Shell All Chaps Annotated-1 PDF

CENG 7503 - Analysis and Design
of Shell Structures
Course Instructor: Dr.-techn. Medhanye B.
STRUCTURAL ENGINEERING CHAIR

ETHIOPIAN INSTITUTE OF TECHNOLOGY
MEKELLE UNIVERSITY
SHELL STRUCTURES
Natural shell forms
Natural shell forms
Natural shell forms
Natural shell forms
Eduardo Torroja
• Born August 27, 1899

• Spanish engineer who
was considered a pioneer
in concrete-shell
structures
• Believed identity of form
and structure could be
realized with thin vaults of
concrete
• Graduated in 1923
• Died June 15, 1961
Pierre Lardy
• Born in 1903
• Attended Federal Institute of
Technology in 1923
• Hired Heinz Isler in 1951
• Proponent for model based
designs
• Emphasized the importance of
aesthetics for full-scale
structures
• Died in 1958
Heinz Isler
• Born July 26, 1926 in Zollikon, Switzerland
• Attended Federal Technical Institute
• Graduated in 1950 with a degree in Civil
Engineering
• Only student out of 100+ to choose thin-
shells for senior design
• Worked as an assistant to Pierre Lardy
• Based most of his structures on smaller
models
• Stated there were 3 ways to arrive at shell
shapes, (Freely shaped hills, Rubber Balloon
Membranes, and Reversed Hanging Cloths
Hanging membranes by Heinz Isler
Thin-Shell Roof over gas station in Bern-Zurich
Highway
Indoor Tennis Center in Heimberg, Switzerland
Introduction
•Shells are mostly thin-walled curved
surface structures in which the thickness is
rather small compared to the lateral
dimensions and radii of curvature.
•Shells constitute among the most common

and most efficient structural elements in
nature and technology. They are used
whenever high strength, large spans and
minimum materials are required.
• Shells are classified as membrane, thin and
thick shells depending on the ratio of their
thickness to respective radius of curvature as
shown below.
Membrane Thin shell Thick shell
≈1/1000 < Thickness < ≈1/20
Radius of curvature
• Shell structures have found wide applications in
practice. These include, among others, roofs of various
type, water tanks and reservoirs as well as liquid
containers and storage facilities, boilers, pressure
vessels, bodies of transportation structures including
aircrafts, space crafts, motor vehicles and ships, egg-
shaped sewage digesters, cooling towers at thermal
power stations.
• Thin shells constitute a class of structures where
form, not mass, plays a significant role in
resisting external actions as well as self-weight
of the system. Shells of double curvature are
among the most efficient of known structural
forms.
SHELL ACTION IN RELATION TO BEAM, ARCH,
AND PLATE ACTIONS
• The property that makes both arches and shells
capable of resisting external transverse loadings
by “extensional” or “membrane” action, as
opposed to “flexural” or “bending” action in
beams and plates, is curvature.
• Extensional or membrane actions are any in–

plane direct forces, whether tensile or
compressive. Structural materials are generally
more efficient when resisting loads in the
extensional (membrane) mode rather than the
flexural mode.
• In beams and plates, transverse external loads
do not give rise to any appreciable extensional
action within these elements, provided the
displacements after deformation remain
relatively small in relation to the transverse
dimension of the elements.
• Arches and shells can resist distributed

transverse applied loads almost wholly by
extensional action. This can take place if the
loading pattern is of a certain specific type and
suitable boundary conditions are provided
• In practice, a mixture of extensional and flexural
actions usually occurs in arches and shells,
since the above ideal conditions are not usually
satisfied; for these structural elements, good
design consists of minimizing the flexural
actions.
• For shells, the relative proportions of extensional

and flexural effects at a given point depend on
several factors such as the type of shell surface,
support conditions, loading configuration, and
the proximity of edges and certain
discontinuities.
TYPES OF SHELL STRUCTURES
Shells can be classified based on different criteria:
Geometric form
Shell curvature
Geometric developability
Classification of shells based on geometric form
(a) Surfaces of revolution: Surfaces of revolution are generated
by rotating a plane curve, called the meridian, about an axis
that is not necessarily intersecting the meridian. Circular
cylinders, cones, spherical or elliptical domes, hyperboloids of
revolution, and toroids are some examples.
(b) Surfaces of translation: A surface of translation is defined as
the surface generated by keeping a plane curve parallel to its
initial plane as we move it along another plane curve. The two
planes containing the two curves are at right angles to each
other.
(c) Ruled surfaces: Ruled surfaces are obtained by the
translation of straight lines over two end curves. The straight
lines are not necessarily at right angles to the planes
containing the end curves.
Classification of shells based on shell curvature
(a) Singly curved shells: These shells have a zero Gaussian
curvature. Some shells of revolution (circular cylinders,
cones), shells of translation, or ruled surfaces (circular or
noncircular cylinders and cones) are examples of singly
curved shells.
(b) Doubly curved shells
With Positive Gaussian curvature: Some shells of revolution
(circular domes, ellipsoids and paraboloids of revolution)
and shells of translation and ruled surfaces (elliptic
paraboloids, paraboloids of revolution) can be assigned to
this category of surfaces.
With negative Gaussian curvature: This category of surfaces
consists of some shells of revolution (hyperboloids of
revolution of one sheet) and shells of translation or ruled
surfaces (paraboloids, conoids, hyperboloids of revolution
of one sheet).
Classification of shells based on geometrical developability:
(a) Developable surfaces
Developable surfaces are defined as surfaces that can be
‘‘developed’’ into a plane form without cutting and/or
stretching their middle surface. All singly curved surfaces are
examples of developable surfaces.
(b) Non-developable surfaces

