You are on page 1of 564

1

Introduction
Page
1.1
DBA
Design by Analysis
1 Introduction
1.1 Background
Pressure vessel design has been historically based on Design By Formula. Standard vessel
configurations are sized using a series of simple formulae and charts. In addition to the Design by
Formula route, many national codes and standards for pressure vessel and boiler design do provide
for a Design By Analysis (DBA) route, where the admissibility of a design is checked, or proven,
via a detailed investigation of the structure's behaviour under the external loads (or actions) to be
considered. Nevertheless Design By Formula remains the dominant approach. In an increasingly
technically sophisticated society, it may be asked why this should be the case?
All these DBA routes in the major codes and standards in the pressure equipment field are based on
the rules first proposed in the ASME Pressure Vessel and Boiler Code, which was formulated in the
late 1950s before being released, originally for nuclear applications, in 1964. All these routes lead
to the same well-known problems, especially the stress categorisation problem
[1-6]
, and all are out-
of-step with the continuing development of computer hardware and software. Further, all are
focused on pressure, and possibly, and to a limited extent, temperature, treating other actions in an
inflexible manner, giving them marginal attention only.
The DBA route in the proposal of CEN's unfired pressure vessel standard prEN 13445-3 tries to
avoid these problems:
1. by addressing the failure modes directly
2. by allowing for non-linear constitutive models
3. by applying a multiple safety factor format for the incorporation of actions other than pressure
4. by specifying mainly the principal technical goals of the standard together with some
application rules as possible methods for the fulfilment of these goals.
In the new proposal of a European Standard, two documents are included concerning design by
analysis:
Document prEN 13445-3, Annex B, Direct Route for Design by Analysis
Document prEN 13445-3, Annex C, Stress Categorisation Route for Design by Analysis
For various reasons SG-DC of the Working Group (WGC) of the CEN Technical Committee TC54
decided to use in the new European Standard an approach similar to the one used in Eurocodes (for
steel structures), using the notions of principles and application rules as well as the notion of partial
safety factors.
One reason is that the DBA-approach is flexible and simplifies the incorporation of constructional
requirements (wind, snow, earthquake, etc.), if required, in a consistent manner. Another reason is,
that there has been considerable criticism of the ASME stress classification (or categorisation)
method, which is used in principle in almost all countries:
One solution to the annoying problem of stress classification is to apply limit analysis, as proposed
in the rules for DBA. Limit analysis does not require categorisation into primary and secondary
2
Introduction
Page
1.2
DBA
Design by Analysis
stresses and it gives a unique result (which stress categorisation in general does not). The
calculations can be made using existing software, but no doubt special software could be readily
developed if there were sufficient demand. Nevertheless, part of the usual stress categorisation
approach is included in the Standard, as an application rule.
The DBA route is included in the new European Standard
as a complement to Design by Formula (DBF) for cases not covered there
as a complement for cases requiring superposition of environmental actions - wind, snow,
earthquake, etc.
as a complement for fitness-for-purpose cases where (quality related) manufacturing tolerances
are exceeded
as a complement for cases where local authorities require detailed investigations, e. g. in major
hazard situations, for environmental protection reasons
as an alternative to the Design by Formula route.
For the time being, this route is restricted to sufficiently ductile steels and steel castings with
calculation temperatures below the creep range.
The main concepts are dealt with here in detail, because
it is a real alternative to DBF, as stated above, with many advantages
many concepts are new in pressure equipment design
it may be used as a yard-stick for DBF solutions, to show possible improvements
some concepts have already influenced the DBF-section, their discussion will shed light onto
some DBF-details
it may lead to an improved design philosophy by indicating more clearly the critical failure
modes, especially of importance for in-service inspections.

1.2 Aims

From the point of view of an analyst or designer, the rules in the new European Standard are quite
general, and in fact as mentioned above this is intentional. In broad terms, in the context of the
Direct Route either an admissibility check, or a check on maximum allowable load, has to prepared
on the basis on either detailed elastic-plastic finite element analysis or some method of estimating
plastic failure loads for gross and progressive plastic deformation. In principle this seems
straightforward, but in practice can be difficult. The aim of this study has been to provide guidelines
on the application of elastic-plastic analysis (in its broadest sense) to the Standard and in doing so to
highlight possible problem areas and suggest methods of resolving these. This has been achieved
using a new collection of ten benchmark problems. These example problems have been chosen to
be typical of cases where design by formula cannot be used. A substantial part of this document
provides detailed, step-by-step, studies of each of the example problems.

This study has been undertaken by experts either in the research and development of design by
analysis itself or in its practical use. This expertise is apparent in the review of the current state-of-
the-art of pressure vessel design by analysis in Section 2, which highlights unexpected, but now
3
Introduction
Page
1.3
DBA
Design by Analysis
well-known, problem areas, and in the detailed description of the analysis procedures and their
application to the Standard in Sections 3 and 7. The solutions reported in Section 7 were carried out
independently, although unusual results were re-checked.

It is intended that this document should be read in conjunction with the Standard, and can be looked
on as a supplement. Particular emphasis is placed on the expected readership, and the expertise and
knowledge required of them. While the Standard itself is fairly simple and transparent, the writers
have also been aware that the current state-of-the-art in finite element analysis technology, and the
expected continuing increase in sophistication, renders elastic-plastic analysis ever more routine.
However the issue here is whether or not the analyst/ designer understands the underlying
mechanics. Many users of elastic-plastic finite element analysis are unaware of the assumptions and
approximations of the Classical Theory of Plasticity, which are embodied in most finite element
software. They generally do not recognise the implications of the neglect of the Bauschinger effect
and hysteresis, the assumptions concerning yield in compression in general, or that the basic
mathematical models of initial and subsequent yield are approximations which are valid in some
situations but not in others. At a more basic level, very few analysts are even aware of the
fundamental assumptions of the engineering yield stress itself, for example it is measured from a
tensile test and arbitrarily used as a reference to develop multiaxial yield criteria. Further, plasticity
in metals is a shear mechanism, yet we use yield measured in tension rather than torsion and the
measured value can be difficult to identify and is usually subjective.

An overview of the contents of this study is given in the next sub-section, followed by some
additional comments on the expected readership and recommendations of how the document should
be used.


1.3 Overview

This document is divided into nine Sections, including this Introduction:

Section 2 provides an overview of the current state of design by analysis, as typified by the ASME
Pressure Vessel and Boiler Code. The ASME Code offers two routes to design by analysis, the so-
called elastic route and an inelastic route which requires the calculation of limit and shakedown
loads these are briefly summarised, together with definitions of basic terminology. Following this
a short discussion of the most common method of analysis, using finite element techniques, is
provided. This is not intended as an introduction to the finite element method applied to pressure
vessels, rather several issues related to choice of element type are raised since they have
implications for code interpretation specifically the two main problem areas of the elastic route:
linearisation and categorisation. These problem areas are then discussed in some detail, to give the
reader an insight into the nature of major difficulties in application of what seems a fairly simple
and straightforward set of design by analysis rules. Following this discussion, application problems
with the inelastic route are then examined, in particular the difficulty of extracting meaningful
plastic design loads from elastic-plastic finite element analysis. This Section then concludes with an
introduction to the novel features of the new European standard in relation to design by analysis.

In Section 3 a description of the various procedures used in this document to satisfy the analysis
requirements is given. Some detail is provided on using the results of elastic-plastic analysis in the
Direct Route for the checks on both gross plastic deformation and progressive deformation. In the
4
Introduction
Page
1.4
DBA
Design by Analysis
case of the latter, problems with estimating shakedown loads when shell elements are used, or when
there are stress singularities are discussed in some detail. The use of deviatoric maps to assist the
shakedown analysis is also described. As an alternative to elastic-plastic analysis a new technique
for directly estimating limit and shakedown loads from elastic finite element analysis alone is also
used. This technique the elastic compensation method is briefly described in the context of the
requirements of the Standard. Also, the treatment of shell elements is discussed. This Section also
reviews various other issues related to the practical use of the Standard in particular wind action,
the stress categorisation route and checks against fatigue and instability.

In Section 4 a simple example a circular plate is used to describe and discuss each step in the
application of the Standard before proceeding to the main examples examined in this document.

Sections 5, 6 and 7 contain the main body of this study the detailed application of the European
Standard to ten benchmark problems. Section 5 gives a specification of each example, followed by
a summary of the results of the analysis and application of the Standard in Section 6. Section 7
provides the detailed results for each benchmark problem using the analysis procedures described in
Section 3.

Finally, in Section 8, recommendations and concluding remarks are given. This covers comments
on the appropriateness and difficulties with the methodology, software requirements, expertise and
knowledge expected by the analyst and various warnings. For example, it is apparent that the
fatigue rules which are used for both design by formula and design by analysis need special
care.

Appended to the report are various Annexes, specifically a bibliography, analysis input files (where
appropriate) and excerpts of the Standard.

1.4 How to read this document
This document is not aimed at the complete novice, but two broad types of reader are envisaged. It
is presumed that anyone starting to read this has a basic familiarity with the concepts of plasticity
theory and the behaviour of structures under plastic strain. In addition, familiarity with the practice
of elastic finite element analysis for pressurised components, preferably with basic experience of
elastic-plastic analysis is suggested. Also it is recommended that the reader should read the
European Standard in some detail beforehand, if necessary. It is then envisioned that the reader will
either already be broadly familiar with pressure vessel design by analysis and elastic-plastic finite
element analysis and is comfortable with the Standard (whom we will call the Expert), or has read
the Standard and has some basic experience of elastic design by analysis (whom we will call the
Novice).
In the case of the Expert, it is anticipated that this reader will begin with Section 5, the specification
of the examples, followed by Section 6, the analysis summary and then initially carry out his own
analysis and code check. It is possible that some reference will have to be made to Section 3 on
procedures if substantial variation from the results reported here are obtained, or if details on
application of the Standard need to be clarified.
In the case of the Novice, it is expected that more or less the whole document will be carefully read,
from Section 2 through to 7 before carrying out his own analyses. (Of course only a few of the
5
Introduction
Page
1.5
DBA
Design by Analysis
benchmarks may be read in detail so that the Novice may test his understanding of the basic
principles and procedures on the remainder).
Finally it is also expected that engineering managers may wish to review Section 8, which deals
with recommendations in particular the discussion of assumed expertise on the part of the
analyst/designer.
1.5 Literature
[1] J. L. Hechmer & G. L. Hollinger, "Considerations in the calculations of the primary plus
secondary stress intensity range for Code stress classification," "Codes & Standards and
Applications for Design and Analysis of Pressure Vessel and Piping Components" Ed. R. Seshardi,
ASME PVP Vol. 136,1988.
[2] A. Kalnins & D. P. Updike, "Role of plastic limit and elastic plastic analyses in design", ASME
PVP-Vol. 210-2 Codes and Standards and Applications for Design and Analysis of Pressure Vessel
& Piping Components, Ed. R. Seshardi & J. T. Boyle, 1991.
[3] A. Kalnins & D. P. Updike, "Primary stress limits ion the basis of plasticity", ASME PVP-Vol.
230, Stress Classification, Robust Methods and Elevated Temperature Design, Ed. R. Seshardi & D.
L. Marriott, 1992.
[4] A. Kalnins, D. P. Updike & J. L. Hechmer, "On Primary Stress in Reducers", ASME PVP-Vol.
210-2, pp. 117-124
[5] D. Mackenzie & J. T. Boyle, "Stress Classification: A Way Forward", IMechE Presentation
5.5.92
[6] T.P. Pastor & J.L. Hechmer: ASME task group report on primary stress Proc. ASME PVP
Conf., 1994, Minneapolis, 277, 67-78.
1
Design by Analysis
Page
2.1
DBA
Design by Analysis
2. Design by Analysis
The aim of this section is to summarise issues related to the current use of design by analysis in
order to put the new European rules in context. The concept of design by analysis was first
formulated in the US ASME Pressure Vessel and Boiler Code in the early 1960s; with almost forty
years of use various critical problem areas have arisen, most of which have been addressed in the
new European rules. These problem areas are discussed in the following since they highlight
implicit difficulties with an apparently simple and straightforward set of design rules. In the
following the approach devised by ASME is briefly summarised, followed by a description of the
usual methods by which the rules are implemented and a discussion of the problem areas which
arise. After this the differences in implementation of design by analysis rules in the European
Standard are described.
2.1 Design by analysis: the current Stress Categarisation route
The design by analysis procedure is intended to guard against eight possible pressure vessel failure
modes by performing a detailed stress analysis of the vessel. The failure modes considered are:
1. Excessive elastic deformation including elastic instability.
2. Excessive plastic deformation.
3. Brittle fracture.
4. Stress rupture/creep deformation (inelastic).
5. Plastic instability - incremental collapse.
6. High strain - low cycle fatigue.
7. Stress corrosion.
8. Corrosion fatigue.
Most of the design by analysis guidelines given in the codes relates to design based on elastic
analysis this is the so-called elastic route. Essentially it was recognised when the rules were being
developed that only elastic stress analysis was feasible. In the 1960s, most designers were restricted
to linear elastic stress analysis, and in the case of pressure vessel design most analysis was defined
in terms of elastic shell discontinuity theory (also known as the influence function method). The
nature of elastic shell analysis impinges significantly upon the way the above failure modes are
treated in the Code. Thus, rules were developed to help the designer guard against the various
failure mechanisms using elastic analysis alone. These guidelines guard against three specific
failure modes - gross plastic deformation, incremental plastic collapse (ratchetting) and fatigue.
These failure modes are precluded by failure criteria based on limit theory, shakedown theory and
fatigue theory respectively. It is essential to appreciate at the beginning, the excessive plastic
deformation and incremental plastic collapse cannot be dealt with simply in an elastic analysis, as
the failure mechanism is inelastic. In addition, the type of loading causing the stress can
significantly affect the level of permissible stress. Ideally, these inelastic failure modes should be
assessed by an appropriate analysis which adequately models the mechanism of failure.
In this approach the designer is required to classify the calculated stress into primary, secondary and
peak categories and apply specified allowable stress limits. The magnitude of the allowable values
assigned to the various stress categories reflect the nature of their associated failure mechanisms,
therefore it is essential that the categorisation procedure is performed correctly.
2
Design by Analysis
Page
2.2
DBA
Design by Analysis
Stress categorisation (sometimes, classification) is probably the most difficult aspect of the design
by analysis procedure and, paradoxically, the problem has become more difficult as stress analysis
techniques have improved. When the design by analysis procedure was introduced, the dominant
analysis technique in pressure vessel design was thin shell discontinuity analysis or the influence
function method. This is reflected in the definitions of stress categories given in the Codes, which
are based on the assumption of shell theory stress distributions; membrane and bending stress. It is
therefore difficult to equate the calculated stresses and the code categories unless the design is
based on shell analysis. The various stress categories are described first in the following:
2.1.1 Stress Categories
The object of the elastic analysis is to ensure that the vessel has adequate margins of safety against
three failure modes: gross plastic deformation, ratchetting and fatigue. This is done by defining
three classes or categories of stress, which have different significance when the failure modes are
considered. These three stress categories are assigned different maximum allowable stress values in
the code: the designer is required to decompose the elastic stress field into these three categories
and apply the appropriate stress limits.
The total elastic stress which occurs in the vessel shell is considered to be composed of three
different types of stress primary, secondary and peak. In addition, primary stress has three specific
sub-categories. The ASME stress categories and the symbols used to denote them in the code are
given below;
(1) Primary Stress
General Primary Membrane Stress, P
m
Local Primary Membrane Stress, P
L
Primary Bending Stress, P
b
(2) Secondary Stress, Q
(3) Peak Stress, F
and depend on location, origin and type. Before we can give a proper definition of these stresses, we
must first give some terminology:

Gross Structural Discontinuity: A gross structural discontinuity is a source of stress or strain
intensification that affects a relatively large portion of a structure and has a significant effect on the
overall stress or strain pattern or on the structure as a whole.

Examples of gross structural discontinuities are:
end to shell junctions,
junctions between shells of different diameters or thickness,
nozzles.

Local Structural Discontinuity: A local structural discontinuity is a source of strain intensification
that affects a relatively small volume of material and does not have a significant effect on the
overall stress or strain pattern or on the structure as a whole.

Examples of local structural discontinuities are:
3
Design by Analysis
Page
2.3
DBA
Design by Analysis
small fillet radii,
small attachments,
partial penetration welds.

Normal Stress: The normal stress is the component of stress normal to the plane of reference; this is
also referred to as direct stress.

Usually the distribution of normal stress is not uniform through the thickness of a part, so this stress
is considered to be made up in turn of two components one of which is uniformly distributed and
equal to the average value of stress across the thickness of the section under consideration, and the
other of which varies with the location across the thickness.

Shear Stress: The shear stress is the component of stress acting in the plane of reference.

Membrane Stress: The membrane stress is the component of stress that is uniformly distributed and
equal to the average value of stress across the thickness of the section under consideration.

Bending Stress: The bending stress is the component of stress that varies linearly across the
thickness of section under consideration.

With this terminology as background, we now can define primary, secondary and peak stresses
properly.

Primary Stresses: A primary stress is a stress produced by mechanical loading only and is so
distributed in the structure that no redistribution of load occurs as a result of yielding. It is a normal
stress or a shear stress developed by the imposed loading, that is necessary to satisfy the simple
laws of equilibrium of external and internal forces and moments. The basic characteristic of this
stress is that it is not self-limiting. Primary stresses that considerably exceed the yield strength will
result in failure, or at least in gross distortion. A thermal stress is not classified as a primary stress.
Primary stresses are divided into general and local categories. The local primary stress is defined
hereafter.

Typical examples of general primary stress are:
The average stress in a cylindrical or spherical shell due to internal pressure or to distributed
live loads,
The bending stress of a flat cover without supporting moment at the periphery due to internal
pressure.

Primary Local Membrane Stress: Cases arise in which a membrane stress produced by pressure or
other mechanical loading and associated with a primary together with a discontinuity effect
produces excessive distortion in the transfer of load to other portions of the structure.

Conservatism requires that such a stress be classified as a primary local membrane stress even
though it has some characteristics of a secondary stress. A stressed region may be considered as
local if the distance over which the stress intensity exceeds 110% of the allowable general primary
membrane stress does not extend in the meridional direction more than 0.5 times (according to
BS5500 - 1 time according to ASME and CODAP) the square root of R times e and if it is not
4
Design by Analysis
Page
2.4
DBA
Design by Analysis
closer in the meridional direction than 2.5 times the square root of R times e to another region
where the limits of general primary membrane stress are exceeded. R and e are respectively the
radius and thickness of the component.

An example of a primary local stress is the membrane stress in a shell produced by external load
and moment at a permanent support or at a nozzle connection.

Secondary Stresses: Secondary stresses are stresses developed by constraints due to geometric
discontinuities, by the use of materials of different elastic moduli under external loads, or by
constraints due to differential thermal expansion. The basic characteristic of secondary stress is that
it is self-limiting. Local yielding and minor distortions can satisfy the conditions that cause the
stress to occur and failure from one application of the stress is not to be expected.
Examples of secondary stresses are the bending stresses at dished end to shell junctions, general
thermal stresses.

Peak stresses: Peak stress is that increment of stress which is additive to the primary-plus-
secondary stresses by reason of local discontinuities or local thermal stress including the effects (if
any) of stress concentration.
The basic characteristic of peak stresses is that they do not cause any noticeable distortion and are
only important to fatigue and brittle fracture in conjunction with primary and secondary stresses. A
typical example is the stress at the weld toe.

2.1.2 Stress intensity
Pressure vessels are subject to multiaxial stress states, such that yield is not governed by the
individual components of stress but by some combination of all stress components. Most Design by
Formula rules make use of the Tresca criterion but in the DBA approach a more accurate
representation of multiaxial yield is required. The theories most commonly used to relate multiaxial
stress to uniaxial yield data are the Mises criterion and the Tresca criterion. ASME chose the Tresca
criterion for use in the design rules since it is a little more conservative than Mises and sometimes
easier to apply.
For simplicity we will consider a general three-dimensional stress field described by its principal
stress components, which we will denote
1
,
2
and
3
, and define the principal shear stresses:

1 2 3
1
2
= ( )
2 3 1
1
2
= ( )
3 1 2
1
2
= ( )
According to the Tresca criterion yielding occurs when
= max( , , )
1 2 3
1
2
=
Y
where
Y
is the uniaxial yield stress obtained from tensile tests.
5
Design by Analysis
Page
2.5
DBA
Design by Analysis
In order to avoid the unfamiliar (and unnecessary) operation of dividing both calculated and yield
stress by two, a new term called "equivalent intensity of combined stress" or simply Stress Intensity
was defined:
Stress differences, S
12
, S
23
and S
31
are equated to twice the principal shear stress given above, such
that:
3 2 1 12
) ( = = S
1 3 2 23
) ( = = S
2 1 3 31
) ( = = S
The Stress Intensity, S is then defined as the maximum absolute value of the stress differences, that
is S = max (|S
12
|, |S
23
|, |S
31
|), so that the Tresca criterion reduces to:
S =
Y
Once an analysis has been performed, the Stress Intensity for each stress category is evaluated and
used in the design stress limits.

2.1.3 Stress limits

The primary stress limits are provided to prevent excessive plastic deformation and provide a factor
of safety on the ductile burst pressure (ductile rupture) or plastic instability (collapse). The primary-
plus-secondary stress limits are provided to prevent progressive plastic deformation leading to
collapse, and to validate the application of elastic analysis when performing the fatigue analysis.
The allowable stresses in the Codes are expressed in terms of design stress S
m
. The tabulated values
of S
m
given in the Code are based on consideration of both the yield stress and ultimate tensile
strength of the material. S
m
is notionally two-thirds of the "design" yield strength
Y
. Code
allowable stresses for primary and secondary stress combinations are shown in the following table
in terms of both S
m
and
Y
.
ALLOWABLE STRESS STRESS INTENSITY
General primary membrane, P
m
k Sm
2/3 k Y
Local primary membrane, P
L
1.5 k Sm
k Y
Primary membrane plus bending
(P
m
+ P
B
) or (P
L
+ P
b
)
1.5 k Sm
k Y
Primary plus secondary
(P
m
+ P
B
+ Q) or (P
L
+ P
b
+ Q)
3 Sm
2 Y
6
Design by Analysis
Page
2.6
DBA
Design by Analysis
In addition to these allowables, when fatigue is considered relevant the total sum of (P
L
+P
b
+Q+F)
should be less than an allowable fatigue stress intensity range, S
a
. The value of the k factor depends
on the load combinations experienced by the vessel. For load combinations including design
pressure, the dead load of the vessel, the contents of the vessel, the imposed load of the mechanical
equipment and external attachment loads the k factor has a value of 1. When earthquake, wind load
or wave load are added to the above, a k value of 1.2 is used. Special limits are also stipulated for
hydraulic testing. Under design load conditions k = 1 and the maximum value of the primary stress
combinations is yield the yield stress of the material. Primary stress is yield limited to ensure gross
plastic deformation does not occur. The primary plus secondary stress combinations have a much
higher allowable stress: twice the yield stress of the material. Primary plus secondary stress is
limited to ensure shakedown of the vessel.
Because of the different allowable values for primary and primary plus secondary stress, it is
essential that the calculated elastic stress is correctly decomposed into the various categories. This
is one of the most difficult problems encountered in DBA and has potentially critical effect on the
final design. If primary stresses are classified as secondary the design may be unsafe, whilst if
secondary stresses are classified as primary the design will be over-conservative. The code provides
explicit classification guidance for certain typical vessel geometries and load through Table 4.120.1
Classification of stresses for some typical cases. In situations other than these cases the designer
must rely on the basic code definitions of primary, secondary and peak stress and his own
judgement to properly classify the elastic stress. In fact some of the stress classifications
recommended in Table 4.120.1 have been in doubt for some time, and must be used with care.
2.2 Design by Analysis: the ASME inelastic route
The ASME VIII Division 2 rules for inelastic analysis are given in Appendix 4-136 Applications of
Plastic Analysis. These rules provide guidance in the application of plastic analysis and some
relaxation of the basic stress limits which are allowed if plastic analysis is used.
The rules for inelastic analysis considered here pertain to calculation of permissible loads for gross
plastic deformation only. Rules are given in the Code for shakedown analysis but in practice
shakedown analysis is difficult and it is simpler to apply the 3S
m
limit to an elastic analysis.
Two types of analysis may be used to calculate allowable loads for gross plastic deformation: limit
analysis and plastic analysis.
Limit analysis is used to calculate the limit load of a vessel. By definition, the analysis is based on
small deformation theory and an elastic-perfectly plastic (or rigid-perfectly plastic) material model.
Plastic analysis is used to determine the plastic collapse load of a vessel. The analysis is based on a
model of the actual material stress-strain relationship and may assume small or large deformation
theory as required. Material models can vary in complexity (or degree of approximation) from
simple bilinear kinematic hardening models to more complex curves defining the actual stress-strain
curve in a piecewise continuous manner.
Including strain hardening in the analysis may give a higher plastic collapse load than the limit load
but in the design by analysis procedure the allowable load is dependent on the criterion of plastic
7
Design by Analysis
Page
2.7
DBA
Design by Analysis
collapse used. Including large deformation effects in the analysis may increase or decrease the
calculated allowable load depending on the geometry of the vessel. Some structural configurations
exhibit geometrical strengthening when non-linear geometry is considered whilst others exhibit
geometric weakening.
The expression plastic collapse load is to some extent a misnomer, as a real vessel may not
physically collapse at this load level, hence the plastic collapse load is often referred to simply as
the plastic load.
2.2.1 Limit analysis
ASME VIII Division 2 Appendix 4-136.3 Limit Analysis states:
The limits on general membrane stress intensity ...local membrane stress intensity ... and primary
membrane plus primary bending stress intensity ... need not be satisfied at a specific location if it
can be shown by limit analysis that the specified loadings do not exceed two-thirds of the lower
bound collapse load. The yield strength to be used in these calculations is 1.5S
m
.
Thus allowable load P
a
is
Lim a
P P
3
2
= where P
Lim
is the limit load of the vessel.
Clearly, if the limit load can be calculated this procedure is much simpler to apply than the elastic
analysis stress categorisation procedure. However, there are two additional requirements that must
be satisfied when applying this approach. Firstly, the effects of plastic strain concentrations in
localised areas of the structure such as points where plastic hinges form must be assessed in light of
possible fatigue, ratchetting and buckling failure. Secondly, the design must satisfy the minimum
wall thickness requirements given in the design by rule section of the Code. In effect, the design by
rule formulae for wall thickness have priority over design by analysis calculations.
2.2.2 Plastic analysis
ASME VIII Division 2 Appendix 4-136.5 Plastic Analysis states:
The limits of general membrane stress intensity ...local membrane stress intensity ... and primary
membrane plus primary bending stress intensity ... need not be satisfied at a specific location if it
can be shown by limit analysis that the specified loadings do not exceed two-thirds of the plastic
analysis collapse load determined by application of 6-153, Criterion of Collapse Load (Appendix 6)
[Mandatory Experimental Stress Analysis], to a load deflection or load strain relationship obtained
by plastic analysis.
Thus allowable load P
a
is P P
a P
=
2
3
, where P
P
is the plastic load of the vessel.
Calculating plastic loads is more problematic than calculating limit loads as no rigorous definition
of what constitutes a plastic load is given. Instead, the twice elastic slope criterion as used in
experimental analysis is prescribed.
8
Design by Analysis
Page
2.8
DBA
Design by Analysis
2.3 Analysis methods for the ASME approach
Design by analysis procedures do not specify particular implementation tools: it has been left to the
analysts to choose the technique they feel most appropriate. Shell discontinuity analysis was the
primary tool in the early days of design by analysis, where stresses could easily be categorised in
terms of shell-type membrane and bending stress. By now analysis techniques have developed
significantly and although shell discontinuity analysis is still used very often in structural analysis,
it is replaced more and more by computer based numerical methods. The most widely used
technique in contemporary pressure vessel design is the finite element method, a powerful
technique allowing the detailed modelling of complex vessels. Shell discontinuity analysis and the
finite element method are discussed in relation to pressure vessel design by analysis in the
following sections.
2.3.1 Shell discontinuity analysis
Shell discontinuity analysis was the primary means of stress analysis in the early days of design by
analysis procedures. Although largely replaced by finite element analysis, shell discontinuity
analysis remains a useful tool for simple geometries, and indeed many engineering software
companies still supply programs for discontinuity analysis.

