Finot Mezgebu

ADDIS ABABA INSTITUTE OF TECHNOLOGY
SCHOOL OF GRADUATE STUDIES

INSTITUTE OF THECNOLOGY
DEPARTMENT OF CIVIL ENGINEERING
COMPUTERIZED ANALYSIS AND DESIGN OF UNDERWATER

CYLINDRICAL SHELL STRUCTURES USING FIBER GLASS
By
Finot Mezgebu Wubete
April 2016
A THESIS SUBMITTED TO THE GRADUATE SCHOOL OF THE ADDIS ABABA

UNIVERSITY IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE
OF MASTER OF SCIENCE IN CIVIL ENGINEERING
2
ADDIS ABABA UNIVERSITY
SCHOOL OF GRADUATE STUDIES
INSTITUTE OF TECHNOLOGY
DEPARTMENT OF CIVIL ENGINEERING
COMPUTERIED ANALYSIS AND DESIGN OF UNDERWATER CYLINDRICAL SHELL

STRUCTURES USING FIBER GLASS
By
Finot Mezgebu Wubete
Approved by Board of Examiners
Dr. -Ing Bedilu Habte _____________ ____________
Advisor Signature Date
Dr. Asnake Adamu _____________ ____________
External Examiner Signature Date
Dr. Shifferaw Taye _____________ ____________
Internal Examiner Signature Date
_____________ ____________
Chair man Signature Date

Declaration
I declare that this thesis is my original work. This thesis has not been presented for any
other university and is not concurrently submitted in candidature of any other degree, and
that all sources of material used for the thesis have been duly acknowledged.
Name: Finot Mezgebu
Signature: __________
Place: Addis Ababa University
Institute of Technology
I
Acknowledgements
Many contributors both direct and indirect to this research I will acknowledge all for the supply of
this knowledge and this achievement. Firstly, I would like to thank God in giving me health and
everything that I need for this research. Secondly, without hesitation thanks for all the reference and
research paper writers which helped in showing new and applicable way for this research. Thirdly,
thanks for my advisor in helping this tedious and wide idea to come to a specific and productive
way. Thanks for all the knowledge and time you poured in this research. Fourthly, I would like to
thank my class mates and friends in helping this research by giving time for discussion and giving
ideas on it. Fifthly, I would like to thank the school of civil and environmental engineering for
allowing me to do this research and providing help. Lastly, I would like to thank my family in
supporting me in every way.
II
Abstract
The computerized design of underwater cylindrical shell will be addressed in this research by using
different mechanisms. Since it should be done more accurately and approximately the combination
of different programs will be needed. The design will include both analytical and finite element
design method. The design will be done for different dimension of the cylindrical shell which can
support the load applied. The most applicable and time saving way will be chosen for the design.
There will be some exposure to see the behavior of the cylindrical shell with respect to its structural
behavior.
III
TABLE OF CONTENET
Declaration……………………………………………………………………………………I
Acknowledgement…………………………………………………………………………....II
Abstract………………………………………………………………………………………III
List of figures……………………………………………………………………………….VIII
List of tables…………………………………………………………………………………IX
Nomenclature………………………………………………………………………………..X
Chapter 1Introduction
1.1 General background of the problem……………………………………………………….1

1.2 Objectives……………………………………………………………………………...…..2
1.3 Methodology……………………………………………………………………………….2
1.3.1 Procedure…………………………………………………………………………….2
1.4 Limitation…………………………………………………………………………………..2
1.5 Scope……………………………………………………………………………………….3
Chapter 2 Litrature review
2.1 Shell Structures Fundamentals……………………………………………………………..4
2.1.1 Introduction……………………………………………………………………………4
2.2 History of design and construction of shells…………………………………………….....4
2.3 History of analysis methods………………………………………………………………..5
2.4 Classification of shells……………………………………………………………………..5
2.4.1 Classification based on thickness…………………………………………………..5

2.4.2 Classification according to Gaussian curvature…………………………………....6
2.4.3 Classification based on surface of geometric behavior…………………………….7
IV
2.4.4 Classification according to manner of generation………………………………….7
2.5 Advantage and disadvantage of shell structures……….………………………………..…7
2.5.1 Advantages……….……………………………………………………………....…7
2.5.2 Disadvantage……….……………………………………….……………………….8
2.6 Cylindrical shell…………………………………………………………………………….8
2.6.1 Classification of cylindrical shell……………………………………………………..8
2.6.1.1 Classification based on radius and longitudinal length…………………………9
2.6.1.2 Classification based on the magnitude of edge disturbance…………………….9
2.6.1.3 Classification based on geometric dimension and thickness……………………9
2.7 Fiber glass……………………………...…………………………………………………....10
2.7.1 Properties…………………………...………………………………………………....10
2.7.2 Types of fiber glass…………………...…………………………………………….....11
2.8 Design of shell structure ………………………………………………………………….....13
2.8.1 Loading………………………………………………………………………………..14
2.8.2 Load combination……………………………………………………………………..15
2.8.3 Safety Factor…………………………………………………………………………..16
2.8.4 Design consideration…………………………………………………………………...16
2.8.5 Method of analysis ……………………………………………………………………18
2.8.5.1 Loading balancing method……………………………………..……………….18
2.8.5.2 Beam theory…………………………………………………………………….19
2.8.5.3 The Arch analysis……………………………………………………………….21
2.8.5.4 Membrane Theory………………………………………………………………21
V
2.8.5.5 Bending theory………………………………………………………………….27
2.8.6 Analysis procedure……………………………………………………………………...28
2.8.6.1 Thin Cylindrical Shell Design…………………………………………………..31
2.8.6.2 Thick Cylindrical Shell Design………………………………………………....34
Chapter 3 Design of the structure using analytical method
3.1
Introduction…………………………………………………………………................................38
3.2
Modeling……………………………………………………………………………..…………...38
3.3 Program development for analytical design of underwater cylindrical shell ……….……......44
3.3.1 Flow chart of the program………………………………………………………………44
3.3.2 Design using C++ program……………………..………………………………………46
3.3.2.1 Input of the program……..………………….……………………………………49
3.3.2.2 Process of the program..…………………….……………………………………49
3.3.2.3 Output of the program…………………….……………………………………...50
Chapter 4 Analysis of the structure using finite element analysis
4.1 Introduction…………………………………………………………………………………...53
4.2 Analysis procedure…………………………………………………………………………….53
4.3 Analysis result…………………………………………………………………………………54
Chapter 5 comparison of analytical design and finite element analysis
5.1 Introduction…………………………………………………………………………………….57
5.2 Comparison………………………………………………………………………………….....57
5.3 Result and discussion…………………………………………………………………………..57
Chapter 6 Conclusion and Recommendation
VI
6.1 Conclusion…………………………………..…………………………………………………59
6.2 Recommendation………………………….…………………………………………………..60
Reference…………………………………….…………………………………………………....61
Appendix I-Program code for thin shell…….…………………………………………………….63
Appendix II-Program code for thick shell….……………………………………………………..74
Appendix III-Loading for analysis of thin shell using structural analysis program………………86
VII
LIST OF FIGURES
Figure 2.1 Stress strain relation of fiber glass in a tensile test…………………………..………14
Figure 2.2 Shear, bending moment and twisting moment…………………………………..…...22
Figure 2.3 Compatibility of boundary conditions with the membrane theory……………..…….23
Figure 2.4 Cylindrical shell with edge beam………………………………………………..……24
Figure 2.5 Edge reaction and load applied…………………………………………………..…....28
Figure 2.6 Analysis procedure………………………………………………………………..…...29
Figure 2.7 Barrel cylindrical shell under uniformly distributedload …………………………..…30
Figure 2.8 Membrane Stresses, Moment and Shear Forces Acting on Infinitesimal Element of
Shell……………………………………………………………………………………...………...31
Figure 2.9 Shell under uniformly distributed load……………………………………...………….32
Figure 2.10 Thick cylindrical shell subjected to internal pressure……………………...………….35
Figure 2.11 Thick cylindrical shell closed at both ends and subjected to internal pressure………..35
Figure 3.1 Under-water cylindrical shell: (a) design model and (b) analysis model……………….39
Figure 3.2 Design variables………………………………………………………………................39
Figure 3.3 Dimension and usable area……………………...………………………………………43
Figure 3.4 Flow chart of the program………………………………………………………………44
Figure 4.1 Maximum stress…………………………………………………………………………54
Figure 4.2 Minimum stress…………………………………………………………………………55
Figure 4.3 Deflection in the Z-direction……………………………………………………………55
Figure 4.4 Deflection in the y-direction…………………………………………………………….56
VIII
LIST OF TABLE
Table 2.1 Common fiberglass types……………………………………………………………..13
Table 3.1 Result of analysis……………………………………………………………………..48
Table 3.2 Input values for the program………………………………………………………….49
Table 3.3 Input data for different types of fiber glass…………………………………………...50
Table 3.4 Dimension of output of the program ………..…….…………………………………..51
Table 3.5 Output result for different types of fiberglass…………………….…………………...51
Table 4.1 Maximum values of finite element analysis……………………..…………………....56
Table 5.1 Comparison of the Result of Finite Element Analysis and Analytical Design………..57
IX
NOMENCLATURE
R-Radius (m)
L-Length (m)
Ψk-Angle of the shell (radian)
t-Thickness of the shell (mm)
E-Young‟s modulus of concrete (kN/m2)
ν -Poisson‟s ratio of fiber glass
Ψ-Generic angle in shell (radian)
Υ-Specific weight (KN/m3)
Mx-Bending moment in plane x-z
MΨ-Bending moment in plane r-Ψ
MXψ-Torque moment in plane x-z
MΨx-Torque moment in plane r-Ψ
Nx -Axial force in x-direction (KN/m)
NΨ(Nh)-Axial force in Ψ-direction (KN/m)
NΨx-Shear force in x –direction(KN/m)
Qx -Shear force in radial direction along longitudinal edges
QΨ-Shear force in radial direction along transversal edges
εx-Normal strain in x-direction
εΨ-Normal strain in Ψ -direction
γXψ-Shear strain in plane x-Ψ
X
Ψx-Curvature in plane x-z
Ψφ-Curvature in plane r-Ψ
Ψφx-Curvature in plane x-z
Pu-Factored distributed surface load (KN/m2)
px-Surface load on the shell in x-direction (KN/m2)
pΨ-Surface load in Ψ-direction (KN/m2)
pr-Surface load on the shell in radial direction (KN/m2)
pz-Surface load on the shell in z-direction (KN/m2)
ux-Displacement component in x-direction (mm)
uΨ(uh)-Displacement component in Ψ-direction(mm)
ur(uv)-Displacement component in radial direction(mm)
uz-Displacement component in z- direction(mm)
Rck-Compressive cube strength of concrete (KN/m2)
Lr-Roof load
R-Rain load
F-Loads due to weight and pressures of fluids with well-defined densities and controllable
maximum heights, or related interval moments and forces
T-Cumulative effect of temperature, creep, shrinkage, differential settlement, shrinkage-

compensating concrete
XI
COMPUTERIZD ANALYSIS AND DESIGN OF UNDERWATER CYLINDERICAL SHELL STRUCTURE USING FIBER GLASS
CHAPTER1 INTRODUCTION
1.1 GENERAL BACKGROUND OF THE PROBLEM
In this world there are different structures, which are built to make the life of a human being more
easy and comfortable. Even though there are different structures all of them have different
purposes. Man is building a structure on the ground, in the ground, on the water and even in the
water. As the science of engineering grows the type and the location of the structure also changes.
Engineering is a big and wide field. As the name implies it is a work of creativity. At the end of any
engineering field there is always a need of creativity and the means to introduce new technological
development. Structural engineering is one of those fields which accommodates one of the basic
need of human being that is shelter and additional recreational structures, water structures, bridges,
etc.
In Ethiopia there are different kind of structures built. The thesis is an attribute to adding to the
structurescurrently existing. Every civilization has its own mark. For Ethiopia the old civilization in
century of the civilization of Axum left us a mark by building those three sculpture from one stone.
And in king Lalibela ruling time those eleven structurally amazing and detailed structures were
built. So from the century we are living now there are different new findings and new technology.
Based on that this will be a step forward.
One of the problems found in Ethiopia is introducing new materials. It is almost taken asunreliable
way to construct a structure. Most people think the stability and durability of the structure depends
on its being constructed with stone or not. But if this material is used practically since it would be a
large structure and carry the load of the water it will be a stepping stone for the society and helps us
to be open for changes that are scientifically proven.
1
1.2 OBJECTIVES
The main objective is to do computerized design of underwater cylindrical shell structure using
fiber glass.There are many options of dimensions and also fiber glass type for the design. But as the
structure is being designed and redesigned the final stable design can be mate. Stability is checked
based on ultimate limit state design for compressive and tensile stress of fiber glass. Serviceability
limit state design for the deflection limitation of cylindrical shell.
1.3 METHODOLOGY
1.3.1 Procedure
A three step methodology can be used for this study:
i) Study about under water cylindrical shell structure, the analysis, the load and the
material that is going to be used.
ii) Analytical Modeling:
a) Modeling the design problem by differentiating the constraint and taking in to
consideration the limitation of stress and deflection.
b) Design of the cylindrical shell structure using the above constraints by
developing appropriate software program.
c) Verification of the design using structural analysis program (SAP).
iii) Conclusion ,recommendation and limitation
1.4 Limitation
This thesis covers computerized design of underwater cylindrical shell structure. It is
specifically limited to:
o The function of the design is for qua zoo so it considers only minimum volume per
usable area.
o Basic design formula for thin and thick cylindrical shell.
o The place of construction is not taken in to consideration.
o Doesn‟t consider the arrangement of fiber glass layers and the diameter of fiber glass is
not also considered.
2
1.5 Scope
The scope of the thesis is deign of cylindrical shell structure using a C++ programing language.
Formulas for design of thin and thick cylindrical shell structure are used for the analytical design.
Structural analysis program is used for the finite element analysis. The design considers ultimate
limit state design for the type of fiber glass used, serviceability limit state design for the deflection
of the cylindrical shell structure and local stability of the structure is taken in to consideration.
3
CHAPTER 2 LITERATURE REVIEW