A non-developable surface is a surface that has to be cut
and/or stretched in order to be developed into a planar form.
Surfaces with double curvature are usually nondevelopable.
CENG 7503 - Analysis and Design
of Shell Structures
Course Instructor: Dr.-techn. Medhanye B.
STRUCTURAL ENGINEERING CHAIR

ETHIOPIAN INSTITUTE OF TECHNOLOGY
MEKELLE UNIVERSITY
COURSE OBJECTIVES
Conferring an in-depth knowledge and

understanding of the analysis methods and
design rules of shell structures with a focus on
shells of revolution.
COURSE CONTENTS:
1. In-plane and out-of-plane effects in shells:
Membrane and Bending theories
2. Bending theory of cylindrical shells
3. Membrane theory of shells of revolution
volution
Axis of revo
4. Bending theory of shells of revolution under
axisymmetric loads
volution
Axis of revo
5. Membrane theory for shells of general shapes
6. Analysis of multi-shell structures for bending
and membrane stresses
7. Stability problems in shell structures
TEACHING & LEARNING METHOD
(Delivery of skills)
Through lecture (Video conferencing) with
integrated practical exercises
Group home page:

http://groups.yahoo.com/group/mushell
Group email address:
mushell@yahoogroups.com
Personal email address:
medhanye.biedebrhan@mu.edu.et
REFERENCE MATERIALS
• Timoshenko and Woinowsky-Krieger. “Theory of plates

and shells”
• Eduard Ventsel & Theodor Krauthammer. “Thin Plates
and Shells: Theory, Analysis, and Applications”
• Calladine. “Theory of Shell structures”
• Flugge. “Stresses in Shells”
1. In-plane and out-of-plane effects in shells:
Membrane and Bending theories
Shell theory:
Consider some possible types of state of stress for thin
shells:
1. If the flexural stresses are negligible compared with the
membrane stresses then such a type of state of stress is
called a membrane or momentless state of stress.
2. If, to the contrary, the membrane stresses are negligible in
comparison with the flexural stresses, then such a type of
state of stress is referred to as a pure flexural or moment
state of stress.
3. If the flexural and membrane stresses are of the same
order, then such a state of stress is called a mixed state of
stress, or edge effect. The term ‘‘edge effect’’ is associated
with the fact that the above-mentioned mixed state of
stress often occurs near edges of the shell.
Example:
Membrane theory:
• A shell mainly balances an applied transverse load,
much like an arch, by means of tensile and
compressive stresses, referred to as the membrane
or direct stresses.
These stresses are uniformly distributed over the shell
thickness. Such a state of stress is called the momentless
or membrane state of stress.
• Although the shear force and bending and twisting
moments are still present in the general case of
loading, the efficiency of the shell form rests with the
presence of the membrane stresses, as the primary
means of resistance with the bending stress
resultants and couples are minimized.
Membrane theory:
• Under appropriate loading and boundary conditions,
however, the resulting bending and twisting
moments are either zero, or so small that they may
be neglected.
Such a state of stress is referred to as the membrane state
of stress because of the analogy to membranes that
cannot support bending and twisting moments.
• The corresponding theory of thin shells that deals
with this state of stress is called the membrane
theory of shells.
Membrane theory:
• The membrane stress condition is an ideal state at
which a designer should aim.
• It should be noted that structural materials are
generally far more efficient in an extensional rather
in a flexural mode because:
1. Strength properties of all materials can be used completely
in tension (or compression), since all fibers over the cross
section are equally strained and load-carrying capacity
may simultaneously reach the limit for the whole section
of the component.
2. The membrane stresses are always less than the
corresponding bending stresses for thin shells under the
same loading conditions.
Membrane theory:
It can be shown that the following conditions are
necessary and sufficient for an existence of the
momentless state of stress in thin shells:
(a) The boundaries of a shell are free from transverse shear
forces and moments. Loads applied to the shell boundaries
must lie in planes that are tangent to the middle surface of
the shell.
(b) The normal displacements and rotations at the shell edges
are unconstrained: that is, these edges can displace freely in
the direction of the normal to the middle surface.
(c) A shell must have a smoothly varying and continuous surface.
(d) The components of the surface and edge loads must be also
smooth and continuous functions of the coordinates.
If these conditions are violated, fully or partially, flexural
stresses occur.
Bending theory:
• Note that the pure bending conditions have no
advantages and should be avoided because shells, in
view of their small thickness, possess a low strength
for this deformation.
However, sometimes bending conditions cannot be
avoided. It turns out that strong severe bending conditions
are localized only in a small domain near some
discontinuities in loading and geometrical conditions, as
well as near supports, etc.
As we move away from such a disturbance zone, the
bending stresses will diminish rapidly and a considerable
part of the shell will be in the momentless stress condition.
Therewith, the thinner the shell the faster this decrease of
bending stresses.
Bending theory:
• At locations where the deflections are restricted, or
there is a change in geometry, such as the cylindrical-
to-spherical shell junction, the membrane theory is
inadequate to maintain deflection and rotation
compatibility between the shells.
• At these locations, discontinuity forces and moments
are developed that result in bending and shear
stresses in the shell.
• It should be noted that these bending and shear
stresses are localized over a small area of the shell,
and they dissipate rapidly along the shell.
Bending theory: Edge bending effects
Basic equations for 3D elastic continuum:
• Equilibrium (Kinetic) relationships:
• Geometric (compatibility) relationships:
• Material (constitutive) relationships:

Solution Procedure:
Stiffness differential operator matrix
CHAPTER
END!
2. Bending theory of cylindrical shells
The general concept of elastic analysis will comprise of
the following basic components:
• Equilibrium condition should be fulfilled