Shell discontinuity analysis is primarily used to evaluate shell membrane and bending stresses for
axisymmetric vessels under internal pressure. It makes use of the fact that typical vessel
configurations are composed of regular parts - spheres, cylinders, cones and flat ends in particular.
For pressure loading, simple regular shapes exhibit mainly membrane stress. However, at junctions
local bending (and additional membrane) stresses are generated. These stresses are called
discontinuity stresses for obvious reasons. Shell discontinuity analysis allows these junction

Semi-infinite cylinder Hemisphere
Cone Flat end
H
V
u
v
q
Forces
Displacements
Figure 2.1.: Shell discontinuity forces and moments.
9
Design by Analysis
Page
2.9
DBA
Design by Analysis
stresses, and their effect in the vessel, to be readily calculated using a simple engineering force
method.
This force method uses analytical solutions for the local bending and shear stress close to junctions
which allow so-called edge forces and moments to be related to edge displacements and rotations,
Figure 2.1. These edge relations are evaluated for each part of the vessel and then assembled at
junctions. Continuity of displacement and rotation between parts then allows the edge forces and
moments at the junction to be derived and finally the stresses in the various parts can be calculated.

2.3.2 Finite Elements for Pressure Vessel Analysis

In creating a model, element selection and mesh definition are crucial aspects of finite element
analysis. The type of element used in a finite element analysis for pressure vessel design can greatly
influence the design procedure, so a brief overview is given here. Most commercial programs
include large finite element libraries, however, in pressure vessel design the most common element
types are 3-D solid, axisymmetric and shell elements.

2.3.2.1 3-D Solid Elements
Solid (or continuum) elements are based on the mathematical theory of elasticity, which describes
the behaviour of a deformable component under load assuming small deformation and strain. The
most general theory is three dimensional, but under specific circumstances certain two dimensional
reductions are possible. 3-D solid elements are used to model real three-dimensional structures
such as the part model of a nozzle-vessel intersection shown in Figure 2.2.


Figure 2.2: 3-D solid model of a nozzle-vessel intersection.
10
Design by Analysis
Page
2.10
DBA
Design by Analysis
Elastic 3-D solid elements are based on 3-D elasticity theory. A general system of forces acting on a
three dimensional elastic body sets up internal forces within the body, which vary with position
throughout the body. The state of stress at a point in the body is fully defined by six components:
Direct stresses:
x
,
y
,
z
Shear stresses:
xy
,
yz
,
zx
as illustrated in Figure 2.3.
Three degrees of freedom are defined at each node
of a 3-D solid element: orthogonal displacements
u
x
, u
y
and u
z
. Displacement throughout the domain
of the element is defined in terms of these nodal
displacements by the interpolation functions used in
the element formulation. Most commercial finite
element packages offer solid elements based on two
different orders of interpolation:

8 node linear element: Figure 2.4.
Each element has 24 (8 node x 3) associated degrees of freedom.
Edges/sides must be straight/ plane before deformation, as the geometry is defined by linear
equations.
When the element is loaded it deforms such that the sides remain straight, as the displacements of
the element are also defined by linear equations.



X
Y
Z
XY
XY
YZ
YZ
ZX
ZX
X
Y
Z

Figure 2.3: Stresses acting on a differential cube


of 3-D elastic continua.

8 NODE LINEAR
ISOPARAMETRIC
3-D SOLID BRICK
ELEMENT
ELEMENT DEFORMATION
Original
geometry
Deformed geometry
X
Y
Z
ux
uy
uz
Figure 2.4: Linear 3-D solid brick element
11
Design by Analysis
Page
2.11
DBA
Design by Analysis
20 node quadratic element: Figure 2.5.
Each element has 60 (20 nodes x 3) associated degrees of freedom.
Edges/sides may be defined as quadratic curves/surfaces, as the geometry of the element is defined
by quadratic interpolation. This means that the 20 node brick can more closely model the true
shape of a curved body than the simpler 8 node element.
When the 20 node element is loaded the sides may deform quadratically, as the displacements of the
element are also defined by quadratic interpolation.
Both 8 node and 20 node brick elements may be
degenerated to give elements of wedge and
tetrahedron shape by defining two or more nodes
at the same position, as shown for the 8 node
element in Figure 2.6. In general, degenerate
elements do not perform as well as the brick
elements, however they can be used to model areas
of a structure which cannot be meshed using brick-
shaped elements.
Solid elements based on tetrahedral geometry (as
opposed to being degenerate bricks) are also
available in many finite element programs.
3-D solid elements can, hypothetically, be used to
model any type of structure but well-designed 3-D solid models are usually large in terms of
computing requirements. Hardware limitations etc. tend to restrict their use to situations where
simplified models (as discussed below) are not viable; for example, thick non-axisymmetric vessels,
thick axisymmetric vessels under non-axisymmetric boundary conditions (loading and support) or
perhaps thinner vessels with unusual or significant geometric details.
Modern finite element programs make it relatively simple (although perhaps time-consuming) to
create complex 3-D models of pressure vessels. The most significant problem in practical design by
analysis using solid models is not in model creation but interpretation of the results of the analysis
in the light of code requirements.

20 NODE QUADRATIC
ISOPARAMETRIC
3-D SOLID BRICK
ELEMENT
ELEMENT DEFORMATION
Original
geometry
Deformed geometry
ORIGINAL GEOMETRY
QUADRATICALLY CURVED
SIDES
ORIGINAL GEOMETRY
STRAIGHT SIDES
Figure 2.5: Quadratic 3-D solid brick element.

2 NODES
4 NODES
BRICK
WEDGE
OR
PRISM
TETRAHEDRON
2 NODES
Figure 2.6: Degenerating brick geometry to give
wedge and tetrahedron geometry
12
Design by Analysis
Page
2.12
DBA
Design by Analysis
As stated above, solid elements are based on elasticity theory, in which stress at a point is defined in
terms of six stress components:
x
,
y
,
z
,
xy
,
yz
,
zx
. These stresses vary continuously throughout
the body and in thick pressure components the through-wall distribution is non-linear. This form of
stress distribution is significantly different from that envisaged when design by analysis was first
implemented, which implicitly assumed a linear through thickness, shell-type stress distribution that
could be decomposed into membrane and bending constituents. This incompatibility in format
between the stresses calculated in the solid model and those required for design by analysis
procedures often makes it extremely difficult for the designer to classify the calculated stresses as
primary, secondary and peak and apply the appropriate category stress limits.

2.3.2.2 Axisymmetric Elements
Whilst, ideally, all three dimensional structures can be
modelled using 3-D solid elements to a greater or lesser
degree of accuracy, it is not always necessary to perform a
complete three dimensional analysis. In certain classes of
structure, advantage can be taken of the structure geometry
and loading to reduce the problem to two dimensions. Such
structures can be analysed using much simpler and smaller
finite element models.
The most useful class of three-dimensional vessel which can
be analysed using two-dimensional elements is the body of
revolution. Here geometric and material properties are
symmetric about the symmetry axis subject to loading
symmetric about the symmetry axis, as illustrated for an
example structure in Figure 2.7.
This type of structure is called an axisymmetric structure. In finite element practice the component
geometry is defined so that the Y axis is the axis of rotational symmetry. For convenience, the
geometry, loading, stresses and strains of the component are defined in polar co-ordinates: distance
r in the radial direction R from the origin, circumferential or hoop position in the circumferential
direction and meridional or height position y in the axial direction Y.
When a rotationally symmetric body is loaded
symmetrically about the symmetry axis - that is,
loads are applied radially or vertically and
uniformly with respect to circumferential position,
as illustrated in Figure 2.8, points in the body can
move radially in or out and vertically up or down.
The material does not move sideways, in the
(hoop) direction, nor rotate as there are no loads
causing the body to deform in this manner.

Y
Z
X
r
Figure 2.7: Axisymmetric structure.

ELEVATION
PLAN
Figure 2.8: Axisymmetric loading
13
Design by Analysis
Page
2.13
DBA
Design by Analysis
As the points on any R-Y plane can move only in that
plane, they have two degrees of freedom: radial
displacement u
r
and vertical displacement u
y
. As the
behaviour will be the same at all R-Y planes in the
body, the behaviour of the entire component is fully
defined if the behaviour of any such plane is defined.
Consequently, in an axisymmetric analysis only a
single 2-D section of body need be analysed. In finite
element practice, axisymmetric models are created in
the global X-Y plane, as shown in Figure 2.9.
The state of strain at a point in an axisymmetric body
may be defined by considering the deformation defined
above. As deformation is symmetric about the Y axis,
no strains are present which would give rise to non-
symmetrical deformation. Thus, no shear strains can
arise perpendicular to the X-Y plane. Under this
condition, the number of stresses at a point reduces
from six (3-D) to four:
x
,
y
,

,
ry
, as illustrated in
Figure 2.10.
Clearly, the form of stress distribution calculated in an
axisymmetric solid analysis is similar to that calculated in 3-D analysis. Consequently, the same
incompatibility exists between the form of calculated stress results and the form required by design
by analysis rules as discussed for 3-D analysis above.
A range of axisymmetric solid elements are available in most commercial finite element programs,
as illustrated in Figure 2.11.


Y
u
x
u
y
X
Figure 2.9: Axisymmetric degrees of freedom

u
u
1
2
Y
X
ry
y

r
y
r


Figure 2.10: Axisymmetric element stresses.
14
Design by Analysis
Page
2.14
DBA
Design by Analysis
The linear quadrilateral element has four nodes, each with two degrees of freedom. Thus, each
linear element has 8 associated degrees of freedom, compared with 24 for a linear 3-D solid. The
quadratic element has eight nodes with two degrees of freedom. Thus, each quadratic element has
16 associated degrees of freedom, compared with 60 for a linear 3-D solid. Clearly, the use of
axisymmetric elements leads to smaller models in terms of degrees or freedom or, if preferred,
permits a finer mesh for the same model size.
An axisymmetric model of the 3-D nozzle
intersection shown in Figure 2.2 is shown in
Figure 2.12. The use of axisymmetric
elements allows the analyst to produce a finer
mesh through the thickness of the vessel wall
without creating an excessively large model.
Care must be taken when defining loads for
axisymmetric models. Forces may be defined
as a total applied force or on a per radian
basis. The program users manual should be
consulted to check the situation for the
software in use.
It is worth noting that many pressure vessel
problems relate to axisymmetric structures
under non-axisymmetric loading. For linear
elastic analysis, it is possible to treat this as an
axisymmetric problem, and model the loading
using Fourier series around the circumference. Some commercial finite element software offers this
capability through modifications to the basic axisymmetric element, known as an harmonic element.
Solid elements are extremely versatile, powerful, and suitable for a wide range of applications.
However, there are two significant factors which limit the use of solid elements in practical finite
element analysis, particularly for pressure vessel problems. These are:

L I NEAR TRI ANGL E
( CONSTANT STRAI N TRI ANGL E)
L I NEAR QUAD
QUA DRA T I C QUA D
Y
Z
X
r
QUADRAT I C T RI ANGL E
Figure 2.11: Typical elements for axisymmetric
analysis.

X
Y
Figure 2.12: Axisymmetric model of nozzle intersection.
15
Design by Analysis
Page
2.15
DBA
Design by Analysis
The aspect ratio of solid elements should ideally be 1 - that is, the element is a cube - but in
practice limited to 2 in the case of linear elements and 5 in the case of quadratic elements.
Solid elements do not respond well to bending loads, and at least three linear or two quadratic
elements must be used through thickness when bending is present.
Taken together, it becomes clear that these factors often make it impracticable to model thin shell
structures such as the longitudinally supported vessel shown in Figure 2.13 using solid elements.
In general, shell structures are thin in one direction and carry both membrane (in-plane) and
bending (out-of-plane) loads. The load-carrying capacity of a shell mainly is derived from its
membrane strength, but it is impossible to construct real shell structures, such as pressure vessels,
without inherent bending during loading - for example, the junction between a cylinder and a
spherical end cap gives a discontinuity in curvature which induces a bending stress (called a
discontinuity stress). Bending stresses can also result from mechanical and thermal loading - for
example, piping forces on a nozzle.
Several solid elements are required through the shell thickness to adequately represent bending
behaviour but these elements cannot themselves be thin or they will violate aspect ratio
requirements of the formulation. Consequently, a large number of solid elements are required in
order to model even simple shell structures. These shell analysis problems are avoided by using
special shell elements, which incorporate assumptions about the nature of the bending in the
formulation.






Figure 2.13: Shell model of vessel.

16
Design by Analysis
Page
2.16
DBA
Design by Analysis
2.3.2.3 Thin Shell Elements
The traditional method of analysing shell structures is to simplify the behaviour of the structure by
assuming an appropriate thin shell theory in which the behaviour of the three dimensional structure
is described in terms of the deformation of a doubly curved reference surface. This reduces the
number of degrees of freedom required to model the real behaviour of the structure as only one
element is used through thickness. In addition, the definition of the finite element model is also
considerably simplified, as only the mid-surface has to be defined and meshed with surface or area
elements. However, the reduction of a real three dimensional shell structure to a reference surface
model is considerably more complicated than reducing three dimensional elasticity to axisymmetry
or plane strain, as discussed earlier.

In general, shell structures are doubly curved
and the radii of curvature are not constant
throughout the body. Shell structures support
both membrane forces, which act in the plane
of the shell, and bending forces or moments
which arise due to out-of-plane loading.
These are shown diagrammatically in Figure
2.14.
The membrane and bending forces are
coupled throughout the shell, but their relative
magnitude differs with position in the
structure. In certain locations membrane action may predominate, whilst in other locations, most
notably at structural supports and discontinuities, bending becomes more significant. In flat plates
and in the theory of shallow shells, membrane and bending actions can be sensibly de-coupled - this
simplification allows a simpler element formulation, and for this reason flat plate elements are
commonly used in the analysis of general shell structures.
It is possible to simplify the real three-dimensional problem if certain assumptions are made about
how the thin plate or shell deforms, particularly during bending. The most significant assumption
of thin plate or shell theory is that straight lines initially perpendicular to the mid-surface remain
straight during deformation, and follow either the Kirchhoff hypothesis or the Mindlin hypothesis.
In the Kirchhoff hypothesis, when the shell deforms, straight lines normal to the mid-surface rotate,
but so that they remain straight and normal to the deformed mid-surface, as illustrated in Figure
2.15. In the less restrictive (but more complex) Mindlin Hypothesis these lines are not required to
remain normal during deformation, such that the influence of shear strain can be better represented.


MEMBRANE MODES
BENDING MODES
Figure 2.14: Thin shell bending & membrane modes.
17
Design by Analysis
Page
2.17
DBA
Design by Analysis

KIRCHHOFF HYPOTHESIS
PLANE SECTIONS REMAIN PLANE
AND NORMAL TO THE NEUTRAL AXIS
MINDLIN HYPOTHESIS
PLANE SECTIONS REMAIN PLANE
BUT NOT NECESSARILY NORMAL
TO THE NEUTRAL AXIS
Figure 2.15: Kirchhoff and Mindlin Hypotheses

The Kirchhoff hypothesis is common in shell theory, and most published results for thin pressure
vessels are based on this. However, some modern finite element formulations for thin shells use the
Mindlin hypothesis, (essentially since numerical analysis is used and there is no need to be so
restrictive). In addition, the Mindlin hypothesis represents shell behaviour more accurately in the
vicinity of discontinuities and restraints, where transverse shear effects are more significant.
The effect of using one of these simplifying hypotheses in the finite element method is that the
deformation at any point can be defined if the displacement and rotation of the mid-surface are
known. The rotation at any point on the mid-surface is defined by interpolating between rotational
degrees of freedom defined at the nodes. Therefore beam, plate and shell elements have
translational and rotational degrees of freedom. Only these bending elements have rotational
degrees of freedom: solid elements do not need rotational degrees of freedom to define the element
deformation (although it is possible to formulate solid elements which include rotational degrees of
freedom to enhance performance but at the cost of increasing the number of degrees of freedom per
element).
A great deal of work has been undertaken on formulating shell elements, and it is an indication of
the complexity of the problem that no single type of formulation has been universally accepted as
being the best. Classical shell theory produces equations which are difficult to solve and which are
remarkably sensitive to slight variations in shape (which are common in the approximate finite
element method). A large number of different approaches have been developed over the years but
basically only three types of shell element are used in practice:
Facet (flat) shell elements, formed by combining membrane and plate bending
elements
Curved shell elements, based on classical shell theory
Reduced (or degenerate) solid (continuum) iso-parametric elements which directly
take account of thinness and the Mindlin hypothesis in their formulation
The most popular are flat elements and reduced solid elements, both of which appear in commercial
software.

18
Design by Analysis
Page
2.18
DBA
Design by Analysis

Flat (Plate) Elements
The geometry of a doubly curved shell surface can be approximated by a faceted surface formed by
connecting flat triangular elements together at their vertices. A flat three-noded triangular shell
element would have six degrees of freedom per node: in an element co-ordinate system (x,y,z), as
shown in Figure 2.16, there would be three translational degrees of freedom, (u
x
, u
y
, u
z
), and three
rotational degrees of freedom, (
x
,
y
,
z
), giving a total of eighteen degrees of freedom per
element.
This element can be used to represent a shell by including both membrane, (u
x
, u
y
,
z
), and bending,
(u
z
,
x
,
y
), degrees of freedom. Membrane stiffness is derived from simple plane stress conditions,
with the added drilling degree of freedom,
z
. The most significant aspect of the derived shell
element is that the membrane and bending stiffness are uncoupled, although there is a degree of
coupling when the elements are assembled.

I
J
K
X
Y
Z
ux
uy
uz
X
Y
Z
Figure 2.16: Triangular flat plate element

Used on their own, it has been found that triangular shell elements based on plate bending elements
do not perform very well, having an artificially high bending stiffness and spurious torsion modes.
Other flat shell elements have been formulated, among the most common of which is the Batoz-
Razzaque element. This is a quadrilateral element formed from four flat shell elements such that the
diagonals are continuous, Figure 2.17. This formulation works well and can be found in many
commercial programs.

Figure 2.17: Batoz Razzaque quadrilateral flat shell element

19
Design by Analysis
Page
2.19
DBA
Design by Analysis
Reduced Continuum Elements
The reduced continuum shell elements are similar to the 3-D solid elements discussed above but
with the Mindlin hypothesis (plane sections remain plane or linear interpolation) applied through
the thickness. Under this condition, if the deformation of the element mid-surface is known, the
deformation at any other point can be defined through the element shape functions. In this way, the
3-D element is reduced to a shell-type area element. Such elements are usually called reduced
continuum, or degenerate solid, or sometimes thick shell, elements. In practice, the elements
curved mid-surface is approximated from the given nodal co-ordinates and this can affect the
performance and accuracy of this type of element if it has a poor shape.

2.3.2.4 Discussion
At first sight the most appropriate choice of finite element may seem obvious for a given vessel
under consideration. However when the choice is examined in the light of the pressure vessel design
by analysis elastic route, where limits are placed on membrane and bending stress and stresses, or
indeed parts of stresses, must be categorised, various well-known difficulties arise. These problem
areas lie at the heart of criticism of the ASME design by analysis rules and consequently are
discussed in more detail in the following:

2.4 Implementation Problems of the Stress Categorisation Route
2.4.1 Overview of Problems
Once the linear elastic analysis of a part is complete and the immediate results for stresses and
strains obtained, there is the need to satisfy the design by analysis rules. As mentioned previously
this is not necessarily as straightforward as it may at first seem. Specifically, there is a requirement
to obtain membrane and bending components of primary stress and the calculated stresses must be
categorised. This does not present a problem in cases where the analysis utilises thin shells.
However, for analysis (in particular finite element analysis) utilising solid models (2 or 3
dimensional) where the calculated stress can not be easily identified as membrane, bending or peak
the problems of linearisation and categorisation become apparent. Difficulties implementing this
area of the design by analysis rules have become increasingly evident to both designers and
analysts
[1]
. This section examines the practical problems associated with the implementation of
these design by analysis rules.
2.4.2 Linearisation
The design by analysis criteria, as formulated nearly thirty years ago, is based on the behaviour of
thin shells and includes the notion of membrane and bending stress. Inherent in this understanding
is the assumption that membrane and bending stress act on a plane under the Kirchhoff hypothesis
that plane sections remain plane during bending. The shell type membrane and bending stresses
cause gross distortions under primary loads and strain enhancement under secondary loads. Most of
our understanding of basic pressure vessel geometry and components come from our knowledge of
their behaviour as thin shells. A consequence of this understanding is the possibility that portions
of total stress, identified as membrane or bending (or peak) can be categorised as primary or
secondary.
20
Design by Analysis
Page
2.20
DBA
Design by Analysis
For example, in the case of a nozzle in a
spherical shell (Figure 2.18) with area
compensated reinforcement, only the
membrane stress is primary, despite
significant bending in the shell close to the
nozzle. The bending stress is secondary
(since only the membrane stress and the hoop
stress in the nozzle are required to satisfy
equilibrium with the internal pressure). In
this case, it is essential to consider membrane
and bending stress for the correct
categorisation.
If the analysis is based on thin shell finite
elements, then there is no difficulty in identifying
membrane and bending stress, as they are part of
the underlying theory, Figure 2.19.


Difficulties arise when thin shell analysis is not
used and the finite element analysis is based on
axisymmetric or three-dimensional solid elements.
In general, unless the section is indeed thin, the
stresses on a through thickness line are not linear,
and further plane sections do not remain plane
during bending. Over the years it has become common practice to linearise the calculated through
thickness stresses in order to separate membrane and bending components.
A technique for linearising stress was first suggested by Kroenke
[2,3]
, and has been adopted in
several finite element postprocessors. A stress classification line (or plane) or supporting line
segment is chosen and the stresses are linearised along this line. The supporting line segment (SLS)
or classification line is the smallest segment joining the two sides of the wall where the stress is to
be linearised. Outside of gross structural discontinuity regions, the SLS is normal to the wall mean
surface, i.e. its length is equal to the thickness of the wall in the analysis. There are difficulties with
this procedure which seems straightforward, but again is a fundamental difficulty - this will be
discussed in more detail in the following.
Pressure vessel design codes are not particularly helpful on the problem of linearisation. ASME III
& VIII admit a non-linear bending stress, but also contains some ambiguities: bending stress is
described as a normal stress - and it is bending stress that may need linearisation. In Paragraph
NB-3215 a note is provided to the effect that .. membrane stress intensity is derived from the stress
components averaged across the thickness of the section. The averaging shall be performed at the
component level .... This implies that only stress components may be linearised (by definition this
could include shear stress), and not derived principal values. However, through omission from the

Rei nf orcement
p
r
BENDI NG
MEMBRANE
M
R
t
P
m
pr
t
=
P
m
M
R
i s PRI MARY
i s SECONDARY
Figure 2.18: Stress categories for a nozzle.

i
j
k
l
m
n
o
p
TOP
BOTTOM
x
y
z
y
x
TOP
BOT
MID
TOP
BOT
MID
X
Y
Z
Figure 2.19: Shell membrane and bending stress
21
Design by Analysis
Page
2.21
DBA
Design by Analysis
code, it may be argued that shear stress should not be linearised. Inclusion of shear stress
linearisation will mostly affect the surface stress: in practice linearisation of the normal stress only
is adopted to modify the surface stress in application of the design criteria.
The basic procedure for stress linearisation is the
selection of a stress classification plane or
supporting line segment on which shell type
stress will be evaluated.
A stress classification line (or plane of reference-
see page 31) is identified through a section and
the non-linear stress distribution along this line is
linearised in order to extract membrane and
bending stress, as shown in Figure 2.20. The
stress classification line (CL) lies along a local
axis X
3
; the origin is located at the mid-point of
the CL (i.e. at radius R
c
); the abscissa of a point
on the supporting line segment is designated x
3
.
In practice, the linearisation procedure is performed automatically by special postprocessors. For
simplicity, some basic postprocessors (in particular self written ones) may require the finite element
mesh to be created so that a line of nodes lie along the chosen classification line, making it
relatively simple to extract stress results. In
postprocessors that are more complex, the
classification line need not pass through a line
of nodes.
The classification line in Figure 2.21 is defined
from node N
i
at the inner surface to N
o
at the
outer. The path of the classification line does
not pass through a line of nodes: it cuts
through the elements. Advanced linearisation
postprocessors use the location of the surface
nodes to define the path through the elements
and then apply interpolation functions to the
appropriate nodal stresses to calculate the
stress along the path. Two possible procedures
for linearisation have been suggested, Kroenke
which has been discussed by numerous
analysts
[4]
and a more refined version by
Gordon
[5]
.