2.1SHELL STRUCTURES FUNDAMENTALS
2.1.1 INTRODUCTION
Structural shells are curved-surface systems that are usually capable of transmitting loads in
more than two directions to supporting elements. They exhibit structural efficiency when they
are so shaped, proportionedand supported that they provided resistance to and transmit the loads
without little bending or twisting. They constitute among the most common and most efficient
structural elements in nature and technology. They are used whenever high strength, large pans
and minimum material content are required. Shells of double curvature, for example, are among
the most efficient of known structural forms.[7]
2.2HISTORY OF DESIGN AND CONSTRUCTION OF SHELLS

The theory and practice of shell design goes back to the 1920‟s when Dyckerhoff and Widmann
contractors from Munich, Germany, together with water Bauersfeld and Franz Discharger,
developed the famous Cupoia for the Zeiss-Planetarium in Germany. [7]
Anton Tadesko‟ summarizes his experience with concrete shells as follows.There are three
important points which the success of shells depends on. [7]
1. Designers learn from having experience with full-scale structures.

2. Engineers should visualize the actual construction during the design stage ,the economy
of shells depends on the close collaboration of design and builder
3. Even in working with architects of prestige, the engineer must stand firm, making it clear
as to what can be done and what should not be done.
The thin reinforced-concrete shell, as we know it today, had its beginning in Germany in the
1920s. Most of the early shells built were cylindrical barrels. In 1924, the first concrete shell roof
was designed by Carl Zeiss and built in the Zeiss work in Jena, Germany. [1]
Shell structures have found wide applications in practice. These include, among others roof of
various types, water tanks and reservoirs as well as liquid contains and storage facilities,
4
foundations, egg-shaped sewage digesters cooling towers at thermal power stations, one of the
major structural elements of nuclear power plants,boilers, pressure vessels, bodies of
transportation structures including aircrafts, spacecraft, motor vehicles and ships.[7]
2.3HISTORY OF ANALYSIS METHODS

The first analytical approach to the design of shells was presented G. Lame‟ and E. Clapeyron,
who in 1826 produced the “membrane analogy” in which a shell was considered capable of
resisting external loads by direct stresses unaccompanied by any bending .The next important
contribution in this field was made in 1892 by A.E.H. Love, who developed mathematical
conceptions which made possible a more accurate analysis than could be achieved by membrane
analogy. Around the year 1923 u. Finsterwalder and F. Dischinger were the first to develop a
theoretical analysis applicable to reinforced concrete cylindrical shells. In the united State, H.
Schorer further simplified the derivation of Finsterwalder (1936). Until about 1940, the
cylindrical shell more or less dominated the scene. The beam theory for shell analysis was
developed by H. Lundgren of Denmark in 1949.This consists of separate analyses in which the
shell is considered first as a beam and secondly as an arch. The load balancing method, which
was developed by T. Y. Lin, for prestressed concrete member offers a new approach and greatly
simplifies the design of prestressed shells. [1]
2.4CLASSIFICATION OF SHELLS
Shell structures can be classified into number of categories based on several factors.
2.4.1 CLASSIFICATION BASED ON THICKNESS
The thickness of a shell is important parameter upon which the behavioral response of a shell
depends in its ability to resist and transmit the effects loads. The use of shells as structural
systems is motivated primarily for two reasons- economy and aesthetics. In either case, however,
the cost factors for shells may vary considerably with parameters such as shape and support
conditions. In concrete roof-shell structures for example, the high cost of formwork and
scaffolding can noticeably restrict the size of spans that are considered economical. On the other
hand, form reuse many significantly lower the cost for the shell. [7]
5
Shells are then classified as membrane, thin and thick shells depending on the ratio of their
thickness to respective radius of curvature. Criteria for thinness vary, depending on the
magnitude of errors that can be tolerated in regarding a particular shell as thin, and hence suited
to study by thin-shell theories rather than by the more general (but less convenient) thick-shell
theories which, by their nature are closer to general theory of elasticity. A precise classification
may not be possible as many shell structures could be classified either way; thenumerical values
presented earlier are only indicative values for the classification. A thin shell is a curved surface
structure whose thickness is generally much, much smaller than the other dimensions and in
which the geometry activates axial internal stress result to become significant load-resisting
stress systems. The effort in planning and design of thin shells to make the shell as thin as
practical requirements permit so that the own weight is reduced and the structure functions as a
member free from large bending stress. This implies that a minimum material volume is used to
maximum structural advantage.[7]
1. Membrane shells : thickness to radius of curvature is

t 1
 ………………………………………………………………………Eq.1.1
rmin 1000
2. Thin shell: thickness to radius of curvature is

1 t 1
  ………………………………………………………………….Eq.1.2
1000 rmin 50
3. Thick shell: thickness to radius of curvature is
t 1
 …………………………………………………………………………..Eq.1.3
rmin 50
2.4.2 CLASSIFICATION ACCORDING TO GAUSSIAN

CURVATURE
Gauss curvature, the product of two principal radii of curvature at any point on the surface of the
Shell. Based on this method of classification, thin shells fall into three distinct groups. [7]
i. Synclastic surface: A surface for which Gauss curvature is positive.
ii. Developable surface: A surface for which Gauss curvature is zero.
6
iii. Anticlastic surface: A surface for which Gauss curvature is negative
2.4.3 CLASSIFICATION BASED ON SURFACE OF GEOMETRIC

BEHAVIOR
Shells are classified based on this method, whether the surfaces are ruled or not.
I. Shells with ruled surface: Surfaces that can be obtained entirely by straight lines. The
surfaceis said to be singly ruled if at every point on the surface a single straight line can
be ruledand doubly ruled when at every point on the surface two straight lines can be
ruled.[7]
II. Shells with no ruled surface: are those that fall outside the above category of shells.
SomeCommon examples are dome, ellipsoid and elliptic paraboloid.[7]
2.4.4 CLASSIFICATION ACCORDING TO MANNER OF

GENERATION
Shells are generally classified based on their mode of formation.
I. SHELL OF REVOLUTION: shell that are obtained when a plane curve is rotated about
an axis of symmetry.[4]
II. SHELLS OF TRANSLATION: shells that are generated when one curve moves parallel
to itself alonganother curve, the plane of the two curves being at right angle to each
other.[4]
III. CYLINDRICAL SHELLS: Singly curved shell in which the generatrix (or directrix) is
a straight line. The surface generated by moving the straight line against a plane curve.[4]
2.5ADVANTAGE AND DISADVANTAGE OF SHELL

STRUCTURES
2.5.1 ADVANTAGES
1. The efficiency in its load carrying behavior(being treated as a membrane)
2. High degree of reserved strength and structural integritY
7
3. High strength: weight ratio( which is the main criteria in measuring a structure
efficiency)
4. Very high stiffness
5. Containment of space
6. High aesthetic value
2.5.2 DISADVANTAGE
1. It‟s impossible to build a story above the story that has a shell roof, thus shells are
always used as a “terminating” roof.
2.6CYLINDRICAL SHELL
A cylindrical shell surface is generated by translating a straight line generatorover a plane curve
directrix. An arc of a circle, semi ellipse, parabola, cycloid, and catenary are some ofthe most
common directrices to form cylindrical surface. Shells for which the plane curve is arc of acircle
are called circular cylindrical shells. All other which have their generating plane curves otherthan
circular arc are collectively referred to as non-circular cylindrical shells. [7]
Cylindrical shells are developable surfaces, a property that influences both their structural and
architectural aspects .Accordingly, large transverse moments occur in cylindrical shells when the
span becomes large, their performance calls for special attention by the way of increasing
reinforcement and thickness in region adjacent to springing in reinforced concrete systems or
provide sufficient stiffeners for other materials such as steel shells. [7]
Open cylindrical shells are mostly used for the construction of roof for industrial building, sport
andexhibition halls, garages and hangars etc...[7]
2.6.1 CLASSIFICATION OF CYLINDRICAL SHELL
Cylindrical shells can be classified as long, short, and sometimes intermediate shells. The object
ofthis classification, principally, is to help the designer to determine whether a shell of
particulardimension is susceptible for an approximate method of analysis. There are three
classificationmethods.
8
2.6.1.1 CLASSIFICATION BASED ON RADIUS AND

LONGITUDINAL LENGTH
A cylindrical shell are usually described as either long or short depending upon the ratio of
thelength (L) and radius (R). If the ratio [7]
r
i.  0.2 without edge beam
l
r
ii.  0.33 with edge beam
l
r
& it is called short shell if the ratio  0.2
l
2.6.1.2 CLASSIFICATIONBASED ON THE MAGNITUDE OF EDGE

DISTURBANCE
Based on the extent to which the disturbances arising from the springing edges penetrate into
thebody of the shell itself. Based on this consideration, once again the radius (r) and the
longitudinallength (l) of the shell are indicative parameter. [7]
r
i.  0.63 ; Long shell
l
r
ii.  0.63 ; Short shell
l
2.6.1.3 CLASSIFICATION BASED ON GEOMETRIC DIMENSION
AND THICKNESS
b
This method is based on the numerical value of the parameter   such that
0.25(12rt )
i. Ф < 3; Long shell

ii. 3< Ф < 5; Intermediate shell
iii. Ф < 5; Short shell[7]
9
2.7FIBER GLASS
Fiber glass is the common name for glass-reinforced plastic or alternatively glass-fiber
reinforced. Fiberglass is also known as GFK, for German: Glasfaserverstärkter Kunststoff
[10].Fiberglass is a fiber reinforced polymer made of plastic reinforced by glass fibers,
commonly woven into a mat. The plastic may be a thermosetting plastic most often epoxy,
polyester- or vinyl ester or a thermoplastic. The glass fibers are made of various types of glass
depending upon the fiberglass use. These glasses all contain silica or silicate, with varying
amounts of oxides of calcium, magnesium, and sometimes boron. Boron-containing fiber glasses
consume half the global production of boron minerals, and are the largest commercial consumer
of the element.[16]
Fiberglass is a strong lightweight material and is used for many products. Although it's not as
strong and stiff as carbon fiber it is less brittle, and its raw materials are much cheaper. Its bulk
strength and weight are also better than many metals, and it can be more readily molded into
complex shapes. Applications of fiberglass include aircraft, boats, automobiles, bath tubs and
enclosures, hot tubs, septic tanks, water tanks, roofing, pipes, cladding, bridges, outdoor aquatics
stage and stair cases.[12]
2.7.1 PROPERTIES
An individual structural glass fiber is both stiff and strong in tension and compression that
is, along its axis. Although it might be assumed that the fiber is weak in compression, it is
actually only the long aspect ratio of the fiber which makes it seem so; i.e., because a typical
fiber is long and narrow, it buckles easily. On the other hand, the glass fiber is weak in shearthat
is, across its axis. Therefore, if a collection of fibers can be arranged permanently in a preferred
direction within a material, and if they can be prevented from buckling in compression, the
material will be preferentially strong in that direction. [12]
Furthermore, by laying multiple layers of fiber on top of one another, with each layer oriented in
various preferred directions, the material's overall stiffness and strength can be efficiently
controlled. In fiberglass, it is the plastic matrix which permanently constrains the structural glass
10
fibers to directions chosen by the designer. With chopped strand mat, this directionality is
essentially an entire two dimensional plane; with woven fabrics or unidirectional layers,
directionality of stiffness and strength can be more precisely controlled within the plane. [12]
The most common types of glass fiber used in fiberglass is E-glass, which is alumino-
borosilicate glass with less than 1% w/w alkali oxides, mainly used for glass-reinforced plastics.
Other types of glass used are A-glass (Alkali-lime glass with little or no boron oxide), E-CR-
glass (Electrical/Chemical Resistance; alumino-lime silicate with less than 1% w/w alkali oxides,
with high acid resistance), C-glass (alkali-lime glass with high boron oxide content, used for
glass staple fibers and insulation), D-glass (borosilicate glass, named for its low Dielectric
constant), R-glass (alumino silicate glass without MgO and CaO with high mechanical
requirements as Reinforcement), and S-glass (alumino silicate glass without CaO but with high
MgO content with high tensile strength).[12]
2.7.2 TYPES OF FIBER GLASS
GRP (glass reinforced plastic) is a composite of tough resilient, durable plastic resin, and glass
fibers of remarkable strength. The resin is a thick, treacly substance which when activated by an
appropriate catalyst, sets to a hard but brittle solid. It can be used alone for small castings, or
with a variety of fillers, but, when reinforced with glass fibers, becomes a material of exceptional
strength and versatility. [16]
The Glass used commonly for GRP is a calcium-alumina borosilicate with an alkali content of
less than one per cent. It is commonly known as „E‟ type glass, since it was originally developed
for use in electrical insulation systems.[16]
Glassfibres are produced by running molten glass from a direct melt furnace into a platinum
alloy bushing containing a large number of small holes, from each of which a glass filament is
drawn. Filaments for commercial use are normally between 9 and 15 microns in diameter. The
filaments are “dressed” with an emulsion before being gathered into fibers. The fibers are
remarkably strong-the tensile strength being particularly high. They also exhibit good chemical
and moisture resistance, have excellent electrical properties, are not subject to biological attack
11
and are non-combustible with a melting point around 1500oC all excellent qualities in a plastic
reinforcement.[16]
Provided the glass reinforcement has been thoroughly impregnated with resin, the result, after
curing, is a cohesive completely integrated matrix of resin and fibers. The matrix can have a
surprising range of properties, depending on the type of glass material and the formulation of the
resin. In general, the GRP laminate will display excellent tensile and compressive strength,
acceptable thermal conductivity, a low coefficient of linear expansion, reasonable chemical
resistance and good dielectric properties. Compared to other materials of equivalent strength, it
will be light durable, moisture-resistant, non-rusting and economic.[16]
There are different kinds of fiber glass with different compression and tensile strength. This is set
by East Coast Fiberglass Supplies Ltd 2013
Material Specific Tensile strength Compressive
12
gravity (Mpa) strength (Mpa)
Polyester resin (Not reinforced) 1.28 55 140
Polyester and Chopped Strand Mat 1.4 100 150

