23,30000,NODE
PLPAGM,ETABSm11,20000,NODE
PLPAGM,ETABSt23,30000,NODE
PLPAGM,ETABN22,20000,NODE
PATH,Syn2
PLPAGM,ETABSm22,20000,NODE
PLPAGM,UX,20000,NODE
PLPAGM,ETABN12,20000,NODE
PLPAGM,UY,20000,NODE
PLPAGM,ETABSm12,20000,NODE
PLPAGM,UZ,20000,NODE
PLPAGM,ETABM11,30000,NODE
PLPAGM,UZCartesian,20000,NODE
PLPAGM,ETABSb11,30000,NODE
PLPAGM,ETABN11,20000,NODE
PLPAGM,ETABM22,30000,NODE
PLPAGM,ETABSm11,20000,NODE
PLPAGM,ETABSb22,30000,NODE
PLPAGM,ETABN22,20000,NODE
PLPAGM,ETABM12,30000,NODE
PLPAGM,ETABSm22,20000,NODE
PLPAGM,ETABSb12,30000,NODE
PLPAGM,ETABN12,20000,NODE
PLPAGM,ETABQ13,30000,NODE
PLPAGM,ETABSm12,20000,NODE
PLPAGM,ETABSt13,30000,NODE
PLPAGM,ETABM11,30000,NODE
PLPAGM,ETABQ23,30000,NODE
PLPAGM,ETABSb11,30000,NODE
PLPAGM,ETABSt23,30000,NODE
91
PATH,Syn4
/ANG,1
PLPAGM,UX,15000,NODE
/ZOOM,1,SCRN,0.304745,-0.320498,0.353033,0.352682
PLPAGM,UY,15000,NODE
/PBC,U1
PLPAGM,UZ,15000,NODE
/AUTO,1
PLPAGM,UZCartesian,15000,NODE
/REP,FAST
PLPAGM,ETABN11,15000,NODE
/ZOOM,1,SCRN,0.367126,-0.732219,0.393614,0.76664
PLPAGM,ETABSm11,15000,NODE
PLPAGM,ETABN22,15000,NODE
/PBC,U0
PLPAGM,ETABSm22,15000,NODE
/SHOW,CLOSE
PLPAGM,ETABN12,15000,NODE
/DEVICE,VECTOR,0
PLPAGM,ETABSm12,15000,NODE
/SOLU
PLPAGM,ETABM11,20000,NODE
ANTYPE,1
PLPAGM,ETABSb11,20000,NODE
PLPAGM,ETABM22,20000,NODE
BUCOPT,LANB,50,0,0,CENTER
modes will be generated
PLPAGM,ETABSb22,20000,NODE
SOLVE
PLPAGM,ETABM12,20000,NODE
FINISH
PLPAGM,ETABSb12,20000,NODE
/SOLU
PLPAGM,ETABQ13,20000,NODE
EXPASS,1
PLPAGM,ETABSt13,20000,NODE
PLPAGM,ETABQ23,20000,NODE
PLPAGM,ETABSt23,20000,NODE
SOLVE
/VSCALE,1,1,0
FINISH
/VIEW,1,1
/POST1
/AUTO,1
/UIS, MSGPOP,3
PLVECT,S, , , ,VECT,ELEM,ON,0
/VIEW,1,1
/ANG,1,-90,XS,1
/AUTO,1
/AUTO,1
/RGB,INDEX,100,100,100, 0
PLVECT,S, , , ,VECT,ELEM,ON,0
/VIEW,1,1
92
!50
/RGB,INDEX, 0, 0, 0,15
*DO,I,1,50 ! For I = 1 to 50
/GRAPHICS,POWER
SET, , ,I
/SHADE,ALL,2
/REPLO
*ENDDO
/SHOW,PNG0
/VIEW,1,1
PNGR,COMP,1,-1
/ANG,1
PNGR,ORIENT,HORIZ
/AUTO,1
PNGR,COLOR,2
/VIEW,1-1
PNGR,TMOD,1
/ANG,1
/GFILE,1200,
/ANG, 1 ,3.000000,XS,1
/EDGE,1,0,45
/ANG, 1 ,3.000000,XS,1
/GLINE,1,-1
/ANG, 1 ,3.000000,XS,1
SHELL,TOP
RSYS,0
AVRES,4
*DO,J,1,50 ! For J = 1 to 50
/VIEW,1,1
SET, , ,J
/AUTO,1
/DIST,1,1.08222638492,1
*ENDDO
/DIST,1,1.08222638492,1
SET,LIST
/DIST,1,1.08222638492,1
FINISH
/DIST,1,1.08222638492,1
/SHOW,CLOSE
RSYS,0
/DEVICE,VECTOR,0
93
Bibliography
[SBP ]
URL http://www.sbp.de/en#build/category/83-Roofs_and_Fa%C3%A7ades
[ArchHello ]
URL http://www.archello.com/sites/default/files/imagecache/
header_detail_large/96008N4.jpg
[SolidWorksHelp ]