N
i
N
o
STRESS
CLASSI FI CATI ON
LI NE
Figure 2.21: Interpolation of a classification line
X
3
Stress classification line or
supporting line segment of
length h.
R
c
Figure 2.20: Stress classification line
22
Design by Analysis
Page
2.22
DBA
Design by Analysis
2.4.2.1 Kroenkes Procedure
Kroenkes procedure makes reference to familiar beam bending stress - a uniaxial stress - and
attempts to define an equivalent linear stress distribution on the classification line (CL). Consider a
typical stress distribution along a classification line as in Figure 2.22.
If x
3
measures local distance along the classification line then the equivalent linearised stress is,
( ) b ax
L
ij
+ =
3
.
The membrane stress component is given by the formula
( )
3
2
2
1
dx
e
b
e
e ij m ij

= = .
The membrane force per unit length of the membrane stress component is equal to that from the
calculated FE stress component.
The bending stress component is given by
( )
3 3
2
2
3
3
3
12
dx x
e
x
x a
e
e ij
b
ij

= = .
The maximum and minimum bending stresses can then be evaluated (for 2 /
3
e x = )
nonlinear stress
distribution
linearised
stress
e/2 e/2
x
3
X
3

b
a

m
p
classification
line
Figure 2.22: Typical Stress Distribution
23
Design by Analysis
Page
2.23
DBA
Design by Analysis
( )
3 3
2
2
2 ,
6
dx x
e
e
e ij
s b
ij

=
The bending moment per unit length of the calculated FE stress component is equal to
( )
3 3
2
2
2
,
6
dx x
e
e
e ij
s b
ij

=
The linearised stress ( )
L
ij
is found by adding membrane and bending stresses
( ) ( ) ( )
b
ij
m
ij
L
ij
+ =
The bending stress of this equivalent linear stress distribution vanishes at 0
3
= x .
The peak value of stress at a point is the difference between the total stress and the sum of the
membrane and bending stresses
( ) ( ) ( ) ( ) ( ) ( ) [ ]
b
ij
m
ij ij
L
ij ij
p
ij
+ = = .

2.4.2.2 Gordons Procedure for axisymmetric problems
In Kroenkes procedure for axisymmetric problems, the shell wall is assumed (locally) straight in
the meridional direction. In some circumstances the meridional curvature is finite. Gordon
suggested a modification to Kroenkes procedure to allow for this.
Gordons procedure for an axisymmetric case is the same, in principle, as the case above, except for
the fact that there is more material at a greater radius than at a smaller radius. The neutral axis is
shifted radially outward to accommodate for this.
Consider the axisymmetric section of a vessel wall as shown in Figure 2.23. is defined as the
radius of curvature of the mid-surface of the shell. In the case of an axisymmetric straight section
such as a cylinder or cone, = .
Adopting the following notation:
radius of meridional curvature
R
1
radius of circumferential curvature
z axial co-ordinate
r radial co-ordinate
angle in hoop direction
angle in meridional direction
X
3
local co-ordinate containing the classification line
X
2
local co-ordinate normal to classification line
x
3
co-ordinate along classification line
e shell thickness
24
Design by Analysis
Page
2.24
DBA
Design by Analysis
R radial co-ordinate of a point of the classification line
R
c
radial co-ordinate of mid-surface point
From an axisymmetric analysis the following stresses would be obtained in the local classification
line co-ordinates:

X2
- (local) meridional stress

X3
- (local) radial stress

- hoop stress

X2X3
- (local) shear stress
The other shear stresses would be zero in an torsion-free axisymmetric analysis.
The aim is to obtain membrane and bending components of these stresses, denoted by subscripts m
and b respectively, evaluated from the average stress across the section and the beam type bending
stress.
The membrane component of the (local) meridional stress on the classification line is given by
( )
( )



=

+
= =


e R
dx R
e R
dx x R
A
F
c
e
e X
e
e X
X
X
m X
2
2
3
1
2
2
3 3 1
2 2
2
2
2
X
3
X
2
e
R
1

z
r

r
z
Axis of
symm.
centreline
R
R
c
Figure 2.23: Geometry for finite curvature
25
Design by Analysis
Page
2.25
DBA
Design by Analysis
where the area
2
X
A of a small segment extending over the angle in hoop direction is given by
= e R A
c X
2
.
In this linearisation the bending stress component of the (local) meridional stress on the
classification line vanishes at
f
x x =
3
, where
f
x is the
3
x - co-ordinate of the resultant of a constant
stress distribution
2
X
and of the centroid of the considered area.
f
x is given by
c
f
R
e
R
e
x
12
cos
12
2
1
2

= = ,
thus the bending stress component is given by
( )
m
f X
b
X
I
x x M ) (
3
2
2

= ,
where
( )
3
2
2
3
2 2
dx R x x M
X
e
e f X
=


and

=
2
2
12
f c m
x
e
e R I ,
which leads to
( ) ( )
3
2
2
3
2
2
3
2 2
12
dx R x x
x
e
e R
x x
X
e
e
f
f c
f
b
X

.
The hoop stress is evaluated in a similar manner to the above; however in this case the meridional
curvature, must be taken into account:
The membrane component of the hoop stress is given by
26
Design by Analysis
Page
2.26
DBA
Design by Analysis
( )
( )
3
2
2
3
2
2
3 3
1
1
dx
x
e e
dx x
A
F
e
e
e
e
m

+ =

+
= =

where the area

A of a small segment extending over die angle in meridional direction is given


by

= e A .
In this linearisation the bending stress component of hoop stress on the classification line vanishes
at
h
x x =
3
, where
h
x is the
3
x - co-ordinate of the resultant of a constant stress distribution

and
of the centroid of the considered area, where
h
x is given by

=
12
2
e
x
h
.
Thus the bending stress component is given by
( )
h
h
b
I
x x M ) (
3

=

,
where
( ) ( )
3 3
2
2
3
dx x x x M
e
e h


+ =

,
and

=
2
2
12
h h
x
e
e I ,
which leads to
( ) ( )
3
3
2
2
3
2
2
3
1
12
dx
x
x x
x
e
e
x x
e
e
h
h
h
b



.
27
Design by Analysis
Page
2.27
DBA
Design by Analysis
(Local) radial stress on the classification line is treated in a special way: in most situations the radial
stress will equal the applied pressure at the internal surface and be free (zero) at the outer surface.
Therefore, membrane stress may be evaluated,
( )
3
2
2
3 3
1
dx
e
e
e X
m
X

=
but it is questionable whether a bending stress should be evaluated. Either this should be taken as
zero
( ) 0
3
=
b
X

or as the simple difference between actual and averaged value


( ) ( )
m
X X b X
3 3 3
=
which may not be linear.
Similarly, an average membrane shear stress can be determined along the classification line
( )
3
2
2
3 2 3 2
1
Rdx
e R
e
e X X
c
m
X X

=
Since the shear stress would be expected to be nearly parabolic (from basic elasticity theory), and
zero at the surface, the bending stress should be taken as zero
( ) 0
3 2
=
b X X

The surface values of shear therefore only contribute to peak stress.


The development of Gordons procedure given here is in terms of stress components in a local co-
ordinate system (X
3
, , X
2
). In practice these would be transformed into the global (r, , z) co-
ordinate system, to give global linearised stress components. Once the global stress components
have been linearised, the principle stresses and stress intensity can then be evaluated, as total values
and as averaged membrane and surface bending stresses.
2.4.2.3 Discussion
As mentioned previously, several finite element programs contain post processing options to
directly calculate the equivalent linearised stresses on any prescribed classification line. Usually all
stress components are linearised. A short consideration of the linearisation procedure immediately
brings several possible problem areas to mind.
.
28
Design by Analysis
Page
2.28
DBA
Design by Analysis
Selecting the stress classification line. This should be a line through the vessel wall, which
would be expected to yield shell type deformations, namely straight lines remaining straight.
However close to discontinuities some
warping and shear would be expected (and
indeed observed in the finite element
calculations) and the concept of averaged
membrane and linearised bending stress is
tenuous. Ideally, inner, outer surfaces,
and, thus, the mid-surface should be
parallel, with the stress classification line
perpendicular to these surfaces. Of course
in some situations ambiguities can arise, as
illustrated in Figure 2.24.
Selecting which stress components
should be linearised. The global stress
components, including shear, may be
linearised - but what about the principal
stresses? With principal stresses, there is the obvious problem that principal directions can alter
from point to point through the thickness unless the classification line is far from a discontinuity.
As mentioned previously, the ASME Code itself implies that the linearisation should be performed
on the global component stresses: ... membrane stress intensity is derived from the stress
components averaged across the thickness of the section. The averaging shall be performed at the
component level .... There is the related question as to whether linearising the component stresses
then calculating the principal stresses from the linearised components is consistent with linearising
the principal stresses - plane sections remaining plane should provide this consistency. BS5500
implies that all stress components should be linearised.
Selection of the classification line has received little attention in the literature. However, selection
of which stress components to linearise has been examined, on behalf of the ASME Code
committee, by G L Hollinger and J L Hechmer
[6]
.
Hechmer & Hollinger analysed a representative axisymmetric vessel problem and examined several
different methods of stress linearisation. Two methods appear which were identified as being both
conservative and consistent: either linearise the two normal stress components on a line (in the hoop
and meridional directions) and use the total normal radial and total shear stress at the surface, or
linearise the meridional principal stress and use the total stresses for the other principal direction
(the exact technique is not wholly clear from the paper). Neither of these would appear to be
common practice.
2.4.2.4 Three Dimensional Problems
Three-dimensional solid finite element analysis poses a significant problem for stress linearisation.
In 3-D analysis it is necessary to find a consistent stress classification plane, which again could
cause problems near the very features the designer is concerned with (fillets and gross structural
discontinuities). There is the added problem of defining exactly what should be meant by plane
sections remaining plane in this case. Three possibilities arise: firstly the stress components at a
point are directly used to evaluate the stress differences and stress intensity; this is easy, but the
Normal t o
Out er Surf ace
Normal t o
I nner Surf ace
LI NE 1
LI NE 2
Figure 2.24: Ambiguous classification lines
29
Design by Analysis
Page
2.29
DBA
Design by Analysis
subsequent categorisation of these stress intensities is not. Secondly, stresses are linearised along
radial lines to obtain beam type membrane and bending stresses; this suffers from the same
problems mentioned above. Thirdly, selected planes are specified and two sets of stresses on
distinct lines (on the plane) are used to evaluate shell type direct and bending stress on a plane. To
the writers knowledge, no commercial post processors offer three dimensional stress linearisation
capability over a plane, only along a line, only one by a vessel manufacturer
[7]
.
In the case of the three dimensional problem, Hechmer & Hollinger
[8]
analysed a complex nozzle
shell assembly using brick elements and examined the consequences of the three different
assessment methods described above to calculate the stress intensity. As expected their study
demonstrated a wide variation in the calculated results for the various methods (mainly because
there are many possibilities open to the analyst) - a variation in stress intensity of over 35% was
noted in this example. The results are indeed inconclusive: stress at a point calculation is easiest to
apply but the results are not always conservative while stress along a line calculation is more
advantageous with respect to Code rules.
2.4.2.5 Linearisation Guidelines
It has been apparent for some time that there are deficiencies in the rules for design by analysis
when the finite element method is used. In particular this has highlighted problems with the design
criteria and the underlying philosophy of assessment. Over the past few years, the US Pressure
Vessel Research Council (PVRC) has funded a project to consider recommendations for updating
the ASME Code. It is worthwhile reviewing some of these recommendations; a summary has been
given by Hechmer & Hollinger
[9]
.
The short term recommendations consisted of six sections. The second, fourth, fifth and sixth
recommendations are related to linearisation problems for primary stress and three-dimensional
problems. The first and third recommendations are of a more fundamental implication since they
relate to the use of finite element methods for design by analysis using the existing ASME - Code
criteria. The project members have been very careful with the wording of the recommendations, and
some interpretation is required. These recommendations consider essential pressure vessel
components, which are basic structural
elements:
Shells of revolution and circular plates
with either constant or variable thickness
(transition elements) - normally
connecting one structural element to
another.
Smooth junctions - where the model
represents the actual geometry for
example connecting fillet or blend
radius.
Sharp junctions - where the model does
not represent actual geometry, such as
sharp corners or notches, as shown in
Figure 2.25.
Basic
Structural Element
Sharp Junction
Transition Element
Smooth Junctions
Fillet
Blend
Sharp Junction
Basic
Structural Element
Figure 2.25: Pressure vessel elements
30
Design by Analysis
Page
2.30
DBA
Design by Analysis
The first and third recommendations are summarised below:
First recommendation: This relates to the use of finite element analysis (FEA) in pressure vessel
design by analysis. It is recommended that for the majority of pressure vessel components, which
are basic structural elements, FEA is inappropriate. P
m
stresses should be calculated using general
equilibrium considerations, with P
m
+P
b
evaluated by hand calculations for conditions where P
m
is
small (for example in flat plates). FEA is appropriate for calculating P
L
+P
b
stresses near
discontinuities (see third recommendation below) and for the calculation of P+Q stresses in general.
Notably it is only in complex components where basic structural analysis does not exist that FEA is
recommended as appropriate for P
m
and P
m
+P
b
stress evaluation. ... the thrust is that the designer
should be applying his ingenuity to calculating equilibrium stresses, not to extracting stresses from
a general finite element model ....
Third recommendation: This relates to the locations in a pressure vessel where stress evaluations for
Code compliance should be considered. It is recommended that it is appropriate to perform P
m
+P
b
(P
L
+P
b
) and P+Q evaluations in basic structural elements, but inappropriate in discontinuity type
transition regions. If there is a smooth junction then the stresses should be evaluated in the row of
elements adjacent to the junction (or the line of nodes at the junction). When there is a sharp
junction, the evaluation must be far enough from the junction so that the stresses are not affected by
the notch behaviour. This recommendation should eliminate the need to linearise erratic stress
distributions; ... the thrust ... is that plastic collapse and gross strain concentration will not occur in
stiff transition regions; they will occur in the more flexible shell elements ... the purpose of the
P+Q limits is to validate the fatigue analysis by precluding strain concentration and ratchet. It is
highly unlikely that ratchet could occur in a transition element ...
The first recommendation is rather subtle. In the light of the ASME Code (as it stands), finite
element analysis is only appropriate in certain special cases in primary stress calculation - in
general, equilibrium and shell discontinuity analysis are to be preferred. However, FEA is
appropriate for secondary (and peak) stress evaluation. In the context of the discussion given this
may be interpreted further as follows: finite element analysis may be used to evaluate the overall
stress distribution for shakedown and fatigue assessment but the analyst should use simple
calculations and strength of materials arguments to extract the primary stress components. In other
words, elastic finite element analysis should not be used as the basis for categorisation or evaluation
of primary stress.
The third recommendation also needs careful interpretation and is the most intriguing of all those
provided by the PVRC project. The implication to the writers is clear - ignore the calculated stresses
in sharp transition regions, since they will not affect the post yield failure mechanisms.
The mid term recommendations aim to provide additional tools and procedures to assist the
designer in making better use of the existing ASME Code rules, specifically to address the problems
of categorisation and linearisation directly through finite element analysis.
Finally the long-term recommendations aim for a more fundamental assessment of the ASME Code
philosophy and criteria and require extensive new research. It is felt that new rules should be based
on specific quantities required to prevent a failure mechanism, perhaps moving away from simple
elastic analysis and stress evaluation. For example, the limits based on shell type membrane and
31
Design by Analysis
Page
2.31
DBA
Design by Analysis
bending stress are difficult to understand and often misinterpreted, while the secondary limits are
probably oversimplified and over-conservative, particularly in the presence of combined load.
Considerable research on shakedown and ratchetting over the past twenty five years has confirmed
this.
2.4.3 Problems with categorisation
The process of stress categorisation (or classification) is difficult, as stress may be composed of
both primary and secondary parts as seen in Figure 2.18 for the nozzle reinforcing pad. It is not
sufficient just to categorise a particular stress corresponding to a given load condition, but also to
categorise segments of the stress. This prospect is not inviting, and indeed rarely done in practice
unless specified in Code rules (as in the case of the nozzle).
We have reached a familiar problem - how should finite element (or otherwise) calculated stress be
categorised? This is usually left to experience or strength of materials type arguments if this is
possible. It is usually possible with simple strength of materials analyses or shell discontinuity
analysis to separate primary and secondary stress with the understanding of the fundamental failure
mechanisms that the Code addresses, since the equilibrium calculations were done manually. This is
not obvious with finite element results, and in particular with the results of using continuum
elements. The question is what can be done to ease this problem.
An obvious solution would be to provide additional Code rules. While this is likely to be the case
in the long term, it does not help the designer who must carry out pressure vessel design with the
current rules.
Briefly, the evaluation rules in this route can be summarised. Membrane and other primary
membrane stresses are not allowed to approach yield since beyond yield there is the possibility of a
catastrophic plastic collapse for example bursting under internal pressure. The total (membrane
plus bending) stress can increase fifty percent above the membrane limit since there is some safety
margin here, but is still yield limited. Discontinuity and thermal stresses (or strain controlled
stresses) must be limited to ensure shakedown under cyclic load; thus the range of secondary stress
is limited to twice yield (or some smaller proportion for particular components). The peak stress
must be limited to ensure a sufficient fatigue life, and certain other failure criteria may need to be
addressed depending on the operating temperature - for example creep rupture at high temperature,
fast fracture at low temperature. At this level categorisation is straightforward: any sustained stress
that, subject to overload, would lead to plastic collapse is primary. The remaining stress (or indeed
proportion of stress) can be classified as secondary and is subject only to the shakedown criterion
(and fatigue limit).
The problem arises because this design by analysis route relies upon elastic analysis. Elastic
analysis on its own cannot characterise the nature of the stress since it is not clear what failure
mechanisms can arise; it is left to the designer to do this. In addition, this approach does not make
use of the ductility of pressure vessel steels, resulting in a wholly inconsistent (conservative) margin
of safety
[10]
. In the absence of any meaningful information the designer is led to classify all stresses
as primary and base redesign on this.
32
Design by Analysis
Page
2.32
DBA
Design by Analysis
It is useful to view the categorisation problem as part of the basic requirement to avoid failure by
the various failure mechanisms: the problem of categorisation should then have a different
interpretation.
The stress system in the component should be such that shakedown is achieved and the fatigue
limits satisfied for all stresses. In fact, these are the basic design requirements. The categorisation
problem can then be interpreted as the need to isolate those stress systems that could cause gross
plastic collapse - that is the primary stresses. The distinction here is subtle - there is no real need to
identify a calculated stress as being primary or secondary; it is only necessary to identify the
primary stresses.
One solution to this difficulty is to calculate the limit load of the vessel by inelastic analysis. Limit
load assessment and calculation of principal stress has been discussed by several authors using a
variety of methods, notably Marriott
[11]
, Kalnins & Updike
[12]
, Mackenzie & Boyle
[13]
, Seshadri
[14]
Ponter & Carter
[15]
and Zeman et. al.
[16]
.
2.5 Implementation problems of the ASME inelastic route
Inelastic finite element analysis
[17,18]
is more difficult than linear analysis and requires considerably
greater computing resources. Essentially, the non-linear problem is solved in a piecewise manner
using incremental solution techniques. The procedure usually requires the analyst to define an
appropriate number of load steps, equilibrium iterations within load steps and convergence criteria
defining the required accuracy of the solution. Poor choice for any of these parameters can lead to
lack of convergence or indeed convergence to the wrong answer. In addition, it is difficult to
make a priori engineering estimates of the inelastic response and to verify results of the analysis
through simple calculations. There is also a shortage of non-linear benchmarks, which the analyst
can use to assess the accuracy of the analysis procedures.
There are two types of inelastic analysis methods, which may be used to guard against gross plastic
deformation: limit analysis and plastic analysis.
Limit analysis is based on an elastic-perfectly plastic material model and small deformation theory.
The assumption of perfect plasticity sometimes causes convergence problems in non-linear analysis
and in practice a bilinear hardening material model with a low value of plastic modulus E
p
(1/10000 of E) is often used. This analysis determines the limit load P
L
of the vessel. The allowable
load P
a
then is defined as a specified fraction of the limit load.
Plastic analysis is based on the actual non-linear stress-strain relationship of the vessel material,
including non-linear geometry effects if desired. This analysis determines the plastic collapse load
P

. However, determination of the plastic collapse load is not straightforward to understand this,
some basic concepts are required.
33
Design by Analysis
Page
2.33
DBA
Design by Analysis
2.5.1 Plastic design loads
The aim with the ASME inelastic route is to estimate limit and shakedown loads directly, which can
then be used to characterise an allowable load (Sec. 2.2). To begin with these are defined:
The First Yield Load P
y
The first yield load P
y
is defined as the load for which the material of the pressure vessel first yields
(from the virgin stress-free state) at the most highly stressed point. Because only one point of the
material is at yield, the surrounding elastic material restrains the vessel from plastic deformation as
a whole.
The Limit Load P
0
The classical definition of a limit load P
0
according to limit analysis is an idealized one, a
mathematical one. This theoretical limit load is the maximum load solution to an analytical model
of the structure which embodies the following conditions:
the strain-displacement relations are those of small displacement theory (first order);
the material response is rigid plastic or elastic-perfectly-plastic (Fig.2.26),
the internal stresses and applied forces are related by the usual linearised equations of
equilibrium which ignore changes in geometry due to deformations.
rigid-perfectly plastic
strain (extension)
s
t
r
e
s
s

(
f
o
r
c
e
)
elastic-perfectly plastic
strain (extension)
s
t
r
e
s
s

(
f
o
r
c
e
)
Fig 2.26 : Elastic-perfectly-plastic and rigid-plastic deformation curves
34
Design by Analysis
Page
2.34
DBA
Design by Analysis
A (sufficiently) small region in an elastic-perfectly plastic material behaves either elastically (if
stressed below yield), or plastically (for stresses at yield). At loads above the first yield load, P >
P
y
, but less than the limit load, P < P
0
, a region of material may have stresses at yield, but this
region is still restrained by the remaining rigid portions of material in the vessel. When the load is
increased to the limit value P
0
, the plastic region has grown to an extent such that the rigid region
has either disappeared or has become insufficient to restrain the plastic region from motion. The
load for which overall plastic deformation of the vessel occurs is called the limit load. According to
limit analysis theory, it is impossible to have loads greater than the limit load for a perfectly plastic
material.
The Plastic Collapse Load P
c
The plastic collapse load P
c
is applied to the actual structure or vessel consisting of an actual strain
hardening material. It includes the effects of geometry change due to large deformations. At this
load, significant plastic deformation occurs for the structure or vessel as a whole (un-contained
plastic flow). The cause is the plastic region in the vessel, who now has grown to a sufficient extent
such that the surrounding elastic regions no longer prevent overall plastic deformation from
occurring. When this occurs, it may constitute a real failure, in the sense that the structure then can
no longer fulfil its intended function. The plastic collapse load can be used as a realistic basis for
design; an efficiently designed structure will be proportioned so that the external (operational)
actions would have to be increased by a specified factor (safety factor) in order to produce failure.
The limit load for an idealised structure then can be an approximation for the plastic collapse load
for the actual vessel, when it is largely plastic at small deflections.
The Ultimate Load P
u
At the plastic collapse load, the vessel does not necessarily collapse. Therefore, the adjective,
collapse, is unfortunate. The terminology of plastic deformation load or just plastic load would be
more meaningful. The load at which the vessel actually collapses is the ultimate load P
u
. An
example of an ultimate load is the burst pressure for a cylindrical vessel of sufficient ductility.
The Plastic Instability Load P
pi
Plastic instability loads can be of two types:
of the material instability type, and
of the structural instability type.
Plastic material instability corresponds for example to necking of a tensile specimen at the ultimate
load. The plastic structural instability load, depends upon the yield strength of the material, and is
accompanied by significant changes in shape of the structure or vessel. The plastic instability load is
important because its value is often less than the limit load.
The Shakedown Load P
S
All the above load definitions are for monotonic increasing loads. The shakedown load refers to
cyclic loading and is considered briefly because it is important to know the relative margin of safety
on shakedown of a design.
35
Design by Analysis
Page
2.35
DBA
Design by Analysis
If upon loading the structure beyond yield into the plastic range to a load value P > P
y
, and upon
unloading, a residual stress distribution is produced in the structure such that further cycles of load
to value P produce only elastic changes in stress, the structure is said to shakedown. The highest
value of P for which shakedown occurs is called the shakedown load P
s
. Failure to shakedown, i.e.
P > P
s
, leads to either progressive plastic flow called ratchetting, or to low cycle fatigue failure.
2.5.2 Limit Analysis Theory as Applied to Pressure Vessels
Consider a typical pressure vessel loaded by internal pressure, a perfectly-plastic material, small
deflections and increasing the pressure p.
At small values of p, the vessel material will be elastic and deformation of the vessel will increase
in proportion to p. However, as the pressure is continually increased, a region of the vessel becomes
plastic and the rate of deformation begins to increase, but deformation of the vessel as a whole is
usually still restrained by the surrounding elastic material. Finally, upon further increase in pressure,
a limit pressure or (in this case) a plastic collapse pressure is reached, where the plastic zone has
grown sufficiently large so that the deformation has suddenly begun to increase with little or no
additional increase in pressure. The problem then is this: What will be the magnitude of the limit
pressure of a particular pressure vessel? This is an important question in designing a vessel with a
sufficient margin of safety.
As described above, the problem from the beginning of loading involves initially elastic, then
elastic-plastic, and finally largely plastic behaviour. This is an involved and complicated loading
process. The theory of limit analysis, an idealised theory, enables the limit pressure to be found by
considering:
a rigid ideal-plastic, or a linear elastic ideal-plastic material, characterised by a sharply defined
yield limit (no strain hardening material),
the small displacement theory (any effect of geometry change of the shell due to deformation is
neglected).
These limitations must be kept in mind when applying limit analysis theory to certain problems
where the effects of strain hardening and geometry change may be important.
2.5.3 Elastic-Plastic Theory as Applied to Pressure Vessels
If effects of strain hardening and geometry change are important, an elastic-plastic analysis is to be
applied. Their influence on the load-deflection curvature is discussed.
Geometry Effects
Fig 2.27 shows a comparison between:
a) a small-deflection rigid-plastic limit load analysis,
b) a small-deflection elastic-plastic analysis,
c) a large-deflection elastic-plastic analysis, with geometrical weakening,
d) a large-deflection elastic-plastic analysis, with geometrical strengthening,
36
Design by Analysis
Page
2.36
DBA
Design by Analysis
e) a large-deflection elastic analysis, with geometrical weakening,
f) a large-deflection elastic analysis, with geometrical strengthening.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0 0.5 1 1.5 2 2.5
deflection (mm)
l
o
a
d

(
b
a
r
)
a
b
c
d
e
f
c
b
d
a
e
f
Fig 2.27 : Influence of geometrical effects
The small deflection elastic-plastic solution b) approaches the small-deflection rigid-plastic limit
load solution a) as expected. The large-deflection elastic-plastic solution, with geometrical
strengthening, gives a value higher than the limit load. The large-deflection elastic-plastic solution,
with geometrical weakening, gives a value lower than the limit load.
Effect of Strain Hardening
The effect of strain hardening is to increase the pressure capability above the limit load predicted by
the perfectly-plastic analysis, including the large deflection effect. Fig 2.28 shows that a higher
slope of the plastic part of the load-deflection curvature corresponds to a higher strain hardening
effect.
37
Design by Analysis
Page
2.37
DBA
Design by Analysis
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0 0.5 1 1.5 2 2.5
deflection (mm)
l
o
a
d