Laminate 30% E-glass
Polyester and Woven Rovings Laminate 1.6 250 150

45% E-glass
Polyester and Satin Weave Cloth Laminate 1.7 300 250

55% E-glass
Polyester and Continuous Rovings 1.9 800 350

Laminate 70% E-glass
E-glass 2.58 3445 1080
S-2 glass 2.46 4890 1600
TABLE 2.1Common Fiberglass Types[16]
The fibers can be used in a variety of ways-chopped into short lengths(“chopped strands”);
gathered together into loosely bound ropes (“rovings”); woven into a variety of fabrics, produced
from yarn made by twisting and doubling continuous strands. In the UK, the most widely used
glassfibre material is chopped strand mat, which consists of glass strands chopped together in
short lengths (approx. 50mm) and held together in mat form by a polyvinyl acetate or polyester
binder. The mat is available in a range of weights, from 225gm2 to 1200gm2, and is a useful
general purpose reinforcement.[16]
Some experiments have been done about fiber glass and this tensile and compressive behavior of
fiber glass shows the capacity of fiber glass to be used as a structural material. The behavior of
13
this carbon fiber composite material is that they are brittle and break at their maximum load
capacity even though it is used in different countries. From the tests done it is concluded that the
material is not ductile.[16]
FIGURE 2.1Stress Strain Relation of Fiber Glass in a Tensile Test[16]
2.8DESIGN OF SHELL STRUCTURE

2.8.1 LOADING
Loading used in a design of a structure depends on the serviceability and also the ultimate
capacity of the structure. If the structure doesn‟t stand to support these two major divisions of
load then it must be revised to do so. There are different kinds of loads this are live load, dead
load, wind load and earthquake loading. These loads are the major loads depending on the
country snow load and temperature consideration may be added. To know what kind of load that
should be considered in the Euro Code check this1990:2002 [13].
2.8.2 LOADING COMBINATION
Required strength, U shall be at least equal to the effects of factored loads in Eq. 2.6.2.1up toEq.
2.6.2.7.The effect of one or more loads not acting simultaneously shall be investigated.
U  1.4( D  F ) …………………………………………………………………………Eq. 2.6.2.1
14
U  1.2( D  F  T )  1.6( L  H )  0.5( LrorSorR) ……………………………………Eq. 2.6.2.2
U  1.2D  1.6( LrorSorR)  (1.0Lor0.8W ) ……………………………………………Eq. 2.6.2.3
U  1.2D  1.6W  1.0L  0.5( LrorSorR) ……………………………………………...Eq. 2.6.2.4
U  1.2D  1.0E  1.0L  0.2S …………………………………………………………Eq. 2.6.2.5
U  0.9D  1.6W  1.6H ……………………………………………………………….Eq. 2.6.2.6
U  0.9D  1.0E  1.6H ……………………………………………………………….Eq. 2.6.2.7
Except as follows:
 The load factor on the live load L in Eq. (2.6.2.3) to (2.6.2.5) shall be permitted to be
reduced to 0.5 except forgarages, areas occupied as places of public assembly, and all
areas where L is greater than 488.0408kg/m2.
 Where wind load W has not been reduced by a directionality factor, it shall be permitted
to use 1.3W in place of 1.6W in Eq. (2.6.2.4) and (2.6.2.6).
 Where E, the load effects of earthquake, is based on service-level seismic forces, 1.4E
shall be used in place of 1.0E in Eq. (2.6.2.5) and (2.6.2.7).
 The load factor on H, loads due to weight and pressure of soil, water in soil, or other
materials, shall be set equal to zero in Eq. (2.6.2.6) and (2.6.2.7) if the structural action
due to H counteracts that due to W or E. Where lateral earth pressure provides resistance
to structural actions from other forces, it shall not be included in H but shall be included
in the design resistance.
 Where Lr is Roof load of the structure. R is rain load, S snow load, F is Loads due to
weight and pressures of fluids with well-defined densities and controllable maximum
heights, or related interval moments and forces and T is cumulative effective effect of
temperature, creep, shrinkage, differential settlement, shrinkage compensating
concrete.[13]
15
If resistance to impact effects is taken into account in design, such effects shall be included with
estimations of differential settlement, creep, shrinkage, expansion of shrinkage-compensating
concrete, or temperature change shall be based on a realistic assessment of such effects occurring
in service.
Generally the factor assigned to each load is influenced by the degree of accuracy to which the
load effect usually can be calculated and the variation that might be expected in the load during
the lifetime of the structure. Dead loads, because they are more accurately determined and less
variable, are assigned a lower load factor than live loads. Load factors also account for
variability in the structural analysis used to compute moments and shears.
Consideration should be given to various combinations of loading to determine the most critical
design condition. This is particularly true when strength is dependent on more than one load
effect, such as strength for combined flexure and axial load or shear strength in members with
axial load.
2.8.3 SAFETY FACTOR
Safety factor is used in various design equations were chosen to prevent first deformation. The
first deformation is the first visible deformation. The recommended value for this kind of
structure is 2.5. [16]
2.8.4 DESIGN CONSIDERATION
While designing different kinds of structure it is a must to know which things should be taken in
to consideration. This helps in efficient design of a structure. Therefore the following points
should be taken in to consideration while designing cylindrical shell structure.
A. Selection of shell type. For covering very large areas, for hangars, warehouses, etc., short
shells are usually economical. The span of the short shell may be chosen as between one-
sixth and one – third of the chord width. A practical limit on the span of long reinforced-
concrete shells is about 30.48m .For longer shells, prestressing will prove economical.[2]
16
B. The radius of cylindrical shell has to be chosen keeping acoustic considerations in view. It
is desirable to see that the center of curvature does not lie at the working level. [2]
C. Semi central angle
The practice is to keep the semi central angle between 300 and 450. If the angle exceeds
450, concreting becomes difficult without the use of top forms .If the angle is below 40 0,
wind load can be ignored, because it causes only a suction on the shell. [2]
D. Thickness
The minimum thickness of reinforced-concrete cylindrical shells is governed by
practical considerations such as accommodating reinforcement and providing adequate
cover. According to a Dutch report the usual recommended thickness is between 0.07
and 0.08 m .A minimum of 0.04 m is recommended by the institute for the Typification
of the German Democratic Republic at Berlin.[2]
E. Width of Edge Beam
A width of two to three times the thickness of shell would usually suffice .A minimum
of 0.1524m is demanded by practical considerations. [2]
F. Design of Reinforcement
When fiber glass is used as a construction material
a) Chopped Strands Short (6mm or 12mm) lengths of glassfibre. Can be used to
make a resin dough, stronger than that made by mixing resin with filler powder.
Chopped Strand Mat a popular and economical form of glass reinforcement for
polyester resins. Short strands of glass are bonded with a powder or emulsion into
a mat available in a variety of thickness.[16]
b) Filaments for commercial use are normally between 9*10^-6 and 15*10^-6m in
diameter. [16]
When reinforced concrete (non prestressed)
a) The ratio of steel to concrete in any portion of the tensile zone should not be less
than 0.35% of the cross sectional area of concrete.[2]
b) The minimum temperature and shrinkage steel should not be less than 0.14% of
the cross sectional area of concrete.[2]
17
c) The maximum spacing of bars should not exceed 40 bar diameters nor five times
the thickness of the shell. [2]
G. Finishing for fiber glass shell
Glassfibre materials once hardened can be polished, sanded, drilled, sawn or filed.
Diamond carborundum or metal finishing tools generally are required. Since the dust
produced can be extremely hazardous to eyes and lungs protective goggles and breathing
masks should be worn at all times when machining hardened resins, with or without
glassfibre reinforcement. [16]
2.8.5 METHOD OF ANALYSIS

2.8.5.1 LOAD BALANCING METHOD
Consider the cylindrical shell without edge beams. The cables can be post-tensioned along the
shell surface so that the vertical component of a cable will balance the gravity load. The
prestressing forces, and its vertical component of a cable will balance the gravity load. The
prestressing forces, and its vertical component Wv and horizontal component Wh are given by [2]
Wl 2 1
H ……………………………………………………………………….…...Eq. 2.8.5.1.1
8 fv 2
8 fv
Wv  H …………………………………………………………………………...Eq. 2.8.5.1.2
l2
8 fh
Wh  H ……………………………………………………………………………Eq. 2.8.5.1.3
l2
In these expression W is the uniform load of the shell per linear feet along the X axis , l is the
span of the shell ,fv is the projected vertical sag of the parabolic cable and fhis the projected
horizontal sag of the parabolic cable and fh is the projected horizontal sag of the parabolic
cable.[2]
18
2.8.5.2 BEAM THEORY

The beam theory involves two analyses:
A) The beam analysis
I. The longitudinal stresses at any cross section of the shell are computed on the basis
of the simple flexural theory
Mc
Nx  …………………………….…………………………………………Eq. 2.8.5.2.1
I
Where I is the moment of inertia of the shell cross-section about the axis yy .It may be shown
that
  2 sin c 
I  R 3t c  sin c (  cos c   ………………………………………………Eq. 2.8.5.2.2
  c 
The shearing stresses are computed by the corresponding simple expression
………………………………………………………………Eq. 2.8.5.2.3
Where Q is the first statically moment of the cross section up to the point under consideration
about the axis yy and is [2]
 sin c
Q  2 R 2t (sin   ) ……………………………………………..Eq. 2.8.5.2.4
c
II. For the case of the prestressing force acting on the shell the vertical component of the cable
force is equal to some gravity load , therefore , the beam shear force and bending moment
are zero for this load .
The longitudinal force is
2H
Nx   ……………………………………………………………Eq. 2.8.5.2.5
A
19
Where A is the cross-sectional area of the shell.
The following equations for transverse forces and moments have been developed by using free
body. The force Fv was obtained by summing vertical forces, and MФ by taking moments about
the intersection of a radial line at angle Ф and the arc of shell.[2]
A. Forces due to Dead Load
Fv  gR(c   ) ………………………………………………………………Eq. 2.8.5.2.6
M   gR 2 cos   cos c  sin  (c   )…………………………………..Eq. 2.8.5.2.7
N  gR(c   ) sin  ………………………………………………………..Eq. 2.8.5.2.8
Q   gR(c   ) cos  ……………………………………………………….Eq. 2.8.5.2.9
B. Force due to water load
Px  0 ……………………………………………………………………….Eq. 2.8.5.2.10
Pn   h  a(1  cos  ) ……………………………………………………Eq. 2.8.5.2.11
P   h  a sin   ………………………………………………………….Eq. 2.8.5.2.12

Its own stress
N  ha cos  ..…………………………………………………………….Eq. 2.8.5.2.13
20
h l 2 2
Nx   (  x ) cos  ............……………………………………………Eq. 2.8.5.2.14
a 4
N x  Nx  2hx sin  ……………………………………………………..Eq. 2.8.5.2.15
2.8.5.3 THE ARCH ANALYSIS

