Você está na página 1de 20

Rotter, J.M.

(1998) Shell structures: the new European standard and current research needs,
Thin-Walled Structures, Vol. 31, pp 3-23.

SHELL STRUCTURES: THE NEW EUROPEAN STANDARD


AND CURRENT RESEARCH NEEDS

J. Michael Rotter
School of Civil and Environmental Engineering, Division of Engineering,
University of Edinburgh, Edinburgh EH9 3JL, UK

ABSTRACT
Shell structures are widely used in a great variety of applications from space rockets to domestic food and
drink containers. Civil engineers are principally concerned with steel shell structures as silos, tanks,
pipelines, chimneys, towers and masts, though other examples may be found in offshore structures and
stadium roofs.
This paper describes the treatment of terrestrial shell structures in Eurocode 3: Steel structures. It
outlines the principles which are guiding the development of the standard, the range of applications
covered, and some details of the current proposals. The axially compressed cylindrical shell is then
chosen as an example illustrating the range of real problems which need to be addressed, and the paucity
of current data on many aspects of these problems. This example is also used to outline the complexity
involved even in this one area, recent progress and current needs.

KEYWORDS
Buckling, cylinders, design philosophy, European, shell structures, silos, standards, tanks.

INTRODUCTION
Shell structures offer the most efficient usage of structural material in many applications. They may
sometimes be a little more difficult to construct than other forms, but if this constraint can be overcome,
they provide a natural choice for many applications. They are widely used in aircraft, space rockets,
motor vehicles, nuclear containment vessels, submarines, domestic food and drink containers, cooling
towers, piles, shell roofs and domes, dams, pressure vessels, cryogenic tanks, biodigesters, silos, tanks,
pipelines, chimneys, towers and masts. Civil engineers have a primary responsibility for some
"applications": in metal structures these are principally silos, tanks, pipelines, chimneys, towers and
masts.

This paper describes the proposed treatment of these structures in Eurocode 3: Steel structures (EC3,
1994). It outlines the underlying the structure of the standard, the range of applications covered, and
some details of current proposals. The axially compressed cylindrical shell is then used as an example
illustrating the range of real problems and the paucity of current data on many aspects of these problems.
It is clear from the above that many different structural forms are built as shell structures by civil
engineers. Each has a different geometry, typical slenderness, loading condition, and set of practical
constraints. Each is, in general, designed by a different community of design engineers specialising in
that structural form, and constructed by companies who often specialise in building that form alone. As a
result, each application has its own historical precedents, and many successful structures have been built
according to the specialised historical rules for each application. The task of unifying design practice
amongst these groups is considerable, and even that of persuading the groups that practice should be
unified or harmonised is not simple. The task therefore requires imaginative and forward-looking
strategies which are not normally required for standards development in fields with a longer history of
association with a single profession.
The range of metal structures described above as the responsibility of civil engineers has been divided
between two Project Teams within the structure of the Eurocode 3 drafting organisation: PT3 covers
Chimneys, Towers and Masts, and is convened by Mr Brian Smith of Flint & Neill Partnership, London:
PT4 covers Silos, Tanks and Pipelines and is convened by the author. The division of tasks into only two
groups is not ideal, but has historical origins. Professor Jacques Brozzetti of CTICM in France is
Chairman of the drafting committee with overall responsibility for the Eurocodes on steel structures.
The division of the work between different project teams means that coordination, and acceptance of
quite different specialised needs for different groups, is very necessary. These differences require some
time for their resolution.
In addition to outlining these general considerations, the paper illustrates the current technical position by
means of an example: the axially compressed cylindrical shell. These shells are widely found in many
applications, so the choice is a natural one. The variety of problems which must be addressed, the current
state of knowledge and a few problems needing urgent solution are described.

EUROCODES FOR STEEL SILOS, TANKS AND PIPELINES, CHIMNEYS, TOWERS AND
MASTS
As noted above, steel shell structures are used in many different applications, and the design engineers
involved are often barely aware of each other's existence, let alone the nature of the problems they face.
However, shells of revolution (axisymmetric shells) abound in different applications, from soft-drink
containers to cooling towers, and from space rockets to farm slurry tanks. Current knowledge of shell
structures is dominated by shells of revolution, so the standards naturally exclude other shell forms,
though general shells are very common and economically important, for example, in motor cars.
Steel silos, tanks, pipelines, chimneys, towers and masts all contain, or may contain, shells of revolution.
The pipelines, chimneys, towers and masts are predominantly cylindrical shells, with steep conical
transitions (near cylinders), whilst the tanks and silos often include conical, spherical or toroidal
components. The aim is to define a single set of generic rules which can be used to define the structural
strength of all these applications.