URL http://help.solidworks.com/2012/English/SolidWorks/
cworks/Buckling_Load_Factor.htm
[DIN-18800 ] 7.2 Ideal buckling stresses. In: DIN 18800-4:2008-11 (E) Steel structures Part 4: Stability - Safety against buckling of shells
[Structurae 1935] Orvieto Hangars, Terni, Umbria, Italy. URL http://structurae.net/
structures/orvieto-hangars, 1935
[Paaskerk 1963]
Augustinuspark, Amstelveen, Netherlands.
URL http://www.
herein-system.eu/sites/dev-hkp/files/Paaskerk%20Amstelveen.jpg, 1963
[Florence 2008] Santa Maria del Fiore, Florence, Italy. URL http://photoartkalmar.
com/Photoart%20Kalmar%20high%20res/Florence_Pisa/slides/IMG_3844%20Santa%
20Maria%20del%20Fiore.html, 2008
[ANSYS 2009]
[Blogspot
2011]
URL
http://4.bp.blogspot.com/-EYkJ011EUkQ/TdV0_Tz9-CI/
AAAAAAAAsA0/ATNEBQQEp0Q/s1600/26%2BSkin%2Bof%2Bgiraffe%2B2.jpg, 2011
[Apple 2012] Apples iconic fifteen glass panel cube store in Manhattan, USA. URL http:
//jameshowephotography.com/wp-content/uploads/2012/06/DSC0538-Edit-full.jpg,
2012
94
[Dsktpbckgrnd 2013]
URL http://www.desktopbackgrounds.me/wp-content/uploads/
2013/11/Leaf-Pattern.jpg, 2013
[Pantheon 2013] The Pantheon, Rome, Italy. URL http://www.travelplusstyle.com/
wp-content/gallery/cache/10674__870x_pantheon.jpg, 2013
[Nytimes 2014]
The Reichtag building, Berlin, Germany. URL http://www.nytimes.
com/times-journeys/travel/two-spy-capitals-of-europe/?action=click&module=
history-and-context-viewer&version=Two%20Spy%20Capitals%20of%20Europe&
region=body-main, 2014
[Aanhaanen 2008] Aanhaanen, Jaap: The stability of a glass facetted shell structure, Masters
thesis, The Netherlands: Delft University of Technology, Dissertation, 2008. URL http:
//repository.tudelft.nl/assets/uuid:b3d35f24-c072-4bef-8b11-5872da7a72af/
The_stability_of_a_facetted_glass_shell_structure.pdf. Zugriffsdatum: 2014-1004
[Adriaenssens u. a. 2014]
Adriaenssens, Sigrid; Block, Philippe; Veenendaal, Diederik;
Williams, Chris (Hrsg.): Shell Structures for Architecture: Form Finding and Optimization. London ; New York : Routledge, Juni 2014. ISBN 9780415840606
[Bagger und DTU Byg 2010] Bagger, Anne; DTU Byg: Plate shell structures of glass:
studies leading to guidelines for structural design : Ph.D. Thesis. [Lyngby] : DTU Civil
Engineering, Technical University of Denmark, 2010. ISBN 9788778773005 8778773008
[Belis u. a. 2007] Belis, Jan; Verhegghe, Benedict; De Beule, Matthieu; Van Impe, Rudy:
Evaluation of Glass Domes using pyFormex. In: Shell and Spatial Structures: Structural
Architecture-Towards the future looking to the past (2007). URL https://biblio.ugent.