When forces acting on the shell are balanced by prestressing forces, there is no shear flow
produced at all, which means no arch action in the shell, but when the acting forces are
unbalanced by the prestressing forces, shear flow will appear and the arch action will have an
effect on the shell response. The vertical components of the shear flow balance the load on
the shell arch; the horizontal components of shear flow which are symmetrically disposed
about the crown balance themselves. The resulting internal forces are [2]
c
gR 2c Q( )
Fv   I
sin d  gR(c   ) …………………………………………..Eq. 2.8.5.3.1
c
gRc Q( )
Fh    cos Rd …………………………………………….......…...Eq. 2.8.5.3.2
 I
Q   Fv cos   Fh sin  ………………………………………………………...Eq. 2.8.5.3.3
N  Fv sin   Fh cos  …………………………………………………………..Eq. 2.8.5.3.4
c
M    gR 3c Q( ) sin  (sin   sin  )d

…………….……Eq. 2.8.5.3.5
c
  gR 3c Q( ) cos  (cos   cos  )d  gR 2 cos   cos c  sin  (c   )

2.8.5.4 MEMBRANE THEORY
The equations of equilibrium for the shell element generally contain ten unknowns: the stress
resultants and stress couples – and six equations. The system is therefore statically indeterminate.
21
As a result, deformation must be considered in order to obtain a solution. The Membrane Theory
avoids the complexities involved with solving the statically indeterminate system by reducing the
number of unknowns from ten to four. This results in a statically determinate system that can be
solved directly. The Membrane Theory accomplishes this by neglecting all normal shears,
bending moments, and twisting moments in the shell.[7]
FIGURE 2.2Shear ,Bending Moment And Twisting Moment[3]
The equations of the equilibrium then reduced to:

N xay   N y y  N xy x   N yxax   pxaxay  0 ……………………Eq. 2.8.5.4.1
 x y y

N yax   N x ax  N xy y   N yxay   pyaxay  0 ……………………Eq. 2.8.5.4.2
y x x x
N x N xy N yx N y
    pz  0 ……........................................................................Eq. 2.8.5.4.3
rx rxy rxy ry
The validity of the results obtained by the Membrane Theory depends upon a number of
conditions. First, the boundary conditions must be compatible with the conditions of equilibrium.
Second, the application of the loading must be compatible with the conditions of equilibrium.
22
FIGURE 2.3Compatibility of Boundary Conditions with the Membrane Theory [8]
Figure 2.7 shows the shell edge with different boundary conditions. The system depicted in(a) is
free to translate horizontally. There is therefore a component of the reaction that acts normal to
the shell surface at the edge. Yet the Membrane Theory neglects all forces that act out of plane,
so the element is out of equilibrium. The boundary condition displayed in (b) supplies a reaction
that acts only in the plane of the shell. Therefore, the support is compatible with both the
Membrane Theory and the conditions of equilibrium.
Now consider the case of a shallow shell subjected to a smoothly distributed load. As the radius
of curvature of the shell approaches infinity, the shell becomes a plate. Under Membrane Theory,
this shell possesses no internal force system capable of contributing to the vertical equilibrium.
Clearly, either normal shear of bending must occur in order for the plate to carry the loading.
Thus the Membrane theory is not applicable to plates.
As the previously discussed examples have shown, the Membrane Theory is not suitable for all
shell analysis. While the assumptions of the Membrane Theory clearly simplify the mathematics
involved in the solution of shell problems, they do not fully explain the behavior of shells. Under
certain circumstances the Membrane Theory may provide the complete solution to a shell
problem. Under different circumstances, however, serious incompatibilities may occur and a
more rigorous analysis method must be employed.
23
FIGURE 2.4Cylindrical Shell with Edge Beam[12]
A cylindrical shell is formed by the motion of straight line, called the generator, parallel to itself
such that its end describes a plane curve which may be circle, ellipse, parabola, etc. Or may be of
arbitrary shape. A section normal to the generator is called profile.
A point p on the middle surface of the shell is defined by the distance x along the generator,
measured from a profile of reference, and the angle θ which the normal n to the surface makes
with a certain reference axis .Thus, a system of right handed orthogonal axes at point p consists
of the tangent to the surface, where the positive directions of such axes are taken as shown in the
figure.
Consider now an infinitesimal element (fig) of the cylindrical shell bounded by the two profiles
at x and x+dx and the two generators at θ and θ+dθ, where all the external loads and internal
actions are represented as components along the chosen axes. The surface loadings have the
components px, pθ and pn per unit area of surface .In the membrane analysis, the only internal
actions considered are the normal forces Nx and the shearing force Nxθ, acting per unit length of
the side x=constant, whereas a unit length of the side θ = constant is acted upon by the normal
force Nθ and the shearing force Nθx as indicated in the figure
24
Now if r denotes the radius of curvature of the shell profile at any point, then, the length of the
side x= constant, of the infinitesimal element of fig., is rdθ. Taking moment about the normal to
the surface of the infinitesimal element, it follows that:
N x  Nx ……………………………………………………………………………Eq. 2.8.5.4.4
Resolving forces acting in the element in the direction of the axes x, θ and n, one obtains,
respectively:
N x 1 N x
dxrd   Px rddx  0 ………………………………………………..Eq. 2.8.5.4.5
x r 
N N x
ddx  dxrd  p rddx  0 …………………………………………..Eq. 2.8.5.4.6
 x
 N dxd  Pn rddx  0 ……………………………………………………….Eq. 2.8.5.4.7
Dividing the above equation by r dθdx and reducing, one obtains:

N x  1 N x 
   Px  ……………………………………………………………Eq. 2.8.5.4.8
x  r  
N x  1 N 
   P  …………………………………………………………...Eq. 2.8.5.4.9
x  r  
N  Pn r …………………………………………………………………………Eq. 2.8.5.4.10
It, thus, appears that Nθ depends only on the local intensity of the normal load component pn
(Eq. 2.7.5.4.5) and is independent of the boundary conditions. This is not so important in the
case of shells having closed profiles; however, this condition puts practical limitations on the
validity of the membrane analysis for shells with open profiles due to the difficulty of
prescribing the required values of Nθ along the straight edge.[7]
25
In the special case when px=0 and pθ, pn are independent of x (same loading over all profiles),
it will be expected that Nθ is also independent of x and Eq. 2.7.5.4.6, 7&8 reduce to:
N x  1 N x 
  ………………………………………………………………..Eq. 2.8.5.4.11
x  r  
N x  1 N 
   P    F ( ) ………………………………………………...Eq.
  r  
2.8.5.4.12
N  Pn r ………………………………………………………………….………Eq. 2.8.5.4.13
Substituting fromEq. 2.7.5.4.7 into Eq. 2.7.5.4.6 and integrating, one has:
Nx   xF ( )  f1 ( )  0 ……………………………………………………...Eq. 2.8.5.4.14
1 d
F ( )   rPn   P  ………………………………………………………Eq. 2.8.5.4.15
 r d 
Where the prime denotes differentiation with respect to θ and f1, f2 are arbitrary functions to be
determined from the support conditions over an edge x= constant. Such edge supports are often
provided by stiffening members, known as diaphragms or traverses, which may be arches,
trusses, beam plate, etc. The diaphragm is usually assumed to be very stiff in its own plane, but
flexible out of this plane. In other words, a diaphragm carries loads (Nxθ) in its own plane, but
cannot receive any loads applied normal to its plane. Hence, a usual boundary condition is Nx= 0
along the edge x=constant.[2]
The corresponding displacement is
26
Transverse displacement:
8g x  2 1 
V sin  cos  2  2 4  …………………………………………………..Eq. 2.8.5.4.16
Et l K R K 

K ………………………………………………………………………………Eq. 2.8.5.4.17
l
Normal displacement
v 8g x  2 1 
W  cos  cos  2  2 4  …………………………………………Eq. 2.8.5.4.18
 Et l K R K 
Longitudinal displacement
1 8g x
U  cos  sin ………………………………………………...………Eq. 2.8.5.4.19
Et RK 3
l
2.8.5.5 BENDING THEORY
The more general and rigorous approach for solving shell problems is the so-called bending
theory, also referred to as the shallow-shell theory as it is often called, and it considers the effects
of the longitudinal moment and transverse shear (in the direction of the axis of the cylinder)
The bending theory is based on two assumptions; these are:
i. The shell is so flat that the out-of-plane translational displacements w are much
greater than the in-plane ones, u and v.
ii. The shell is steep enough so that in-plane stress resultants Nx, Ny and Nxy contribute to
resisting the loads that do the out-of-plane stress resultants Qx and Qy.
Both the membrane theory and the bending theory are very important and widely
applicable to solving various kinds of thin shell structures. The edge-load solutions (bending
27
solutions), together with the membrane solution, will yield the complete solution of a shell
problem.[7]
For solving shell problems, therefore, the general procedure will be to obtain membrane
forces under a given loading, and then superimpose on these the bending theory solutions for
edge loads. The edge loads are determined so that the membrane and edge load solutions
together satisfy the boundary conditions; i.e., the edge loads are obtained by solving
equations of compatibility at the boundaries.[7]
FIGURE 2.5Edge Reactions and Load Applied
2.8.6 ANALYSIS PROCEDURE
After defining design assumptions, the general theory of shell can be formulated in the following
five steps [8]
28
Determine the equilibrium equations for the shell

Element( Five equations with eight unknown)
Establish the strain-displacement

relationships(six equations with three
unknowns)
Establish the stress-strain relationships by assuming material

properties (three equations with six unknowns) and then
deriving forces-strain equations (six equations with three
unknowns)
Transform the force-strain relationship in to force-

displacement equations (six equations with six
unknowns)
Obtain the complete formulation by

combining the force-displacement equation
with equilibrium equations(eleven equations
with eleven unknowns)
FIGURE 2.6 Analysis Procedure
a) Forces affecting the equilibrium of the shell element are set up

N x 1 N x
  0 ………………………………………………………………Eq. 2.8.6.1
x R 
N N
 R x  0 ………………………………………………………………Eq. 2.8.6.2
 x
Qx Q
R   N  0 …………………………………………………………..Eq.
x 
2.8.6.3
29
M x M 
R   RQ  0 ………………………………………………………Eq.
x 
2.8.6.4
M x M x
R  RQ  0 ……………………………………………………….Eq. 2.8.6.5
x x
N x  Nx  0 …………………………………………………………………...Eq. 2.8.6.6
Let us consider an open barrel cylindrical shell loaded by uniform distributed load, as seen in
Figure.
FIGURE 2.7Barrel cylindrical shell under uniformly distributed load [4]
The boundaries of the shell along its length are considered as fixed support in the analysis.
Stress functions will depend only on one variable, namelyϕ leading to the following differential
equations system:
30
FIGURE 2.8Membrane Stresses, Moment and Shear Forces Acting on Infinitesimal

Element of Shell. [4]
2.8.6.1 THIN CYLINDERICAL SHELL DESIGN

Let us consider a cylindrical coordinate system (x,ϕ,r) and a cylindrical shell with
thickness.Thincylindrical shell structure subjected to uniform loading responds by equilibrium
equation. A stress element with its faces parallel to the axis of the cylinderis shown on the wall
of the shell. The normal stresses σ1 and σ2 acting on the side faces of the element. No shear
stresses act on these faces because of symmetry of the vessel and its loading. Therefore, the
stresses σ1 and σ2 are principal stresses. Because of their directions, the stress σ1 is called the
circumfrancial stress or the hoop stress, and the stress σ2 is called the longitudinal stress or axial
stress. Each of these the stresses can be calculated from static equilibrium equations.[15]
Several assumptions have been made to derive the following equations for circumfrancial and
longitudinal stresses:
1. Plane sections remain plane
r
2.  10 with t being uniform and constant
t
3. Material is linear-elastic, isotropic and homogenous
4. Stress distributions throughout the wall thickness will not vary
5. Weight of the fluid is considered negligible[15]
31
FIGURE 2.9Shell under Uniformly Distributed Load [1]
 hoop 
 P * R  ………………………………………………………………………...Eq.2.8.6.7
2*t
p
 radial  ……………………………………………………………………………..Eq.2.8.6.8
2
But while deriving the longitudinal stress the shell is considered to be closed at both sides.
 p*d
 axial  …………………………………………………………………………..Eq.2.8.6.9
4*t
Strain of the shell can be calculated as:
32
Hoop strain:
h 
1
 h  r t  ……………………………………………………………......Eq.
E
2.8.6.10
Longitudinal strain:
t 
1
 t  r  h  ………………………………………………………….......Eq. 2.8.6.11
E
Radial strain:
r 
1
 r  h  t ……………………………………………………………….Eq.2.8.6.12
E
Change in diameter can be calculated from the strain on diametral, i.e. the diameter diametral
strain.
change in diameter
Diametral strain  ……………………………………………..Eq.2.8.6.13
original diameter
Now the change in diameter may be found from a consideration of the circumferential change.
The stress acting around a circumference is hoop or circumferential stress σH giving rise to the
circumferential strain εH. [15]
New circumference   * d   * d *  H ……………………………………………....Eq.2.8.6.14
 d * 1   H 
But this is circumference of a circle of diameter d(1+εH)
New diameter  d * 1   H 
33
change in diameter  d * H
d H
Diameteral strain  D    H …………………………………….Eq.2.8.6.15
d
i.e the diametrical strain equals the hoop or circumferential strain
d
change in diameter  d H  [ H   L: ]
E
………………………………………….Eq.2.8.6.16
pd 2
 2  
4tE
Deflection in the radial direction:
F  L3
 …………………………………………………………………….E.q.2.8.6.17
384  E  I
Assuming the cylindrical shell to be fixed at both ends of the longitudinal direction. The above
equation can be used to find the maximum deflection.
2.8.6.2 THICK CYLINDERICAL SHELL DESIGN
The theoretical treatment of thin cylindrical shells assumes that hoop stress is constant across the
thickness of the shell wall and also that there is no pressure gradient across the thickness of shell
wall and also that there is no pressure gradient across the wall. The Lame theory for
determination of the stress in the walls of thick cylindrical shells considers a mutually
perpendicular, triaxial, principal-stress system consisting of the radial hoop or tangential, and
longitudinal stresses acting at any element in the wall. For the case of the shell depicted in figure
2.14, subjected to internal pressure only, radial stress σr, and hoop stress σh are given by[17]
r 
pri 2 
 r 2  ro2