However, there are considerable differences between the typical example structures of each of these
applications. The shell slenderness, characterised by the radius to thickness ratio, may be of the order of
50 for a pipe, but 3000 for a tank. Such a wide slenderness range, with strong behavioural changes in
between, is not generally seen in building or bridge structures. For most of these shells, the axis of the
structure is generally vertical, but pipelines are commonly horizontal, so that the support and loading
conditions are very different. The dominant loading conditions often also differ widely, from wind and
dynamic conditions on towers and chimneys to differential settlement in buried pipes or biaxial
membrane tension in a silo hopper.
Standards may be of two types: "applications" standards, relevant to a particular structural type, such as a
tower or tank, and "generic" standards, intended to be relevant to a whole class of structures with a
common form, such as a shell or box.
There are several generic standards which may be regarded as relevant to metal shells, though some of
these are now rather old (NASA, 1968; CRCJ, 1971; Baker et al, 1972; ESDU, 1974; ECCS, 1988; DIN
18800, 1990). The best generic standards for metal shells of revolution predating the Eurocode
development are probably the ECCS code (ECCS, 1988) and the German standard (DIN 18800, 1990).
One reason for the reduced effort apparently being put into developing these standards is probably the
advance of very powerful computer programs. The strong effort put into shell strength assessment using
standards for space applications in the 1960s and 1970s was no longer seen as necessary when each
individual component could be accurately assessed using advanced software in an environment of
plentiful resources. Pressure vessel codes (e.g. BS5500, 1996) may have a wide application, but the rules
are generally couched in a form which is not consistent with civil engineering standards.
In the field of "applications standards", there is great diversity in the current situation. There are many
standards for different types of tank, and most of these contain provisions which are far wider than
structural strength considerations. Similarly there are many standards for pipelines, and it is necessary to
limit the scope of the work here to those aspects which are not offshore and are the primary business of
civil engineers: this definition of a boundary is not straightforward. Silos represent a special case, for
which no major standard exists yet in the world, the best metal silos standards being the Japanese
aluminium silos standard (JIS, 1987) and the British Materials Handling Board/ BSI standard (BMHB,
1987). The Australian AISC guide (Trahair et al, 1983) and textbooks (e.g. Gaylord and Gaylord, 1984;
Martens, 1988; Hampe, 1991) provide the best current design advice.
Each application has developed its own specialised rules, relevant to its own conditions and omitting such
complexity as is unnecessary to its needs. Thus, the imposition of a general set of rules of wider
application and also necessarily of greater complexity will not often be viewed by practitioners.
Principles and Constraints
The following principles have been defined as a basis for the Eurocode development:
a) Uniformity of rules for different applications: The idea that all these structural forms should be
designed to the same rules stems from the proposition that the structure does not know what the role
of the structure might be, and the strength of a chimney of a given geometry in a given stress state
should be indistinguishable from that of a tank under the same conditions. Where unusual structures
are built, whose role as a tank or a silo is unclear (e.g. petuka tanks), or with both tank and tower
parts in one structure, it is clear that there should not be economic benefit by treating it as one form
or the other. Thus, there is justification for treating all shells of revolution with common rules.
b) Retention of existing rules for many structures: Many successful structures have been constructed to
existing rules, and these generally capture the essential needs of the application. They should be
retained.

c)

Incompatibility between rules for different applications: Different structure types need different
specialised rules. The generic rules must be broad enough to address the needs of all structures.

Structure of the Planned Standards


A set of draft standards has been produced within Eurocode 3 as follows. There will be two generic
standards to cover structural forms. These are
Strength and stability of shell structures, Eurocode 3 Part 1.6
Strength and stability of plated box structures, Eurocode 3 Part 1.7
In addition, the following "applications" standards have been drafted, referring to the above, but which
have additional specialised rules within them for each application.
Chimneys, Eurocode 3 Part 3.1
Towers and masts, Eurocode 3 Part 3.2
Silos, Eurocode 3 Part 4.1
Tanks, Eurocode 3 Part 4.2
Pipelines, Eurocode 3 Part 4.3
The remainder of this paper outlines the nature of the generic shells standard (EC3 Part 1.6, 1998),
describes something of the proposed standard on silos (EC3 Part 4.1, 1998) and examines technical issues
which need to be addressed in the development of new rules.

STANDARD FOR STRENGTH AND STABILITY OF SHELLS


Scope
The generic shells standard is concerned only with the strength and stability rules for axisymmetric metal
shells. Whilst the term strength could have been taken to include buckling considerations, stability is so
dominant in shell structures that it was felt that it should receive special mention in the title (the ECCS
code is concerned only with stability). Since it is a Eurocode under the set of structural engineering
standards, it is effectively restricted to problems which are the common province of civil engineers. This
naturally excludes many common shell structures.
Format
Three features have governed the choice of format for the standard:
a) the need to contain all specialised rules as subsets of the generic rules;
b) the desirability of producing a forward-looking standard which addresses the methods of design
currently being used and those which are likely to grow in usage; and
c) the need for engineers to be able to use their own preferred approaches to design.
The first of these is not easy to achieve, as any algebraic general rule, which is conservative throughout a
large domain, is likely to become very conservative in certain areas, where the specialised rules of a
particular application with restricted scope may well be quite accurate. This problem is addressed by
using a hierarchy of methods, so that a simple rule which is good and accurate for a restricted class of
problems in a given application may still be based on the generic standard in the sense that rather
complicated procedures in the generic standard may demonstrate its conservatism within the scope of the
applications standard. Thus, the generic standard (Part 1.6) includes sophisticated computer analyses
which can justify simple rules of limited scope in the applications standards.