• Kinematic relationships should hold true
• Constitutive behavior (linear isotropic elasticity)
GENERAL SHELLS OF
REVOLUTION
Components of displacements and rotations:
Notation for shells of revolution: membrane state
Notation for shells of revolution: membrane state
Notation for shells of revolution: bending state
Notation for shells of revolution: bending state
GENERAL SHELL THEORY
without
NORMAL HYPOTHESIS
for shells of revolution
Basic equations: General shells of revolution
(without normal hypothesis)
GENERAL SHELL THEORY
with
NORMAL HYPOTHESIS
(with normal hypothesis)
CYLINDRICAL SHELLS
Circular Cylindrical Shells
• In general very complicated equilibrium equations will result
for general shells. However, we can obtain relatively simple
equilibrium equations for a circular cylindrical shell.
• Then equilibrium equations are found as a combination of
effects from membrane action and that of a flat plate subject
to bending loads
Introducing the non–dimensional meridional parameter

PARTICULAR
EXAMPLE
Consider a ring-radially-loaded long cylindrical shell
with and without axial loading and internal pressure:
Geometry, loading and section forces
The following assumptions are taken into account:
• Axisymmetric loading conditions are considered.
• isotropic & homogeneous material behavior
• long cylinder, i.e. no edge-restraint effect
• Normal hypothesis, i.e., a straight fiber
perpendicular to the thickness before deformation
remains straight and penpendicular after
deformation and hence transverse shear strains are
neglected
Equilibrium relationships
Equilibrium of forces:
Equilibrium of moments:
When written in matrix form, a summary of the
equilibrium equations will be given as
Kinematic (geometric) conditions
• The general relationship between the strains and
deformations of thin-walled cylindrical shells with
normal hypothesis is given as follows,
0000000000000000000
• For cylindrical shells under axisymmetric conditions,
all the derivatives with respect to φ will be zero
because of symmetry. The kinematic relationships
will then reduce to:
BENDING THEORY
OF
CYLINDRICAL SHELLS
Decomposition of the total solution into particular and
homogeneous parts
Basic equations: General cylindrical shell case
THIN SHELL THEORY
without
NORMAL HYPOTHESIS
Thin-walled cylindrical shells under general loading
conditions
Assumptions:
• General loading conditions are considered.
• long cylinder, i.e. enough decay length
Generally loaded thin cylindrical shells:
THIN SHELL THEORY
+
NORMAL HYPOTHESIS
Thin-walled cylindrical shells under general loading
conditions
Assumptions:
• General loading conditions are considered.
remains straight after deformation and hence
transverse shear strains are neglected

THIN SHELL THEORY
+
NORMAL HYPOTHESIS
+
AXISYMMETRIC LOADING
Thin-walled cylindrical shells under axisymmetric loading
conditions
Assumptions:
• Axisymmetric loading conditions are considered.
remains straight after deformation and hence
transverse shear strains are neglected
Axisymetrically loaded thin cylindrical shells:

with
The material (constitutive) relationships can also be written as:
with
Stiffness differential operator matrix:
Reducing the three basic equations into one will give
the stiffness differential operator matrix:
The stiffness differential operator matrix for the current

problem will be given by:
Decoupling the stiffness differential operator matrix:
• The second row of the matrix can be decoupled and will give
• We can see that a stress in the circumferential direction

produces only a shift in the circumferential direction, which
produces torsion; and a pure longitudinal effect arises as a
result of an external moment acting in the longitudinal
direction.
Since these types of loading are not significant for our problem,
namely, the edge-bending problem, we will not go into detail.
The stiffness differential operator matrix equation will
then reduce to:
Decomposition of the total solution into particular and

homogeneous parts
The axial section force nx will result in pure membrane
effect and hence not important in the edge bending
discussion. The meridional normal strain will be given
by the product of the Poisson's ratio and the
circumferential normal strain.
Then we have:
• The equation in the radial direction for the homogenous
loading case with px = 0 and pz = 0 will be
Then
with
By transformation, we finally get a dimensionless form

of the desired differential equation of 4th order in W
Matrix-oriented solution method:
With
From the differential equation, let’s identify the fundamental

solutions of the radial displacement as a product of a
fundamental solution vector and the integration constants,
which are summarized in a vector form as:
We now introduce a new dimensionless length
parameter to the later notion of effective Length to
obtain:
with
i.e.
Then, we get
or
The above equation can be re-written as
Where
and
With ν = 0.3 for steel,

Fundamental solution vector:
The solution of the characteristic equation are the four

roots:
Then, with
The fundamental solution vector will be given by
Modified fundamental solution vector:

Modified fundamental solution vector:
The fundamental solution vector at the two edges of

the cylindrical shell, i.e. for and will be
given by:
HOMOGENEOUS EDGE DISPLACEMENTS VECTOR:
One can see that the derivatives of the fundamental
solution vector f are given by
with
And
The meridional rotation βφ = β can be multiplied by the

radius R to give
The displacement vector will then be:
with
and
The displacement vector at the two edges A & E will
then be:
where
and
The integration constants vector will be calculated

using the edge displacements vector as follows
Applying unit edge displacements, for example, for
The integration constant c1 can be obtained as follows:
Following the same procedure, one can find

Then, the integration constants matrix will be,
with
HOMOGENEOUS EDGE FORCES VECTOR:
One can see that the edge forces can be given in terms
of the radial displacement W by
i.e.,
So that
and
When written in compact form, the homogeneous
section forces vector can be expressed in terms of the
fundamental solution vector f as:
At the two edges A & E:

Sign convention for global edge forces and displacements:
Global edge displacements Global edge forces

Shell stiffness matrix:
where
Back calculation of the displacements in the shell

interior:
Back calculation of the section forces in the shell
interior:
and
Stiffness relationship of a short cylindrical shell
Stiffness matrix of a short cylindrical shell, including
normal hypothesis, under axisymmetric loading and
support conditions is given by:
Where the stiffness matrix is given as follows in terms
of non-dimensional parameters:
With
The following table summarizes the expressions for the

different matrix coefficients used in the stiffness matrix
for both short and long cylindrical shells.
Section forces of a short cylindrical shell
Section forces of a short cylindrical shell, including
normal hypothesis, under axisymmetric loading and
support conditions is given by:
with
Section forces at the starting end A of a short cylindrical shell:
Section forces at the end E of a short cylindrical shell:
Where the new parameter f is given by:

Special case for an infinitely long cylindrical shell:
Where
with
with
Example No. 1: Calculate the deformations and section forces at
a shell junction (I), where two cylindrical shell segments with
differing wall thickness are joined together centrically. The
loading of both of the shell segments is hydrostatic pressure
caused by water filling.
We need to have
Membrane solution
first!
This will lead us to the
next chapter
CHAPTER
END!
3. Membrane theory of shells of revolution
volution
Axis of revo
Membrane theory?
Necessary and sufficient conditions for the existence of
the membrane (moment-less) stress state in thin shells:
must lie in planes tangent to the middle surface of the shell.
stresses occur.
membrane theory of shells deals with shells under membrane
state of stress.
Membrane theory: Basic Equations
• We need to modify/reduce the general fundamental
equations of SOR for the membrane state of stress.
• The fundamental equations for thin shell theory with normal
hypothesis will be used with
and
• Since, in practice loading and boundary conditions are very

often axisymmetric, e.g. wall friction loads, centrifugal loading,
weight at vertical mounting, internal gas pressure, hydrostatic
pressure, etc., special case of axisymmetric loading and
boundary conditions will be treated and as a result, analytical
solutions will be possible.
• The following diagram shows the coverage outline of this

chapter.
General shells of revolution, SOR
General membrane shells of

revolution
Axisymmetric Loading & Boundary

conditions
Straight meridian Curved meridian
Cylindrical Conical Spherical

membrane shells membrane shells membrane shells
GENERAL MEMBRANE
SHELL THEORY
with
NORMAL HYPOTHESIS
General membrane shells of revolution
with
MEMBRANE THEORY
+
NORMAL HYPOTHESIS
+
AXISYMMETRIC LOADING & BCs
where is derivative w.r.t. the arc length

and
with
• As can be seen from the kinetic relations for
axisymmetrically loaded membrane shells, only three
unknown shell section forces are present, these can
be calculated from the equilibrium conditions alone.
• The determination of deformations will then follow,
i.e.
Equilibrium (Kinetic) relationships:
Material (constitutive) relationships:
Geometric (compatibility) relationships:

Membrane shells with straight meridian:
(Cylindrical and Conical membrane shells)
Cylindrical membrane shells:
We can use n and p instead of and . Then,

Cylindrical membrane shells: Section force Components
Symmetric loading: when px and pz are considered,

which will result in normal section forces nx & nφ only
Unsymmetrical loading: when pφ is considered, which
will result in torsion-type loading case and section force
nxφ. This loading case will not be discussed anymore.
Strain Components:
where
When written without considering the unsymmetrical
loading and resulting strain γxφ and section force nxφ:
Displacement Components:
The two integration constants C1 and C2 will be found
from the boundary conditions given by, say,
and
SUMMARY:
Example: (Cylindrical membrane shell with uniform
internal pressure)
Loading:
Solution:
Loading:
Section forces:
Displacements:
BC:
Cylindrical membrane
shell under
axisymmetric conditions
Conical membrane shells:
Conical membrane shells: Section force components
Considering the symmetric part only:
For a general equation of the form:
the general solution is given by:

The solution for the equation:
will then be
Strain components: Once the section forces are

obtained, the strain components can be found using:
Displacement components:
SUMMARY:
Example: (Conical membrane shell with uniform
internal pressure)
Loading:
Solution:
Loading:
Section forces:
BC:
Displacements:
BC:
Solution:
Conical membrane shell
under axisymmetric
conditions
Membrane shells with curved meridian:
(Spherical membrane shells)
Once again, the unsymmetrical

force & displacement components,
and
will not be considered.
Spherical membrane shells: Section force components
Multiplying by cos θ will give

Spherical membrane shells: Section force components
Integrating gives
With
Spherical membrane shells: Strain components
Spherical membrane shells:
Displacement components:
with
SUMMARY:
Example: (Spherical membrane shell with uniform
internal pressure)
Loading:
Solution:
Loading:
Section forces:
BC:
Displacements:
BC:
Spherical membrane
shell under
axisymmetric conditions
SUMMARY
Membrane shells of revolution
• The basic equations of the kinematic, kinetic (static
equilibrium) and constitutive behavior of the membrane
cylindrical, conical and spherical shells are represented in clear
and compact form as shown on the next page.
• These equations are valid for arbitrary loading pattern over
the shell surface which are compatible with the membrane
assumptions:
• For the membrane cylindrical and conical shells the symbol ‘θ‘
should be replaced by the symbol ‘X’.
constitutive behaviour:
• The constitutive equations for isotropic material hold
irrespective of the type of geometry of the shell
CHAPTER
END!
4. Bending theory of shells of revolution under
axisymmetric loads
volution
Axis of revo
General
• Our goal is to provide an analogy of the differential
equation of the circular cylindrical shell and the
general shell of revolution (internal analogy) and an
analogy of the corresponding stiffness relationships
(external analogy) approach.
• Thin shell theory with normal hypothesis will be used
in the discussion on bending theory of shells of
revolution under axisymmetric loads and boundary
conditions.
• As a result of the normal hypothesis, the equation
obtained from setting the transverse shear strains to
zero, will be used as an additional equation along with
the existing ones.
Basic equations for general SOR under axisymmetric
conditions
conditions
conditions
Decomposition of the basic equations into symmetric
and asymmetric parts:
• Asymmetric part:
The asymmetric part, shaded gray in the above equations, will
not have an effect on the edge bending phenomena, hence, will
not be considered.
• Symmetric part:
Setting the asymmetric part to zero will reduce the equations to
the symmetric part only; which will lead to the solution of the
edge bending problem as has been done in the case of
cylindrical shells.
After neglecting the derivatives with respect to the
circumferential angle φ and the asymmetric part, the expressions
for the modified section forces and stresses will be:
where is derivative w.r.t. the arc length