  …………………………………………………………......Eq. 2.8.6.19

ro2  ri 2   r
2 

34
r 
pri 2 
 r 2  ro2

  ……………………………………………………………...Eq. 2.8.6.20

ro2  ri 2   r
2 

FIGURE 2.10 Thick Cylindrical Shell Subjected to Internal Pressure [17]
Where ri and ro is inside and outside radii of the shell, respectively, and r is any radius such that
ri<r<ro. [17]
In order to get an expression for the longitudinal stress σt, the shell is considered closed at both
ends, as shown in figure 2.15.Simple consideration of the force equilibrium in longitudinal
direction yields [17]
FIGURE 2.11Thick Cylindrical Shell Closed at Both Ends and Subjected to Internal
Pressure [17]
35
pri 2
 t  2 2 ………………………………………………………………………..….Eq. 2.8.6.21
ro  ri
The changes in the dimensions of the cylindrical shell can be determined by the following strain
formulas:
Hoop strain:
h 
1
 h  r t  ……………………………………………………………......Eq. 2.8.6.22
E
Longitudinal strain:
t 
1
 t  r  h  …………………………………………………………..........Eq. 2.8.6.23
E
Radial strain:
r 
1
 r  h  t ……………………………………………………………...Eq. 2.8.6.24
E
It can be shown easily that diametral stress and the circumferential or hoop stress are equal. It is
seen that the inside diameter-to-thickness ratio di/t is an important parameter in the stress
formulas. For di/t valuesgreater than 15, the error involved in using the thin-shell theory is within
5 percent. If, however, the mean diameter dm is used for calculation of thin-shell values instead
of inside diameter, the error reduces from 5 percent to approximately 0.25 percent at
dm/t=15.0.[17]
When the cylindrical shell is subjected to both the external pressure po and the internal pressure
pi, p is the pressure on the the radial and hoop stresses can be expressed by
ri 2 pi  ro2 po  ri 2 ro2  po  pi  / r 2
r  ……………………………………………….Eq. 2.8.6.25
r02  ri 2
36
ri 2 pi  ro2 po  ri 2 ro2 ( po  pi ) / r 2
h  ……………………………………………….Eq. 2.8.6.26
r02  ri 2
It is observed from the preceding equations that the maximum value of σh occurs at the inner
surface. The maximum radial stress σr is equal to the larger of the two pressure Pi and Po. These
equations are known as the Lamé solution. The sum of the two stress remains constant, which
indicates that the deformation of all element in the axial direction is the same, and the cross
sections of the cylinder remain plane after deformation. [17]
The maximum shearing stress at any point in the wall of cylinder is equal to one-half the
algebraic difference between the maximum and minimum principal stress at that point. The axial
or longitudinal stress is usually small compared with the radial and tangential stresses. The shear
stress τ at any radial location in the wall is given by[17]
h r
 …………………………………………………………………...………Eq. 2.8.6.27
2

 pi  po ri2 ro2 ……………………………………………………………………....Eq. 2.8.6.28
r
o
2

 ri 2 r 2
Deflection in the radial direction can be calculated as:
 P0 r02 r  ri 2 
Ur       ………………………………………………….Eq. 2.8.6.29
E (ro2  ri 2 )  r 2 
(1 ) (1 )
37
CHAPTER 3DESIGN OF THE STRUCTURE USING

ANALYTICAL METHOD
3.1 INTRODUCTION
In this section, structural design of underwater cylindrical shell structure is done. There are five
methods of design for thin cylindrical shell design. They are thoroughly explained in chapter
two. But combination of bending analysis and membrane analysis is chosen for the design.
Because combination of this two analysis gives a result that describes the reaction of the
cylindrical shell.
There are multiple of software‟s that can be used to calculate the analysis. Microsoft excel is one
of those software‟s. When using this software by substituting the equations and trying to find a
stable structure it is tedious as well as takes time. This software is not effective for this purpose.
Writing a program using C++ language was tried. While doing this it was important to choose
what kind of program was needed. Structural design by is designing the structure to its stable
dimension and material type depending on the purpose of the structure.so this softer should do
just that for the design.
The design considers minimum volume per usable area of the structure. This is considered
because the structure is needed for qua zoo. And much as possible the usable area should be
maximum. The volume should be minimum because the cost will increase as the volume
increases. The design considers this function and compares some combination of dimensions.
The stability is checked based on ultimate limit state design and serviceability limit state
design.The initial design concept is based on a cylindrical shell with edge beams described by
Billington. The design model is presented for two different modeling.
38
3.2 MODELING
The underwater cylindrical shell has edge beam. Furthermore, an ideal load bearing wall is added
at the end of the cylindrical shell so that the water won‟t flood the space inside the shell.
Fiber glass is used as a construction material and the material is fabricated as per the design
results. The thickness is limited between 4cm and 8cm which is recommended by Typification of
the German Democratic Republic at Berlin. This new material offers possibilities for the design
of efficient shell structures that can be installed easily, which will be confirmed by production
companies. There are different kinds of fiber glass so the design is done for different type of
fiber glass and different dimension. The length of the fiber glass is limited from 20meters to 40
meters. The radius is limited from 4meters to 9meters. The last two dimensions are limited as per
the judgment of the designer. So for each type of fiber glass there will be a search for stable
structure.
FIGURE 3.1Underwater Cylindrical Shell: (A) Design Model and (B) Analysis Model [2]
39
Figure 3.2 Design Variables.
Five parameterizations of the underwater cylindrical shell are presented. The comparison of the
design results for each alternative will clearly demonstrate the importance of the
parameterization. In figure 3.2, the design variables are indicated on the structure. In the first
parameterization, is the first type of fiber glass is presented. The next parameterization the value
of the next stronger fiber glass is presented.The type of fiber glass is chosen based on the
stability of the structure. The shell thicknesst is assumed to be constant over the shell surface.
CONSTRAINTS
In summary the constraints are
1. DESIGN CONSTRAINTS
  compressive   min ( x)   max ( x)   tensile.


Design constra int u hmax  H t / 200
 max
u v  B / 300
40
The horizontal deflection is limited to uhmax  H t / 200 and the vertical deflection is limited to
uvmax  B / 300 . [14]
The behavior constraints are related to stresses and displacements. For the ultimate limit state
(ULS) load combinations, the maximal and minimal principal stress σ max (x) and σ Min (x) are
taken from the table that is provided by East coast manufacturers. Which are the ultimate
compression and tension strength σcompressive and σtensile. For the serviceability limit state (SLS)
load combinations, the maximal horizontal displacement Uhmax is limited to a fraction of the
total height Ht/200=(R-RcosΨk)/200 m, the maximal vertical displacement Uvmax is limited to a
fraction of the total height B /300=2*RsinΨk m/200.
2. SIDE CONSTRAINTS
A. For thin cylindrical shell
1. The length(L) ranges from 20m to 40m
2. The radius(R) ranges from 12m to 17m
3. The thickness(t) is in the interval of 0.04m to 0.266m
4. Angle of curvature(Ψk ) is between450 to 900
5. Height of the shell above the ground(Ht) is 3.52m to 17m
B. For thick cylindrical shell
1. The length(L) is ranges from 20m to 40m
2. The radius(R) is free from 19m to 24m
3. The thickness (t) is in the interval of0.38m to 0.424m.
4. Angle of curvature(Ψk ) is between300 to 450
5. Height of the shell above the ground(H) is 2.545m to 7.029m
MATERIAL PROPERTY
1. Polyester and Woven Rovings Laminate 45% E-glass
Tensile stress allowable is 250Mpa
Compressive stress allowable is 150Mpa
41
Modulus of elasticity is 2.48Gpa
2. Polyester and Satin Weave Cloth Laminate 55% E-glass
3. Polyester and Continuous Rovings Laminate 70% E-glass
4. E-glass
Tensile stress allowable is 3455 Mpa
Modulus of elasticity 2.58Gpa
5. S-2 glass
Tensile stress allowable is 4890 Mpa
Compressive stress allowable is 1600 Mpa
Modulus of elasticity 2.46Gpa
LOADING
Depends on the material property and the geometry of the structure it changes according to this
data‟s.
Load combination for ultimate limit state design
PL=1.35DL+1.5LL(Eurocode basis of structural design sec 6.4.3.2 Equation 6.9a)
42
Load combination for serviceable limit state design
PL=DL+LL (Eurocode basis of structural design sec 6.4 Equation

6.14)
Dead load=self-weight +water load
=   t   w  h0   w * r  r * cos( ) ;
 γis the specific gravity of the fiber glass

 t is the thickness of shell
 γw is the specific gravity of water
 ho the height of the water above the cylindrical shell which is 2meters.
 r is radius of curvature
 ϕ is design variable that changes from zero to ϕk
Live load =220KN as recommended on American committee institution code(ASCE 7-05

Table4-1) to use the load asking experts on the field the value load of largest shark is 220KN
when distributed on the whole area.
ADJUSTMENT REGARDING VOLUME PER USABLE AREA
The purpose of the structure is for aqua zoo. The structure needs to be serviceable. But the cost
must be as minimum as possible. If the function is only volume then when the volume is zero or
minimum is when it is efficient but this can‟t be used. If only usable area is used then this will
result in infinite answer. So there is a need for deriving a formula which will take in to
consideration both the functions.
If minimizing volume per usable area is used. This function will be minimum when the volume
is as minimum and the usable area is maximum. Which means that the structure will be modified
to achieve maximum efficiency by adjusting the ratio of volume per usable area to be minimum.
Volume of the cylindrical shell can be calculated as:
43
Volume    R  t  L
Figure 3.3 Dimension and usable area
usable Area  Area of sec tor  Area of triangle ………………………….…Eq.3.2.1

r  2   k r cos  k  2  r sin  k
2
 
2 2
 r   k  r  cos  k  sin  k
2 2
Therefore volume per usable area can be calculated as:
V ( k , R, L, t )  k  r  t  L 
 2  ……………………………….Eq.3.2.2
usable A( k , R)  ((r   k )  (r  cos( k )  sin( k ))) 
2
3.3 PROGRAM DEVELOPMENT FOR ANALYTICAL DESIGN

OF UNDERWATER CYLINDRICAL SHELL STRUCTURE
3.3.1 FLOW CHART OF THE PROGRAM
The flow chart is a simple way to analyze and describe the program used for the optimization
process.
44
45
Start
Declaration of variables and

arrays
Initialization of PSIk,PSI,thickness,radius,Length
Input the values of modules of elasticity, specific weight,

Poission‟s ratio, stress minimum, stress maximum
Do the coming loop after initializing

the value of thickness
Do the coming loop after initializing

PSIk
Do this loop after initializing

the value of radius
Calculate the volume per usable area

=(t*r*PSIk*L)/((r*r*PSIk)/2)
Do Calculation for the perspective

values
Using the provided formulas to calculate
the stress, deflection vertical and
deflection horizontal.
Does the deflection vertical and deflection horizontal satisfy Increase the
No value of PSI
serviceability limit state? Is the stress in the material limit?
yes
Multiply the value of deflection

vertical, deflection horizontal and
NO
stress if negative.
Increase
Is the value of PSI is maximum No PSI NO
Yes NO
Compare the value of deflection and stress and NO

write the maximum value. And the respective
PSIk, radius, thickness, Length, volume per usable
area.
compare the volume per usable
Save the first value in the areas and save the minimum
minimum volume per usable value on minimum volume per
NO Is the loop different from the first loop or YES
area and save the respective usable area.And the respective
more?
dimension, stress, deflection. dimension, stress, deflection.
Is the value of radius

maximum?
Yes
Increase value of PSIk
Is the value of PSIk

maximum?
YES
Increase the vlaue of

thickness
Is the value of thickness

maximum?
Increase the value of Length
Is the Length maximum?
Out put and write the minimum value and

respective parameters, stress and
deflections.
End
FIGURE 3.4 flow chart of the analytical method of design
46
3.3.2 DESIGNUSING C++ PROGRAM
There are different method of analysis which is used to find the stress and deflection of
cylindrical shell structure. This methods are load balancing method, Beam theory, arch analysis,
membrane theory and bending theory. From this methods both membrane analysis and bending
analysis are combined together to find accurate result of the actual behavior of the thin
cylindrical shell structure. The detail program is attached APPENDIX I.
3.3.2.1 EXAMPLE HOW THE PROGRAM WORKS.
TRIAL ONE
STEP 1: Taking the assumption of dimension from the lower bound of thin cylindrical shell
Length=20meters
Radius=12meters
Thickness=0.04meters
Half of the central angle (ϕk) =450=0.7854rad
Height of the cylinder from the ground level to the selling 3.52 meters
STEP 2: Material property
Polyester and Woven Rovings Laminate 45% E-glass
250Mpa
Factored tensile stress   100Mpa
2.5
47
150Mpa
Factored compressive stress   60Mpa
2.5
STEP 3: Load
PL=1.35DL+1.5LL for ultimate limit state design
PL=DL+LL for serviceable limit state design
Dead load=self -weight +water load
=   t   w  h0   w * r  r * cos( )
 15.696KN / m3 * 0.04m  9.81KN / m3 * 2m  9.81KN / m3 * (9m  9m * cos(450 ))
=46.107KN/m2
2.224
Live load   0.01573KN / m 2
9m * 20m * 0.7854rad
PL=1.35*46.107KN/m2+1.5*0.01573KN/m2
=62.268KN/m2
PL=46.107KN/m2+0.01573KN/m2
=46.123KN/m2
STEP 4: Analysis
Then the change in diameter and stress of the lower bound of design variables are calculated
using the following equations Eq.2.8.6.7, Eq.2.8.6.8, Eq.2.8.6.14, and Eq.2.8.6.15.
48
Deflection Deflection radius Radial Hoop Volume per

direction(m) Stress(KN/m2) Stress(KN/m2) usable area(m)
horizontal
Direction(m)
0.0002 0.00002 32.715 9814.38 0.183
TABLE 3.1 Result of Analysis