The second and third considerations mean that the standard should address modern methods which make
extensive use of computers as well as simple methods for those who have only a morning to design a
structure. These matters, and similar issues concerned with safety classification, were discussed in detail
by Rotter (1997a) and Greiner (1997).
One of the key problems with the design of shell structures is that an increased effort in analysis often
leads to a more conservative design, if the same criteria are used to assess the strength. For this reason, it
is necessary to define different strength assessment criteria according to the method of analysis, and thus
it is also necessary to define the analysis methods rather carefully.
Limit States
The standard defines four limit states, which in general the design should address. These are: Plastic
collapse and rupture; Cyclic plasticity; Buckling; and Fatigue. For each, the types of analysis which may
be used and the criteria of failure are defined. The limit state of cyclic plasticity, limiting the extent to
which local stresses may induce local plastic deformations and induce a low cycle fatigue failure, is novel
for structural engineers.
Different Analytical Approaches
The standard recognises that the structure may be analysed using membrane theory, linear shell bending
theory, linear eigenvalue analysis, geometrically nonlinear analysis or geometrically nonlinear analysis
including modelling of imperfections. Each of these analyses is used by a significant group of engineers
at present, so the conditions under which each may be used, and a codification of the assumptions is
necessary. The criteria of failure must be defined relative to each analysis type, since all analyses
(however sophisticated) require an additional factor on the outcome before they can be used in design.
To clarify these matters, and the different analysis procedures applied to an interesting problem, Greiner
and Wallner (1997) presented a number of example calculations.
Cheap and Expensive Structures: Simple and Complex Rules
A large number of small structures of each type are built with relatively low-level engineering
calculations and narrow profit margins. For these, a quick inexpensive approach to the analysis is
necessary. For the few expensive, complicated or novel installations, it is necessary to permit much more
sophisticated analyses. It could be argued that sophisticated analyses do not need to be codified, but there
are two special reasons why they should be. First, there is the problem that the specialised rules of the
"applications" standards should be conservative relative to the general rules of the "generic" code. This
means that it is necessary to permit the very maximum real strength to be determined in some manner
within the scope of the generic code. Secondly, shell buckling is extremely sensitive to some geometric
imperfections (form & amplitude), and the use of sophisticated computer tools to these problems should
not be undertaken without definitive requirements on either the appropriate knock-down factor to account
for imperfections, or the appropriate choice of imperfections to ensure a safe outcome. These matters are
worthy of codification as they become more and more common practice in the professional community.

Rules for Problems Not Yet Identified


In addition to the above advantages of codifying sophisticated calculation methods, there is an additional
gain. As noted in the later parts of this paper, many elementary problems in shell buckling have not been
investigated at all, or only to a very limited extent. Codifying the methods by which shells can be safely
designed using computational methods has the great advantage that the designer can tackle many
situations which have never been researched and the researcher can undertake secure investigations in the
knowledge that the conditions of the standard have been fully met. Moreover, by showing that the
requirements of the standard have been met, the acceptability of research results for immediate adoption
as algebraic rules in standards is greatly facilitated.
The Problem of the Textbook
The chief criticism which has been levelled at the plan for a standard which includes a hierarchy of
methods of analysis and strength evaluation is that the standard must not be a "text-book". Precisely what
is meant by a text-book is not defined, but the term is clearly intended to mean a document which is
informative rather than prescriptive. For the most part, this is not a difficulty, since the rules are
concerned with the manner in which calculations must be performed, but some informative material is
needed to instruct the user on the significance of the calculations. A general discussion on the definition
of analysis methods within a structural standard is certainly needed.

AN EXAMPLE: BUCKLING OF CYLINDRICAL SHELLS UNDER AXIAL COMPRESSION


In many aspects of shell structures, there is a mismatch between the needs for structural design and
current codified provisions. This section sets out current needs and describes rules which have been
implemented in EC3 Part 1.6 (1998): an extensive description of the practical needs for silo construction
was given by Rotter (1985). The best existing rules for cylindrical shell buckling, assessed according to
the breadth of the scope and the authority with which they are underpinned, are probably the ECCS code
(1988) and the new German standard (DIN 18800, 1990). For axial compression, the ECCS code
includes uniform compression, compression combined with global bending, the effect of internal
pressure, and an imperfection tolerance measurement which is, in some way, related to strength. It does
not include collapse near boundary conditions, longitudinal stress gradients, local supports, local stress
concentrations, explicit relationships between imperfection amplitude and strength, or many other factors.
The basic loading condition is essentially that of a uniformly stressed cylinder. Whilst this may be the
worst case, it is not satisfactory to justify long-established practical construction in current use. The
German standard is very similar in coverage, but omits the important imperfection-reducing effect of
internal pressure. It has achieved some notable advances in integration and scope, and it has made
important progress in addressing non-uniform stress states. The notation used here follows that of the
Eurocode, which is aligned with other structural Eurocodes.

0.9

Dimensionless axial stress Rk/Rc

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0

500

1000

1500

2000

2500

3000

3500

Radius to thickness ratio r/t

Fig. 1 Experimental buckling strengths of unpressurised cylinders


Buckling of Cylinders Under Uniform Axial Compression Alone
Uniform axial compression is the best-known strength condition for a cylindrical shell, and perhaps that
leading to the greatest variability (Fig. 1). The strength is commonly related to the classical elastic
critical stress Rc (Resistance, elastic critical value)
Rc =

E
t
t
. r = 0.605 E r
2
3(1- )

(1)

in which E is Young's modulus, r is the cylinder radius and t is the wall thickness. The strength under
uniform compression is highly sensitive to geometric imperfections, so that the elastic characteristic
resistance Rk is given by
Rk = x Rc

(2)

where x is the knock-down factor for axial compression and may be subscripted to indicate the
conditions (x for unpressurised uniform compression, xpe for elastic buckling with internal pressure
and xpp for plastic instability under high internal pressure). The amplitudes of the imperfections are
such that it is not possible to draw a single strength curve of x against shell slenderness and apply it to
all construction. There is increasing agreement that cylinders should be classed into strength groups
according to the quality of construction, as first proposed by Odland (1980), then by Rotter (1985, 1995)
and now more widely accepted (Poggi, 1996; Kndel and Ummenhofer, 1996). The version now in EC3
follows the proposal of Rotter (1995, 1998). The design value of imperfection (larger than the maximum
permitted value by a factor), is
wk
1
t = Q

r
t

(3)

10

where wk is the amplitude of the largest notional imperfection and Q is the fabrication quality parameter.
For normal construction, Q=16; for high quality Q=25; and for excellent quality, Q=40. The relation
between strength x and imperfection amplitude is given by
x =

0.83
w 1.44
1 + 1.91 tk

for all slendernesses

(4)