be/publication/383469/file/460813.pdf. Zugriffsdatum: 2014-10-20
[Billington 1965] Billington, D. P.: Thin Shell Concrete Structures. McGraw-Hill Inc.,US,
September 1965. ISBN 9780070052710
[Bischoff u. a. ] Bischoff, M.; Ramm, E.; von Scheven, M.: Modelling of shells with finite
elements. In: Computational Methods for Shell Analysis - lecture notes, COMMAS course
E3-03, Winter term 2013/2014. Institut fr Baustatik und Baudynamik, Uni Stuttgart
[Blandini 2005] Blandini, Lucio: Structural use of adhesives in glass shells. Grauer, 2005.
ISBN 9783861864868
[Calladine 1989] Calladine, C. R.: Theory of Shell Structures. Cambridge University Press,
Februar 1989. ISBN 9780521369459
[Dallard u. a. 2001] Dallard, Pat; Facer, Mark; Hikone, Shigeru; Hirose, Ryoichi; Oguri,
Arata; Sasaki, Jin: Osaka Maritime Museum. In: ARUP JOURNAL 36 (2001), Nr. 1, S. 21
27. URL http://www.arup.com/_assets/_download/download39.pdf. Zugriffsdatum:
2014-10-18
[Dillard 2010] Dillard, D.: Advances in Structural Adhesive Bonding. Elsevier, Mrz 2010.
ISBN 9781845698058
[Dimcic 2011] Dimcic, Milos: Structural optimization of grid shells based on genetic algorithms. Stuttgart : Inst. for Tragkonstruktionen u. Konstruktives Entwerfen, Univ., 2011.
ISBN 9783922302322 3922302327
[Flgge 1962]
[Orton 2013] Orton, Andrew: The Way We Build Now: Form, Scale and Technique. Taylor
& Francis, Juni 2013. ISBN 9781136737091
[Ramaswamy 1984] Ramaswamy, G. S.: Design and construction of concrete shell roofs.
R.E. Krieger, Oktober 1984. ISBN 9780898740011
[Ramm und Wall 2004] Ramm, Ekkehard; Wall, Wolfgang A.: Shell structuresa sensitive
interrelation between physics and numerics. In: International journal for numerical methods
in engineering 60 (2004), Nr. 1, S. 381427. URL http://onlinelibrary.wiley.com/
doi/10.1002/nme.967/abstract. Zugriffsdatum: 2014-10-13
[Ramos und Hoogenboom 2013] Ramos, Nathalie; Hoogenboom, P. C. J.: Concentrated
loads on anticlastic shells. In: Delft university of technology (2013). URL http://
homepage.tudelft.nl/p3r3s/BSc_projects/eindrapport_ramos.pdf. Zugriffsdatum:
2014-10-02
[Rice 1998] Rice, Peter: An Engineer Imagines. New edition edition. London : Ellipsis
London Pr Ltd, September 1998. ISBN 9781899858118
[Robbin 1996] Robbin, Mr T.: Engineering a New Architecture. 1ST edition. New Haven :
Yale University Press, April 1996. ISBN 9780300061161
[Schittich 2007] Schittich, Christian: Glass Construction Manual. 2nd, revised and expanded
ed. edition. Basel; London : Birkhuser Architecture, Juni 2007. ISBN 9783764382902
[Veer u. a. 2003]
Veer, F. A.; Wurm, J.; Hobbelman, G. J.: The Design, Construction and Validation of a Structural Glass Dome. In: Proceedings of Glass Processing