STEP 5: Check for stability
As the program starts to calculate the result taking in to consideration the design constraints the
structure is stable. Because the deflection is much less than serviceability limit state design
consideration. The radial stress and hoop stress are below the compressive and tensile strength of
the fiber glass.
2.636m
umax  H t / 200   0.013m  0.0002m
200
2 * r * sin( k ) 2 * 9 * sin(45 0 )
u rmax  B / 300    0.042m  0.00002m
300 300
Allowable Compressive stress=-60000KN/m2
Allowable Tensile stress=100000KN/m2 from Eq.2.8.6.7and Eq.2.8.6.8 shows both the hoop and
radial stress are compressive stress. Therefore greatest of the radial stress and hoop stress is hoop
stress.
 compressive hoop  9814.38KN / m 2  60,000KN / m 2  compressive stress
This just the first trial but the global efficient result using the function derived above can be
calculated as follows:
49
3.3.2.2 INPUT OF THE PROGRAM
The input of the program can be written on a note pad created in a folder named file. The note
pad should consist first modulus of elasticity after a space Poisson‟s ratio space factored
compressive strength space factored tensile strength of the material.Where the factor of safety
suggested by the Manuel for fiber glass is for flexural member is 2.5.
Polyester and Polyester and Polyester and E-glass S-2 glass

Woven Satin Weave Continuous (fiberglass (fiberglass
Rovings Cloth Rovings 4) 5)
Laminate Laminate Laminate
45% E-glass 55% E-glass 70% E-glass
(fiberglass 1) (fiberglass2) (fiberglass3)
Modules of 2.48*106 2.968*106 7.846*106 76*106 97*106

elasticity (E)
(KN/m2)
Gamma(Υ)(KN/m 15.696 16.677 18.639 25.506 24.426
3
)
Poisson's ratio 0.3 0.3 0.3 0.3 0.3
factored -6*104 -1*105 -1.4*105 -4.32*105 -6.4*105
compressive
strength(KN/m2)
Factored tensile 1*105 1.2*105 3.2*105 1.378*106 1.956*106
strength(KN/m2)
TABLE 3.2Input Values forthe Program
3.3.2.3 PROCESS OF THEPROGRAM

The program takes the input values as described in the flow chart the program has different loops
which will find the best result. The program uses Brute force method to do this. Brute force
method is trying every combinations of values and choosing the best suited answer for that
function. The loops take in to consideration the design variables, the design constraints and
volume per usable area. The following table shows the arrangement of the input file. The
program starts by reading this.
Material type Modulus of Specific Poisson’s Factored Factored tensile
50
elasticity weight ratio compressive strength

strength
fiberglass 1 2.48*106 15.696 0.3 -60000 100000
fiberglass 2 2.968*106 16.677 0.3 -100000 120000
fiberglass 3 7.846*106 18.639 0.3 -140000 320000
fiberglass 4 7.6*107 25.506 0.3 -432000 1378000
fiberglass 5 9.7*107 24.426 0.3 -640000 1956000
Table 3.3 Input Data for different types of fiber glass
The program all in all works by changing different design variable and choosing one value
between intervals with lower volume per usable area. Volume per usable area is a function
embedded in the program. The program will take the input value and after running different
iteration the final result will be written on a black page and after stroking any key on the
keyboard the result will be saved on a different note pad in the same folder the C++ program is
saved. If the program uses the first type of fiber glass then the output file name will be
fiberglass1.
The first program used for the analysis is the program written for thin shell structure. As
mentioned above the particular constraints for thin shell design is included in the program.
3.3.2.4 OUTPUT OF THE PROGRAM

To find the final result for the design of the shell different dimensions and their respective
structural responses should be checked. The dimension with minimal volume perusable area will
be chosen form the first iteration.
51
The output of the program for different combination using the boundaries as:
1. The length(L) ranges from 20m to 21m interval of iteration 0.1m
2. The radius(R) ranges from 12m to 13m interval of iteration 0.1m
3. The thickness(t) is in the interval of 0.04m to 0.05m interval of iteration 0.001
4. Angle of curvature(Ψk ) is between0.7854radian to 0.955radian interval of iteration
0.017radians
Based on the function volume perusable area the best result in these interval is:
Length(m) Thickness(m) PSIk(rad) Radius (m)
20 0.04 0.935 17
Table 3.4 Dimension of the Output of the Program
Material Deflection Deflection Radial Hoop Volume per

2 2
psi- radius Stress(KN/m ) stress(KN/m ) usable area(m)
type direction(m) direction(m)
fiberglass 1 0.0002 0.00002 51.489 21882.9 0.095
fiberglass 2 0.0002 0.000016 51.516 21894.2 0.095
fiberglass 3 0.00007 0.000006 51.5686 21916.7 0.095
fiberglass 4 0.000008 0.0000006 51.754 21995.5 0.095
fiberglass 5 0.000006 0.00000006 51.725 21983.1 0.095
Table 3.5 Output Result for Different Type of Fiber Glass
52
Best result of the analysis is fiberglass 5 but the cost of this fiber glass is 20dollar per kilogram.
The next best result is fiberglass 4 which costs 2 dollar per kilogram. Therefore fiberglass 4 is
the best design result.
If the designer didn‟t find a stable structure it can be revised. And changing the design of the thin
shell in to thick shell by changing the design variables. The first thing that was takenin to
consideration is the ratio of thickness to radius of the cylindrical shell. Look Eq.1.2 and
Eq.1.3.As described in the equations the ratio of thickness to radiusshould be increased. To do
this the thickness of the shell should be increased.The program for thick cylindrical shell
structure is attached APENDEX II.
53
CHAPTER 4ANALYSIS OF THE STRUCTURE USING FINITE

ELEMENT ANALYSIS
4.1 INTRODUCTION
Structural analysis program is one of finite element analysis software. This software is used for
different structural design and analysis. Structural analysis program is like any other software.
The output values depend on the input values. The input values are the final result of the
analytical design.
4.2 ANALYSIS PROCEDURE

1) Type of structure designed is thick shell
2) The design variables are:resulting dimension on table 3.3
3) Support condition
The support condition on the length direction is gone be fixed at both sides.
4) Defining material type
All the properties of E-glass is used except the weight per volume is zero. Because
factored combination of both live load and dead load are gone be used.so self-weight will
not be considered twice. The factored compressive stress and factored tensile stress are
defined using non-linear analysis. The structure is elastic so both the compressive strain
and tensile strain can be calculated using modulus of elasticity and respective stresses.
5) Defining section property
The section is defined as thick shell and both the membrane and bending thickness are
the same thickness which is the whole thickness of the shell.
6) Define load type
Dead load factored by 1
7) Define load combination
The factor for the dead load is 1 because the load is already factored. The load applied as
uniform shell load in the gravity direction.
54
8) Loading
Since the load varies for different values of PSI the average load of the edges of the row of
shell elements are taken. Loading table is found APPENDIX III.
4.3 ANALYSIS RESULT

The analysis result for
FIGURE 4.1 Maximum Stress (KN/m2)
55
FIGURE 4.2 Minimum Stress (KN/m2)
FIGURE 4.3 Deflection in the Z-Direction (m)
56
FIGURE 4.4 Deflection in the Y-Direction
Maximum values Finite element Maximum allowable deflection and

analysis result stress
Maximum deflection y-direction(mm) 8.4 13
Maximum deflection z-direction(mm) -9 42
Maximum compressive stress(KN/m2) -47,966.364 -60,000
Maximum tensile stress(KN/m2) 28,911.23 100,000
Table 4.1 Maximum Values of the Finite Element Analysis
57
CHAPTER 5 COMPARISON OF ANALYTICAL DESIGN AND

FINITE ELEMENT ANALYSIS
5.1 INTRODUCTION
Both analytical method of design using different theories of design and finite element analysis
using structural analysis program can be used in design and analysis. The particular dimension
and the material type is taken form the analytical design result. Second the loading is
approximated by using average method. For each strip in the X axis average of the highest and
the lowest of the segment is taken. The support condition is the same for both the analytical
design and the finite element analysis
5.2 COMPARISON
Maximum values Analytical results Finite element analysis results
Maximum deflection y-direction(mm) 0.008 8.4
Maximum deflection z-direction(mm) 0.0002 -9
Maximum compressive stress(KN/m2) -21,995.5 -47,966.364
Maximum tensile stress(KN/m2) - 28,911.23
TABLE 5.1 Comparison of the Result of Finite Element Analysis and Analytical Design
5.3 RESULT AND DISCUSSION

The values of the structural analysis program and the numerical formula were not the same
because there were different assumptions when deriving the formulas. When deriving
longitudinal stress the cylindrical shell is considered as closed at both ends. Which brings the
58
difference in the horizontal displacement. In case of vertical deflection the equation used
assumed the shell as a beam with both ends fixed in the longitudinal direction. The more accurate
result of deflection response is the result of the finite element analysis method.
Both hoop stress and radial stress are negative values which means compressive stress. Since the
structure is under uniform load in the gravity direction the equation used shows only the
compressive stress of the shell. Compressive stress comparison the analytical result is almost half
of the finite element analysis result. But while designing the finite element analysis the loading
was taken average as shown in APPENDIX III. Due to this different assumptions the results
differ.
It is recommended to take the finite element analysis result because it takes in to consideration
the behavior of the structure both in the hoop direction and in the longitudinal direction.
Even though the result of the finite element analysis method is greater than the analytical result,
both the ultimate limit state and serviceability limit state stability criteria are satisfied. The
structure is stable.
59
CHAPTER 6CONCLUSION AND RECOMMENDATION

6.1CONCLUSION
The analysis and the steps taken to get the stable design of underwater cylindrical shell structure
can be summarized as follows. The first step is to choose the type of cylindrical shell structure
needed for the specific purpose. The first choice for design was thin cylindrical shell structure.
The reason for this is because its cost is less than thick shell.
Using this information formulating the analytical formula for the design of this shell structure is
searched and simplified for use. Since there are many combinationsMicrosoft Excel for manual
calculation takes longer time.so a program was developed to design by taking different intervals
between the constraints. Even though developing a program which is able to calculate the values
of the functions was great. It was also complex and took a lot of time to consider different
combination in between the constraints. The program failed repeatedly one because of the
complexity of the formula, second because of the arrangement of the loops. After extensive
formulation and thinking how the program can be developed the program started working.
Although, the program started working there was no result for many trials. Which was tedious
and repetitive.
In order to get an expression for longitudinal stress the shell is considered closed at both ends.
Since this is assumed in the formula for the longitudinal stress the value of the deflection in the
horizontal direction is not exact. While deflection in the radial direction is not the same because
the shell was assumed as a hollow beam fixed in both direction. Even though determination of
the response of the structure is different, the results shows that the structure is stable.
While trying to find the dimension of the structure efficient for aquarium the results showed the
behavior of cylindrical shell structure. Which is whenvolume per usable area decreases, the
internal stress increases, the cost decreases and deflection increases.
60
6.2 RECOMMENDATION
Every research needs further study and recommendation on how to do that for this paper is:
 To develop a more easier C++ program