(
b
a
r
)
0%
2%
4%
6%
0%
2%
4%
6%
Fig 2.28: Influence of strain hardening effect
2.5.4 Estimation of plastic loads
Using an elastic-plastic analysis including strain hardening and large deflections or equivalently
considering experimental analysis of an actual vessel, one is confronted with the problem of
defining a realistic measure of plastic loads. A number of estimations have been used. These are
reviewed next. The discussion refers to pressure loading, but the same definitions can be applied to
other types of loadings.
The Limit Pressure p
0
Characteristic for the limit pressure definition according to the rigid perfectly-plastic theory is (with
p = pressure and = deflection)
dp/d = or d/dp = 0 for p < p
0
,
and
dp/d = 0 or d/dp = for p = p
0
.
Characteristic for the limit pressure definition according to the elastic perfectly-plastic theory is
dp/d > 0 for p < p
0
,
38
Design by Analysis
Page
2.38
DBA
Design by Analysis
and
dp/d = 0 or d/dp = for p = p
0
.
These definitions only apply for small-deflection analyses.
The Tangent-Intersection Pressure p
ti
(Fig 2.29)
Pti = 1.35 bar
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0 0.5 1 1.5 2 2.5
deflection (mm)
l
o
a
d

(
b
a
r
)
Fig 2.29 : Tangent-Intersection method
The tangent-intersection pressure is the pressure at the intersection of the two tangents, drawn to the
elastic and plastic parts of the pressure-deflection curves. The value of the pressure obtained by this
method is sensitive to the localisation of the tangent-point in the plastic range
The 1% Plastic Strain Pressure p
1
The plastic pressure is defined as the pressure with an equivalent plastic strain of 1%. Methods
based upon an absolute maximum strain not only will depend on the material assumed, but more
significantly on the geometry:
Material: e.g. a 1% plastic strain is ten times the yield point strain if the yield point stress is 150
MPa, but five times the yield point strain if the yield point stress is 300 MPa. Consequently, the
relative size of the elastic and plastic zones will differ and the shape of the pressure-deflection
response curves will differ.
Geometry: Ellipsoidal heads have been found to deform less than torispherical or toriconical
heads. Whereas a torispherical vessel may reach a 1% strain at a certain pressure, the ellipsoidal
vessel may reach the same pressure at a lower strain.
39
Design by Analysis
Page
2.39
DBA
Design by Analysis
At a yield hinge location, strains will be larger than at other locations. Consequently, the selection
of a strain gauge location on an experimental vessel presents a variable when yield hinge locations
are not known precisely or a priori.
Thus, in summary, a strain basis for defining a plastic pressure may be subject to error in locating
the exact location of maximum strain. Also, strain is a local phenomenon that is not indicative of
plastic work.
The Twice-Elastic-Deformation Pressure p
2y
A plastic pressure is defined to be the pressure at which the deflection or strain reaches twice the
value of the elastic deflection or elastic strain at the first yield pressure p
y
. Thus, p
2y
depends upon
p
y
. Exact determination of p
y
using a computer analysis should not be a problem. In experiments
however, determining the elastic limit on the load deflection curve may be subject to error.
The Twice-Elastic-Slope Pressure p

A plastic pressure is defined to be the value at the intercept of a line drawn from the origin of a
pressure-deformation curve at a slope of twice the slope of the elastic portion of the curve (see Fig
2.30).
y = 2.2x
y = 1.1x
P = 1.39 bar
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0 0.5 1 1.5 2 2.5
deflection (mm)
l
o
a
d

(
b
a
r
)
Fig 2.30 : Twice-Elastic-Slope method
40
Design by Analysis
Page
2.40
DBA
Design by Analysis
The 0.2% Offset Strain Pressure p
.2
The 0.2% offset strain pressure is a test pressure that causes a permanent strain of 0.2%.
The Proportional Limit Definition p
pl
P
pl
is a test pressure defined as the pressure causing the displacement versus pressure curve to
deviate from linearity. The displacement of the vessel is to be measured at the weakest point, the
most highly stressed point, giving the lowest value of p
pl
.
Analytical calculations can determine this pressure correctly. It will not necessarily be equal to the
first yield pressure p
y
. Experimental measures are subject to error in determining the point of
deviation from linearity. Values of p
pl
up to 30% greater than p
y
can be estimated from an
experimental curve.
This method of determining a plastic pressure will generally give a lower bound to the plastic
pressure found by most other methods.
The Plastic-Instability Pressure p
pi
This is an actual plastic collapse pressure and not just an estimate of a plastic pressure. It may be
identical to the limit pressure if large deflection effects are small, e.g. when the vessel is relatively
thick. However, the plastic-instability pressure may be less than the small-deflection limit pressure
as in the case of a large-deflection elastic-plastic solution, with geometrical weakening (see Fig
2.27, curve c). The plastic instability is defined by a zero slope on the pressure-deflection curve.
A large-deflection elastic-plastic analysis is required to detect p
pi
. It will also be detected in
experiments on actual vessels and it is possible to have plastic instability pressures less than lower-
bounds to the limit pressure where the latter are based on small-deflection analyses. It may occur
that some of the above estimations of the limit pressures will be non-conservative estimates of the
real plastic collapse pressure, an instability pressure, if the estimates were applied to small-
deflection theoretical results.
2.5.5 Inelastic Progressive Plastic Deformation - Shakedown
Within the DBA approach the determination of the limit load, for a given constitutive law, is one
step. Proving that Progressive Plastic Deformation (PD) will not occur, or more stringent that
neither PD nor Accumulating Plasticity (AP) will occur, in other words, proving that the structure
under consideration will shake down to pure elastic behaviour under cyclic varying actions, is
another step. Considering this proof, which is to be obtained through numerical simulation, the
following information may be useful:
- In this proof the constitutive law may be, but needs not to be the same law as used in the
determination of the limit load. Normally the structure shakes down under cyclic actions which are
to be specified as functions of a single parameter. This parameter determines the sequence of the
actions and quite often this parameter is time. The proof of shakedown is easier to perform than the
proof against PD. Being conservative this approach yields the proof that neither PD nor AP occurs.
41
Design by Analysis
Page
2.41
DBA
Design by Analysis
- In the inelastic simulation the proof of shakedown can be performed by applying the action cycle
repeatedly.
- The proof may also be performed using Melan's shakedown principle:
Using an equivalent linear elastic structure, a given cyclic action results in a corresponding
cyclic stress field. Additionally a time-independent self-equilibrating stress field should be
found, such that, using superposition of both stress fields, the stress intensity does not exceed
the yield limit at any time in the cycle.
This approach is especially attractive in those particular cases where an appropriate self-
equilibrating stress field is already known. A thermal stress field may serve as an example, as
well as the difference between a purely elastically determined stress field and the corresponding
field using plastic constitutive laws.
In many cases the proof may be performed using the check of primary + secondary stresses used
in a linear elastic DBA route, against the so-called 3f-criterion. The fulfilment of this criterion
is a necessary condition for shakedown. It is considered accurate enough for most cases
especially in combination with some other checks. However care should be taken whenever the
cyclic action contains a non-negligible time-invariant part e.g. a large contribution of self-
weight.
2.5.6 Discussion
Again it can be seen that apparently simple requirements of the inelastic route can be problematic.
Limit or shakedown analysis could be used to directly estimate the limit and shakedown loads, but
until recently this was difficult if not impossible for complex structures. If elasto-plastic finite
element analysis is used there remains the problem of defining the plastic load there are various
estimations as described above. The twice-elastic-slope method recommended by ASME has been
shown to give inconsistent results. The European standard aims to remove some of these problem
areas. In the following an overview of the new rules is given:
2.6 Design by Analysis in the European Standard
2.6.1 General
The European Standard has introduced the possibility of satisfying the requirement to avoid various
failure mechanisms directly through the detailed rules embodied in the new Direct Route, while
retaining the conventional elastic route which uses stress categorisation. In addition it also
introduces several new concepts to help overcome the known difficulties with the current design by
analysis approach and to assist the formulation of the Direct Route. In particular the notion of an
action rather than a force, and the inclusion of partial safety factors is a novel and welcome
addition to the area of pressure vessel design by analysis.
In the following some background to these new concepts is provided, followed by a summary of the
required design checks, with some explanation if required.
42
Design by Analysis
Page
2.42
DBA
Design by Analysis
2.6.2 New concepts
2.6.2.1 Principles and application rules
Like in the Eurocode (for steel structures) distinction is made between principles and application
rules. Principles comprise general statements, definitions and requirements for which there is no
alternative, and requirements and analytical models for which no alternative is permitted (unless
specifically stated). Application Rules are generally recognised rules which follow the principles
and satisfy their requirements; alternatives are allowed provided it is shown that they accord with
the relevant principle.
Typical examples are the primary and the primary & secondary stress criteria of the stress
categorisation approach, which are stated here, in slightly modified forms, as application rules.
2.6.2.2 Actions
This term, which replaces the old term loadings, denotes all thermo-mechanical quantities imposed
on the structure causing stress or strain, like forces (including pressure), temperature changes and
imposed displacements.
Actions are classified by their variation in time:
permanent actions (G)
variable actions (Q)
exceptional actions (E)
operating pressures and temperatures ( T p, ) - although these are variable actions, they are
considered separately to reflect their special characteristics (variation in time, random
properties, etc.).
The notion variable actions encompasses actions of quite different characteristics from those
actions which are deterministically related to pressure and/or temperature, via actions not correlated
with pressure or temperature but with well defined (bounded) extreme values, to actions which can
be described only as stochastic processes not correlated with pressure or temperature, like wind
loads. Actions with a deterministic relationship with pressure and/or temperature shall be combined
in the pressure/temperature action and the relationship, exact or approximate, shall be used.
The characteristic values of actions describe the regime of actions which envelops all the actions
that can occur under reasonably foreseeable conditions. The characteristic values are used in
determining the design values of the actions, and they depend on the actions' (statistical) properties.
The characteristic values of permanent actions are usually their mean values (or credible extreme
values). The characteristic values of variable actions are defined as mean values, or p% -
percentiles, of extreme values, and values specified in relevant codes for wind, snow, earthquake
may be used; usually they are adapted to Eurocode concepts anyway. The upper characteristic value
of pressure shall not be smaller than the lesser of the set pressure of the protecting device or the
highest credible pressure that can occur under normal and upset conditions (reasonably
foreseeable), and the upper characteristic value of the temperature not smaller than the highest
credible temperature (under the same conditions). Therefore, the (limited) pressure excursion
43
Design by Analysis
Page
2.43
DBA
Design by Analysis
(overpressure) that occurs if a safety valve opens need not be included in the (maximum)
characteristic value of pressure; it is taken care of in the partial safety factors.
2.6.2.3 Partial safety factors
To allow for an easy, straightforward combination of pressure action with environmental ones, and,
at the same time, to give the flexibility, expected from a modern code, to adjust safety margins to
differences in action variation, likelihood of action combinations, consequences of failure,
differences of structural behaviour and consequences in different failure modes, uncertainties in
analyses, a multiple safety factor format was introduced, using different partial safety factors for
different actions, different combinations of actions, different failure modes and corresponding
resistances of the structure. Examples of partial safety factors are given in the following Table. The
corresponding combination rules for e.g. Design Check GPD-OC Global Plastic Deformation
Operating Conditions are:
all permanent actions shall be included in each load case
each pressure action shall be combined with the most unfavourable variable action
each pressure action shall be combined with the corresponding sum of variable actions;
stochastic actions may be multiplied by the combination factor.
favourable actions shall not be considered.
The partial safety factors of pressure and resistances are calibrated with respect to the DBF results;
no attempt has been made to justify the partial safety factors by probabilistic investigations or
decision theory under uncertainty; if pressure is the only action the approach can be transformed to
a nominal design stress one.
Partial safety factors
Design check
Actions GPD-OC GPD-HT
Permanent
G

Unfavourable 1.35 1.35


Favourable 1.0 1.0
Pressure
P

1.2 (1.0) 1.0


Variable
Q

1.5 (1.0) -
Combination factor 0.9
1
1.0
1
(stochastic actions)
Resistance
R

1.25 1.05
(Temperature
T
)
(1.0) (1.0)
1
If not specified differently in the relevant code of environmental actions.
2.6.2.4 Design checks effects of actions
Design checks are investigations of the structure's safety under the influence of specified
combinations of actions - the design load cases - with respect to specified limit states (representing
44
Design by Analysis
Page
2.44
DBA
Design by Analysis
one or more failure modes). Characteristic values of the actions are multiplied by the corresponding
partial safety factors to obtain their design values and their combined design effect (on the
structure) is evaluated:
E G p Q a
d G p Q d
( , , , . . . , , . . . )
In the design checks these design effects are compared with the corresponding design resistances,
obtained by dividing the resistance of the structure, corresponding to the action's combination, by
the relevant partial safety factor of the resistance:
E R R G p Q a
d d d R
= ( , , , . . . , , ) /
This comparison can, in general, be performed in actions, in stress resultants (generalized stresses)
or in stresses.
The resistances are related to the limit states - states beyond which the part no longer satisfies the
design performance requirements.
2.6.2.5 Design checks resistances
Design checks are designated by the failure modes they deal with. The following ones are
incorporated in the first issue of the standard:
gross plastic deformation (GPD), with corresponding failure modes ductile rupture and, for
"normal" designs, also excessive local strains
progressive plastic deformation (PD)
instability (I)
fatigue (F)
static equilibrium (SE).
Checks against gross plastic deformation
The design resistances are given by the lower-bound limit loads for
proportional increase of all actions
a linear-elastic ideal-plastic material (or a rigid ideal-plastic one)
first-order theory
Tresca's yield criterion and associated flow rule
specified design strength parameters.
Design strength parameters RM and partial safety factors of the resistances
R
are chosen such that
for the simplest structures and pressure action only DBA and DBF results agree. The only exception
are steels, other than austenitic ones with A
5
30% , where the design strength parameter RM is
given by R
eH T ,
or R
p T 0 2 . ,
and
R
= 1 25 . for R R
eH m
/ . 0 8 and
R eH m
R R = 1 5625 . / otherwise.
45
Design by Analysis
Page
2.45
DBA
Design by Analysis
If the procedure used to determine the limit action does not give an (absolute) maximum in the
region with maximum absolute values of principal strains less than 5%, the boundary maximum, for
which the maximum absolute value of the principal strains equals 5%, shall be used.
As an application rule the "usual" primary stress criterion is given, formulated in stresses and - for
structures where the concept of stress resultants is applicable - in stress resultants and local
(technical) limit loads.
These checks (against GPD) are considered also to encompass Excessive Yielding, provided
"usual" design details (with not too severe strain concentrations) exist.
Checks against progressive plastic deformation
On repeated application of specified action cycles PD shall not occur for
a linear-elastic ideal-plastic material
first-order theory
Mises' yield condition and associated flow rule
specified design strength parameters RM.
A slight modification of the "usual" f 3 criterion is given as application rule; it is noted that this
application rule, which is derived from shakedown considerations, is only a necessary condition for
the fulfilment of the principle, but is considered, together with all the other checks, to be sufficient
to achieve the principle's goal - avoidance of ratchetting in the structure.
Check against fatigue failure
Reference is made to the Fatigue Assessment section of the Standard.
Instability
Static equilibrium
The usual checks against overturning and (rigid body) displacement are stated explicitly, using the
partial safety factors given in the other checks.
2.6.3 Application remarks
Whether the Direct Route or the Stress Categorisation Route is followed, it is imperative that all
stated checks are considered:
in the Direct Route:
At least the five given checks sometimes it may be necessary to include additional ones, like
excessive deformation (leakage). Not all of the checks will require calculations, but all must be
considered e. g. it may be obvious that instability can be excluded.
in the Stress Categorisation Route:
46
Design by Analysis
Page
2.46
DBA
Design by Analysis
Check of Primary Stresses
Check of Primary + Secondary Stresses
Check of Total Stresses (Primary + secondary + peak stresses) - Fatigue
Usually it is required to perform each of these checks for different load cases for different
combinations of coincident actions, as well as for different characteristic values of actions, e. g.
different pressure temperature pairs.
The Design by Analysis route may be chosen to prove conformity of a design also for a part of a
component, suitably selected and limited; and with appropriate boundary conditions.
47
Design by Analysis
Page
2.47
DBA
Design by Analysis
2.7 REFERENCES
[1] W. C. Kroenke, J. L. Hechmer, G. L. Hollinger & A. J. Pedani, Component evaluation using
the finite element method. In Pressure Vessel & Piping Design: A Decade of progress, 1970-
1980, Chap. 2.11, ASME, 1981.
[2] W. C. Kroenke, Classification of finite element stresses according to ASME Section III stress
categories, Proc 94th ASME Winter Annual Meeting, 1973.