The above procedure allows the fabrication tolerance quality to be explicitly included in the design and
incorporated into the contract specification. It is interesting to note that in a recent comprehensive survey
of shell buckling codes, the tolerances defined in different standards were not found to relate to specified
strength assessments at all (Chryssanthopoulos et al, 1996), though this reflects a scatter in the implied
safety margin rather than a lack of mechanical relationship.
Imperfections: Sensitivity, Forms, Amplitudes, Measurement, Characterisation
The strength is sensitive to both the amplitude and form of the imperfections, and there has been a long
debate about the "worst" form of imperfections. The serious imperfection form certainly varies with the
loading to which the shell is subjected, so a different measure is needed depending on the critical design
state. However, a sinusoidal imperfection, as used originally by Koiter (1945, 1963) is clearly the worst,
but is quite impractical. Seeking a practical imperfection which can be used to model buckling
phenomena is not easy, and surveys of real structures in service (Ding, 1992; Coleman et al, 1992) show
that the biggest challenge lies in interpreting measurements into tolerance specifications. A small piece
from the results of Ding's study is shown in Fig. 2.

20

10

Profile number
8000

6000

4000

Vertical coordinate mm

11

10000

-10

Imperfection
mm

Port Kembla Silo B Imperfections: Zone 1, Profiles 1-12

2000

Fig. 2 Measured imperfections in a 24m diameter cylindrical shell


Despite these difficulties, there now appears to be relatively clear agreement (Kndel and Ummenhofer,
1996) that an axisymmetric local imperfection (Fig. 3), first studied by Amazigo and Budiansky (1972),
and later by Bornscheuer et al (1983), Rotter and Teng (1989), Teng and Rotter (1992) and Rotter (1996,
1997b) is the best choice for practical design calculations in relation to full scale structural

11

measurements. The problematical question of the imperfection which is not completely axisymmetric
was recently addressed by Berry and Rotter (1996), who showed that a circumferentially invariant local
defect could model an axisymmetric imperfection if it extended over 40, or about 8 linear bending halfwavelengths. This indicates that an imperfection which has an aspect ratio of about 5 is effectively
axisymmetric. Indeed, the result showed that this locally circumferentially invariant defect was
marginally weaker than the truly axisymmetric defect (Fig. 4).
A huge body of work remains to be done in identifying forms of imperfection to which the strength of a
shell is sensitive under certain loadings, seeking them in maps of real measurements, determining the
amplitude with respect to these forms, identifying the most serious forms from a spectrum of possible
choices and characterising them for use in design or code development calculations. In addition, the
imperfection forms which may be expected to arise from manufacturing processes should be studied.
Some interesting results have recently been found by Holst et al (1997), who examined mis-fits of plates
or shrinkage of welds and deduced the resulting imperfection forms and strengths.
1700

Dimensionless
amplitude wk/t

1650

0.2
1.0

Axial coordinate

1600
1550
1500
1450
1400
1350
1300
999.0

999.2

999.4

999.6

999.8

1000.0

Radial coordinate

Fig. 3 Axisymmetric local imperfection

1000.2

12

Buckling strength / classical elastic strength

0.7

0.6

wk/t = 1.0
0.5

0.4

0.3

0.2
0

60

120

180

240

300

360

Circumferential extent (degrees)

Fig. 4 Strength of a partial axisymmetric imperfection


Buckling of Axially Compressed Cylinders with Internal Pressure
Under internal pressure, the imperfection-sensitivity of the cylinder declines and the strength increases
from the unpressurised value towards the classical elastic critical stress Rc. This important effect is
included in the ECCS (1988) code. Another reason for choosing the axisymmetric imperfection as the
characterising one for pure axial compression is that its strength gain with internal pressure is rather
smaller than for non-symmetric imperfections (Hutchinson, 1965). Different wavelengths and amplitudes
of imperfection were studied by Teng and Rotter (1992), and these, together with additional more recent
calculations have led to the following strength calculation proposal. Rational analyses (Hutchinson,
1965; Calladine, 1983; Rotter and Seide, 1987) indicate that the strength gains with pressure should be
pr
governed, in the elastic range, by the parameter
p=
, but the ECCS (1988) standard uses instead
tRc
p r3/2
= E t . This appears to have arisen from an empirical fit to experimental results. Rotter and Teng

(1989) showed that the ECCS rule is unconservative for some practical imperfections. Following many
further analyses, the following rule is proposed for the elastic strength gain due to internal pressure in
Eurocode 3 (Fig. 5). The value of xpe for pressurised cylinders is related to that for unpressurised
cylinders as
xpe = x + (1-x)

(5)
0.3
p +
x

This relation includes the changing form of the relationship with the value of the initial strength x.

Axial stress / classical elastic critical stress

13

1.0
0.9
0.8
0.7

w/t=0.2

0.6

Eqn. 5
0.5

w/t=0.5

0.4

Eqn. 5

0.3

w/t=1.0
Eqn. 5

0.2

w/t=2.0

Type A weld depression

0.1

Eqn. 5

0.0
0

0.5

1.5

2.5

Dimensionless internal pressure pr/tRc

Fig. 5 Increase in elastic buckling strength with internal pressure


Cylinders under High Internal Pressure: Elephant's Foot
Under high internal pressures, local collapse of the shell can occur. This is best known at a base
boundary condition, where is has been often observed in tanks which have been subjected to earthquake
loading, but it can also occur on the cylinder wall where a ring, thickness change or local imperfection
occurs. The stress levels associated with this effect are considerably lower than the von Mises membrane
yield envelope, as a result of the effect of amplified displacements near the disturbance. The first
scientific study of this problem appears to have been that of Rotter (1990). In that study, it was assumed
that the local collapse and elastic buckling had little interaction, without proper justification. However,
this has now been shown to be substantially true, by considering the local collapse at an axisymmetric
weld depression (Fig. 6). The strength at an "elephant's foot" failure at the boundary condition is
somewhat lower than internal failures at local imperfections, but much more work is needed to establish a
less conservative evaluation. The standard has adopted the approximate relationship