 To do chart for design of different volume per usable area and their stresses, deflection
graph. For different range of dimensions.
 To prepare charts even for both thin and thick shell
 To study further more advanced materials in designing this kind of underwater structure
 To design the structure using structural analysis program with a more exact model
61
REFERENCE
1. Srinivasan Chandrasekaran ,S.K.Gupta,Federico Carannante, Design aids for fixed
support reinforced concrete cylindrical shells under uniformly distributed loads,
International Journal of engineering science and technology ,Multicraft limited ,2009
2. Wain-Lai Kuo, Analysis and design p/c cylindrical shells, A masters report, Kansas state
university,1968
3. C. R. Calladine, Theory of shell structures, Cambridge university press, digitally printed
version,2007
4. Eduard Ventsel and Theodor Krauthammer, Thin plates and shells, The Pennsylvania
state university ,Marcel Dekker,Inc,2001
5. Ralph E.Byrne,Jr.,The Analysis of stress in a thin cylindrical shell of circular cross-
section, California institute of thecnology,1940
6. American concrete institute committee, Building Code Requirements for Structural
Concrete and Commentary, 2008.
7. Shifferaw Taye, Lecture Notes on shell structures, (unpublished) School of Civil and
Environmental Engineering, AAIT, AAU 2013/14
8. David B. Farnsworth JR, Behavior of Shell structures, M.sc. Thesis in structural
engineering, Masschusettus institute of technology, 1999
9. Structural analysis program SAP2000, Integrated Finite Element analysis and design of
structure, computer and structures, Inc, Berkeley, California, USA, 2000.
10. Euro CodeBasis of structural design, Europe, 29 November 2001.
11. S. Arnout, G. Degrande,G. Lombaert and G. De Roeck, An optimal barrel vault design in
the conceptual design stage,2ndInternational Conference on Engineering
Optimization,September 6-9, 2010, Lisbon, Portugal
12. Eng.Mohammed M.Jahjouh , Analysis of cylindrical shells, The Islamic university of
Gaza, Faculty of post graduate studies\civil engineering,2010
13. Rodney, structural fiber glass Design Manuel,Rev.3,January 2004,modified January 2007
14. John Blaauwendraad and Jeroen H.Hoefakker, Structural Shell analysis understanding
and application, University of waterloo, Department of civil Engineering, 2014.
62
15. Myriam Iturgiaz Elso,finite element studies on the stability behavior of cylindrical shells
under axial and radial uniform and non-uniform loads,Hochschlue Niederrhein university
of applied science,department of mechanical and process engineering, May 2012
16. Loewenstein,K.L,The manufacture of technology of continuous Glass
Fiber,Newyork,1993
17. Sachindranarayan Bhaduri,Ph.D, pressure cylinders, university of Texas, Mechanical and
industrial engineering department, The McGraw-Hill companies,2004
63
Appendix I: program code for thin shell

#include <iostream>
#include <cstdlib>
#include <math.h>
#include <conio.h>
#include <fstream>
using namespace std;
void loadFromFile(double& E,double& gamma,double& poisson,double& compressivestrength,

double& tensilestrength);
int main() {
//Initialization of i which helps in keeping the out put in array,radius PSIk,thickness,Length
//declaring variables
int i = 0,k=0,m=0;
double r = 12, PSIk = 0.7854 , t =0.04 , L=20 ;
double
p,E,gamma,poisson,compressivestrength,tensilestrength,ro,ri,PSI,volumeperusablearea,Uvmax,U
hmax,Nhmaxfinalr,Nhmaxfinalt,Lminm,tminm,PSIKminm,rminm,UVminm,Uhminm,Nhminmr,
Nhminmt,Volumeperusableareaminm;
double Uv[12];
double Uh[12];
double Nhr[12];
64
double Nht[12];
double Nhl[12];
double er[12];
//Taking the value of modules of elasticity(E),poisson's ratio,stress maximum and stress

minimum of the material
loadFromFile(E,gamma,poisson,compressivestrength,tensilestrength);
ofstream write("optimization result of fiberglass 1out.txt",ios::out);
// the outer loop is the Loop where the length increases since when the loop inters the outer loop
and increase the Length
//the thickness has reached maximum so it should be initialized
cout <<"\t" <<"No"<< "\t" <<"Length"<< " "<<"Thickness"<< " "<<"PSIk"<< "
"<<"Radius" <<" "<<"Uvmax"<<" "<<"Uhmax"<<" "<<"Nhmaxr"<<"
"<<"Nhmaxt"<<" "<<"Volumeper usable area" <<endl;
write <<"\t" <<"No"<< "\t" <<"Length"<< "\t\t\t"<<"Thickness"<< "\t"<<"PSIk"<<

"\t\t"<<"Radius"
<<"\t\t"<<"Uvmax"<<"\t\t"<<"Uhmax"<<"\t\t"<<"Nhmaxr"<<"\t\t"<<"Nhmaxt"<<"\t\t"<<"Volu
me per usable area"<<"\t\t"<<endl;
do {
t=0.04;
//the this loop is where the thickness increases since when the loop inters this loop and increase
the thickness
//the PSIk has reached maximum so it should be initialized
65
do {
PSIk=0.7854;
//the this loop is where the PSIk increases since when the loop inters this loop and increase the
PSIk
//the radius has reached maximum so it should be initialized
do {
r=12;
//if the loop didn't have a result the output no length,thickness,PSIk,radius

Uhmax,Uvmax,Nhmax
do {
//assignment operation for volume per usable area
volumeperusablearea= (t*r*PSIk*L)/(((r*r*PSIk))-(r*r*sin(PSIk)*cos(PSIk)));
//This loop initializes PSI and increases it from zero to twice PSIk to know the value stress and
deflection in each interval and take the maximum.
PSI=0,k=0,Uhmax=0,Uvmax=0,Nhmaxfinalr=0,Nhmaxfinalt=0;
do{
66
//assignment of operation for the variables necessary to calculate the stress and deflection
where p is the combination of dead load and live load.
//Uv is vertical deflection,Nhr is radial stress,Nht is tangential stress
//use the following load combination to check if the structure is stable for ultimate limit state
design criteria
//p=(1.35*(2*9.810+9.810*(r-r*cos(PSI))+gamma*t)+1.5*(2.224/(r*L*PSIk)));
//activate the following load combination while check if the structure is stable for
serviceability limit state design criteria
p=(1*(2*9.810+9.810*(r-r*cos(PSI))+gamma*t)+1*(2.224/(r*L*PSIk)));
Nhr[k]=(-p/2);
Nht[k]=((-p*r)/(2*t));
Nhl[k]=((-p*r)/(4*t));
er[k]=(1/E)*(Nhr[k]-poisson*Nht[k]-poisson*Nhl[k]);
Uv[k]=((p*PSIk*(L*L*L*L))/(384*E*r*r*r*3.14*t));
Uh[k]=((4*(2*r*sin(PSIk))/(4*t*E))*((2-poisson)/2));
//checking whether the deflection and the stresses are in their maximum and minimum interval
if ( (((r*(-1+cos(PSIk))/(200))<=Uh[k]) && (Uh[k]<= (r*(1-cos(PSIk))/(200)))) && ((((-

2*r*sin(PSIk))/300)<= Uv[k]) && (Uv[k]<=((2*r*sin(PSIk))/300)) ) &&
((compressivestrength<=Nhr[k]) && (Nhr[k]<=tensilestrength)) &&
((compressivestrength<=Nht[k]) && (Nht[k]<=tensilestrength)))
// if the values are negative absolute value is needed to assist the comparison.
67
if(Uh[k]<0)
Uh[k]=Uh[k]*-1;
if(Uv[k]<0)
Uv[k]=Uv[k]*-1;
if(Nhr[k]<0)
Nhr[k]=Nhr[k]*-1;
if(Nht[k]<0)
Nht[k]=Nht[k]*-1;
68
k++;
PSI=PSI+(PSIk)/10;
}while(PSI<=PSIk);
//since Uvmax ,Nhmaxfinalr and Nhmaxfinalt has a random number before comparing anything
the initial values are assigned to it and comparison starts.
if(k>0)
Uhmax=Uh[0];
Uvmax=Uv[0];
Nhmaxfinalr=Nhr[0];
Nhmaxfinalt=Nht[0];
for(int l=1;l<k;l++)
if(Uhmax<Uh[l])
69
Uhmax=Uh[l];
if(Uvmax<Uv[l])
Uvmax=Uv[l];
if(Nhmaxfinalr<Nhr[l])
Nhmaxfinalr=Nhr[l];
if(Nhmaxfinalt<Nht[l])
Nhmaxfinalt=Nht[l];
70
if (i=0)
Volumeperusableareaminm=volumeperusablearea;
Lminm=L;
tminm=t;
PSIKminm=PSIk;
rminm=r;
Uhminm=Uhmax;
UVminm=Uvmax;
Nhminmr=Nhmaxfinalr;
Nhminmt=Nhmaxfinalt;
else
if(Volumeperusableareaminm > volumeperusablearea)
71
Lminm=L;
tminm=t;
PSIKminm=PSIk;
rminm=r;
Uhminm=Uhmax;
UVminm=Uvmax;
// the output is saved and written in the note pad.
i++;
// if the designer wants to see all the combination of different dimension and their
respective response the following output command can be activated
//cout <<"\t" << i << "\t" <<L << " "<<t<< " "<<PSIk<< " "<<r <<"
"<<Uvmax<<" "<<Uhmax<<" "<<Nhmaxfinalr<<" "<<Nhmaxfinalt<<"
"<<volumeperusablearea << endl;
//write <<"\t" << i << "\t" <<L << "\t\t\t"<<t<< "\t\t\t"<<PSIk<< "\t\t\t"<<r
<<"\t\t\t"<<Uvmax<<"\t\t\t"<<Uhmax<<"\t\t\t"<<Nhmaxfinalr<<"\t\t\t"<<Nhmaxfinalt<<"\t\t\t"
<<volumeperusablearea << endl;
72
r+=0.5;
}while(r<=17);
PSIk+=0.017;
}while(PSIk<=0.955);
t+=0.004;
}while(t<=0.08);
L+=0.5;
}while(L<=25);
// if the designer wants the final best result the following out put command can be used
cout <<"\t" << i << "\t" <<Lminm << " "<<tminm<< " "<<PSIKminm<< " "<<rminm
<<" "<<UVminm<<" "<<Uhminm<<" "<<Nhminmr<<" "<<Nhminmt<<"
"<<Volumeperusableareaminm<<endl;
write <<"\t" << i << "\t" <<Lminm << "\t\t"<<tminm<< "\t\t"<<PSIKminm<< "\t\t"<<rminm
<<"\t\t"<<UVminm<<"\t\t\t"<<Uhminm<<"\t\t"<<Nhminmr<<"\t\t"<<Nhminmt<<"\t\t"<<Volu
meperusableareaminm<<endl;
73
void loadFromFile(double& E, double& gamma,double& poisson,double&

compressivestrength,double& tensilestrength) {
ifstream fin("file/fiberglass 1in.txt");
if(fin.fail()) {
system("cls");
cout << endl << "ERROR OPENING FILE !!!";
getch();
exit(1);
} else {
while(!fin.eof()) {
//coping from the file to the variables
fin >> E;
fin >> gamma;
fin >> poisson;
fin >> compressivestrength;
fin >> tensilestrength;
74
Appendix II: program code for thick shell

#include <iostream>
#include <cstdlib>
#include <math.h>
#include <conio.h>
#include <fstream>
using namespace std;
void loadFromFile(double& E,double& gamma,double& poisson,double& compressivestrength,

double& tensilestrength);
int main() {
//Initialization of i which helps in keeping the out put in array,radius PSIk,thickness,Length
//declaring variables
int i = 0,k=0,m=0;
double r = 19, PSIk = 0.524 , t =0.38 , L=20 ;
double
ro,ri,p,E,gamma,poisson,compressivestrength,tensilestrength,PSI,Uhmax,volumeperusablearea,U
vmax,Nhmaxfinalr,Nhmaxfinalt,Lminm,tminm,PSIKminm,rminm,rominm,riminm,Uhminm,UV
minm,Nhminmr,Nhminmt,Volumeperusableareaminm;
double Uh[12];
double Uv[12];
double Qh[12];
double Nhr[12];
75
double Nht[12];
double eh[12];
//Taking the value of modules of elasticity(E),poisson's ratio,stress maximum and stress

minimum of the material
loadFromFile(E,gamma,poisson,compressivestrength,tensilestrength);
ofstream write("optimization result of fiberglass 1out.txt",ios::out);
// the outer loop is the Loop where the length increases since when the loop inters the outer loop
and increase the Length
//the thickness has reached maximum so it should be initialized
cout <<"\t" <<"No"<< "\t" <<"Length"<< " "<<"Thickness"<< " "<<"PSIk"<< "
"<<"Radius" << " "<<"Uhmax" <<" "<<"Uvmax"<<" "<<"Nhmaxr"<<"
"<<"Nhmaxt"<<" "<<"Volumeper usable area" <<endl;
write <<"\t" <<"No"<< "\t" <<"Length"<< "\t\t\t"<<"Thickness"<< "\t"<<"PSIk"<<