[3] W. C. Kroenke et al, Interpretation of finite element stresses according to ASME III, ASME
Tech. Paper 75-PVP-63, 1975.
[4] N.V.L.S. Sarma, G. L. Narasaiah & G. Subhash, A computational approach for the
classification of FEM axisymmetric stresses as per ASME Code, Proc ASME Pressure Vessel &
Piping Conf, Pittsburgh, 1988.
[5] J. L. Gordon, OUTCUR: An automated evaluation of two dimensional finite element stresses
according to ASME, ASME Paper 76-WA/PVP-16, 1976.
[6] J. L. Hechmer & G. L. Hollinger, Considerations in the calculations of the primary plus
secondary stress intensity range for Code stress classification, Codes & Standards and
Applications for Design and Analysis of Pressure Vessel and Piping Components Ed R. Seshadri,
ASME PVP Vol.136, 1988.
[7] B. W. Leib, An automatic surface element generator for calculating membrane and bending
stresses from three dimensional finite element results, Proc 4th Int Conf on Structural Mechanics
in Reactor Technology, San Francisco, 1977.
K. H. Hsu & D A McKinley SOAP - a computer program for classification of three
dimensional finite element stresses on a plane, Proc ASME Pressure Vessel & Piping Conference,
Nashville, 1990.
[8] J. L. Hechmer & G. L. Hollinger, Three dimensional stress criteria - a weak link in vessel
design and analysis, ASME Special Publ. PVP 109 A Symposium on ASME Codes and Recent
Advances in Pressure Vessel and Valve Technology Ed J. T. Fong, 1986.
J. L. Hechmer & G. L. Hollinger, Three dimensional stress criteria -application of Code rules,
ASME Special Publ. PVP 120 Design and Analysis of Piping, Pressure Vessels and Components
Ed W. E. Short, 1987.
J. L. Hechmer & G. L. Hollinger, Code evaluation 3D stresses on a plane, Codes & Standards
and Applications for Design & Analysis of Pressure Vessels & Piping, ASME PVP-Vol.161, 1989.
[9] J.L. Hechmer & G.L. Hollinger, Three dimensional stress criteria, ASME PVP-Vol.210-2
Codes and Standards and Applications for Design and Analysis of Pressure Vessel & Piping
Components, Ed R. Seshadri & J.T. Boyle, 1991.
G. Hollinger, Summary of three dimensional stress classification, Proc Int Conf on Pressure
Vessel Technology, Dusseldorf, 1992.
48
Design by Analysis
Page
2.48
DBA
Design by Analysis
[10] R.L. Roche, Practical procedures for stress classification, Int Journ Press Vess & Piping,
Vol.37, 27-44, 1989.
[11] D.L. Marriott, Evaluation of deformation or load control of stresses under inelastic conditions
using elastic finite element analysis, Proc ASME Pressure Vessel & Piping Conf, Vol.136,
Pittsburgh, 1988.
[12] A. Kalnins & D.P. Updike, Role of plastic limit and elastic plastic analyses in design, ASME
PVP-Vol.210-2 Codes and Standards and Applications for Design and Analysis of Pressure Vessel
& Piping Components, Ed R. Seshadri & J.T. Boyle, 1991.
A. Kalnins & D.P. Updike, Primary stress limits on the basis of plasticity, ASME PVP-
Vol.230, Stress Classification, Robust Methods and Elevated Temperature Design, Ed R. Seshadri
& D.L. Marriott, 1992.
[13] D. Mackenzie & J. T. Boyle, A computational procedure for calculating primary stress for the
ASME B&PV code, Trans ASME, Jrn Pressure Vessel Tech, Vol. 116, No. 4, 1994.
D. Mackenzie, J. Shi, R. Hamilton & J. T. Boyle, "Simplified lower bound limit analysis of
pressurised cylinder-cylinder intersection Shells using a generalised yield criteria", Int Jrn of
Pressure Vessels & Piping, 67, pp. 219-226, 1996.
J. T. Boyle, R. Hamilton, J. Shi, & D. Mackenzie, "A simple method of calculating Limit Loads
for thin axisymmetric shells", Trans. American Society of Mechanical Engineers (ASME), Jrn
Pressure Vessel Technology, Vol. 119, No.2, pp. 236-242, 1997.
[14] R. Seshadri & C.P.D. Fernando Limit loads of mechanical components and structures using
the GLOSS r-node method, Proceedings of ASME PVP, Vol. 210-2, pp. 125-134, 1991.
[15] A.R.S. Ponter, K.F. Carter, Limit state solutions, based upon linear elastic solutions with a
spatially varying elastic modulus, Jrn of Computer Methods in Applied Mechanics and
Engineering, Vol.140, No.3-4, pp.237-258, 1997.
A.R.S. Ponter, K.F. Carter, Shakedown state simulation techniques based on linear elastic
solutions, Jrn of Computer Methods in Applied Mechanics and Engineering, Vol.140, No.3-4,
pp.259-279, 1997.
[16] T. Seibert, J. L. Zeman, Analytischer Zulssigkeitsnachweis von Druckgerten, Techn.
berwachung, Bd. 35 (1994) Nr. 5, 222-228.
W. Poth, J. L. Zeman, Grenztragfhigkeit der Zylinder-Kegel-Verbindung unter
Innendruckeinwirkung, Konstruktion 48 (1996), 219-223.
J. L. Zeman, Ratcheting limit of flat end cylindrical shell conections under internal pressure,
Int J Pres Ves & Piping 68 (1996), 293-298.
R. Preiss, F. Rauscher, D. Vazda, J. L. Zeman, The flat end to cylindrical shell connection
limit load and creep design, Int J Pres Ves & Piping, 75 (1998),715-726
[17] D. Mackenzie, J. T. Boyle & R. Hamilton, Application of Inelastic Finite Element Analysis to
Pressure Vessel Design, International Conference on Pressure Vessel Technology, Volume 2,
ASME 1996.
49
Design by Analysis
Page
2.49
DBA
Design by Analysis
[18] J. C. Gerdeen, A critical Evaluation of Plastic Behaviour Data and a United Definition of
Plastic Loads for Pressure Components, WRC Bulletin 254, November 1979, ISSN 0043-2326.
[19] A. Kalnins, D. Updike & J.L. Hechmer, On Primary Stress in Reducers, ASME PVP-Vol.
210-2, pp. 117-124
[20] D. Mackenzie, J.T. Boyle, J. Spence, "Some Recent Developments in pressure vessel Design
by Analysis" Proc IMechE, Part E, Journal of Process Mech Eng, 1994, 208, 23-30.
1
Procedures
Page
3.1
DBA
Design by Analysis
3 Procedures
3.1 General
Procedures, used in section 7 in the various checks, which are of a more general nature and which
would have to be repeated in the design checks frequently, are collected here in this section. Also
included are general remarks on the various routes and possibilities within one route.
3.2 The Two Routes
There are two approaches for carrying out a DBA that cover both the direct route method and the
stress categorisation method. If in a design the values of the actions are defined, a check on the
admissibility of the actions, i.e. an adequacy check, can be made. Alternatively, if only the
geometry is defined but not the magnitude of the actions, DBA may be used to calculate the
maximum allowable values of the actions.
3.2.1 Admissibility check for defined actions
For a given design where the actions are defined, DBA may be used to check if the defined actions
are admissible. It is not necessary in this case to calculate an upper limit of the actions, although by
doing so one will automatically check the admissibility of the defined actions. In the admissibility
check it is only required to show that the defined actions do not exceed the limits defined in the
applicable section of the code. For example in the direct route check against GPD, for admissibility
it is only necessary to show that (for the defined actions with appropriate safety factors) the
resulting elasto-plastic stress field is an equilibrium stress field where the absolute maximum total
principal strain does not exceed 5%. (For the GPD-check using elastic compensation admissibility
of the actions is shown if the maximum stress in the redistributed equilibrium stress field does not
exceed the design resistance). In the case of a stress categorisation route, admissibility is shown if
the linearised stress categories do not exceed the limits defined in the code rules, prEN 13445-3
Annex B.
3.2.2 Determination of maximum allowable actions
For a given design where only the type of action is specified, DBA may be used to calculate the
maximum allowable actions. The maximum allowable actions are given by the actions that place
the structure at the defined limits as specified in the applicable section of the code. In the direct
route the maximum allowable action according to GPD is usually calculated using limit analysis. In
elasto-plastic analysis the actions are increased until the limit is reached, where loss of equilibrium
occurs or the absolute maximum total principal strain exceeds 5%. The value of the action at this
limit with the appropriate partial safety factors applied is the maximum allowable action according
to the GPD-check. (In elastic compensation, the redistributed equilibrium stress field is scaled,
along with the applied action to the design resistance; the value of the scaled action with the
appropriate safety factors applied is the maximum allowable action). In the check against PD, a
lower bound of the maximum allowable actions can be calculated by finding the maximum actions
for which the structure will shake down (see the procedure for Melans theorem below).
2
Procedures
Page
3.2
DBA
Design by Analysis
For the stress categorisation approach the maximum allowable action can be found by scaling the
applied action (by evoking the linear proportionality of the elastic solution) to a point where the
linearised stress categories are at the limits defined in the code rules, prEN 13445-3 Annex B.
3.3 Direct Route (using elasto-plastic calculations)
3.3.1 Check against Gross Plastic Deformation (GPD)
Generally, two different ways of performing calculations according to the GPD-check are possible:
First, if the actions for a given structure are specified, a real check can be performed, to show that
the actions are admissible (or not) under application of prEN 13445-3 Annex B. This procedure was
applied within the GPD-check of the examples 2, 5 and 6. The second possibility is to calculate the
limit actions for a given structure, and using these limit values then to determine the maximum
admissible actions according to prEN 13445-3 Annex B. The latter procedure will be useful, if the
structure is to be used with extreme actions. This procedure was applied within the GPD-check of
the examples 1.1, 1.2, 1.3, 1.4, 3.1, 3.2 and 4.
As stated in the application rule in prEN 13445-3 Annex B.9.2.2, which fulfils the principle in the
check against GPD - prEN 13445-3 Annex B.9.2.1 the limit load has to be determined using
Trescas yield condition with associated flow rule and first order theory. Since there is no
standardised subroutine for Trescas yield condition in the ANSYS software, a made-to-order
subroutine of an ANSYS distributor was used. Unfortunately, for some structures the subroutine
showed bad convergence and/or long computation times. The cause may be the edges in Trescas
yield surface.
If no subroutine for Trescas yield condition is available or if it shows too bad convergence, Mises
yield condition can be used instead. Since the maximum ratio of the Mises equivalent stress to the
Tresca equivalent stress for the same load is 3 2 , a multiplication of the design resistance with
2 3 will always lead to conservative results.
Furthermore, if the result of the check against PD (where Mises yield condition is allowed) is used
in the check against GPD, instead of a separate calculation, and if there is only one partial safety
factor
R
for the considered structure, multiplication of the limit pressure from the check against
PD with 2 3 leads to the same result as the multiplication of the material strength parameter. Of
course, since no partial safety factors are used in the check against PD they have to be taken into
consideration by scaling down the PD-check results. Of course, if the check against GPD is
performed first, the results can be used in the check against PD, using the relevant factors.
As stated in the application rule in prEN 13445-3 Annex B.9.2, the maximum absolute value of the
principle strains must not exceed 5%. To fulfil this demand in cases where the results from the PD-
check are used, such a value of the pressure from the check against PD shall be used that the
maximum absolute value of the principle strains calculated with this pressure (and the material
strength parameter used for the check against PD) does not exceed 5 %. If this procedure leads to
too conservative results (because of the restraint by the 5 % - limit) an additional FE-calculation
using the scaled down material strength parameter is required to determine the exact 5% - limit
pressure.
Because of this strain limitation in the GPD-check, there is now in general a difference between
3
Procedures
Page
3.3
DBA
Design by Analysis
the design resistance, defined as ratio of the resistance (of the specified model) and the
appropriate partial safety factor, where the resistance is determined with (specified)
material strength parameters
the resistance of a corresponding model but with material strength parameters replaced by
the corresponding design material strength parameters, obtained by dividing the material
strength parameters by the partial safety factors of the resistance.
Usually the difference is very small, and it is recommended that the resistance in the second route
may be used as design resistance; this recommendation has already been used in the examples.
Performing a real GPD-check only, e.g. no limit actions are determined, the admissibility of
specified actions acting on a specified structure is shown, if the maximum absolute value of the
principal strains does not exceed 5%, under usage of a linear-elastic ideal-plastic material law in the
FE-calculations.
If the FE-model consists of shell elements, usually the mid-surface of the structure is modelled.
Therefore, the practical relevance of the results in points (nodes) on the intersection curve of two
shells depends on the kind of geometry of the structure.
For example, for (cylindrical) main shell
(cylindrical) nozzle intersections the results in
nodes of the intersection curve should not be
used, since they do not correspond to points of
the real structure with the real geometry under
consideration of the reinforcement due to the
weld - see Figure 3.1. Therefore, if usage of
solid elements or of submodelling is not
possible, the results in the so called
evaluation cross-sections should be used for
the determination of the 5% principal strain
limit.
Figure 3.1: Shell intersection
As stated in the principle in prEN 13445-3 Annex B.9.2.1, the design resistance (limit action)
should be obtained from calculations with proportional increase of all design actions. The limit
action is independent of the action history, but with strain limitation, i.e. if the strain limitation
governs, the limit will depend on the actions history. In the case of constant moment load and
varying internal pressure load (examples 3.1 and 3.2), where the strain limitation does govern, the
deviation from the standards procedure proportional loading is the only sound one, the moment
being constant during all action cycles.
3.3.2 Check against Progressive Plastic Deformation (PD)
3.3.2.1 General
Again, two different ways of performing calculations to check PD are possible: First, if the action
cycles for a given structure are specified, a real check can be performed, showing that the actions
are admissible (or not) under application of prEN 13445-3 Annex B. This procedure was applied
within the PD-check of the examples 2, 5 and 6. The second possibility is to calculate the limit
4
Procedures
Page
3.4
DBA
Design by Analysis
actions (in the sense of PD) for a given structure, and afterwards using these limit actions to
determine the maximum admissible actions according to prEN 13445-3 Annex B. The latter
procedure will be useful, if the structure is to be used with extreme actions. This procedure was
applied within the PD-check of the examples 1.1, 1.2, 1.3, 1.4, 3.1, 3.2 and 4.
Principally, the application rule in prEN 13445-3 Annex B.9.3.2 corresponds to the well-known
criterion for the sum of the primary and secondary stresses in stress categorisation the (often) so-
called 3fcriterion (where f stands for the allowable stress). This criterion is an upper bound
criterion for shakedown, and, therefore, the requirement given in this application is only a necessary
and not a sufficient condition for the fulfilment of the principle prEN 13445-3 Annex B.9.3.1
[1],
[2]
. Usage of this application rule could be the easiest way of applying the check against PD if only
one action is considered, but if more than one action and/or additional thermal stresses have to be
considered, its usage could be difficult and uncertain. Therefore, and for guideline purposes, usually
another possibility of fulfilling the principle Melans shakedown theorem is employed.
Melans shakedown theorem states
[1], [2]
:
The structure will shake down for a given cyclic action, if a time-invariant self-equilibrating
stress field can be found such that the sum of this stress field and the cyclically varying
stress field determined with the (unbounded) linear-elastic constitutive law for the given
cyclic action is compatible with the yield condition the equivalent stress nowhere and at no
time exceeds the (yield) material parameter.
Using this theorem the principle is fulfilled, since Progressive Plastic Deformation (PD) and
Alternating Plasticity (AP) are the two possible inadaption modes if a structure does not shake
down under a cyclic load set.
One advantage of using Melans theorem, in comparison with the application rule, is, that the
admissibility is shown for all points of the structure, if the determined self-equilibrating stress field
is superposed onto the linear-elastic stress fields of the different states using the postprocessor of
the FE-software.
A problem arises if the maximum allowable action given by the shakedown limit is lower than the
one resulting from the check against GPD. In this case, detailed examination of the structures
behaviour under a cyclic loading with the maximum allowable action according to the check against
GPD, i.e. determination of the structures inadaption mode progressive plastic deformation and/or
alternating plasticity - would be necessary. If it can be shown that the inadaption mode is given by
alternating plasticity only, the action would be admissible. Unfortunately, the possibility of such
examinations is restricted, on one hand because of the hardware and time limits for performing
cyclic calculations and on the other hand because of a lack of generally applicable theorems.
Determining generally applicable theorems in this field is a present topic of research, and, therefore,
they should be available in the future.
3.3.2.2 Problems in performing the shakedown check using shell elements in the FE-model
[3]
Usage of stress resultants of technical theories of structures, i.e. generalised stresses, to verify that
progressive plastic deformation (PD) does not occur is often not appropriate, because
the validity of the corresponding theorem
[4]
seems to be restricted to passive (unloading)
processes
[2], [5], [6]
;
5
Procedures
Page
3.5
DBA
Design by Analysis
(local) interaction surfaces are available for rotational-symmetric shells under rotational-
symmetric loading only.
Therefore, a shakedown check using stresses has to be carried out quite often. However, if shell
elements are used, the only stress results available in the postprocessor are surface stresses (top /
bottom of the shell, i.e. outer / inner surface).
Usage of Melans shakedown theorem and corresponding self-equilibrating stress states to ensure
that a structure shakes down under a given cyclic load path could then be non-conservative, because
the fulfilment of the theorem can only be proven for surface stresses but not in the interior of the
shell model.
In fact, when the absolute maximum of the self-equilibrating stress distribution in a cross-section is
located in the interior, the above stated non-conservatism is possible.
In the elasto-plastic simulation of a structure's behaviour during the loading or unloading half cycle,
three phenomena can occur:
(1) if the surface equivalent stress in at least one point of the structure remains, or becomes equal to
the material strength parameter, yielding occurs during the half cycle, on the surface and/or in
the interior;
(2) if the surface equivalent stress in all points of the structure is less than the material strength
parameter, no yielding occurs on the surfaces during the half cycle, but yielding inside of cross-
sections cannot be excluded;
(3) only if the stress state on the surface after the half cycle is identical to the one resulting from
purely elastic loading or unloading of the structure, no stress redistribution due to plastic
deformation occurs, neither at the beginning nor at the end of the half cycle. Therefore, it is
assured that the half cycle is purely elastic and the structure has shaken down.
There follows that to verify in a strict manner that a structure modelled by shell elements shakes
down, one has to show that condition (3) is fulfilled.
If computation of the half cycle shows that condition (1) is fulfilled, further cycles are necessary
until condition (2) is fulfilled - if actually possible. Afterwards, to verify that the structure shakes
down, one has to show that condition (3) is fulfilled. However, using this procedure can be difficult
if considerable numerical errors, for example due to extrapolations and averaging, are present in the
stress plots.
Alternatively, if the FE-software confirms that no further plastic strains occur during a half cycle
(after some initial load cycles) - requiring a suitable parameter in the computation output -, it is
proven that the structure has shaken down, independently of possible numerical errors in the
postprocessor.
Additionally, the practical relevance of the results in points (nodes) on the intersection curve of two
shells depends on the kind of geometry of the structure, and, therefore, care must be taken when
using these results. Using evaluation cross-sections in a similar manner as in the check against
GPD is not a satisfying solution, since the stationarity of stress cycles cannot be proven if the whole
structure is not considered.
A possibility of showing that Melans theorem is fulfilled by performing the shakedown check at
the shell model, i.e. only at the surface but not in the interior of the structure, is to use submodeling.
Submodeling is a finite element technique to obtain more accurate results in a region; by
performing an analysis of a coarse model (shell model), interpolating the results to the (cut-)
6
Procedures
Page
3.6
DBA
Design by Analysis
boundaries of the fine model (solid element model), which only represents the critical part of the
structure, and by computing the stress distribution in the fine model afterwards. If this stress
distribution confirms that the maximum stresses at the critical parts of the structure are on the
surface of the model, the shakedown check using the shell model fulfils the theorem.
3.3.2.3 Problems concerning stress singularities
Performing linear-elastic calculations as necessary for a shakedown check , stress singularities
[8]
can arise specific points of a weld modelled without fillets.
Since it is stated in the standard see prEN 13445-3 Annex B.3.9.3.2 that the check against PD
can be performed for a stress-concentration-free structure, different possibilities of avoiding the
stress singularities are possible:
modelling the welds with fillets corresponding to the weld influence zones - see Figure 3.2. This
procedure is suitable, if 2-D FE-models are used.
modelling the welds with fillets, which are completely inside of the weld see Figure 3.3. This
procedure is suitable if 3-D FE-models are used, and if modelling of fillets corresponding to the
weld influence zones would be too difficult and time consuming.
Figure 3.3: Fillet inside the weld
3.3.2.4 Use of the deviatoric map (for constant principal stress axes)
[7]
The deviatoric projection is a simple tool in plasticity theory for visualising stress states vis-a-vis to
yield conditions. In principle, it is the projection of the stress point in the (three dimensional,
Figure 3.2: Fillets-weld influence zones
7
Procedures
Page
3.7
DBA
Design by Analysis
Cartesian) space of principal stresses with co-ordinates in the directions of the unit vectors
3 2 1
e e e , , equal to the three principal stresses onto the deviatoric plane, also often called -plane,
i.e. the plane which is normal to the hydrostatic axis, given by equal principal stresses. The co-
ordinates of this projection point in the co-ordinate system
3 2 1
e e e , , equal the principal values of
the stress deviator.
Using this deviatoric projection as a tool, this projection point can be obtained quite simply by
vector addition of
d d
e e
2 1 2 1
, , and
d
e
3 3
, with arbitrary scale, see Figure 3.4; quite conveniently
3 , 2 , 1 , / i RM
i i
, can be used instead of
i
, where RM is the appropriate relevant strength
parameter.
The vector d from the origin of the deviatoric map to a specific stress point in a Cartesian co-
ordinate system, with
y
e =
d
e
2
and
x
e to the right, has the components
2 / 3 ) ( 3 1 in the direction of
x
e
] 2 / ) ( [ 2 3 1 + + in the direction of
y
e
if the standardised principal stresses are used.
If the considered vector d corresponds
to the stress in a specific point of the
structure due to a specific action, the
location of its end point vis-a-vis a limit
curve, e. g. the unit circle with centre in
the origin which corresponds to Mises'
yield condition, visualises clearly the
instantaneous behaviour of the structure
under the considered action: If the end
point is not outside of the limit curve,
the stress state is compatible with the
corresponding yield condition; if it is on
the limit curve plastic deformation may
occur "at the structural point".
If the action changes - and the
orientation of the principal axes does
not change -, the stress state may
change, and the vector d may change.
If a cyclic action results in a cyclic stress state, the vector will describe a closed curve in this
deviatoric map. A curve completely inside of the limit curve corresponds to only elastic stress
states, portions on the limit curve correspond to plastic flow, portions outside are possible only in
case of hardening, or for fictitious stress states, e. g. for a fictitious unlimited linear elastic
constitutive law.
e e
e
e
e
d
2
d
=
y
d
3
x
d
1
2
e
d
e
d
1
d
e 3
Figure 3.4: Deviatoric projection
8
Procedures
Page
3.8
DBA
Design by Analysis
If two actions A
1
and A
2
act simultaneously and both result in stress states (at a specific point of the
structure) with the same principal axes, the vector property of the deviatoric mapping can be used:
If
1
d corresponds to
1
A , and
2
d to
2
A (determined with a fictitious unlimited linear elastic
constitutive law), then d =
1

1
d

+

2

2
d corresponds to the stress state due to
2 2 1 1
A A + , where
1
and
2
may be functions of time.
If
3
d corresponds to A for a specific constitutive law, e. g. a linear elastic ideal plastic
one or one used in an elastic compensation procedure, and
4
d corresponds to A for a
different constitutive law, e. g. a linear elastic one, then d =
3
d -
4
d corresponds to a
self equilibrating stress state
Melan's shakedown theorem can be reformulated in the vectors of the deviatoric mapping:
If a value can be found such that the end point of the vector
(
3
d -
4
d ) + ) (
1
t
1
d + ) (
2
t
2
d
for the prescribed cyclic action
2 2 1 1
) ( ) ( A t A t + is never outside the limit curve, then the structure
will, at the considered structural point P, shake down to purely elastic action.
Even the necessary condition for shakedown, resulting immediately from Melan's shakedown
theorem and usually designated as 3f-criterion, can be easily visualised:
If the largest diameter of the path of the vector's end point the stress path -,
) (
1
t 1 d + ) (
2
t 2 d , corresponding to a cyclic action
2 1
) ( ) ( A t t + in a specific
, 2RM
then the structure, with the specified RM, cannot shake down under this cyclic action.
The extension of this procedure to more than two actions, or to more than one self-equilibrating
stress, including, of course, those due to thermal stresses, is obvious and straightforward.
3.3.2.5 Shakedown Analysis for a single varying action (internal pressure)
In the following procedure, it is assumed that the only action acting on the structure under
consideration is internal pressure, and that it varies between zero and the maximum admissible
pressure for shakedown PS
max SD
.
The problem in using Melans theorem is to find an optimal self-equilibrating stress field. Often, the
optimal, or a near optimal, stress field can be found from the stress fields at the limit load:
The difference of the linear-elastic stress field at (or near) the limit pressure of the structure
l le ij ,
) (
and the elasto-plastic stress field at (or near) the limit pressure
l ep ij ,
) ( is a self-equilibrating stress
field
res ij
) ( :
l le ij l ep ij res ij , ,
) ( ) ( ) (
9
Procedures
Page
3.9
DBA
Design by Analysis
Since one endpoint of the considered linear-elastic load cycle is the point PS = 0, the self-
equilibrating stress field used in Melans theorem must not violate the yield condition itself.
Therefore, the self-equilibrating stress field
res ij
) ( has to be scaled with a factor such that it
does not violate the yield condition:
res ij co res ij
) ( ) (
,
.
If the uncorrected self-equilibrating stress field according to the limit load
res ij
) ( does not violate
the yield condition, the shakedown load is equal to the limit load.
After this correction of the self-equilibrating stress field, the linear-elastic stress field with the
possible greatest value of internal pressure has to be determined, such that the superposition with
the corrected self-equilibrating stress field does not violate the yield condition. Because of the
linearity, this can be done exactly for a fixed point of the structure. The stress field at a lower bound
shakedown limit
SD ij
) ( is found as
l le ij co res ij SD ij , ,
) ( ) ( ) ( + .
Note: The scaling factors and can be determined easily using the equivalent stress plots (or
listings).
Note: Another possibility of finding a self-equilibrating stress state is given by elasto-plastic
unloading of the structure within an FE-calculation.
3.3.2.6 Shakedown analysis for one constant action (nozzle moment) and a single varying
action (internal pressure)
In the following procedure, the structures under consideration are cylinder-cylinder intersections,
where a constant moment load M is acting at the nozzle and the cyclic load is given by the internal
pressure, which varies between zero and the maximum admissible pressure for shakedown PS
max SD
.
Again, the problem in using Melans theorem is the determination of an optimal self-equilibrating
stress field.
The difference of the elasto-plastic stress field
) ( ,
) (
p M ep ij +
, which corresponds to a loading state
with the constant moment M and an internal pressure near the limit state of the structure, and the
linear-elastic stress field at this state
) ( ,
) (
p M le ij +
, the stress field
) ( , ) ( , ) ( ,
) ( ) ( ) (
p M le ij p M ep ij p M res ij + + +

is a self-equilibrating stress field.
One endpoint of the considered linear-elastic load cycle is the point (
SD
PS p M
max
, ), the other
endpoint is given by ( 0 , p M ), where
SD
PS p
max
is the maximum admissible pressure for
10
Procedures
Page
3.10
DBA
Design by Analysis
shakedown of the structure. Therefore, to fulfil Melans theorem, the following conditions have to
be met:
0 ] ) ( ) ( ) ( [
, , ) ( ,
+ +
+ le p ij le M ij p M res ij
,
0 ] ) ( ) ( [
, ) ( ,
+
+ le M ij p M res ij
,
where is the yield condition,
le M ij ,
) ( is the linear-elastic stress field corresponding to the
moment , M
le p ij ,
) ( is the linear-elastic stress field corresponding to an arbitrary value of internal
pressure p , and and are the factors which have to be determined such that the conditions are
fulfilled. Thus, the maximum admissible internal pressure according to shakedown is given by
p PS
SD

max
.
Defining and using load cases in ANSYS, the superposition of stress fields can be done easily.
Unfortunately, in the examples quite often no combination of stress states, i.e. no factors and ,
could be found, such that the two conditions above were fulfilled. Often, the Mises equivalent
stresses of the combined load cases were too high either at the outer surface of the weld-fillet or at
the inner edge of the nozzle shell intersection.
In this case, the conclusion is, that the chosen equilibrating stress field was not an appropriate one.
Therefore, a linear combination of self-equilibrating stress fields the one according to the limit
state
( ) p M res ij + ,
) ( and the one according to moment load only
M res ij ,
) ( is used to fulfil Melans
theorem. This procedure is permissible because of the following attributes of self-equilibrating
stress fields:
The sum of two self-equilibrating stress fields is a self-equilibrating stress field.
The multiple of a self-equilibrating stress field is a self-equilibrating stress field.
Using this procedure, the necessary conditions are given by
0 ] ) ( ) ( ) ( ) ( [
, , , 2 ) ( , 1
+ + +
+ le p ij le M ij M res ij p M res ij
,
0 ] ) ( ) ( ) ( [
, , 2 ) ( , 1
+ +
+ le M ij M res ij p M res ij
,
where the maximum admissible internal shakedown pressure is now given by
p PS
SD

max
.
The self equilibrating stress field according to the moment only,
M res ij ,
) ( , is given by the
difference of the elasto-plastic stress field at this state
M ep ij ,
) ( and the linear-elastic stress field at
this state
M le ij ,
) ( :
M le ij M ep ij M res ij , , ,
) ( ) ( ) ( .
For the determination of the factors
2 1
, and the deviatoric maps of the stress states, i.e. the
co-ordinates of a stress point given by its principal stresses, at the critical locations of the structure
11
Procedures
Page
3.11
DBA
Design by Analysis
are used, since due to the increased number of factors and the different critical locations load case
operations using the FE-software directly are not feasible. Nevertheless, the validity of the
determined self-equilibrating stress field should be checked by superposition with the linear-elastic
stress fields in the postprocessor of the FE-software.
3.3.2.7 Shakedown Analysis for problems including thermal stresses
Since thermal stress fields are self-equilibrating stress fields they can be used in Melans theorem
directly e.g. multiplied with a suitable factor as part of a self-equilibrating stress, which is given
by the sum of different self-equilibrating stress fields. A suitable factor for a thermally induced self-
equilibrating stress field can be 0.5, since states with and without thermal stresses (e.g. the zero
stress state) have to be considered within the cycle.
If one part of the structure is highly influenced by the thermal stresses and another part by non-
thermal action induced stresses, use of the sum of two self-equilibrating stress fields can be suitable
one being a thermally induced self-equilibrating stress field and the other one being induced by
the non-thermal action.
3.3.2.8 Literature
[1] Zeman, J.L.: Repititorium Apparatebau Grundlagen der Festigkeitsberechnung;
Oldenbourg, 1992.
[2] Gokhfeld, D.A., Cherniavsky, O.F.: Limit Analysis of Structures at Thermal Cycling;
Sijthoff & Noordhoff, 1980.
[3] Preiss, R.: On the Shakedown Analysis of Nozzles Using Elasto-Plastic FEA; Intl. J. Pres.
Ves. & Piping 76 (1999), 421-434.
[4] Knig, J.A.: Shakedown of Elastic-Plastic Structures; Elsevier, 1987.
[5] Burgreen, D.: Design Methods for Power Plant Structures; C.P. Press, 1975.
[6] Preiss, R.: Das Einspielverhalten des Balkens mit Rechteckquerschnitt bei Biege- und
Normalkraftbelastung; Bericht Nr. 13, Institut fr Apparate- und Anlagenbau (A&AB), TU
Wien, 1998.
[7] Zeman, J.L., Preiss, R.: The Deviatoric Map A Simple Tool in DBA. Intl. J.Pres. Ves. &
Piping 76 (1999) 339-344.
[8] Rammerstorfer, F.G.: On the Modeling of Cracks by Finite Elements; Scand. J. of
Metallurgy, 12 (1983) p. 293 298.
3.4 Wind actions
If a wind load is specified as an action on a structure, its effects have to be considered in the checks
against GPD and PD. Since wind loading is three-dimensional, or at least not rotational-symmetric,
the corresponding FE-model has to be three-dimensional too. A possibility to avoid a computation
time intensive 3-D model in the elasto-plastic calculation is to use a 3-D model only for the
12
Procedures
Page
3.12
DBA
Design by Analysis
calculation of the linear-elastic stresses caused by the wind. Furthermore, the GPD check and the
PD (or SD) check can be performed using an axial-symmetric model without consideration of the
wind effects (which would not be possible within this model), with decreased design resistance (or
material strength parameters) given by the difference of the original design resistance (or material
strength parameters) minus the maximum linear-elastic equivalent stresses in the structure due to
the wind load. This procedure is admissible due to the positive definiteness of the equivalent stress,
i.e. the sum of the maximum equivalent stresses of two stress tensors is greater or equal to the
maximum equivalent stress of the sum of the two stress tensors. Of course, the usefulness of this
approach is limited by the margin of the stresses caused by the wind action.
In general, the local wind load per unit area is given by the product of the stagnation pressure times
the local drag coefficient. Figure 3.5 shows a typical distribution of the (standardised) pressure
distribution for a cylindrical structure with smooth surface as a function of the angle from the
stagnation point.
The corresponding global drag coefficient is given by