xpp = 1

2
2
p 2
1
s + 1.21. x
3/2
4 1
1.12 + s s(s + 1)
x

(6)

in which s = (1/400)(r/t) and


x2 = fY/Rc, where fY is the yield stress.
Whilst the above expression contains some empiricism, other methods of dealing with this type of failure
(ECCS, 1988; Bandyopadhyay et al., 1995) involve complicated iterative calculations with no better
foundation in rational mechanics.
Where the shell is subjected to global bending, the ductility of this failure mode permits higher stresses to
be achieved. The range of strength gains has been recently assessed and appears to lie in the range 30 to
50%, so further work on elephant's foot failures under combined axial compression and global bending
would be useful.

14

Dimensionless axial buckling stress Rk/Rc

1
0.9
pb

0.8

pc

0.7

eb

pc

0.6

pb

eb

0.5
0.4
0.3

1/2 wall thickness amplitude


1 wall thickness amplitude

0.2

Elephant's foot

0.1

von Mises

0
0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Dimensionless internal pressure pr/tfy

1.0
0.8
0.6
0.4
0.2

0.02
0.06
0.1

0.0

Dimensionless column 0.14


width d/r
0.18
0.2

0.4

0.6

0.8

1.2

1.4

Dimensionless buckling stress


Rk / fy

Fig. 6 Elastic plastic interaction of pressurised cylinders at a local weld depression

1.6

Dimensionless shell slenderness

Fig. 7 Elastic plastic buckling strength above a local support (after Guggenberger, 1996)

15

Locally Supported Cylinders


The limited rules currently available for axial compression cannot be applied with any rationality to
conditions near local supports. Early attempts to produce empirical rules (Gorenc, 1985; Samuelson,
1987) made radical assumptions about both the stress distribution and the conditions which might lead to
buckling. Much research work on this topic has been undertaken in recent years (Teng and Rotter, 1990;
Guggenberger, 1992; Guggenberger et al, 1995), and the most recent work from Guggenberger (1996)
indicates that a very thoroughly researched formulation should be available for inclusion in the standard.
The behaviour must be divided between narrow supports, in which the plastic failure is effectively forcelimited, and wider supports where the mean axial stress above the support is related to the classical elastic
critical stress (Fig. 7). The behaviour is imperfection-sensitive, but an axisymmetric imperfection and a
local imperfection have very similar effects. This work has not yet been adopted into the standard, but is
expected to be developed fully during the period of trial usage.
Residual Stresses and Fabrication Methods
Residual stresses in steel beams and columns are generally regarded as uniformly deleterious to strength.
This is because the stresses are one-dimensional, induce early yield and so reduce the local stiffness.
However, in shells, the stress state is two-dimensional, and the effect of realistic residual stresses is less
clear. The residual stresses depend on the fabrication technique, which also produces geometric
imperfections. As the latter have a marked effect on shell strength, it is more difficult to determine the
effect of the former experimentally.

Dimensionless buckling stress, x = Rk/Rc

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3

Local expansion with residual stress retained


Local expansion with residual stress removed
Local shrinkage with residual stress retained
Local shrinkage with residual stress removed

0.2
0.1
0
0

Dimensionless imperfection amplitude, |wk/t|


Fig. 8 Effect of residual stresses on shell buckling strength
Theoretical investigations of the problem are not straightforward either. First, it is necessary to devise a
consistent residual pattern which is everywhere in equilibrium (a non-trivial task), next it is necessary to
investigate the geometric imperfections which are associated with the residual stress, which will have
their own separate effect on the strength, and finally a nonlinear shell buckling analysis is needed. The

16

effect of the residual stresses can only be determined by comparing two shells with the same geometric
imperfections, one with the consistent residual stress field and one without (Fig. 8). Very few such
rigorous studies exist to date (Rotter, 1988; Holst et al, 1996; Rotter, 1996; Guggenberger, 1996), but the
general conclusion is that residual stresses are generally, but not quite always, beneficial to shell strength.
Patch Loading on Silo Walls: Notional Symmetry
Silos are subjected to a wide variety of different loading conditions, but the assumption is normally made
in design that the stresses are axisymmetric. Unfortunately this is often not the case even in extremely
carefully arranged symmetrical tests (Ooi et al, 1990, Zhong and Rotter, 1996). The commonest cause of
failure under symmetrical conditions may well be unsymmetrical normal pressures on the wall during the
discharge of solids, which can lead to high axial compressive membrane stresses (Fig. 9), but with a
relatively local range. The condition depends strongly on the flow pattern, but unsymmetrical semi-mass
flow (Fig. 10) is a primary culprit. This condition is addressed within Eurocode 1 Part 4 (1995) by use of
a patch of local pressure (Fig. 10), but the calibration of the patch for steel silos has been very weak,
mainly because the task is very great and the phenomenon poorly understood. One of the major
difficulties in proposing a rational treatment of the loads is that there is no shell buckling criterion against
which to assess the predicted stresses. The German code (DIN 18800, 1990) does address non-uniform
stress states, but is so conservative that many designers may conclude that their designs are
uncompetitive and a poor match with existing practice. The most plausible current treatment is that of
Rotter (1986), but this currently lacks comprehensive investigation. Extensive work on the effect of local
high stresses, including the size of the zone over which they are high and the effect of different forms and
amplitudes of imperfections is needed to provide a high quality codification of these phenomena.
TEST M1CCA

Level F

10

F0

Vertical stress in wall (MPa)

Vertical stress
F10

F20
6
F30
4

F45
F60

F90

F135
-2

Filling

Discharge

Storing

F270

-4

F315
1

10

11

12
F340

Time (hours)