"\t\t"<<"Radius" << "\t\t"<<"Uhmax"
<<"\t\t"<<"Uvmax"<<"\t\t"<<"Nhmaxr"<<"\t\t"<<"Nhmaxt"<<"\t\t"<<"Volume per usable
area"<<endl;
do {
t=0.38;
//the this loop is where the thickness increases since when the loop inters this loop and increase
the thickness
//the PSIk has reached maximum so it should be initialized
do {
76
PSIk=0.524;
//the this loop is where the PSIk increases since when the loop inters this loop and increase the
PSIk
//the radius has reached maximum so it should be initialized
do {
r=19;
//if the loop didn't have a result the out put no length,thickness,PSIk,radius
Uhmax,Uvmax,Nhmax
do {
//assignment operation for volume per usable area
volumeperusablearea= (t*r*PSIk*L)/(((r*r*PSIk))-(r*r*sin(PSIk)*cos(PSIk)));
//This loop initializes PSI and increases it from zero to twice PSIk to know the value stress and
deflection in each interval and take the maximum.
PSI=0,k=0, Uhmax=0,Uvmax=0,Nhmaxfinalr=0,Nhmaxfinalt=0;
do{
//assignment of operation for the variables necessary to calculate the stress and deflection
where p is the combination of dead load and live load.
77
//Uh is stress horizontal,Uv is stress vertical,Nhr is radial stress,Nht is tangential stress.
ro=r + ( t / 2 );
ri=r - ( t / 2 );
//use the following load combination to check if the structure is stable for ultimate limit
state design criteria
p=(1.35*(2*9.810+9.810*(r-r*cos(PSI))+gamma*t)+1.5*(2.224/(r*L*PSIk)));
//activate the following load combination while check if the structure is stable for
serviceability limit state design criteria
//p=(1*(2*9.810+9.810*(r-r*cos(PSI))+gamma*t)+1*(2.224/(r*L*PSIk)));
Nhr[k]=-p;
Nht[k]=((-2*p*ro*ro)/(ro*ro-ri*ri));
Qh[k]=(((-p)*ri*ri*ro*ro)/((ro*ro-ri*ri)*r*r))/(PSIk*r);
eh[k]=((1/E)*(Nht[k]-(poisson*Nhr[k])-(poisson*Qh[k])));
Uh[k]=eh[k]*PSI*r;
Uv[k]=(-((p*ro*ro*r)/(E*(ro*ro-ri*ri))))*((1-poisson)+((1+poisson)*((ri*ri)/(r*r))));
//checking whether the deflection and the stresses are in their maximum and minimum interval
if ( ( ((r*(-1+cos(PSIk))/(200))<=Uh[k]) && (Uh[k]<= (r*(1-cos(PSIk))/(200)))) && ((((-

2*r*sin(PSIk))/300)<= Uv[k]) && (Uv[k]<=((2*r*sin(PSIk))/300)) ) &&
((compressivestrength<=Nhr[k]) && (Nhr[k]<=tensilestrength)) &&
((compressivestrength<=Nht[k]) && (Nht[k]<=tensilestrength)))
// if the values are negative absolute value is needed to assist the comparison.
78
if(Uh[k]<0)
Uh[k]=Uh[k]*-1;
if(Uv[k]<0)
Uv[k]=Uv[k]*-1;
if(Nhr[k]<0)
Nhr[k]=Nhr[k]*-1;
if(Nht[k]<0)
Nht[k]=Nht[k]*-1;
79
k++;
PSI=PSI+(PSIk)/10;
}while(PSI<=PSIk);
//since Uhmax ,Uvmax Nhmaxfinalr and Nhmaxfinalt has a random value before comparing any
thing the initial values are assigned to it and comparison starts.
if(k>0)
Uhmax=Uh[0];
Uvmax=Uv[0];
Nhmaxfinalr=Nhr[0];
Nhmaxfinalt=Nht[0];
for(int l=1;l<k;l++)
if(Uhmax<Uh[l])
80
Uhmax=Uh[l];
if(Uvmax<Uv[l])
Uvmax=Uv[l];
if(Nhmaxfinalr<Nhr[l])
Nhmaxfinalr=Nhr[l];
if(Nhmaxfinalt<Nht[l])
81
Nhmaxfinalt=Nht[l];
if (i=0)
Lminm=L;
tminm=t;
PSIKminm=PSIk;
rminm=r;
rominm=rminm+(tminm/2);
riminm=rminm-(tminm/2);
Uhminm=Uhmax;
UVminm=Uvmax;
82
else
if(Volumeperusableareaminm > volumeperusablearea)
Lminm=L;
tminm=t;
PSIKminm=PSIk;
rminm=r;
rominm=rminm+(tminm/2);
riminm=rminm-(tminm/2);
Uhminm=Uhmax;
UVminm=Uvmax;
83
// the output is saved and written in the note pad.
i++;
// if the designer wants to see all the combination of different dimension and their
respective response the following out put command can be activated
//cout <<"\t" << i << "\t" <<L << " "<<t<< " "<<PSIk<< " "<<r << "
"<<Uhmax <<" "<<Uvmax<<" "<<Nhmaxfinalr<<" "<<Nhmaxfinalt<<"
"<<volumeperusablearea << endl;
//write <<"\t" << i << "\t" <<L << "\t\t\t"<<t<< "\t\t\t"<<PSIk<< "\t\t\t"<<r <<
"\t\t\t"<<Uhmax
<<"\t\t\t"<<Uvmax<<"\t\t\t"<<Nhmaxfinalr<<"\t\t\t"<<Nhmaxfinalt<<"\t\t\t"<<volumeperusable
area << endl;
r+=0.5;
}while(r<=24);
PSIk+=0.02618;
}while(PSIk<=0.786);
t+=0.0044;
84
}while(t<=0.424);
L+=1;
}while(L<=30);
// if the designer wants the final best result the following out put command can be used
cout <<"\t" << i << "\t" <<Lminm << " "<<tminm<< " "<<PSIKminm<< " "<<rminm
<< " "<<Uhminm <<" "<<UVminm<<" "<<Nhminmr<<" "<<Nhminmt<<"
"<<Volumeperusableareaminm<<" "<<endl;
write <<"\t" << i << "\t" <<Lminm << "\t\t"<<tminm<< "\t\t"<<PSIKminm<< "\t\t"<<rminm
<< "\t\t"<<Uhminm
<<"\t\t"<<UVminm<<"\t\t"<<Nhminmr<<"\t\t"<<Nhminmt<<"\t\t"<<Volumeperusableareamin
m <<"\t\t"<<endl;
void loadFromFile(double& E, double& gamma,double& poisson,double&

compressivestrength,double& tensilestrength) {
ifstream fin("file/fiberglass 1in.txt");
if(fin.fail()) {
system("cls");
cout << endl << "ERROR OPENING FILE !!!";
getch();
exit(1);
} else {
while(!fin.eof()) {
//coping from the file to the variables
85
fin >> E;
fin >> gamma;
fin >> poisson;
fin >> compressivestrength;
fin >> tensilestrength;
86
Appendix III: Loading for the analysis of thin shell using Structural analysis
program
Ultimate limit state load combination
Gamma(K poisso
E(KN/m^2) N/m^3) n
76000000 25.506 0.3
No. gamm
of Υ water*(r- a*t
divisi PSI r 2*Υ water r*cos(PSI)) (KN/m 1.35*DL 1.5*LL Pd load avg
on (rad) (m) t(m) (KN/m2) (KN/m2) 2
) (KN/m2) (KN/m2) (KN/m2) (KN/m2)
17 0.04 19.620 0 1.020 27.864 0.006 27.871 28.362
1 0
0.094 17 0.04 19.620 0.728 1.020 28.848 0.006 28.854 30.325
2
0.187 17 0.04 19.620 2.907 1.020 31.789 0.006 31.796 34.233
3
0.281 17 0.04 19.620 6.518 1.020 36.663 0.006 36.670 40.052
4
0.374 17 0.04 19.620 11.528 1.020 43.427 0.006 43.434 47.731
5
0.468 17 0.04 19.620 17.895 1.020 52.022 0.006 52.029 57.204
6
0.561 17 0.04 19.620 25.562 1.020 62.373 0.006 62.379 68.387
7
0.655 17 0.04 19.620 34.463 1.020 74.389 0.006 74.395 81.183
8
0.748 17 0.04 19.620 44.519 1.020 87.965 0.006 87.971 95.480
9
0.842 17 0.04 19.620 55.644 1.020 102.983 0.006 102.989 111.154
10
0.935 17 0.04 19.620 67.739 1.020 119.312 0.006 119.318
11
87
Serviceability limit state load combination
Gamma poisso
E(KN/m2) (KN/m3) n
76000000 25.506 0.3
No. Υ water
of *(r- gamm DL
divisi PSI r 2*Υ water r*cos(PSI)) a*t(KN (KN/m2 LL Pd load avg
on (rad) (m) t(m) (KN/m2) (KN/m2) /m2) ) (KN/m2) (KN/m2) (KN/m^2)
1 0 17 0.04 19.620 0 1.020 20.640 0.004 20.644 21.009
2 0.094 17 0.04 19.620 0.728 1.020 21.369 0.004 21.373 22.462
3 0.187 17 0.04 19.620 2.907 1.020 23.548 0.004 23.552 25.357
4 0.281 17 0.04 19.620 6.518 1.020 27.158 0.004 27.162 29.667
5 0.374 17 0.04 19.620 11.528 1.020 32.168 0.004 32.173 35.356
6 0.468 17 0.04 19.620 17.895 1.020 38.535 0.004 38.539 42.373
7 0.561 17 0.04 19.620 25.562 1.020 46.202 0.004 46.206 50.657
8 0.655 17 0.04 19.620 34.463 1.020 55.103 0.004 55.107 60.135
9 0.748 17 0.04 19.620 44.519 1.020 65.159 0.004 65.164 70.726
10 0.842 17 0.04 19.620 55.644 1.020 76.284 0.004 76.288 82.336
11 0.935 17 0.04 19.620 67.739 1.020 88.379 0.004 88.383
88

Finot Mezgebu

Uploaded by

Copyright:

Available Formats

Finot Mezgebu

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

Finot Mezgebu

Uploaded by

Copyright:

Available Formats

ADDIS ABABA INSTITUTE OF TECHNOLOGY

SCHOOL OF GRADUATE STUDIES

COMPUTERIZED ANALYSIS AND DESIGN OF UNDERWATER

A THESIS SUBMITTED TO THE GRADUATE SCHOOL OF THE ADDIS ABABA

SCHOOL OF GRADUATE STUDIES

DEPARTMENT OF CIVIL ENGINEERING

COMPUTERIED ANALYSIS AND DESIGN OF UNDERWATER CYLINDRICAL SHELL

Finot Mezgebu Wubete

Approved by Board of Examiners

Dr. -Ing Bedilu Habte _____________ ____________

Advisor Signature Date

Dr. Asnake Adamu _____________ ____________

External Examiner Signature Date

Dr. Shifferaw Taye _____________ ____________

Internal Examiner Signature Date

Chair man Signature Date

Name: Finot Mezgebu

Place: Addis Ababa University

1.1 General background of the problem……………………………………………………….1

Chapter 2 Litrature review

2.1 Shell Structures Fundamentals……………………………………………………………..4

2.2 History of design and construction of shells…………………………………………….....4

2.3 History of analysis methods………………………………………………………………..5

2.4 Classification of shells……………………………………………………………………..5

2.4.1 Classification based on thickness…………………………………………………..5

2.5 Advantage and disadvantage of shell structures……….………………………………..…7

2.6 Cylindrical shell…………………………………………………………………………….8

2.6.1 Classification of cylindrical shell……………………………………………………..8

2.6.1.1 Classification based on radius and longitudinal length…………………………9

2.6.1.2 Classification based on the magnitude of edge disturbance…………………….9

2.6.1.3 Classification based on geometric dimension and thickness……………………9

2.7 Fiber glass……………………………...…………………………………………………....10

2.7.2 Types of fiber glass…………………...…………………………………………….....11

2.8 Design of shell structure ………………………………………………………………….....13

2.8.2 Load combination……………………………………………………………………..15

2.8.3 Safety Factor…………………………………………………………………………..16

2.8.4 Design consideration…………………………………………………………………...16

2.8.5 Method of analysis ……………………………………………………………………18

2.8.5.1 Loading balancing method……………………………………..……………….18

2.8.5.2 Beam theory…………………………………………………………………….19

2.8.5.3 The Arch analysis……………………………………………………………….21

2.8.5.4 Membrane Theory………………………………………………………………21

2.8.6 Analysis procedure……………………………………………………………………...28

2.8.6.1 Thin Cylindrical Shell Design…………………………………………………..31

2.8.6.2 Thick Cylindrical Shell Design………………………………………………....34

Chapter 3 Design of the structure using analytical method

3.3.1 Flow chart of the program………………………………………………………………44

3.3.2 Design using C++ program……………………..………………………………………46

3.3.2.1 Input of the program……..………………….……………………………………49

3.3.2.2 Process of the program..…………………….……………………………………49

3.3.2.3 Output of the program…………………….……………………………………...50

Chapter 4 Analysis of the structure using finite element analysis

4.2 Analysis procedure…………………………………………………………………………….53

4.3 Analysis result…………………………………………………………………………………54

Chapter 5 comparison of analytical design and finite element analysis

5.3 Result and discussion…………………………………………………………………………..57

Chapter 6 Conclusion and Recommendation

Appendix I-Program code for thin shell…….…………………………………………………….63

Appendix II-Program code for thick shell….……………………………………………………..74

Figure 2.2 Shear, bending moment and twisting moment…………………………………..…...22

Dr. -Ing Bedilu Habte _

Dr. Asnake Adamu _

Dr. Shifferaw Taye _