180
0
,
cos
180

d c c
e p
The wind load component normal to the wind direction calculated with the global drag coefficient
equals the one calculated with the local drag coefficients, but local effects due to the real pressure
distribution are neglected in this approach.
To apply this pressure distribution, or a similar non rotational-symmetric one, performing an FE
analysis is time consuming. It is, therefore, customary to use the following approximate procedure:
With respect to the centre of an interesting cross-section, the resultant moment and (shear) force of
the wind action above are calculated. At the upper end of a sufficiently long shell above the section
of interest, the resultant moment and shear force are applied (often the shear force can be
0 25 50 75 100 125 150 175
-1.5
-1
-0.5
0
0.5
1
c
p,e
Figure 3.5: Wind pressure distribution
13
Procedures
Page
3.13
DBA
Design by Analysis
neglected). Of course, at cross-sections in some distance from the reference one, the results are
inaccurate. This approach was followed in example 2 throughout the whole of the checks.
To check the deviation from the non rotational-symmetric approach, the latter one was used also for
the stress calculation due to the wind in example 2, but only for comparison not in the DBA
design checks. The difference in the wind effects is a remarkable 29% - the more accurate one
giving the larger values.
3.5 Direct Route (using Elastic Compensation)
3.5.1 Check against Global Plastic Deformation GPD
Elastic compensation
[1-6]
calculates bounds of the limit load and shakedown load for a - structure for
a given load set by using iterative elastic FE-analysis. This method is a generalisation of the
technique proposed by Marriot
[5]
for estimating lower bound limit loads on pressure vessel
applications. The procedure involves calculating a series of elastic equilibrium stress fields where
the stress is redistributed by altering the elastic modulus of each element based upon the maximum
unaveraged nodal stress from the previous iteration, thus
where E is the elastic modulus, i the iteration number,
nom
is some nominal value, and
emax
the
maximum unaveraged nodal stress in that element from the previous solution. The resulting
redistributed stress fields are equilibrium stress fields. By definition, if the equivalent stress
anywhere in the equilibrium stress field does not exceed the yield stress of the material then that
stress field relates to a lower bound on the limit load. Therefore, scaling the applied loads by the
amount given by maximum stress in the redistributed stress field to the yield stress of the material
will give the limit load, i.e.
where A
L
is the limit load for the action(s), A
ap
is the applied load to the FE-model,
y
is the yield
strength of the material and
max
the maximum unaveraged nodal stress in the model. Due to the
simplicity of this method, it lends itself to application in design checks against GPD according to
the direct route method in prEN 13445-3 Annex B.
As this method of determining limit loads is wholly elastic it is very simple to apply different yield
criteria to the analysis without the convergence difficulties associated with elasto-plastic analysis.
In the rules for the check against GPD, the analysis is required to be based on Trescas yield
condition and associated flow rule, first order theory and an elastic-perfect plastic material model.
This can be performed directly using the elastic compensation procedure for solid models.
For analyses utilising shell elements, elastic compensation cannot be applied directly in the same
method as described above. As a shell has only one element through thickness, it is not possible to
modify the elastic modulus through the thickness. To allow the application of elastic compensation
to shell elements a generalised yield model is adopted in the analysis. Ilyushin's
[6]
generalised yield
model for a doubly curved shell is used which is based upon Mises' condition and associated flow
max
1
e
nom
i i
E E

max

y
ap L
A A
14
Procedures
Page
3.14
DBA
Design by Analysis
rule. A brief overview of Ilyushins generalised yield model is given at the end of this sub-section.
As the code specifies the application of Trescas, some modification of the results is required when
the analysis is based upon Mises' condition. As stated in 3.3.1 the application of a factor of 3/2 to
the design stress will result in a conservative result. This method can also be used as a check on the
results for solid models utilising Trescas condition.
As with the elasto-plastic method above, elastic compensation may be applied in two ways to the
check against GPD. First, for specified actions a check on the admissibility of the load set can be
made by checking that the equilibrium stress fields satisfy the lower bound limit load theorem. If
the maximum equivalent stress anywhere in the equilibrium stress field remains below the design
strength of the material then the specified loading is admissible. Second, the limit on the applied
action(s) may be found using the above procedure and the maximum admissible action(s) can be
determined according to prEN 13445-3 Annex B.
Where there are multiple actions applied, the second case, where the actual limit is calculated,
becomes more complex. For example in problems 3.1 and 3.2 there is a constant moment action
and an internal pressure action. The limit on the pressure has to be found. In elastic compensation,
the applied load set is scaled to give the limit load set. Therefore, in multiple action conditions one
analysis is not sufficient to define the limit load, as the ratio of the loads at the limit is not already
known. In this situation multiple analyses are made for different ratios of applied load and a limit
locus is constructed that describes the limit state for all combinations of load. In the case of
problems 3.1 and 3.2, with the constant moment known, the limit pressure can be found directly
from the limit locus.
As the code rules in prEN 13445-3 Annex B address elasto-plastic analysis and not any simplified
method, some problems arise in applying elastic compensation. The application rule for GPD in
prEN 13445-3 Annex B.9.2 states that the maximum absolute value of principal strain should not
exceed 5%. As elastic compensation is not a displacement-based approach the values of strain are
not accurate in the equilibrium stress fields and cannot be used. Therefore, it is possible for
structures that are stiff well into the plastic range (where the limit according to the code is defined
by the 5% maximum principal strain limit) that elastic compensation non is conservative. In this
situation the elastic compensation result would be close to the limit defined by loss of equilibrium
in the elasto-plastic analysis (for an elastic-perfect plastic material). This can be noted in problem
3.1 where the elastic compensation result is very much higher than the elasto-plastic result defined
by the limit on the principal strain. However, if the elasto-plastic result were defined by the tangent
intersection method with no limit on the principal strain, the results would be similar.
3.5.2 Check against Progressive Plastic Deformation (PD)
The principle in the check against PD according to prEN 13445-3 Annex B.9.3 is fulfilled if the
structure can be shown to shake down, as progressive plastic deformation and alternating plasticity
are the two possible failure modes if the structure fails to shakedown. It is possible using the
elastic compensation procedure to calculate lower bounds on the shakedown load using elastic
compensation. As used in the elasto-plastic check described in 3.3.2 above Melans shakedown
theorem is also used in the elastic compensation procedure. The Mises limit stress field for the
applied action(s) is calculated using elastic compensation as described above. As the zeroth
iteration in elastic compensation is the true elastic stress field (i.e. no modulus modification), the
residual stress field can be found by subtracting the elastic stress field from the redistributed limit
15
Procedures
Page
3.15
DBA
Design by Analysis
stress field. When the limit stress field and applied loads are scaled to give the limit, the maximum
stress in the limit field equals the yield stress of the material. Therefore, if the maximum stress in
the residual stress field is less than the maximum stress in the limit stress field then the shakedown
load is the same as the limit load. If, however, the maximum stress in the residual stress field is
greater than the maximum stress in the limit stress field, then the shakedown load will be lower than
the limit load. The value of the shakedown load in this case can be calculated by invoking the
linear proportionality of the FE-solution, thus
where A
sh
is the shakedown limit on the action, A
ap
is the applied value of the action,
y
the yield
strength of the material, and
res max
is the maximum residual stress in the model.
As with the check against GPD there are two possible ways of performing calculations in the PD-
check. First, by checking for admissibility of the action(s) by checking if the maximum stress in the
residual stress field is lower than the design resistance for the PD-check. Second, by calculating the
maximum shakedown load as described above, and then determining from that the maximum
allowable shakedown load according to the code rules for the check against PD.
As with the GPD calculations for elastic compensation with multiple actions described above, loci
have to be constructed to describe the shakedown limits for all ratios of the actions.
3.5.3 Ilyushins generalised yield function
In the usual classical theory of thin shells, stretching and bending stress resultants, Figure 3.6, are
used, usually defined as
N
T
dz i x y xy
M
T
z dz i x y xy
i i
T
T
i i
T
T

1
1
2
2
2
2
2

/
/
/
/
, ,
, ,
where T is the shell thickness and
x y xy
, , , are in-plane stress components.
max res
y
ap sh
A A


x
y
z
N
x
N
x
M
y
dsx
M
x
ds
y
dsx
N
xydsy
N
yxdsx
M
y
dsx
M
x
dsy
dsy
dsy
N
y dsx
dsy
N
y ds
x
T
Figure 3.6: Thin shell variables
16
Procedures
Page
3.16
DBA
Design by Analysis
The Mises yield function

x x y y xy Y
2 2 2 2
3 + +
where
Y
is the yield stress, can be approximated by a function of the thin-shell stress resultants,
and, in this manner, a generalized yield surface obtained. This function the generalized yield
function is conveniently expressed by means of the functions Q Q Q
N M NM
, and :
Q N N N N N
Q M M M M M
Q N M N M N M N M N M
N x y x y xy
M x y x y xy
NM x x x y y x y y xy xy
+ +
+ +
+ +
2 2 2
2 2 2
3
3
1
2
1
2
3
Initially Ilyushin
[6]
presented an exact generalized yield model, but this was too complex for
practical use (at that time) so a linear approximation was proposed (usually referred to as Ilyushin's
generalised yield model):
+ +

Q Q
Q
N M
NM
IL
Y
3
2
2

3.5.4 Literature
[1] Mackenzie D., Boyle J. T. et al: A simple method of estimating limit loads by iterative
elastic analysis I, II & III, Int. J. Pres. Ves. & Piping 53 (1993) 77-142.
[2] Mackenzie D., Nadarajah C. Shi J. & Boyle J. T.: Simple bounds on limit loads by elastic
finite element analysis, Trans. ASME J. Pres. Ves. Tech 115 1993 27-31.
[3] Mackenzie D., & Boyle J. T.: A simple method of estimating shakedown loads for complex
structures, Proc. ASME PVP, Denver 1993.
[4] Hamilton R., Mackenzie D., Shi J & Boyle J. T.: Simplified lower bound limit analysis of
pressurised cylinder/cylinder intersections using generalised yield criteria, Int. J. Pres. Ves.
& Piping 67 (1996) 219-226.
[5] Marriot D. L.: Evaluation of deformation or load control of stresses under inelastic
conditions using elastic finite element analysis, Proc. ASME PVP Conf., Vol 136, Pittsburgh
1988.
[6] Ilyushin A. A.: Plasticity (Russian), Gostekhizda, Moscow, 1948 and Plasticite (French),
Eyrolles, Paris 1956.
3.6 Stress Categorisation Route
3.6.1 General
DBA based on stress categories is today well established, even if the method is not used as much as
it should be, in order to optimise the design of pressure equipment. During the years the method has
17
Procedures
Page
3.17
DBA
Design by Analysis
been developed, but still some work can be done, for example to improve and increase the table on
typical cases.
The method has been criticised for difficulties with the stress classification, but for thin-walled
structures the classification is usually not a problem. The problems will occur in thick-walled
structures, with complicated geometry, for example big valves and pumps.
3.6.2 Evaluation procedure
The analysis is performed in three steps:
Step 1 Primary stresses
Assure that the primary stresses caused by mechanical design loads will fulfil the requirements. The
allowable design stress f is based on the design temperature, but no thermal effects or
discontinuities are included.
Step 2 Secondary stresses
Assure that the 3f (shakedown) criterion is fulfilled. Here f is based on t*, where t* is a kind of
balanced value taking into account maximum and minimum temperatures during the considered
operating cycle or cycles. The 3f-criterion shall be satisfied by any equivalent stress range resulting
from variation of primary + secondary stresses between two normal operating conditions.
A secondary stress is a stress which is self-equilibrating. It occurs at large discontinuities, but does
not include stress concentrations. A secondary stress can be caused both by mechanical and thermal
loads.
It can be stated that if the secondary stress range, caused by for example thermal effects, is large,
this will limit the pressure bearing capacity of the structure.
Step 3 Fatigue assessment
The fatigue assessment can be directly based on the results from step 2. This is of the utmost
importance as fatigue is the most important failure mode in pressure vessel technology. The fact
that the requirements from step 2 are fulfilled will for most weld classes guarantee a fatigue life of
more than 500 load cycles. For a more exact fatigue assessment, if required, the results from step 2
shall be used in a detailed assessment of fatigue life.
3.6.3 Notations
In the plots, collected in the discussion of examples 1.1, 1.2, 1.3, 1.4, 2, the following notations are
used:
- Gen.Dir. Inside (I), Outside (O)
- Circum Dir I, O
- Membrane (Tresca)
- Shear
- Tresca I, O
18
Procedures
Page
3.18
DBA
Design by Analysis
3.7 Check against Instability
Of the various approaches to perform the instability check three have been used here:
3.7.1 Usage of DBF or well-proven handbook formulae
There are cases where the formulae of the DBF section or well-proven formulae in handbooks can
be used, directly or indirectly, exactly or as an approximation.
Wherever possible, this approach should be used, and if it is only used as a complement to one of
the other approaches, or to obtain a reasonably good starting point for a classical (eigenvalue)
procedure.
It shall always be ascertained that the formulae used for imperfection-sensitive structures are not
simply theoretical results but do contain appropriate reduction (knockdown) factors for
imperfection-induced and for possible plasticity-induced effects.
The partial safety factor for the resistance in the I-check, as specified in the draft standard, does not
incorporate imperfection effects; these have to be taken into account in the determination of the
resistances directly.
In some cases the formulae, of the DBF section or of the handbooks, can be used directly see, for
instance, the buckling of the smaller cylinder of Example 2, or the case radial external pressure on
inner shell of Example 6.
In other cases it may be necessary to use results of FEM calculations, e. g. reactions between
components, thermal stresses, as actions in the DBF or handbook models see, for instance, the
ring buckling check in Example 2, and the axial loading case of the inner shell of Example 6; in the
latter example the axial force (per unit length) can be obtained by simple hand calculations directly
as well.
3.7.2 Classical (eigenvalue) approach
Modern software allows for the determination of classical (birfurcation) buckling loads.
Appropriate reduction factors for both imperfection-induced and plasticity-induced effects have to
be applied these may be taken from the same sources as before if necessary as approximations
only.
It is especially important to ascertain that an eigenvalue problem does exist for the structure and the
load case considered.
It is equally important to ascertain that the model used does allow for the appropriate buckling
deformations - a symmetric model will not allow for non-symmetric buckling modes.
Care is necessary to ascertain that the software used can reach the relevant buckling modes, and that
it does so. For instance in Example 6, the jacket is subject to internal pressure. If pressure is applied
on all walls of the jacket, the internal pressure action on the outside wall rendered non-convergence
or non-usable results the buckling modes of the internal wall, which is under outside pressure and
for which buckling modes do exist, are not obtained.
3.7.3 Fully nonlinear approach
In this approach a geometrically nonlinear analysis with nonlinear constitutive law is used, the
initial imperfections of the structure are applied as initial geometry.
19
Procedures
Page
3.19
DBA
Design by Analysis
In case of the investigation of a real structure these initial imperfections may be the actual
imperfections (in detail) or the relevant buckling shapes obtained via one of the two former results
but scaled-up by some measures of the actual imperfections, like out-of-roundness, flatness,
peaking.
In case of a virtual structure, specified by the drawings and relevant codes and standards, these
initial imperfections may be the relevant buckling shapes obtained via one of the two former
methods but here scaled-up using the allowed deviation (measures).
Unfortunately the constitutive law to be used is not yet specified in prEN 13445-3; it was agreed
that the very same model as in the PD-check should be used.
The maximum action, corresponding to collapse of the structure, is the characteristic value of the
resistance of the structure to be divided by the relevant partial safety factor to obtain the design
resistance. No additional reduction is required.
There are various approaches for this proof:
If the sufficient resistance against instability of a structure for a set of specified actions A
1
, . . ., A
n
shall be proven, the design values of these actions shall be determined first, and then it shall be
shown that the resistance of the structure (the limit carrying capacity) obtained by proportional
increase of all design actions, is large enough, such that the limit values of the actions divided by
the partial safety factor of the resistance are larger than the design values.
With the (limit) multiplication factor MF obtained by means of the fully nonlinear model for the
limit carrying capacity, this requirement can be written symbolically as
. /
.....
.....
....
....
...
...
1 1 1 1 1
R
n An
A
n An
A
nd
d
MF A
MF A
A
A
A
A

1
1
1
1
]
1

1
1
1
1
]
1

1
1
1
1
]
1

If just a specific load case shall be proven to be admissible, it is not necessary to obtain the limit
carrying capacity it is only required to show that MF is not smaller than
R
, i. e. that the set of
actions ,
1 1
A
R A
. . . .,
n R An
A can be carried.
3.7.4 Literature
[1] Samuelson, L. A., Eggwertz, S.: Shell Stability Handbook. Elsevier, London 1992.
[2] Bushnell, D.: Computerized buckling analysis of shells. Martinus Nijhoff Publishers,
Dordrecht 1985.
[3] Baker, E. H., Kovalevsky, L., Rish, F. L.: Structural Analysis of Shells.
Krieger Publishing Comp., Malabar, USA 1986.
[4] Ross, C. T. F.: Pressure Vessels under External Pressure. Statics and Dynamics. Elsevier,
London 1990.
[5] Como, M., Grimaldi, A.: Theory of Stability of Continuous Elastic Structures,
CRS Press, Boca Raton 1995.
20
Procedures
Page
3.20
DBA
Design by Analysis
[6] Schmidt, H., Krysik, R.: Towards Recommendations for Shell Stability Design by Means of
Numerically Determined Buckling Loads. Int. Coll. Buckling. Elsevier, London 1991.
[7] Budiansky, B.: Theory of Buckling and Post-Buckling Behaviour of Elastic Structures.
In: Advances in Applied Mechanics, Vol. 14,
Academic Press, New York 1974.
[8] Hutchinson, J. W.: Plastic Buckling. In: Advances in Applied Mechanics, Vol. 14,
Academic Press, New York 1974
3.8 Check against Fatigue
In the F-check of the DBA section of prEN 13445-3 reference is made to the fatigue calculations of
the DBF route. Unfortunately this section, reprinted in Annex A2, is not easily readable. Some
misprints are still in this version, but they are not always obvious as such, and they are often quite
misleading.
The flow-sheets on the following pages should help, misprints we discovered are corrected, non-
unique approaches are deleted.
The main changes are:
Structural stresses to be determined by quadratic extrapolation in all cases, into the hot
spot or the point of maximum equivalent stress, respectively. The extrapolation shall be
performed as stated in Fig. 18-3 of prEN 13445-3, i. e. with pivot point distances from the
critical point of 0.4e, 0.9e, 1.4e, e being the thickness at the critical point.
The "thickness" to be used in the thickness correction formulae is the shortest distance
from the crack initiation site (critical point) to the other wall to which the crack is likely to
grow. In case of more than one potential crack direction, the largest of these shortest
distances shall be used.

t eff
K K if the equation for
eff
K gives a value greater than
t
K
Simplified procedures for using Class 100 design data for unwelded regions is deleted.
The flow-sheets are for repeated action in form of one and the same cycle, superposition with other
cycles is not directly included, nor is the contribution to the overall damage of cycles with more
than 2.10
6
repetitions in the case of unwelded regions.
The results given for the various examples are for neutral environment environmental assisted
corrosion, including the cracking types, is excluded.
For further reading, the booklet by Niemi, E.: Stress Determination for Fatigue Analysis of Welded
Components. Abington Publ. , Cambridge 1995, and the literature cited there, can be recommended.
The "definition" of structural stress given there differs from the one used here, but the numerical
differences in the results are usually small.
21
Procedures
Page
3.21
DBA
Design by Analysis
Data
t
max
= .. C
t
min
= .. C
t* = 0,75 t
max
+ 0,25 t
min
=.. C
R
m
= .. MPa
R
p0,2/t*
= .. MPa
Effective stress concentration factorK
eff
but not largen than K
t
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors. k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
R
z
= .. m (table 18-8)
e
n
= .. mm

D
= .. MPa (table 18-10 for N 2
.
10
6
cycles)
N = ..(for the first iteration)

R
= .. MPa (allowable stress range for N< 2
.
10
6
cycles)
Stresses

eq,t
(total or notch equivalent stress range) = .. MPa

struc
(structural equivalent stress range) = .. MPa (obtained by quadratic extrapolation)
.....
D

struc

t
K 0,5 1
1)
t
(K 1,5
1
eff
K
+

+
1
1
]
1

Theoretical elastic stress concentration factor K


t
K
t
=
eq,t
/
struc
= ..
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..

total
= k
e
.
eq,t
= .. MPa
Else
total
=
eq,t
= .. MPa
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
l eq,

0,4
0,5
0,7
k

k

= ..

total
= k

.
eq,t
= .. MPa
Else
total
=
eq,t
= .. MPa
) stress equivalent notch (mean MPa .....
eq

18-10-6-2 Temperature correction factor f
t*
For t* > 100 C
f
t*
= 1,03 1,5
.
10
-4
t* -1,5
.
10
-6
t*
2
= ..
Else f
t*
= 1
18-11-1-1 Surface finish correction factor f
s
f
s
= F
s
[0,1ln(N)-0,465]
if N < 2
.
10
6
f
s
= F
s
if N 2
.
10
6
with F
s
= 1- 0,056 [ln (R
z
)]
0,64
[ln(R
m
)] +0,289 [ln (R
z
)]
0,53
= ..
f
s
= ..

eq,struc
=
t
K
total
(for usage 18-11-3)

eqmax
= .. MPa (maximum notch equivalent stress)
22
Unwelded material Ferritic steel
Fatigue calculation
Page
3.22
DBA
Design by Analysis
18-11-1-2 Thickness correction factor f
e
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
e
= F
e
[0,1ln(N)-0,465]
if N < 2
.
10
6
f
e
= 0,7217
[0,1ln(N)-0,465]
if N< 2
.
10
6
f
e
= F
e
if N 2
.
10
6
f
e
= 0,7217 if N 2
.
10
6
with Fe = (25/en)
0,182
= .....
f
e
= 1 f
e
=

.. f
e
= ..
18-11-1-3 Mean stress correction factor f
m
If
struc
< 2 R
p0,2/t*
and
eqmax
< R
p0,2/t*
If
eq
> 0 then
r eq,
= R
p0,2/t*
-
2
t eq,

If
eq
< 0 then
r eq,
=
2
t eq,

- R
p0,2/t*
and
eq
=
r eq,
= .. MPa
For N 2
.
10
6
cycles
See figure 18-14
f
m
= ..
If
struc
>2 R
p0,2/t*
18-11-2-1 Overall correction factor f
u
f
u
=
eff
K
* t
.f
m
.f
e
.f
s
f
= ..
u
f /
struc eq,

18-11-3 Allowable number of cycles N


2
10 6 , 4
5 , 11 63 , 0
4
,
1
1
]
1

m
u
struc eq
R
f
N

if N 2
.
10
6
cycles
if R
p0,2/t*

eq

( ) M 1 2
R

+
then
f
m
=
( )
5 , 0
eq
2
M 1
M 2 M
- 1
R
1
1
]
1

,
_

+
+
= ..
N = if
D u
f /
struc eq,

N = ..
N is obtained by iterations. If the value of N oscillates, a difference equal to 1 % between
two iterations is acceptable. If the values decrease monotonously, the difference must be
less than 0,001 % between two iterations.
If
struc
< 2 R
p0,2/t*
and
eqmax
> R
p0,2/t*
if
( ) M 1 2
R

+

eq
R
p0,2/t*
then
f
m
=

,
_

+
+
R

eq
2
3
M
M 1
3 M 1
= .
For N 2
.
10
6
cycles
M = 0,00035 R
m
0,1 = ..
f
m
= 1
v
23
Welded material Ferritic steel
Fatigue calculation
Page
3.23
DBA
Design by Analysis
Data
t
max
= .. C
t
min
= .. C
t* = 0,75 t
max
+ 0,25 t
min
=.. C
R
m
= .. MPa
R
p0,2/t*
= .. MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= .. mm

D (5
.
10
6
cycles)
= .. MPa (class ..)
equivalent stresses or principal stresses
m = 3 C = .. m = 3 C

= ..
C
//
= ..
m = 5 C = .. m = 5 C

= ..
C
//
= ..
Stresses

struc
= .. MPa (structural equivalent stress range,determined by extrapolation)
[or
struc
= .. MPa and
struc//
= .. MPa]
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..
= k
e

struc
= .. MPa
Else

=
struc
= .. MPa
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
= .....
f
ew
= 1 f
ew
=

.. f
ew
= 0,639
24
Unwelded material Austenitic steel
Fatigue calculation
Page
3.24
DBA
Design by Analysis
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= ..
18-10-7 Allowable number of cycles N
w
f

= .. MPa N = if <
10
8
, else
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,03 1,5
.
10
-4
t* -1,5
.
10
-6
t*
2
= ..
Else f
t*
= 1
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
with C from Table 18-7 of prEN 13445-3,
in dependence of the (weld) class, given by prEN 13445-3, Tables 18.4 and 18.5, respectively.
Data
t
max
= .. C
t
min
= .. C
t* = 0,75 t
max
+ 0,25 t
min
=.. C
R
m
= .. MPa
R
p1,0/t*
= .. MPa
Effective stress concentration factorK
eff
but not larger than K
t
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors. k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
R
z
= .. m (table 18-8)
e
n
= .. mm

D
= .. MPa (table 18-10 for N 2
.
10
6
cycles)
N = ..(for the first iteration)

R
= .. MPa (allowable stress range for N < 2
.
10
6
cycles)
Stresses

eq,t
(total or notch equivalent stress range) = .. MPa

struc
(structural equivalent stress range) = .. MPa (obtained by extrapolation)
.....
D

struc

t
K 0,5 1
1) (K 1,5
1 K
t
eff

+

+
1
1
]
1

Theoretical elastic stress concentration factor K


t
K
t
=
eq,t
/
struc
= ..
18.8 Plasticity correction factor k
e
mechanical loading
If
eq,l
> 2 R
p1,0/t*

1
1
]
1

+ 1
* p1,0/t
R 2

4 , 0 1
e
k
l eq,
k
e
= ..

total
= k
e
.
eq,t
= .. MPa
Else
total
=
eq,t
= .. MPa
Thermal loading
If
eq,l
> 2 R
p1,0/t*
1
1
]
1

* p1,0/t
R
l eq,

0,4
0,5
0,7
k

k

= ..

total
= k

.
eq,t
= .. MPa
Else
total
=
eq,t
= .. MPa
) stress equivalent notch (mean MPa .....
eq

eqmax
= .. MPa (maximum notch equivalent stress)
25
Unwelded material Austenitic steel
Fatigue calculation
Page
3.25
DBA
Design by Analysis
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C
f
t*
= 1,043 4,3
.
10
-4
t* = ..
Else f
t*
= 1
18-11-1-1 Surface finish correction factor f
s
f
s
= F
s
[0,1ln(N)-0,465]
if N < 2
.
10
6
f
s
= F
s
if N 2
.
10
6
with F
s
= 1- 0,056 [ln (R
z
)]
0,64
[ln(R
m
)] +0,289 [ln (R
z
)]
0,53
= ..
f
s
= ..

eq,struc
=
t
K
total
(for usage in 18-11-3)
26
Welded material Austenitic steel
Fatigue calculation
Page
3.26
DBA
Design by Analysis
18-11-1-2 Thickness correction factor f
e
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
e
= F
e
[0,1ln(N)-0,465]
if N < 2
.
10
6
f
e
= 0,7217
[0,1ln(N)-0,465]
if N < 2
.
10
6
f
e
= F
e
if N 2
.
10
6
f
e
= F
e
if N 2
.
10
6
with Fe = (25/en)
0,182
= .....
f
e
= 1 f
e
=

.. f
e
= ..
18-11-1-3 Mean stress correction factor f
m
If
struc
< 2 R
p1,0/t*
and
eqmax
< R
p1,0/t*
If
eq
> 0 then
r eq,
= R
p1,0/t*
-
2
t eq,
If
eq
< 0 then
r eq,
=
2
t eq,
- R
p1,0/t*
and
eq
=
r eq,
= .. MPa
For N 2
.
10
6
cycles
See figure 18-14
f
m
= ..
If
struc
>2 R
p1,0/t*
18-11-2-1 Overall correction factor f
u
f
u
=
eff
K
* t
.f
m
.f
e
.f
s
f
= ..
u
f /
struc eq,
= ..
18-11-3 Allowable number of cycles N
2
10 6 , 4
5 , 11 63 , 0
4
,
1
1
]
1

m
u
struc eq
R
f
N

if N 2
.
10
6
cycles
if R
p1,0/t*

eq

( ) M 1 2
R

+
then
f
m
=
( )
5 , 0
eq
2
M 1
M 2 M
- 1
R
1
1
]
1

,
_

+
+
= ..
N = if
D u
f /
struc eq,

N = ..
N is obtained by iterations. If the value of N oscillates, a difference equal to 1 % between two iterations is
acceptable. If the values decrease monotonously, the difference must be less than 0,001 % between two
iterations.
If
struc
< 2 R
p1,0/t*
and
eqmax
> R
p1,0/t*
if
( ) M 1 2
R

+

eq
R
p1,0/t*
then
f
m
=

,
_

+
+
R

eq
2
3
M
M 1
3 M 1
= .
For N 2
.
10
6
cycles
M = 0,00035 R
m
0,1 = ..
f
m
= 1
Data
t
max
= .. C
t
min
= .. C
t* = 0,75 t
max
+ 0,25 t
min
=.. C
R
m
= .. MPa
R
p1,0/t*
= .. MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


struc
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= .. mm

D (5
.
10
6
cycles)
= .. MPa (class ..)
equivalent stresses or principal stresses
m = 3 C = .. m = 3 C

= ..
C
//
= ..
m = 5 C = .. m = 5 C

= ..
C
//
= ..
Stresses

struc
= .. MPa (structural equivalent stress range,determined by extrapolation)
[or
struc
= .. MPa and
struc//
= .. MPa]
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p1,0/t*

1
1
]
1

+ 1
* p1,0/t
R 2
struc

4 , 0 1
e
k
A
0
= ..
k
e
= ..
= k
e

struc
= .. MPa
Else =
struc
= .. MPa
Thermal loading
If
struc
> 2 R
p1,0/t*
1
1
]
1

* p1,0/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
= .....
f
ew
= 1 f
ew
=

.. f
ew
= 0,639
27
Welded material Austenitic steel
Fatigue calculation
Page
3.27
DBA
Design by Analysis
28
Procedures
Page
3.28
DBA
Design by Analysis
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= ..
18-10-7 Allowable number of cycles N
w
f