Fig. 9 Axial compressive stresses in the wall Fig. 10Unsymmetrical semi-mass flow in a silo
of a test silo during filling, storing and
and patch load representation
discharge
Silo Eccentric Discharge Pressure Regimes
The most notorious cause of silo failures is eccentric discharge. Here, an outlet at the base of the silo is
unsymmetrically placed relative to the circular silo structure (for good operational reasons), and where
this eccentricity is large, a flowing channel of solid often develops against the silo wall. It leads to much
reduced normal pressures locally, which in turn induce high meridional membrane stresses in the wall
and cause buckling failures. Some of the buckles are local and benign, whilst others are catastrophic.
The need to distinguish these two types in design was identified by Greiner (1997), but progress on
defining the buckling strength has been slow. The pressure regime is now becoming quite well defined
(Rotter et al, 1995), though is not yet rationally predictable. For design, the method of Rotter (1986)
appears to have gained wide acceptance.

17

R
R

int

R
Fig. 11 Selection of values of stress resultants for use in interaction formulas
Non-uniform Stress States in Cylinders
The above cases of silo loading and discrete support conditions are special cases of non-uniform stress
states in cylinders. The best codified treatment of the problem at the present time is DIN 18800 (1990),
but the method adopted is so conservative that it is unlikely to be adopted widely unless the structure is
particularly important. The method (Fig. 11) involves the combining of different stress resultants (axial,
circumferential, shear) at points which can be very widely separated in the cylinder into a single buckling
interaction formula. No account is taken of the area of shell wall over which the high stresses extend.
The result is that stresses which are quite unrelated to each other are treated as combining to initiate a
buckling failure, leading to very conservative strength assessments. However, since no better general
method could be found, this procedure has been adopted into the draft European shells standard.

1.6

Dimensionless axial load x/Rc

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0
0.0

0.5

1.0

1.5

2.0

2.5

Dimensionless displacement u/t

Fig. 12 Stress redistribution may permit post-buckling stiffening and repeated bifurcations

18

Local Buckling and Post-buckling Paths


The traditional image of shell buckling is that it is generally catastrophic in nature, because the postbuckling path often descends steeply. Whilst this is often true for states of uniform compression
throughout a large part of a cylinder, it is progressively less true when the buckle is local and the shell
can re-distribute the stresses into adjacent parts which were initially less highly stressed (Fig. 12). Thus,
the area of highly stressed shell, the slope of the post-buckling path and the imperfection sensitivity all
need to participate in the assessment if benign "dents" of minor significance are to be distinguished from
potentially catastrophic buckles. The safety index which should be applied to each form of buckling
failure needs some careful assessment, and there may be cases in which buckles should be treated as an
acceptable feature of the structure in operation (Greiner, 1997). There is much to learn about structural
behaviour in this area, but a philosophical advance is also needed to permit sensible new rules to be
formulated.
This section has reviewed a number of problems in the buckling of cylindrical shells, as an indication of
the range of unsolved technical problems which still await a solution.

CONCLUSIONS
Shell structures present some of the most complicated mechanics which the structural engineer must face.
As a result, conditions seen in other structural forms are almost certain to exist somewhere in shells. The
variety of shell responses means that some development of the philosophy of design is needed in a
standard which addresses shell design thoroughly.
This paper has outlined the current development of European standards for shell buckling and the
distinction being made between generic and applications codes. The rationale behind these standards has
been defined and some issues have been noted which are currently under debate. The development of the
generic standard on shell strength and stability has been discussed, and its range and structure outlined.
As an example of the material which must appear within the standard on shells, the axially compressed
cylinder has been discussed. Many different problems have been outlined, and the current state of
knowledge on each noted. Some have been the subject of extended research for many years, but others
need much new research.

ACKNOWLEDGEMENTS
The author wishes to thank the UK EPSRC for research grant GR/H 41027 and to the European
Concerted Action on Silos research (CA-Silo) under both of which some of the work was undertaken.
Special thanks are given to Richard Greiner, who generously undertook many hard days debating shell
buckling and codes, and to Herbert Schmidt, whose tireless efforts and generous attitude has made the
development of the draft standard progress much more smoothly than might have been expected.