Using the normal hypothesis and considering the last row of the
above matrix,
Or
With and
Transformation of the basic equations
—> use of global deformation and force quantities
• Instead of the meridional displacement component U
and the outward normal displacement W we
introduce the axial (vertical) displacement U* and the
radial (horizontal) displacement W*. For the circular
cylindrical shell case, U = U* and W = W*.
• Instead of the membrane normal force & the local

transverse shear force we introduce the global
transversal (radial) force and the longitudinal (or
axial) force
• Meridional (membrane–type) edge loading
This case is not discussed further here since it represents a
membrane–type loading case.
• Edge bending problem
There is one important condition which defines the isolated
edge bending problem and which states that the resulting
section force in longitudinal (or axial) direction, which for
clarification is the direction parallel to the axis of revolution,
has to be zero.
By introducing the transformed global deformation and force

quantities this condition can be fulfilled in a very easy way, i.e.
simply by setting the longitudinal force to zero (condition
for the edge bending problem).
Transformed deformation and force quantities:
The relationship between the local and global
deformation and force quantities is defined by
introducing a (rotation) transformation matrix as
follows:
Transformation of the kinetic equations (formulation in L and T)
Equilibrium in the vertical direction:
Constant
For edge bending problem:
The transversal/horizontal shear force T can also be

expressed as (using the equations from the previous
slide):
Equilibrium in the horizontal direction:
And from the basic equilibrium equations, one can see that:
Transformation of the compatibility equations (formulation in
U* and W*)
The relationship between the local and global deformations can
be re-written as:
Using the global deformation components, the compatibility

relationship can also be re-written as:
Summary of the globally transformed basic equations


Basic (Reissner–Meisner) equations – meridional
moment equilibrium and normal hypothesis expressed
by βφ and T:
• Moment equilibrium expressed by βφ and T :

• Normal hypothesis expressed by βφ and T :
• Therefore we introduce now the differential operator
L( ) which permits us to write down these equations
in compact form as follows:
• We can now express the meridional moment

equilibrium and the normal hypothesis (= Kirchhoff
assumption) as follows:
• Special case I: Circular cylindrical shell
The Reissner–Meissner differential operator simplifies

considerably as follows:
By elimination of the radial transversal force , i.e. by expressing
it in dependency of βφ with the help of the meridional moment
equilibrium (1) and inserting it into the normal hypothesis (2) we
obtain an ordinary linear differential equation, with constant
coefficients, of 4th order in βφ as follows:
• By introducing the non–dimensional parameter , we

arrive at the following final ODE of 4th order for the system
parameter βφ which also holds for the radial displacement W*:
• By comparison with the formerly derived equation for the

circular cylindrical shell, we can recognize their identity — as
it should be.
• Special case II: Conical shell
The Reissner–Meissner differential operator for the

conical shell now simplifies as follows:
• Special case II: Spherical shell
Here we have
The Reissner–Meissner differential operator for the

spherical shell now simplifies as follows:
Analogy relationship (Geckeler 1926):
shell of revolution vs. cylindrical shell
GECKELER-APPROXIMATION AND ANALOGY MODEL
• We want an analogy of the differential equations for
a circular cylindrical shell and that of a general shell
of revolution.
• To achieve this, we must simplify the basic equations
for the general shell of revolution by means of the
so-called Geckeler approximation.
• The aim of this simplification is simply to reduce the
basic equations of the general shell of revolution to
come to a differential equation, which is similar to
the equation for the circular cylindrical shell.
• If we succeed in an analogy of the differential
equation as stated above, then the so-called 'internal
analogy' will result.
• Subsequently, we can also compare the stiffness of
the circular cylindrical shell with that of the general
shell of revolution, 'external analogy’.
• This comparison of the differential equation of the
type and stiffness relationship goes back to Geckeler
and is therefore called Geckeler analogy.
• We know from chapter 2 that the application of an
axisymmetric out-of-plane shear load to a thin
cylindrical shell of radius R and thickness t yields a
solution in which all relevant quantities vary in the
axial direction in a strongly decaying oscillatory
fashion with a characteristic length of order Sqrt(Rt).
• In Geckeler’s approximation, only the functions
themselves and their derivatives of higher order are
retained and the lower order derivatives are discarded.
• The Geckeler’s approximation is based on the physical
argument that the boundary effects are localized and
thus the higher order derivatives are of greatest
importance.
• Hence, in each equation, only the highest order
derivative of the main function and the zero order
derivative of another, coupling function, shall be
retained.
• On the basis of this argument, the Reissner-Meissner
equations shown below can be re-written as will be
shown later.
Approximation of the transverse force as a function of
global displacement parameters :
using Geckeler’s approximation
and using the relationship obtained from the normal