= .. MPa N = if
w
f

<
10
8
, else:
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,043 4,3
.
10
-4
t* = ..
Else f
t*
= 1
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
with C from Table 18-7 of prEN 13445-3,
in dependence of the (weld) class, given by prEN 13445-3, Tables 18.4 and 18.5, respectively.
1
Illustrative Example
Page
4.1
DBA
Design by Analysis
4 Illustrative Example
4.1 Introduction
A simple illustrative example is described in this Section - a circular plate under the action of a
uniform pressure action which varies between zero and a maximum value.
This Section should be read in conjunction with prEN13445-3, Annex B. The checks which have to
be considered according to this are the check against global plastic deformation (GPD), the check
against progressive plastic deformation (PD), the fatigue (F) check, the check against instability (I),
and the static equilibrium check (SE).
In the following the maximum admissible pressure according to the GPD-check and the maximum
pressure for shakedown of the structure, with a load variation between zero and this maximum, are
determined. If the structure shakes down under a given action cycle, the admissibility of this action
cycle against PD is proven see subsection 3.3.2.1 of section 3 Procedures. For the calculation of
the allowable number of action cycles, in the F-check, an upper value for the pressure equal to 90%
of the maximum allowable pressure (given by the other checks) is used: see also subsection 5.1 of
section 5 Case Specification. For this structure and the specified action cycle, the I-check and the
SE-check are not required.
4.2 Problem Specification
The dimensions of the plate are: diameter 500 mm, thickness 25 mm. The design temperature is
specified as 20C. The plate is assumed to be clamped at the edge.
The material of the plate is the ferritic steel P265GH according to EN 10028-2, the surface is
assumed to be machined with m R
Z
50 . The material strength parameter is MPa RM 255 , it
corresponds to the upper yield strength of P265GH at room temperature for a thickness range of 16
mm to 40 mm according to EN 10028-2.
The modulus of elasticity is given by GPa E 212 (see subsection 5.4 of section 5 Case
Specification).
4.3 Finite element model and boundary conditions
Since the radius to thickness ratio of the plate is relatively small, and since the application of the
checks can be shown more clearly, 2-D axisymmetric solid elements are used instead of plate
elements for the finite element model: see Figure 4.1. The mesh is fairly dense, and would be
expected to yield accurate solutions for elastic and inelastic analysis without further refinement.
The commercial finite element analysis system ANSYS

is used here.
2
Illustrative Example
Page
4.2
DBA
Design by Analysis
Figure 4.1: Finite element model
The boundary conditions require some thought. In the technical theory of structures (beams, plates
and shells) the notion of a clamped edge of a plate has a specific meaning displacement in
thickness direction and tangent rotation of the mid-plane at the plates edge are zero.
In a solid model a clamped edge has to be modelled appropriately and suitable boundary
conditions chosen. In the present example, which of course is chosen just to illustrate DBA and not
modelling, the vertical and horizontal displacements in the nodes at the plates edge were
constrained to zero. These boundary conditions seem to be fairly reasonable, but they create a
localised stress concentration near the clamped edge.
4.4 Determination of the maximum admissible pressure according to the GPD-Check
In general terms, the principles specified in DBA, as given in prEN 13445-3, Annex B, require that
the design effect of actions does not exceed the corresponding design resistance.
The design effect of actions is the relevant response of the model, with a specified material model,
to the relevant design actions. What the relevant effect actually is depends on the design check
under consideration. The design actions are products of specified characteristic values of actions
and corresponding partial safety factors of actions. Which design actions are relevant depends on
the load cases that have to be considered:
In the GPD-check it is the carrying capacity of the model that matters, that is, the carrying capacity
in terms of actions. Therefore, the design effects are the design actions themselves.
In the simple example considered here there is only one action pressure - and only one load case
to be considered maximum allowable pressure. In this specific case the design principle can here
be rephrased to read:
The design action, given by the product of the maximum allowable pressure and the corresponding
partial safety factor, shall not be larger than the design resistance which is given by the ratio of the
maximum pressure the model can carry to the partial safety factor of the resistance.
Following the proposal for change in this check, as stated at the beginning of Annex 2 of this
manual, there is a side-condition to be taken into account when determining this maximum
pressure: the side-condition is that the maximum absolute value of the principal strains for this
maximum pressure must not exceed 5%.
3
Illustrative Example
Page
4.3
DBA
Design by Analysis
4.4.1 Approach using Mises yield criterion
In the GPDcheck, Tresca's yield criterion is specified. Depending on the analysis software being
used, if no subroutine for this yield criterion is available, or if it is available but shows bad
convergence, Mises yield criterion can be used instead:
Since the maximum ratio of the Mises equivalent stress to the Tresca equivalent stress for the same
load is 3 2 , a GPD-check with Mises' yield criterion, with the design material strength parameter
multiplied by 2 3 , will always lead to conservative results.
Furthermore, if the result of the check against PD (where Mises yield criterion is allowed) is used
in the check against GPD (instead of a separate calculation) and if, like in this problem, there is only
one partial safety factor of the resistance
R
, multiplication of the limit pressure, from the check
against PD, with 2 3 leads to the same result as the multiplication of the material strength
parameter. Of course, since no partial safety factors are used in the check against PD, they have to
be taken into account by scaling down the PD-check results.
As stated in the application rule in prEN 13445-3 Annex B.9.2, the maximum absolute value of the
principal strains must not exceed 5%. To fulfil this requirement in cases where the results from the
PD-check are used, a value of the pressure from the check against PD shall be used such that the
maximum absolute value of the principal strains calculated with this pressure (and the material
strength parameter used for the check against PD) does not exceed 5 %.
Since in the final loadstep
the maximum principal
strain of the elasto-plastic
calculation for the PD-
check, which corresponds
to a pressure of 8.48 MPa,
was about 11%, a lower
pressure value had to be
used in the GPD-check.
Figure 4.2 shows the
principal strain distribution
for a pressure of 8.43 MPa
- the maximum value is
approximately equal to the
allowed 5%.
The partial safety factor for the resistance
R
according to prEN-13445-3 Annex B, Table B.9-3 is
25 . 1 , and the partial safety factor for pressure action without a natural limit is given by 2 . 1
P
,
according to prEN-13445-3 Annex B, Table B.9-2.
Figure 4.2: Distribution of
principal strain
4
Illustrative Example
Page
4.4
DBA
Design by Analysis
Thus, the maximum pressure according to the GPD-check is, in this approach, given by
MPa PS
R P
GPD
87 . 4
2
3
25 . 1 2 . 1
43 . 8
2
3 43 . 8
max


.
4.4.2 Approach using Trescas yield criterion
In prEN 13445-3 Annex B, Trescas yield criterion is prescribed. Unfortunately most commercial
software does not include this criterion in elasto-plastic calculations (although this could be
available in future releases). The approach given in Sec.4.4.1 is a simple work-around, but one
which leads to conservative results.
To show the possibilities given by the Standard, an analysis was performed with Trescas yield
criterion the special routine was provided by an ANSYS

distributor:
The design material strength parameter is given by MPa
R
204 25 . 1 / 255 / 255 .
In this analysis, a first order theory and a linear-elastic ideal-plastic material with design material
strength parameter of 204 MPa were used. The pressure was increased until either an absolute
maximum was obtained or the maximum absolute value of the principal strains reached 5%.
In this example the second condition governed at a pressure of 6.07 MPa the maximum absolute
value of the principal strains reached 5%.
Therefore, the maximum allowable pressure according to this approach for the GPD-check is given
by
. 06 . 5 2 . 1 / 07 . 6 / 07 . 6
, max
MPa PS
p T GPD

This value is about 4% larger than the one obtained in Sec.4.4.1.
Finally, it should be noted that since most commercial software do not offer this approach directly,
the result obtained in Sec.4.4.1 is used in the F-check.
4.5 Check against PD
The PD-check was performed by way of a shakedown check using Melans shakedown theorem
see subsections 3.3.2.1 and 3.3.2.5 of Section 3 (Procedures) for further details.
The elasto-plastic finite element analysis was carried out as required in prEN 13445-3 Annex B,
Sec. B.9.3.1, using Mises yield condition and associated flow rule, a linear-elastic ideal-plastic
constitutive law with a design material strength parameter of 255 MPa, and first order theory.
By defining and using load cases in ANSYS, the superposition of stress fields can be done easily.
Therefore, the first load step of the analysis was defined to be at a very low load level (0.1 MPa), so
that there was a linear-elastic response of the structure. All other linear-elastic stress fields can then
be determined easily by multiplication with a suitable scale-up factor.
5
Illustrative Example
Page
4.5
DBA
Design by Analysis
The analysis was carried out using the
arc-length method, using a maximum
vertical displacement of 25 mm in the
middle of the plate as termination
criterion.
The termination criterion was fulfilled
for a pressure of MPa 48 . 8 , and this
pressure was used as limit pressure.
Figure 4.3 shows the elasto-plastic
Mises equivalent stress distribution
for the limit pressure of MPa 48 . 8 .
Figure 4.4 shows the linear-
elastic Mises equivalent stress
distribution for the limit
pressure the stress maximum
is located near the edge of the
plate.
Figure 4.4: Linear-elastic Mises equivalent stress distribution for the limit pressure
Figure 4.3: Elasto-plastic Mises equivalent stress distribution
6
Illustrative Example
Page
4.6
DBA
Design by Analysis
Figure 4.5 shows the Mises
equivalent stress distribution of
the corrected residual stress
field, the scaling factor is
given by 719 . 0 (see subsection
3.3.2.5 of section 3
Procedures).
Figure 4.5: Mises equivalent stress distribution of the corrected residual stress field
Figure 4.6 shows the Mises
equivalent stress distribution at the
lower bound shakedown limit. The
scaling factor is given by 854 . 0
(see subsection 3.3.2.5 of section 3
Procedures).
Figure 4.6: Mises equivalent stress distribution at the lower bound shakedown limit
7
Illustrative Example
Page
4.7
DBA
Design by Analysis
Thus, the shakedown limit pressure is given by
MPa PS
SD
24 . 7 854 . 0 48 . 8
max
.
This value is already well above the value of the maximum allowable pressure of 4.87 MPa, given
in Sec.4.4.1, or 5.06 MPa in 5.2. Thus, a further (complicated) investigation of the PD behaviour is
not required.
A simulation with pressure cycling
between 0 and 8.0 MPa has shown
that the model shakes down under
this cyclic action to steady-state
behaviour after four action cycles,
within the numerical accuracy that
can be expected. Of course, since the
maximum pressure, 8.0 MPa, is
larger than 7.24 MPa, the value
obtained for elastic shakedown, the
model does not shake down to elastic
behaviour, but to a purely cyclic
behaviour, where at the end of each
cycle the stress distribution is equal
to the one at the beginning, and
where in two distinct and non-
connected regions alternating
plasticity occurs such that the strain
increment over one cycle is zero in
every point within the numerical
accuracy.
Figure 4.7 shows this steady-state
Mises equivalent stress distribution for
maximum pressure, and Figure 4.8 the
deviatoric mappings of the stress state
in the node of maximum accumulated
plastic strain, for maximum and
minimum pressure. The connection line
already passes close to the origin an
indication that steady-state behaviour
with alternating plasticity is almost
reached.
Figure 4.8: Deviatoric map
Figure 4.7: Steady state equivalent stress distribution
minimum
pressure
maximum
pressure
hoop

radial
normal

8
Illustrative Example
Page
4.8
DBA
Design by Analysis
Figure 4.9 finally
shows the evolution,
over four cycles, of
the radial strain, for
maximum pressure
and for minimum
pressure,
respectively.
Figure 4.9: Total radial strain versus load history
Note: Within the framework of the technical theory of plates the model does sustain theoretically
pressure cycles from zero to the limit pressure (according to the GPD-check) without progressive
plastic deformation.
4.6 Fatigue (F) Check
The number of allowable action cycles has been determined for a lower value of the pressure equal
to MPa P
op
0
inf ,
and an upper value of MPa PS P
GPD op
38 . 4 87 . 4 9 . 0 9 . 0
max sup ,
.
The location where the linear-elastic
stress maximum occurs is near the
edge of the model. The structural
equivalent stress range at this point is
obtained by quadratic extrapolation.
Figure 4.10 shows the corresponding
pivot points: node 65, node 67 and
node 69, the stress shown in this plot is
Mises equivalent stress for a pressure
of MPa 1 . 0 .
The details of the fatigue calculation
are given on the pages 4.9 and 4.10
the fatigue calculation sheets.
9
Illustrative Example
Page
4.9
DBA
Design by Analysis
Data
t
max
= 20C
t
min
= 20C
t* = 0,75 t
max
+ 0,25 t
min
= 20C
R
m
= 410 MPa
R
p0,2/t*
= 255 MPa
Effective stress concentration factorK
eff
but not largen than K
t
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors. K
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
R
z
= 50 m (table 18-8)
e
n
= 25 mm

D
= 279.3 MPa (table 18-10 for N 2
.
10
6
cycles)
N = 100 000 (for the first iteration)

R
= 407.8 MPa (allowable stress range for N< 2.10
6
cycles)
Stresses

eq,t
(total or notch equivalent stress range) = 344.5 MPa

struc
(structural equivalent stress range) = 312.3 MPa (obtained by quadratic extrapolation)
0957 . 1
D

struc

t
K 0,5 1
1)
t
(K 1,5
1
eff
K
+

+
1
1
]
1

Theoretical elastic stress concentration factor K


t
K
t
=
eq,t
/
struc
= 1.1031
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..

total
= k
e
.
eq,t
= .. MPa
Else
total
=
eq,t
= 344.5 MPa
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
l eq,

0,4
0,5
0,7
k

k

= ..

total
= k

.
eq,t
= .. MPa
Else
total
=
eq,t
= .. MPa
stress equivalent notch (mean MPa 172.75
eq

18-10-6-2 Temperature correction factor f
t*
For t* > 100 C
f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
18-11-1-1 Surface finish correction factor f
s
f
s
= F
s
[0,1ln(N)-0,465]
if N < 2.10
6
f
s
= F
s
if N 2.10
6
with F
s
= 1- 0,056 [ln (R
z
)]
0,64
[ln(R
m
)] +0,289 [ln (R
z
)]
0,53
= 0.7889
f
s
= 0.8539

eq,struc
=
t
K
total
(for usage in 18-11-3 and 18-11-2-1) = 312.3

eqmax
= 344.5 MPa (maximum notch equivalent stress)
Illustrative Example
Page
4.10
DBA
Design by Analysis
Fig.4.10: Pivot points for quadratic interpolation
18-11-1-2 Thickness correction factor f
e
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
e
= F
e
[0,1ln(N)-0,465]
if N < 2.10
6
f
e
= 0,7217
[0,1ln(N)-0,465]
if N< 2.10
6
f
e
> F
e
if N 2.10
6
f
e
= 0,7217 if N 2.10
6
with Fe = (25/en)
0,182
= .....
f
e
= 1 f
e
=

.. f
e
= ..
18-11-1-3 Mean stress correction factor f
m
If
struc
< 2 R
p0,2/t*
and
eqmax
< R
p0,2/t*
If
eq
> 0 then
r eq,
= R
p0,2/t*
-
2
t eq,

= 82.75
If
eq
< 0 then
r eq,
=
2
t eq,

- R
p0,2/t*
and
eq
=
r eq,
= 82.75 MPa
For N 2 10
6
cycles
See figure 18-14
f
m
= ..
If
struc
>2 R
p0,2/t*
18-11-2-1 Overall correction factor f
u
f
u
=
eff
K
* t
.f
m
.f
e
.f
s
f
= 0.7658
u
f /
struc eq,
407.8
18-11-3 Allowable number of cycles N
2
10 6 , 4
5 , 11 63 , 0
4
,
1
1
]
1

m
u
struc eq
R
f
N

if N 2 10
6
cycles
if R
p0,2/t*

eq

( ) M 1 2
R

+
then
f
m
=
( )
5 , 0
eq
2
M 1
M 2 M
- 1
R
1
1
]
1

,
_

+
+
= 0.9826
N = if
D u
f /
struc eq,

N = 81600
N is obtained by iterations. If the value of N oscillates, a difference equal to 1 % between
two iterations is acceptable. If the values decrease monotonously, the difference must be
less than 0,001 % between two iterations.
If
struc
< 2 R
p0,2/t*
and
eqmax
> R
p0,2/t*
if
( ) M 1 2
R

+

eq
R
p0,2/t*
then
f
m
=

,
_

+
+
R

eq
2
3
M
M 1
3 M 1
= .
For N 2 10
6
cycles
M = 0,00035 R
m
0,1 = ..
f
m
= 1
Specification of Examples
DBA
Design by Analysis
Page
5.1
5 Specification of examples
5.1 General
In this section the specifications are given of ten examples, dealt with by the group.
The various design checks had been assigned to the members a priori, differences discussed,
and, where necessary, supplemented by additional investigations and corrected.
The results are summarized in the next section, the details in the section thereafter.
The specifications are complete, but following the examples in details, or using them as
benchmarks, it is necessary to consult standards, like material standards, drafts of other parts
of prEN13455, etc.
As a help, physical properties of materials used in the examples are collected in an annex to
this section.
The geometries specified in the drawings are already those to be used in the analyses, i. e. the
thicknesses given are already analysis thicknesses, allowances for tolerances and, if
relevant, corrosion have already been deducted. Should the corrosion allowance be required
for specifying the weld regions: A value of 1.00 mm was used for ferritic steels, and 0 for
austenitic ones.
The proposal for Detailed Fatigue Analysis states that in the (fatigue) analysis extreme
operating values of actions rather than design values should be used. For the calculation of the
allowable number of action cycles an upper value for the pressure
sup , op
P equal to 90% of the
maximum allowable pressure PS is specified in the examples here.
In cases where the maximum allowable pressure PS can be determined by the Design by
Formulae (DBF) section of the CEN TC 54 proposal of an Unfired Pressure Vessel standard,
PS has not been specified here.
The maximum allowable pressure according to this DBF proposal -
DBF
PS
max
- shall be used
as characteristic value in the design checks for Gross Plastic Deformation (GPD), Progressive
Plastic Deformation (PD) or Shakedown (SD):
DBF c
PS P
max
=
In the other cases values for PS are specified.
Note: Unfortunately, because of the combination of ideas and designations from Euronorm 3
on one hand, and those from the Pressure Equipment Directive (PED) and EN 764 on the
other hand, one has to distinguish between the design values of (the action) pressure
obtained by multiplying the characteristic values of pressure by the relevant partial safety
factors and the design pressure
d
P - the maximum pressure at the top of the equipment
specified by the manufacturer and used for the determination of the calculation pressures,
mainly within the framework of DBF.
Specification of Examples
DBA
Design by Analysis
Page
5.2
To minimize the possibility of confusing the two, the notion design pressure relevant to
DBF is never used here. The specified values for PS or those determined by DBF shall be
used as upper characteristic values; the design value for (the action) pressure shall not be
called design pressure and especially not be denoted by
d
P ! (It could be denoted by
Pd
A ).
In all cases the admissibility of the design shall be checked and proven first by the simplest
means possible for the example to show that quite often DBA can be quite short and simple,
if only the admissibility is to be be proven.
In all examples a normal hydraulic test is presupposed, i. e. the checks against GPD for
testing conditions do not require separate calculations.
Specified pressures and temperatures are to be considered as pairs. If other actions are
specified, they are considered to form, with pressures and temperatures, triplets, etc.
Should the alternative of using any primary stress field be used in the check against GPD, the
principal strains corresponding to this primary stress field shall be limited by 5% - as
specified in the tangent intersection procedure. In general, this requires primary stress fields
obtained by (inelastic) FEM.
5.2 General notations
Where possible, the notations of the CEN proposal for the Unfired Pressure Vessel Standard,
prEN 13445, are used, and shall be used:
Analysis thickness: Nominal thickness minus allowances manufacturing,
e
, and
corrosion, erosion, c :
c e e
e n a
=
Maximum allowable pressure: The maximum pressure (on top of the vessel) specified (by
the manufacturer) for design, for normal operating conditions: PS . This pressure PS
constitutes an upper limit for the set-pressure of the safety valve if there is only one -, or
for the maximum pressure (at the top of the compartment) that can occur under reasonably
foreseeable conditions if no safety valve is required. It shall be used in the design checks
against GPD and PD, or SD.
Maximum operating pressure:
sup , op
P
This value specified directly, or as being equal to 90% of
DBF
PS
max
- shall be used as
upper value (of full pressure cycles) for cases with cyclic pressure.
Maximum allowable pressure according to prEN 13445-3, Annex 5.B:
DBF
PS
max
Maximum allowable pressure according to DBA:
DBA
PS
max
For GPD and PD/SD checks, and, if relevant, for checks against instabilty (I), only.
Characteristic value of moment: Maximum reasonably foreseeable value of external
moment; (in general equal to the "usual" design moment):
c
M
Allowable number of cycles: Number of cycles (for specified actions) allowed by
prEN 13445-3, Section Detailed Fatigue Analysis.
Specification of Examples
DBA
Design by Analysis
Page
5.3
Modulus of elasticity: E
E at 100C, say:
100
E
Mean coefficient of (linear) thermal expansion:
between 20C and 100C, say:
100 , 20

Weld symbols according to EN 22553.