19

REFERENCES
Amazigo, J.C. and Budiansky, B. (1972) "Asymptotic Formulas for the Buckling Stresses of Axially
Compressed Cylinders with Localised or Random Axisymmetric Imperfections", Journal Applied
Mechanics, ASME, E-39, 179-184.
Bandyopadhyay, K., Cornell, A, Costantino, C., Kennedy, R., Miller, C. and Veletsos, A. (1995)
"Seismic Design and Evaluation Guidelines for the Department of Energy High-level Waste Storage
Tanks and Appurtenances", Dept Advanced Tech, Brookhaven Nat. Lab., Associated Universities Inc.,
October, 350pp.
Baker, E.H., Kovalevsky, L. and Rish, F.L. (1972) Structural Analysis of Shells, McGraw Hill, New
York.
Berry, P.A. and Rotter, J.M. (1996) "Partial axisymmetric imperfections and their effect on the buckling
strength of axially compressed cylinders", Proc. Int. Workshop: Imperfections in Metal Silos, CA-Silo,
35-48.
BMHB (1987) "Silos: Draft Design Code for Silos, Bins, Bunkers and Hoppers", British Materials
Handling Board and British Standards Institution, London.
Bornscheuer, F.W., Hafner, L. and Ramm, E., (1983) "Zur Stabilitat eines Kreiszylinders mit einer
Rundschweissnaht unter Axialbelastung", Der Stahlbau, Vol. 52, Heft 10, pp. 313-318.
BS 5500 (1996) "Unfired Fusion Welded Pressure Vessels", British Standards Institution, London.
Bushnell, D. (1972) "Stress, Stability and Vibration of Complex Branched Shells of Revolution: Analysis
and Users Manual for BOSOR4", NASA CR-2116, 1972.
Calladine, C.R. (1983) Theory of Shell Structures, Cambridge University Press.
Chryssanthopoulos, M., Spagnoli, A., Burgan, B.A. and Awad, S. (1996) "Imperfection tolerances in
shells due to manufacturing processes", SCI Doc RT523, Final report to DTI, Imperial College, March,
51pp.
Coleman, R., Ding, X.L. and Rotter, J.M. (1992) "The Measurement of Imperfections in Full-Scale Steel
Silos", Proc., 4th Int Conf Bulk Matls Storage Handling and Transptn, IEAust., Wollongong, June 1992,
pp 467-472.
CRCJ (1971) "Handbook of Structural Stability", Column Research Council of Japan, Corona, Tokyo.
DIN 18800 (1990) "Stahlbauten: stabilittsflle, schalenbeulen", DIN 18800 Part 4, Deutsches Institut fr
Normung, Berlin, November.
Ding, X.L. (1992) "Precise Engineering Surveying - Application to the Measurement of Large Scale Steel
Silos", PhD Thesis, University of Sydney, Australia, July, 252 pp.
EC1 Part 4 (1995) Eurocode 1: Basis of design and actions on structures Part 4: Actions on silos and
tanks, ENV 1991-4, draft, CEN, Brussels.
EC3 (1994) Eurocode 3: Design of steel structures, ENV 1993, draft, CEN, Brussels.

20

EC3 Part 1.6 (1998) Eurocode 3: Design of steel structures, Part 1.6: General rules - Supplementary rules
for the strength and stability of shell structures, draft standard, ENV 1993-1-6, CEN, Brussels.
EC3 Part 1.7 (1998) Eurocode 3: Design of steel structures, Part 1.7: General rules - Supplementary rules
for Planar Plated Structures Loaded Transversely, draft standard, ENV 1993-1-7, CEN, Brussels.
EC 3 Part 4.1 (1998) Eurocode 3: Design of steel structures, Part 4.1: Silos, ENV 1993-4-1, draft
standard, CEN, Brussels.
EC 3 Part 4.2 (1998) Eurocode 3: Design of steel structures, Part 4.2: Tanks, ENV 1993-4-2, draft
standard, CEN, Brussels.
EC 3 Part 4.3 (1998) Eurocode 3: Design of steel structures, Part 4.3: Pipelines, ENV 1993-4-3, draft
standard, CEN, Brussels.
ECCS (1988) European Recommendations for Steel Construction: Buckling of Shells, 4th edition,
European Convention for Constructional Steelwork, Brussels.
ESDU (1974) Engineering Sciences Data Unit, Design Sheets, London, July.
Gaylord, E.H. and Gaylord, C.N. (1984) Design of Steel Bins for Storage of Bulk Solids, Prentice Hall,
Englewood Cliffs, NJ.
Gorenc, B.E. (1985) "Design of supports for steel bins", in Design of Steel Bins for the Storage of Bulk
Solids, Ed. J.M. Rotter, Univ. Sydney, March 1985, pp 184-188.
Greiner, R. (1997) "A concept for the classification of steel containments due to safety considerations",
Containment Structures: Risk, safety and reliability, (Ed. B. Simpson) E & FN Spon, London, pp 65-76.
Greiner, R. and Wallner, S. (1997) "The new part Strength of Shells of Eurocode 3.1", Proc., Int. Conf.
on Carrying Capacity of Steel Shell Structures, Brno, 1-3 October 1997, pp 119-125.
Guggenberger, W. (1992) "Nichtlineares Beulverhalten von Kreiszylindrischalen unter lokaler
Axialbelastung", Dr.techn dissertation, Institut fr Stahlbau, Technische Universitt Graz, Austria.
Guggenberger, W. (1996) "Proposal for Design Rules of Axially Loaded Steel Cylinders on Local
Supports", Proc., International Conference on Structural Steelwork, Hong Kong, December 1996, pp
1225-1230.
Guggenberger, W., Greiner, R. and Rotter, J.M. (1998) "The Behaviour of Locally-Supported Cylindrical
Shells: Unstiffened Shells" to appear in Journal of Constructional Steel Research.
Hampe, E. (1991) "Silos: Band 1 Grundlagen: Band 2 Bauwerke", Verlag fr Bauwesen GmbH, Berlin
1991, Band 1 256pp, Band 2 252pp.
Holst, J.M.F.G., Rotter, J.M. and Calladine, C.R. (1997) "Characteristic imperfection forms for cylinders
derived from misfit calculations", Proc., Int. Conf. on Carrying Capacity of Steel Shell Structures, Brno,
1-3 October 1997, pp 333-339.