hypothesis,
Approximation of the vertical equilibrium equation
Using order of magnitude estimation, the above

equation can approximately be expressed as:
Approximation of the horizontal equilibrium equation
In this case one can see that all the terms are of the
same order of magnitude. Hence, the above equation
can be taken as it is.
Elimination of the vertical displacement:
The last two equations can be re-written in matrix form
as follows:
Using order of magnitude estimation, the matrix
equation can approximately be expressed as:
i.e.,
Let or then,
Thus, an 'internal analogy', that is an analogy of the differential

equation type, between the circular cylindrical shell and the
general rotation shell is obtained.
Once the fourth order differential equation is solved in exactly
the same manner as that of the cylindrical shell, the remaining
displacements, strains and section force components can be
obtained as a function of the horizontal displacement W* as
follows.
Local and global displacement components:
Rotation of the meridional tangent:
Normal strain and curvature components:
Global and normal transverse shear force components:

Normal section force and moment components:
• For cylindrical shells:
with
• For general shells of revolution:
with
• For general shells of revolution, it can also be re-
written as:
with
CHAPTER
END!
5. Membrane theory for shells of general shapes
Membrane theory?
Necessary and sufficient conditions for the existence of
the membrane (moment-less) stress state in thin shells:
must lie in planes tangent to the middle surface of the shell.
stresses occur.
membrane theory of shells deals with shells under membrane
state of stress.
Membrane theory: Basic Equations
• In the classical theory of elasticity, a common procedure is to
use Hooke's law in order to substitute the equilibrium
equations into the compatibility relations (or vice versa) so as
to obtain a single governing equation.
• On the other hand, it is advantageous to obtain the governing
equations for thin-shell structures by a process of formal
reduction from the general equations of the elastic theory of
three-dimensional solids, i.e. reducing the general
fundamental equations of 3D elastic continuum for the
membrane state of stress.
i.e., with
and
• We must emphasize that in membrane theory we are first
concerned with satisfaction only of the equilibrium equations,
and not with complete solutions of shell problems.
Heinz Isler
• Born July 26, 1926 in Zollikon, Switzerland
• Attended Federal Technical Institute
• Graduated in 1950 with a degree in Civil
Engineering
• Only student out of 100+ to choose thin-
shells for senior design
• Worked as an assistant to Pierre Lardy
• Based most of his structures on smaller
models
• Stated there were 3 ways to arrive at shell
shapes, (Freely shaped hills, Rubber Balloon
Membranes, and Reversed Hanging Cloths
Hanging membranes by Heinz Isler
Thin-Shell Roof over gas station in Bern-Zurich
Highway
• The equilibrium equations are concerned with external forces
and internal stress resultants, and involve the shell only
through the geometry of its (undistorted) surface.
• To analyze the internal forces we cut from the shell an
infinitely small element formed by two pairs of adjacent
planes which are normal to the middle surface of the shell
and which contain its principal curvatures.
• In this system, the number of unknowns is equal to the
number of equations, so the problem of the membrane
theory of shells is statically determinate (that is true for the
equilibrium of an infinitely small shell element but is not
always true for the equilibrium of the entire shell).
• That is, if the external load components, px, py, and pz are
known, then the membrane forces & stresses for such a shell
are uniquely determined from the equilibrium equations.
• For shells of arbitrary shapes, the character of solutions to the
equilibrium equations depends much on the geometric form
of these shells, and in particular on the sign of the 'Gaussian
curvature' of the surface.
• Equilibrium in shells of arbitrary form generally is done by
means of generalized curvilinear coordinates.
COURSE PARTICIPANTS ARE ADVISED TO READ
MORE ON THIS CHAPTER!!
SHELL DESIGN
SHELL DESIGN CONSIDERATIONS
A structural shell (metallic, concrete or composite) design
involves the following steps:
Choice of a shell geometry which meets the overall
architectural requirements.
A trial choice of the dimensions of the shell and its
supporting members, based on previous experience and/or
observation of similar shell structures
A preliminary design and analysis of the shell. For this
purpose we can use the membrane theory as well as other
simplified shell analysis schemes.
At this stage, we obtain the initial dimensions of the shell system
including the shell thickness, the pattern of shell reinforcements, and
the tentative values of required materials and reinforcements.
These results are useful for a more comprehensive analysis and also
for the initial material & cost estimates.
The preliminary analysis and design of a shell also
provides an insight into the general behavior of that shell.
Therefore, having performed an initial design-analysis,
the designer would have a free hand in refining the
design without going into extensive calculations and/or
unwanted expenses.
A more comprehensive analysis and detailed design of the
shell using shell theory and/or available computer programs.
This stage is required for a final check of the design. A
logically based preliminary design usually remains valid,
except for some local modifications or perturbations.
Determination of the form, dimensions, and reinforcements
of some shells can also be obtained through experiments on
physical models of the shell. This is particularly so for
complicated shell geometries and/or cases in which there are
no reliable analytical tools.
A good design should always involve considerations related to
construction, shell manufacturing processes, and shell
aesthetics.
Design-construction precautions must be taken to assure the

integrity of the structure as a whole since shell structures are
usually accompanied by strengthening members such as
stiffeners, edge and/or ridge beams, and end diaphragms.
For thin shells, on the other hand, there exist situations in