21

Hutchinson, J.W. (1965) "Axial Buckling of Pressurised Imperfect Cylindrical Shells", AIAA J., 3:8
1461-1466
JIS (1987) "Construction of Welded Aluminium and Aluminium Alloy Cylindrical Silos", Japanese
Industrial Standard B 8511, translated and published by Japanese Standards Association, Tokyo, Japan,
Edition 1, 32 pp.
Koiter, W.T. (1945) "On the Stability of Elastic Equilibrium", (in Dutch) PhD Thesis, Delft University,
(see also Translation AFFDL-TR-70-25 Wright Patterson Air Force Base, 1970).
Koiter, W.T. (1963) "The Effect of Axisymmetric Imperfections on the Buckling of Cylindrical Shells
under Axial Compression", Koninklike Nederlandische Akademie van Wetenshappen, Proc. Ser. B66, pp
265-279.
Kndel, P. and Ummenhofer, T. (1996) "Substitute imperfections for the prediction of buckling loads in
shell design", Proc. Int. Workshop on Imperfections in Metal Silos, CA-Silo, Lyon, France, 19 April, pp
87-102.
Martens, P. (ed) (1988) "Silo-Handbuch", Ernst und Sohn, Berlin.
NASA (1968) Buckling of Thin-Walled Circular Cylinders, NASA SP8007.
Odland, J. (1978), "Buckling Resistance of Unstiffened and Stiffened Circular Cylindrical Shell
Structures", Norwegian Maritime Research, 6:3, 2-22.
Ooi, J.Y., Rotter, J.M. and Pham, L. (1990) "Systematic and Random Features of Measured Pressures on
Full-Scale Silo Walls", Engineering Structures, Vol. 12, No. 2, April, pp 74-87.
Poggi, C. (1996) "Bifurcation and limit points in unstiffened cones under axial compression", Proc. Int.
Workshop on Imperfections in Metal Silos, CA-Silo, Lyon, France, 19 April, pp 141-152.
Rotter, J.M. (1985) "Buckling of Ground-Supported Cylindrical Steel Bins under Vertical Compressive
Wall Loads", Proc., Metal Structures Conference, Institution of Engineers Australia, Melbourne, May, pp
112-127.
Rotter, J.M. (1986) "The Analysis of Steel Bins Subject to Eccentric Discharge", Proc., 2nd Int Conf
Bulk Matls Storage Handling & Trptn, IEAust., Wollongong, July, pp 264-271.
Rotter, J.M. (1988) "Calculated Buckling Strengths for the Cylindrical Wall of 10 000 tonne Silos at Port
Kembla", Inv. Rpt S663, School of Civil and Mining Engineering, University of Sydney, Australia.
Rotter, J.M. (1990) "Local Inelastic Collapse of Pressurised Thin Cylindrical Steel Shells under Axial
Compression", Journal of Structural Engineering, ASCE, 116:7, 1955-1970.
Rotter, J.M. (1995) "Proposal for cylinder axial compression buckling strength to be described in terms of
amplitude of imperfection", ECCS TWG8.4 and CEN TC250/SC3 working paper, May, 20 pp.
Rotter, J.M. (1996) "Elastic plastic buckling and collapse in internally pressurised axially compressed silo
cylinders with measured axisymmetric imperfections", Proc. International Workshop on Imperfections in
Metal Silos: Measurement, Characterisation and Strength Analysis, CA-Silo, Lyon, France, 19 April, pp
119-140.

22

Rotter, J.M. (1997a) "Challenges for the future in the design of bulk solid storages", Containment
Structures: Risk, safety and reliability, (Ed. B. Simpson) E & FN Spon, London, pp 11-34.
Rotter, J.M. (1997b) "Pressurised axially compressed cylinders", Proc., Int. Conf. on Carrying Capacity
of Steel Shell Structures, Brno, 1-3 October 1997, pp 354-360.
Rotter, J.M. (1998) "Development of proposed European design rules for buckling of axially compressed
cylinders", Advances in Structural Engineering, (to appear).
Rotter, J.M. and Seide, P. (1987) "On the Design of Unstiffened Cylindrical Shells Subject to Axial Load
and Internal Pressure", Proc., International Colloquium on Stability of Plate and Shell Structures, Ghent,
April, pp 539-548.
Rotter, J.M. and Teng, J.G. (1989) "Elastic Stability of Cylindrical Shells with Weld Depressions",
Journal of Structural Engineering, American Society of Civil Engineers, Vol 115, No. 5, May, pp 12441263.
Rotter, J.M., Ooi, J.Y., Chen, J.F., Tiley, P.J., Mackintosh, I. and Bennett, F.R. (1995) Flow Pattern
Measurement in Full Scale Silos, British Materials Handling Board, Ascot, UK, 230pp.
Samuelson, L.A. (1987) "Design of Cylindrical Shells subjected to Local Loads in Combination with
Axial or Radial Pressure", Proc. Int. Coll. on Stability of Plate and Shell Structures, Gent, Belgium,
ECCS, pp 589-596.
She, K.M. and Rotter, J.M. (1993) "Nonlinear and Stability Behaviour of Discretely-Supported
Cylinders" Research Report R93-01, Dept of Civil Engrg, University of Edinburgh, March.
Teng, J.G. and Rotter, J.M. (1989) "A Study of Buckling in Column-Supported Cylinders", in Contact
Loading and Local Effects in Thin-Walled Plated and Shell Structures, eds V. Krupka and M. Drdacky,
Proc., IUTAM Symposium, Preliminary Report 1990, Academia Press, Prague, 1992, pp. 52-61.
Teng, J.G. and Rotter, J.M. (1992) "Buckling of Pressurized Axisymmetrically Imperfect Cylinders under
Axial Loads", Journal of Engineering Mechanics, ASCE, Vol. 118, No. 2, Feb., pp 229-247.
Trahair, N.S., Abel, A., Ansourian, P., Irvine, H.M. and Rotter, J.M. (1983) Structural Design of Steel
Bins for Bulk Solids, Australian Institute of Steel Construction, Sydney, Australia.
Zhong, Z. and Rotter, J.M. (1996) The sensitivity of silo flow and wall pressures to filling method,
Proc., 12th Int Congr. Chem. & Process Engrg, CHISA '96, Prague, August, 10 pp.
Professor J. Michael Rotter
Professor of Civil Engineering and Director of the Division of Engineering
Department of Civil and Environmental Engineering
Division of Engineering
University of Edinburgh
The Kings Buildings
Edinburgh EH9 3AX, UK
Email: M.Rotter@ed.ac.uk

Você também pode gostar