which the loading will produce compressive internal stresses
in the shell. These stresses could make the shell elastically
unstable and thus cause buckling failure of the shell
structure.
The buckling strength of shells can be several order of
magnitudes lower than the material strength of the shell.
Furthermore, thin shells have been shown to be among the so
called imperfection-sensitive structures. This means that the
buckling load of the shell is highly dependent on
imperfections in the shell geometry and loading.
As a consequence, the true buckling load of an actual
shell could be several times smaller than the buckling
strength of a corresponding ideal perfect shell.
Premature failure of shells, due to buckling, makes the

stability design of shells important and dimensioning of shell
thickness is usually based on buckling considerations rather
than material strength criteria.
The last chapter of this course will cover the buckling
instability of shells.
The practical design of shell structures are generally based on
Codes of Practice. There are many codes and standards with
shell design recommendations. These codes help the designer
to carry-out practical design of shells and hence, can be used
as guidelines and official criteria in the design process.
Codes of practice offer general design guidelines specifying

limiting design parameters, such as the minimum
reinforcement, maximum stress and strain, and minimum
thickness, etc.
CHAPTER
END!
6. Analysis of multi-shell structures for bending
and membrane stresses
MULTIPLE SHELL ROOFS
Multiple shells are used in the construction of roofs for many
large buildings efficiently and aesthetically such as bus, railway
and airport terminals, sport stadia, exhibition halls and factories
many of which are cylindrical.
(a) A set of parallel shells, connected along their edges,

(b) As (a), but with edge-beams,
(c) and (d) Isolated shell, without and with edge-beams, respectively.
MULTIPLE SHELL ROOFS
(a) Shells with feather edge-beams, (b) Detailing of feather edge-beams,

(c) Multiple shells with small edge-beams,
(d) and (e) Multiple shells with regular edge-beams,
(f) Beam analysis of multiple shells
MULTIPLE SHELL (Barrel) ROOFS
Multiple barrel shells can be any one of the following:
1. Multiple shells with feather edges. In this type, the inner
shells have rounded edges called feather edges but not
real edge-beams. At the outer sides, there are edge-
beams.
2. Multiple shells with inner shells having small edge-
beams which act as a tie between the shells and are
incapable of bending resistance. This case is similar to
case (1), but the weight of the small edge-beam has to be
considered in the design.
3. Multiple shells with all shells having conventional edge-
beams.
In general, we may say that short span multiple shells can be
built with no or small internal edge-beams. For long spans, it is
more convenient to have regular edge-beams for internal shells
as well.
CYLINDRICAL SHELL ROOFS
• As has been mentioned in introduction part of this course, a
cylindrical shell is named after the shape of the profile, i.e., a
circular, elliptical, or parabolic cylinder.
• Of these shells, only the circular cylinder is a shell of
revolution, whereas the others represent shells of translation
since they can be generated by the translation of a straight
line.
• Let us consider a cylindrical shell of a general shape, as shown
below:
where x is the distance measured from

some reference profile to the point of
interest and θ is the angle between the
normal to the shell at a point and a
reference normal at a chosen origin.
• Let us consider a circular-cylindrical shell roof which is
specified by its leading dimensions L, B, t and R=a for length,
breadth, thickness and radius, respectively, as shown below.
(a) Single cylindrical shell, showing dimensions and coordinate

system.
(b) Cross-section of a shallow circular cylindrical shell, here drawn
as a parabolic arc.
• Most of this chapter will be devoted to a single section of the
shell, Figures c and d shown above, which is structurally
separate from its neighbors.
• This isolated shell is conventionally taken to be an adequate

representation of the end-shells of an array; and it is easy in
fact to adapt the analysis to the case of a shell in the interior
of such an array, which is connected to neighboring shells at
its edges.
• Shells both with and without edge-beams will be studied and

in this way we will be able to grasp the advantages associated
with the provision of these auxiliary members, and to discuss
ways of deciding on their dimensions.
The effect of edge-beams
• There are two aspects in the general analysis of the behaviour
of a shell with edge-beams:
– The first is connected with the details of load resistance
over the shell itself.
– The second concerns the gross distribution of load
between the shell and the beams.
Beam-arch method of roof analysis
• The beam-arch method makes use of combined beam and
arch actions of cylindrical shells.
• In this method, beam analysis and arch analysis of the shell

are carried out almost independently by the well-known
simple theories of structural analysis. The results of these
analyses are, then, combined appropriately to yield the spatial
behavior of the cylindrical shells.
Beam Analysis:
The shell is conceived as a beam resting on the end supports
and behaving according to the classical beam theory. The
beam has the profile of the cylindrical shell as its cross-section
and the length of the shell as its longitudinal span.
Arch Analysis:
A unit width of the shell will be considered as an arch
subjected to loading.
• Now one of our objectives is to find how the total shearing
force on any vertical cross-section is shared between the shell
and the beams.
• The easiest way of discovering this is to consider the
equilibrium of one half of the structure, separated into shell
and beam components, as shown below:
Figure: Resultant forces acting on half of the shell and the

associated edge-beams
Other multiple shells
CHAPTER
END!
Assignment problems
Using linear elastic shell analysis, compute and draw
a) deformation (radial deformation w*, meridional
rotation βx) and
b) section force (meridional section force nx,
circumferential section force nφ, transverse shear
force Qx, meridional section moment mx and
circumferential section force mφ)
results of the following cases along the axial direction
of the corresponding shells.
Note: The normal section forces and bending moments should
be plotted normalized with respect to the corresponding uni-
axial yield section force, Npl = t.fy, and section moment, Mpl =
t2.fy/4, respectively
You can download ebooks @
www.gigapedia.com
www.ebookee.com
7. Stability problems in shell structures
INTRODUCTION
• Shells, when in compression, are very sensitive to instability.
• It is, often, very difficult to determine the risk of instability
because the results of an idealized analysis can be very
different from reality.
Comparison of Bar – Plate – Shell
• In the following the Euler-bar under a single load, a Navier
supported quadratic plate and the hinged supported cylindrical
shell subjected to axial load will be compared.
• The table shows the related load-displacement diagrams.
Where,
u or w, respect. longitudinal or normal displacement
P, p axial load
Pkr, pkr theoretical critical load of perfect structure
behavior of ideal structure & load application
imperfect structure or load application
additional material failure (plasticity, cracks,
…)
Comparison of Bar – Plate – Shell
Assessment:
Bar: neutral
Post critical behavior Plate: stable
Shell: unstable
Consequence: the critical load of an imperfect shell is

dramatically reduced compared to the
ideal, perfect structure
Note: Not all types of shells behave sensitive w.r.t. stability, e.g.
negative Gaussian curvature
Imperfection sensitivity
Imperfection may be regarded as deviation from:
geometry
load
support conditions
Material
Problem: Imperfections are not well defined. They depend
on the production process and are of statistical
nature (type and size).
The imperfection sensitivity ( = reduction of critical load)
depends on several parameters:
o geometry
o support conditions
o load
o slenderness
Effect of imperfection amplitude:
Example: experimental buckling results of a tip-loaded
cylindrical shell
axial load
Example: experimental buckling results of a
thin-cylindrical shell under external pressure
external
pressure
• the resultant axial buckling load is independent of the radius
except for the reduction factor
• spherical shell under external pressure and cylindrical shell

under axial load behave similar
• the buckling load is a function of the “material stiffness”
i.e., P cr ~ E
Example: P Steel > P Alumnium > P Rubber
cr cr cr
• support conditions are important, e.g.
Pkr,1 Pkr,2 < Pkr,1

• the buckling load of a cylindrical shell under external pressure
is significantly less than that of the spherical shell
• simultaneous external pressure reduces the axial buckling
load of a cylindrical shell
• simultaneous internal pressure raises the axial buckling load

of a cylindrical shell
reason:
• the stabilizing effect of the ring tension reduces the
imperfection sensitivity
but reduced material resistance under combined
tension-compression
Plastic strength according to von Mises membrane criteria

• For axisymmetric shells under axisymmetric loading and
boundary conditions where 2D principal membrane stresses
exist, the von Mises membrane yield criteria is given by
with an upper limiting (yield) curve given by

FE MNA for both fixed & pinned bottom; and bi-axial membrane
Mises yield strength
Elastic imperfection reduction
factor for cylindrical shells
• Q is fabrication quality parameter and should be taken,
depending on the fabrication tolerance quality class, from
EN1993-1-6 pressurized elastic buckling imperfection reduction
factor
EN1993-1-6 pressurized elastic buckling imperfection reduction
factor
Buckling mode
determination
Linear buckling modes
(LBA)
Strength calculation for shells
(LBA, MNA & GMNA with or
without imperfections)
• Comparison of different buckling strengths in pressure
representation
• Comparison of different buckling strengths in capacity
representation
CHAPTER
END!

Shell All Chaps Annotated-1 PDF

Uploaded by

Copyright:

Available Formats

Shell All Chaps Annotated-1 PDF

Uploaded by

Document Information

Original Title

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

Shell All Chaps Annotated-1 PDF

Uploaded by

Copyright:

Available Formats

CENG 7503 - Analysis and Design

Course Instructor: Dr.-techn. Medhanye B.

STRUCTURAL ENGINEERING CHAIR

• Born August 27, 1899

•Shells constitute among the most common

• Extensional or membrane actions are any in–

• Arches and shells can resist distributed

• For shells, the relative proportions of extensional

(b) Non-developable surfaces

Course Instructor: Dr.-techn. Medhanye B.

STRUCTURAL ENGINEERING CHAIR

Conferring an in-depth knowledge and

Group home page:

• Timoshenko and Woinowsky-Krieger. “Theory of plates

• Equilibrium (Kinetic) relationships:

• Geometric (compatibility) relationships:

• Material (constitutive) relationships:

• Equilibrium condition should be fulfilled

Introducing the non–dimensional meridional parameter

• Geometric (compatibility) relationships:

• Geometric (compatibility) relationships:

The stiffness differential operator matrix for the current

• We can see that a stress in the circumferential direction

Decomposition of the total solution into particular and

By transformation, we finally get a dimensionless form

From the differential equation, let’s identify the fundamental

With ν = 0.3 for steel,

The solution of the characteristic equation are the four

Modified fundamental solution vector:

The fundamental solution vector at the two edges of

The meridional rotation βφ = β can be multiplied by the

The integration constants vector will be calculated

The integration constant c1 can be obtained as follows:

Following the same procedure, one can find

At the two edges A & E:

Global edge displacements Global edge forces

Back calculation of the displacements in the shell

The following table summarizes the expressions for the

Section forces at the end E of a short cylindrical shell:

Where the new parameter f is given by:

• Since, in practice loading and boundary conditions are very

• The following diagram shows the coverage outline of this

General membrane shells of

Axisymmetric Loading & Boundary

Straight meridian Curved meridian

Cylindrical Conical Spherical

where is derivative w.r.t. the arc length

Material (constitutive) relationships:

Geometric (compatibility) relationships:

We can use n and p instead of and . Then,

Symmetric loading: when px and pz are considered,

Considering the symmetric part only:

For a general equation of the form:

the general solution is given by:

Strain components: Once the section forces are

Once again, the unsymmetrical

Multiplying by cos θ will give

• Equilibrium (Kinetic) relationships:

where is derivative w.r.t. the arc length