5.3 Designations
DBA. Design by Analysis
DBF Design by Formulae
GPD Gross Plastic Deformation
PD Progressive Plastic Deformation
SD Shakedown
I Instability
F Fatigue
NLG Non-Linear Geometry
PS Maximum alloxable pressure
PS
maxDBF
Maximum allowable pressure according to prEN 13445-3 Section DBF
PS
maxGPD
Maximum allowable pressure according to Gross Plastic Deformation using
DBA
PS
maxSD
: Maximum allowable pressure according to Progressive Plastic Deformation
using DBA
sup , op
P Maximum operating pressure.
Pd
A Design value for the pressure action
P
ap
Applied pressure to elastic compensation analysis
M
c
Characteristic value of moment
T
c
Calculation temperature
E Modulus of elasticity
Mean coefficient of (linear) thermal expansion
Heat conduction coefficient
h Heat transfer coefficient
a Thermal diffusivity (temperature conductivity)
Stress.....
nom e e i ij
, , , ,
max ,
Strain.....
e ij
,
Specification of Examples
DBA
Design by Analysis
Page
5.4
5.4 Specifications of examples
Example No. 1.1: Thick unwelded flat end
1. Material: P280GH according to EN 10222-2
The relevant heat treatment dimension is specified as
101.6 mm (in deviation from EN 10222-1).
2. Actions: Pressure MPa PS PS
DBF
17
max
= =
x)
Temperature C T
c
= 20
3. Operational cycles: const T = , p varying from 0 to PS P
op
= 9 . 0
sup ,
4. Geometry: See Fig. 5.1
Fig. 5.1
x) A not very reasonable result: The end thickness is large and the ratio of admissible pressure to
nominal design stress is outside the graphs and the scope of DBF. Extrapolation was necessary.
Specification of Examples
DBA
Design by Analysis
Page
5.5
Example No. 1.2: Thin unwelded flat end
1. Material: P280GH according to EN 10222-2
The relevant heat treatment thickness is specified as
101.6 mm (in deviation from EN 10222-2).
2. Actions: Pressure MPa PS PS
DBF
2 . 4
max
= =
Temperature C T
c
= 20
3. Operational cycles: const T = , p varying from 0 to PS P
op
= 9 . 0
sup ,
4. Geometry: See Fig. 5.2
Fig. 5.2
Specification of Examples
DBA
Design by Analysis
Page
5.6
Example No. 1.3: Welded-in flat end without nozzle
1. Material: P265GH according to EN 10028-2
2. Actions: Pressure MPa PS PS
DBF
7 . 12
max
= =
Temperature C T
c
= 20
3. Operational cycles: const T = , p varying from 0 to PS P
op
= 9 . 0
sup ,
4. Geometry: See Fig. 5.3 and 5.4
Fig. 5.3
Specification of Examples
DBA
Design by Analysis
Page
5.7
Fig. 5.4
Example No. 1.4: Welded-in flat end with nozzle
1. Material: Plate and Shell: P265GH according to EN 10028-2
Nozzle: P265 according to prEN 10216-2
2. Actions: Pressure MPa PS PS
DBF
9 . 7
max
= =
Temperature C T
c
= 20
3. Operational cycles: const T = , p varying from 0 to PS P
op
= 9 . 0
sup ,
4. Geometry: See Fig.5.5, 5.6 and 5.7
Specification of Examples
DBA
Design by Analysis
Page
5.8
Fig. 5.5
Fig.5.6 Fig. 5.7
Specification of Examples
DBA
Design by Analysis
Page
5.9
Example No. 2: Storage tank (cone-cylinder junctions)
1. Material: Shell: X6CrNiTi 18-10 (1.4541) according to prEN 10028-7
Reinforcing ring, foot ring: P235GH according to EN 10028-2;
Note: the different thermal expansion coefficients shall be
considered.
2. Actions: Hydrostatic pressure
H
p , medium density
3
1000 m kg
M
=
minimum medium level h
MIN
and maximum medium level
h
MAX
see Fig. 5.8. Note:
no longitudinal stress in the main cylindrical shell caused by
hydrostatic pressure.
Temperature in service C T
c
= 60 ;
Temperature before complete filling of the
vessel 20C.
Internal pressure during draining (see also Fig. 5.9)
MPa PS 06 . 0 = ;
Note: longitudinal stress in the main cylindrical shell caused
by internal pressure acting on the upper end of the vessel.
Dead load (self weight and insulation): Insulation:
2
220 m N q
d
= (weight force / surface of the vessel),
insulation thickness 200 mm; dead weight of roof including
insulation and reinforcing ring 26,15 kN.
Wind load (limit value): stagnation pressure q
W
depending on
height h: : 6 0 m h m
2
/ 81 . 0 m kN q
W
=
: 10 6 m h m <
2
/ 88 . 0 m kN q
W
=
: 15 10 m h m <
2
/ 94 . 0 m kN q
W
=
: 25 15 m h m <
2
/ 02 . 1 m kN q
W
=
Wind force:
i i W i
A q c W =
,
where 44 . 0 = c and
i
A = projection of the surface of
the vessel in wind direction.
3. Detail to be investigated: wide and narrow ends of cone
4. Operational cycles: See Figure 5.9.
Note: It is ascertained that internal pressure can be
increased only if the medium height is below h
MIN
. The
internal pressure increases slowly; for safety reasons both
extremes shall be considered, the very slow (dotted line) and very
fast (full line) pressure increases.
5. Geometry: See Fig.5.8.
Specification of Examples
DBA
Design by Analysis
Page
5.10
Fig. 5.8
Specification of Examples
DBA
Design by Analysis
Page
5.11
Fig. 5.9
Example No. 3.1: Thin-walled cylinder-cylinder intersection
1. Material: P295GH according to EN 10028-2
2. Actions: Pressure MPa PS PS
DBF
28 . 0
max
= =
Nozzle longitudinal moment Nm M
c
4 . 15644 =
(moment vector normal to plane through both cylinder
axes).
Temperature C T
c
= 50
3. Operational cycles: A) const T = , p varying from 0 to , 9 . 0
sup ,
PS P
op
=
const M
c
=
and
B) const T = , M varying from 0 to 26400 Nm,
MPa const p 28 . 1 = =
(for comparison with experimental results).
Crotch corner surface machined: m R
z
50 =
Specification of Examples
DBA
Design by Analysis
Page
5.12
4. Geometry: See Fig.5.10.
Note: Checks against GPD and PD, or SD, to be performed for constant
longitudinal moment only.
Fig. 5.10
Specification of Examples
DBA
Design by Analysis
Page
5.13
Example No. 3.2: Thick-walled cylinder-cylinder intersection
1. Material: Shell: P265GH according to EN 10028-2
Nozzle: 11CrMo9-10 according to prEN 10216-2
2. Actions: Pressure MPa PS PS
DBF
09 . 14
max
= =
Nozzle longitudinal moment Nm M
c
1 . 711 =
(moment vector normal to plane through both cylinder
axes).
Temperature C T
c
= 50
3. Operational cycles: A) const T = , p varying from 0 to , 9 . 0
sup ,
PS P
op
= const M
c
=
and
B) const T = , M varying from 0 to1200 Nm,
MPa const p 24 = =
(for comparison with experimental results).
Crotch corner surface machined: m R
z
50 =
4. Geometry: See Fig 5.11.
Note: See note in Example No. 3.1.
Fig. 5.11
Fig. 5.11
Specification of Examples
DBA
Design by Analysis
Page
5.14
Example No. 4: Dished end with nozzle in knuckle region
1. Material: X6CrNiMoTi 17-12-2 (1.4571) according to prEN 10028-7
2. Actions: Pressure MPa PS PS
DBF
583 . 0
max
= =
Temperature C T
c
= 180
3. Operational cycles: const T = , p varying from 0 to PS P
op
= 9 . 0
sup ,
4. Geometry: See Fig. 5.12 and 5.13.
Fig. 5.12
Fig. 5.13
Specification of Examples
DBA
Design by Analysis
Page
5.15
Example No. 5: Nozzle in spherical end with cold medium injection
1. Material: Shell: 11CrMo9-10 according to EN 10028-2
Nozzle reinforcement: 11CrMo9-10+QT according to
prEN 10216-2
Nozzle: P265 according to prEN 10216-2
2. Actions: Pressure MPa PS PS
DBF
71 . 11 01 . 13 9 . 0 9 . 0
max
= = =
Temperature of medium inside the vessel C T
S
= 325 (constant
in operation).
Temperature of injected cold medium C T
N
= 80 .
Location of different heat transfer coefficient for cold medium
injection see Fig. 5.15. The outer surface of the vessel is
insulated ideally.
Heat transfer coefficients:
-) medium to vessel wall, and to nozzle if there is no injection:
K m kW h
S
2
16 . 1 =
-) cold (injection) medium to nozzle wall during injection:
K m kW h
N
2
8 . 10 = .
3. Operational cycles: See Fig. E 5.15.
Cold medium injection takes place for 10 minutes. The time
between the injection cycles is long enough such that temperature
reaches , stationarity. After 500 injection cycles one shutdown (and startup) should be
considered. At shutdown and startup, pressure and temperature are decreased or
increased in phase, respectively. Temperature changes during shutdown and
startup are slow, and therefore thermal stresses can be neglected.
4. Geometry: See Fig. 5.14
Specification of Examples
DBA
Design by Analysis
Page
5.16
Fig. 5.14
Fig. 5.15
Specification of Examples
DBA
Design by Analysis
Page
5.17
Example No. 6: Jacketed vessel with jacket on cylindrical shell only, and flat annular
end plates
1. Material: X6CrNiTi 18-10 (1.4541) according to prEN 10028-7
2. Actions: Inner space: Pressure MPa PS 3 . 1 / 1 . 0 + =
Temperature C TS T
c
= = 160
Outer space: Pressure MPa PS 5 . 0 / 0 + =
Temperature C TS T
c
= = 160
3. Operational cycles: See Fig. 5.18.
Inner space: MPa P
op
1 . 1
sup ,
= MPa P
op
0
inf ,
=
TS T
op
=
sup ,
C T
op
= 20
inf ,
.
Outer space: MPa P
op
45 . 0
sup ,
= MPa P
op
0
inf ,
=
TS T
op
=
sup ,
C T
op
=10
inf ,
A pressure in the inner space below atmospheric can occur
independently and repeatedly in operation, and an underpressure
will occur concurrently with an outer space temperature of 10C
(whereby an inner space temperature value of 160C shall be
used). This case shall be included as a normal operating
condition in the check against GPD, I, PD, or SD, but not in
the fatigue check. A pressure in the outer space below
atmospheric cannot occur, but a minimum pressure of 0 bar
cannot be excluded. This case shall also be included as a normal
operating condition in the check against GPD, I, PD, or SD
(with temperatures in the inner and outer space of 160C).
Note:
sup , op
T and
inf , op
T are medium temperatures. The wall
temperatures shall be determined using heat transfer coefficients
of K m kW h
i
2
16 . 1 = on inside of inner vessel wall and
K m kW h
o
2
4 . 14 = on all surfaces of the inside of the jacket. Jacket
and main vessel outside of jacket are insulated ideally.
Note: checks against GPD, I, PD or SD shall be performed usingthe PS
values.
Note: only steady state thermal stresses shall be considered.
The maximum allowable out-of-roundness of the inner cylindrical shell
is specified in prEN13445-3 as (D+1250) / 200 = (2780 + 1250) / 200 = 20,15 mm, where D
is the mean shell diameter.
4. Details to be investigated: Jacket and jacketed part of inner vessel
5. Geometry: See Fig. 5.16 and 5.17.
Specification of Examples
DBA
Design by Analysis
Page
5.18
Fig. 5.16
Fig. 5.17
Specification of Examples
DBA
Design by Analysis
Page
5.19
Fig. 5.18
160C
160C
Specification of Examples
DBA
Design by Analysis
Page
5.20
5.5 Appendix: Physical properties of some materials
P 235 GH
P 265 GH
Specification of Examples
DBA
Design by Analysis
Page
5.21
P 295 GH
11CrMo9-10
Specification of Examples
DBA
Design by Analysis
Page
5.22
1.4541
1.4571
1
Analysis Summary
Page
6.1
DBA
Design by Analysis
The following tables summarise the checks carried out for the ten examples, specified in Section 5.
All fatigue results of welded regions are obtained for testing group 1 and, in general, based on the
principal stress range approach; in cases where the equivalent stress range approach is used this is
stated in a footnote.
Results are designated in terms of the routes used as follows:
DBF Design by formulae
DBS Design by analysis using the direct route elasto-plastic approach
DRC Design by analysis using the direct route elastic compensation approach
SC Design by analysis using the stress categorisation approach
NLG For information only: Result for non-linear geometry in the elasto-plastic approach
and in terms of the checks performed:
GPD Check against gross plastic deformation
PD Check against progressive plastic deformation
I Check against instability
F Calculation of allowable number of cycles according to the F-check
SE Check for static equilibrium
The result summaries are tabulated as follows:
Table 6.1 Example 1.1 Thick unwelded flat end
Table 6.2 Example 1.2 Thin unwelded flat end
Table 6.3 Example 1.3 Welded-in flat end without nozzle
Table 6.4 Example 1.4 Welded-in flat end with nozzle
Table 6.5 Example 2 Storage tank (cylinder-cone junction)
Table 6.6 Example 3.1 Thin walled cylinder-cylinder intersection
Table 6.7 Example 3.2 Thick walled cylinder-cylinder intersection
Table 6.8 Example 4 Dished end with nozzle in knuckle region
Table 6.9 Example 5 Nozzle in spherical shell with cold medium injection
Table 6.10 Example 6 Jacketed vessel: jacket on cylindrical shell with flat
annular end plates
Table 6.1: Analysis Summary for Example 1.1: Thick unwelded flat end
Project Member and Analysis Type
St. A A&AB WTCM Strathclyde TKS RWTUV CETIM
Design Check
DBF DRS DRS DRC DRS SC NLG
GPD PS(MPa) 60.87 58.02 60.1 62.8 58 -
PD PS(MPa)
17
[1]
101.45 - 102.1 -
69.7
[2]
57.5
[3]
- -
I Not Required
F (cycles) infinity infinity - - - infinity - infinity
SE Not Required
[1] Conservative result as the geometry is at the limit allowed by DBF.
[2] Thick cylinder and head modelled using shell elements with pressure applied at inner radius (primary stresses are limiting).
[3] Thick cylinder and head modelled using shell elements with pressure applied at mean radius (primary stresses are limiting).
2
Analysis Summary
Page
6.2
DBA
Design by Analysis
Table 6.2: Analysis Summary for Example 1.2: Thin unwelded flat end
Project Member and Analysis Type
St. A A&AB WTCM Strathclyde TKS RWTUV CETIM
Design Check
DBF DRS DRS DRC DRS SC NLG
GPD PS(MPa) 5.7 6.18 4.5 5.7 5.6 -
PD PS(MPa)
4.2
7.9 - 7.25 -
5.6
[1]
- -
I Not Required
F (cycles) 360000 - - - - 309600 - 375200
SE Not Required
[1] Primary stresses are limiting
Table 6.3: Analysis Summary for Example 1.3: Welded-in flat end without nozzle
Project Member and Analysis Type
St. A A&AB WTCM Strathclyde TKS RWTUV CETIM
Design Check
DBF DRS DRS DRC DRS SC NLG
GPD PS(MPa) 12.6 12.41 10.8 12.48 12.4 -
PD PS(MPa)
12.7
13.3 - 13.3 -
12.3
[1]
- -
I Not Required
F (cycles) 13581
[2]
6910 - - - 4665 - 14673
[2]
SE Not Required
[1] Shakedown is limiting
[2] Equivalent stress range approach used.
Table 6.4: Analysis Summary for Example 1.4: Welded-in flat end with nozzle
Project Member and Analysis Type
St. A A&AB WTCM Strathclyde TKS RWTUV CETIM
Design Check
DBF DRS DRS DRC DRS SC NLG
GPD PS(MPa) 9.9 10.0 8.42 9.92 9.9
[2]
-
PD PS(MPa)
7.8
12.2 - 12.8 -
10.0
[1]
- -
I Not Required
F (cycles) 28600
[3]
21010 - - - 14570 - 26947
[3]
SE Not Required
[1] Primary stresses are limiting
[2] 5% maximum absolute value of principal strain exceeded at this load.
[3] Equivalent stress range approach used.
3
Analysis Summary
Page
6.3
DBA
Design by Analysis
Table 6.5: Analysis Summary for Example 2: Storage tank (cylinder-cone junction)
Project Member and Analysis Type
St. A A&AB Strathclyde TKS CETIM
Load
Case
Design Check
DBF DRS DRC SC
GPD PS(MPa) Admissible Admissible -
1
PD PS(MPa) Admissible - -
GPD PS(MPa) Admissible Admissible -
2
PD PS(MPa) Admissible - -
GPD PS(MPa) Admissible Admissible -
3
PD PS(MPa) Admissible - -
- I Admissible Admissible - -
- F (cycles) - - - 830 1984
[2]
- SE - Admissible
[1]
- - -
[1] Bolting required for wind load only.
[2] Equivalent stress range approach used.
Load Cases:
(1) Hydrostatic pressure at maximum medium level, dead weight, wind load, and for PD check only, thermal stresses.
(2) Hydrostatic pressure at minimum medium level, draining pressure, dead weight and wind load, and for PD check only, thermal
stresses.
(3) Draining pressure, dead weight and wind load, and, for PD check only, thermal stresses.
Table 6.6: Analysis Summary for Example 3.1: Thin walled cylinder-cylinder intersection
Project Member and Analysis Type
St. A A&AB Strathclyde WTCM CETIM
Design Check
DBF DRS DRC DRS SC
GPD PS(MPa) 0.39 0.467
[1]
0.383 -
PD PS(MPa)
0.28
0.5 1.34
[2]
-
0.45
-
I Not Required
load A 96500
[3]
- - - - 122770
[3]
F
(cycles)
load B 1710
[3]
- - - - 1042
[3]
SE Not Required
[1] The 5% maximum absolute principal strain rule can currently not be used in elastic compensation, resulting in a much higher load
corresponding to loss of equilibrium between the external applied loads and internal stresses and strains. See Chapter 3 for
more details.
[2] This value exceeds the 2RM limit placed on shakedown at the discontinuity, for more information see the analysis details.
[3] Equivalent stress range approach used.
load (A): constant moment, varying pressure
load (B): constant pressure, varying moment
4
Analysis Summary
Page
6.4
DBA
Design by Analysis
Table 6.7: Analysis Summary for Example 3.2: Thick walled cylinder-cylinder intersection
Project Member and Analysis Type
St. A A&AB Strathclyde Strathclyde WTCM CETIM
Design Check
DBF DRS DRC DRS SC
GPD PS(MPa) 14.79 11.5 15.1 - -
PD PS(MPa)
14.09
17.65 13.7 -
14.25
- -
I Not Required
load A - 70865
[1]
- - - 59494
[1]
55474
[1]
F
(cycles)
load B - 1160 - - - - 790
[1]
SE Not Required
[1] Equivalent stress range approach used
load (A): constant moment, varying pressure
load (B): constant pressure, varying moment
Table 6.8: Analysis Summary for Example 4: Dished end with nozzle in knuckle region
Project Member and Analysis Type
St. A A&AB Strathclyde WTCM RWTUV CETIM
Design Check
DBF DRS DRC DRS SC NLG
GPD PS(MPa) 0.375 0.41
[1]
0.42 0.42
[2]
-
PD PS(MPa)
0.583
0.289
[5]
0.375
[6]
0.271
[3]
-
0.263
- -
I Not Required
F (cycles) 2000
[4]
4072 - - - - 880
[4]
SE Not Required
[1] The 5% maximum absolute principal strain rule can currently not be used in elastic compensation, resulting in a much higher load
corresponding to loss of equilibrium between the external applied loads and internal stresses and strains. See Chapter 3.3 for
more details.
[2] 5% maximum absolute value of principal strain exceeded at this load.
[3] Modelled with shell elements thus simplified geometry and discontinuities.
[4] Equivalent stress range approach used.
[5] Shakedown limit pressure
[6] Allowable number of cycles: 1440. See Chapter 7.9 for more details.
Table 6.9: Analysis Summary for Example 5: Nozzle in spherical shell with cold injection
Project Member and Analysis Type
St. A A&AB JRC Strathclyde WTCM TKS CETIM
Design Check
DBF DRS DRS DRC DRS SC SC
PS
max
[1]
(MPa) 13.013 13.04 13.125 11.1
[2]
11.43
[2]
-
GPD - Admis. Admis. N-A N-A -
PD - N-A
[5]
N-A
[5]
- -
N-A
[3]
N-A
[4]
-
I Not Required
F (cycles) - - - - - - 13 9.8
[6]
SE Not Required
[1] Maximum allowable pressure according to the GPD-check.
[2] Defined internal pressure, PS is 0.9xPS
maxDBF
= 11.71MPa, therefore, load is non-admissible.
[3] No pressure allowable due to the magnitude of the thermal stresses according to the conditions of the SC Route.
[4] The allowable pressure for the primary stress criterion only is 12.8 MPa using ANSYS and 11.7 MPa using BOSOR.
[5] The thermal stress prevents the structure from shaking down at the applied load and is therefore non-admissible according to the
PD-check.
[6] 9.8 full cycles (start up shut down + 500 cold media injection).
5
Analysis Summary
Page
6.5
DBA
Design by Analysis
Table 6.10: Analysis Summary for Example 6: Jacketed vessel: jacket on cylindrical shell with flat
annular end plates
Project Member and Analysis Type
St. A A&AB Strathclyde WTCM RWTUV CETIM
Load
Case
Design
Check
DBF DRS DRC NLG DRS NLG
PS
maxGPD
[1]
(MPa)
1.3
1.35 1.19 2.0 1.26 1.23 -
GPD Admissible Non-
Admissible
Admissible Non-
Admissible
Non-
Admissible
-
1
PD
[3]
Admissible Admissible - - - -
GPD Admissible Non-
Admissible
Admissible Non-
Admissible
Non-
Admissible
-
2
PD
[3]
Admissible Admissible - - - -
GPD Admissible Admissible Admissible Admissible Admissible -
3
PD
Admissible
Admissible - - - - -
4 PD - Admissible - - - - -
GPD Admissible Admissible Admissible Admissible Admissible -
5
PD
Admissible
Admissible - - - - -
- I - Admissible Admissible - - - -
- F (cycles) - 20506 - - 20695 - 67953
[2]
- SE Not Required
[1] Maximum calculated allowable pressure in the dished end (pressure limiting component)
[2] Equivalent stress range approach used
[3] To meet DBF requirements the thickness of upper stiffener should be increased from 13.5 to 28 mm and the dimensions of the
lower stiffener from 20x110 mm
2
to 41x125 mm
2
.
Load Cases:
(1) Inner chamber internal pressure PS
i
= 1.3 MPa (4) Inner chamber internal pressure PS
i
= 1.3 MPa
Outer chamber internal pressure PS
o
= 0.5 MPa Outer chamber internal pressure PS
o
= 0.5 MPa
temperature in both chambers TS
i
= 160
o
C temperature in inner chamber TS
i
= 160
o
C
(2) Inner chamber internal pressure PS
i
= 1.3 MPa temperature in outer chamber TS
o
= 10
o
C
Outer chamber internal pressure PS
o
= 0.0 MPa (5) Inner chamber internal pressure PS
i
= 0.0 MPa
temperature in both chambers TS
i
= 160
o
C Outer chamber internal pressure PS
o
= 0.5 MPa
(3) Inner chamber internal pressure PS
i
= -0.1 MPa temperature in inner chamber TS
i
= 20
o
C
Outer chamber internal pressure PS
o
= 0.5 MPa temperature in outer chamber TS
o
= 10
o
C
temperature in inner chamber TS
i
= 160
o
C
temperature in outer chamber TS
o
= 10
o
C
Analysis Details
Page
7.1
DBA
Design by Analysis
7 Analysis Details
7.1 General
Details of some of the various checks by the members of the group are compiled in this section.
The page numbering is consecutively throughout the whole section.
Since each contribution is a more or less self-contained part, each is preceeded by a summary page,
which shows in the page head the example number, in the Analysis Type box the checks dealt with,
and in the Member box an acronym for the responsible group member.
To ease browsing, a letter is added in brackets to the page number to indicate the responsible group
member, and the paragraph numbering is new for each new contribution.
The usual order is
GPD-Check and PD-Check
Stress Categorisation Route
F-Check
I-Check, where applicable
SE-Check, where applicable
For the GPD-check and the PD-check there are usually two approaches given
Direct route using elastic compensation
Direct route (using non-linear calculations)
The stress categorisation route follows after the PD-check.
The fatigue results are all based on linear-elastic FE-calculations. If the F-check given is from a
group member different from the one whose PD-check is given, these linear-elastic calculations are
separate ones, just for the F-check. The stress components used are stated, but they cannot be
deduced always from other plots exactly.
To avoid unnecessary problems, we state here again that the formulae in the Fchecks differ from
those given in section 18 of prEN 13445-3, which is reproduced in Annex 2.
The formulae used here agree with those in the flow-sheets given in subsection 3.8 of this manual,
and the flow-sheets have been used directly. The main changes in the formulae are stated in
subsection 3.8 of section 3 Procedures, and they are repeated at the beginning of Annex 2.
Since some details on partial safety factors for actions and on characteristic values are missing in
the subsection on the I-check in prEN13445-3 Annex B, some additional material had been agreed
upon, used, and described in subsection 3.7 of section 3 Procedures.
.
Analysis Details
Example 1.1
Page
7.2(S)
DBA
Design by Analysis
Analysis Type: GPD-Check and PD-Check
Direct Route using Elastic Compensation
FE-Software: ANSYS 5.4
Element Types: 8 node, 2-D structural axisymmetric solid.
Boundary Conditions: No vertical displacement in the undisturbed end of the shell remote
from the flat end.
Axisymmetry from elements
Model and Mesh:
Member:
Strathclyde
Number of elements 339
Height of model 500 mm
Results:
Maximum internal pressure according to the GPD-check: PS
max GPD
= 60.1 MPa
Check against PD: Shakedown limit pressure PS
max GPD
= 102.1 MPa
Analysis Details
Example 1.1
Page
7.3(S)
DBA
Design by Analysis
Figure 7.2.1-1: Limit Stress Field (Tresca)
1. Finite Element Mesh
Element divisions for the finite element mesh were defined parametrically to allow modification of
the element density. Analysis was carried out for a coarse mesh density and a much finer mesh
density to allow any alterations on the results to be noted. Finite element models were created using
linear 4-node 2-D axisymmetric solids, the analysis was repeated with higher order 8-node
structural 2-D axisymmetric solids, any alterations this may have on the results could also be noted.
The allowable pressure according to GPD and the shakedown pressure according to PD were
calculated from the model giving the greatest lower bounds. Here, the fine density mesh with 8-
node higher order elements gave the highest lower bounds.
Boundary conditions applied to the model reflect axisymmetry, applied via a key option when
defining the element type in the FE-software (axisymmetry around the vertical axis Y). The nodes
at the undisturbed end of the cylindrical shell have their vertical degree of freedom constrained to
zero to ensure that plane sections remain plane.
2. Material properties
Material strength parameter MPa RM 255 = , modulus of elasticity GPa E 212 = .
3. Determination of the maximum admissible pressure according to the GPD-Check
Using the application rule in prEN 13445-3 Annex B.9.2.2 to check against GPD, the principle is
fulfilled when for any load case the combination of the design actions do not exceed the design
resistance. This may be shown by calculating the limit load. The limit load has to be determined
using Trescas yield condition and associated flow rule. As elastic compensation is based on a
series of elastic equilibrium stress fields, it is a relatively simple procedure to derive a lower bound
limit load directly from the Tresca yield model.
From prEN 13445-3 Annex B, Table B.9-3 the partial safety factor,
R
on the resistance is 1.25.
Therefore, the design material strength parameter is given by RM/
R
= 204 MPa. The analysis was
carried out using the elastic
compensation method conforming to
the direct route rules for GPD in
Annex B: linear elastic - ideal plastic
material law, Trescas yield
condition and associated flow rule
and first order theory.
In each elastic compensation
iteration an equilibrium stress field is
produced where the elastic modulus
of each subsequent iteration is
defined by the previous elastic
solution. In this way some regions
in the FE-model may be
systematically stiffened or weakened
depending upon the stress
magnitudes in the previous solution.
The result is, that every equilibrium
stress field is a lower bound of the
Analysis Details
Example 1.1
Page
7.4(S)
DBA
Design by Analysis
Figure 7.2.1-2: Residual Stress Field (Mises)
limit load, the stress field giving the greatest limit load is taken as the closest to the actual limit state
and defines the limit load in the analysis. Convergence of the equilibrium stress field will occur
typically between 8 and 15 iterations, i.e. no further benefit in the limit load will be noted with
subsequent iterations. The total computing time to run the analysis on a 300 MHz Pentium two
processor with 128 Mb RAM Windows NT workstation was 70 seconds. The stress field was
shown to converge after eight iterations giving a lower bound on the pressure limit of 72.1 MPa.
Figure7.2.1-1 shows the limit stress (intensity) field according to the Tresca criterion. The limit
pressure is given by scaling the limit stress field so that the stress anywhere in the model does not
exceed the design materialstrength, 204 MPa, i.e. the applied pressure is scaled by the factor
(204/28.304) 7.207. According to prEN 13445-3 Annex B, Table B.9-2 for pressure loads (without
natural limit) the partial safety factor
p
is 1.2. Thus, the internal pressure limit according to failure
by GPD is
It is also possible to determine a limit pressure from the check against PD. In this case, the elastic
compensation is based on Mises yield condition. The partial safety factor on the resistance
R
is
not applied for the PD-check. However, as the analysis is wholly elastic it is possible to scale the
stress fields at any time (similarly as was done above). The maximum ratio of Mises' equivalent
stress to Tresca's equivalent stress for the same load is 2/3. Therefore, applying a factor of 3/2 to
the yield stress in the Mises analysis (or to the limit load, as the analysis is elastic) will always lead
to a conservative result. From the Mises analysis the limit load was found to be 102.1 MPa, and
with the partial safety factors
R
= 1.25 and
p
= 1.2, the internal pressure limit according to GPD
can be found as
4. Check against PD
In this check the principle in prEN-
13445-3 B.9.3.1 is fulfilled if the
structure can be shown to shake
down. When a structure has been
shown to shake down, the failure
modes of progressive plastic
deformation and alternating
plasticity can not occur.
In elastic compensation the load at
which the structure will shake down
is simple to calculate. Based on
Melans shakedown theorem, the
self-equilibrating residual stress field
that would result after a loading
cycle can be calculated by
subtracting the linear-elastic stress
field at the limit pressure from the
limit stress field. The residual stress
MPa PS
GPD
1 . 60
2 . 1
1 . 72
max
= =
MPa PS
R p
GPD
95 . 58
2
3 1 . 102
max
=

=

Analysis Details
Example 1.1
Page
7.5(S)
DBA
Design by Analysis
Figure 7.2.1-3: Limit Stress Field (Mises)
Figure 7.2.1-4: Limit Stress Field (non-linear analysis)
field is in effect the resulting stress from an elastic unloading from the limit state back to zero
pressure. If no stress in the residual field violates the yield condition, i.e. if there is no equivalent
stress above the material (yield) parameter, then the shakedown load is equal to the limit load.
Where the residual stress field of a structure does exceed the yield condition, the shakedown limit
can be calculated easily from the stress plots because of the linearity.
The residual stress field of the
compensation analysis using Mises
criterion is shown in Figure 7.2.1-2.
Because the applied load is arbitrary
and the resulting stress fields are scaled
to the yield condition, the maximum
residual stress is then compared to the
maximum stress in the limit stress field,
Figure 7.2.1-3. The maximum residual
equivalent stress of 21.246 MPa is
smaller than the 24.971 MPa for the
limit stress field. The shakedown limit
is therefore the same as the calculated
limit load from the Mises condition,
given by scaling up the load by a factor
of material yield parameter to
maximum stress in the limit field
(255/24.971) = 10.21. With an applied
load of 10 MPa the shakedown limit is
102.1 MPa.
5. Check against GPD Using Non-linear Analysis
A check against GPD was also
performed for the same FE - model
using conventional non-linear
analysis. In this way, a direct
comparison may be made between the
two limit approaches. The FE -
geometry, mesh and boundary
conditions are the same as those used
in the elastic compensation analysis.
Material non-linearities were applied
corresponding to the material strength
parameter, 204 MPa and perfect
plasticity. A ramped load is applied
and the analysis runs until the applied
load is such that convergence can no
longer occur due to unrestrained
displacement Gross Plastic
Deformation. It is assumed that the
last converged solution is the limit
Analysis Details
Example 1.1
Page
7.6(S)
DBA
Design by Analysis
load. Here, the last converged solution was at a load of 87 MPa, using the same method as above
the allowable pressure according to the GPD-Check using Mises' criterion is
The result offers a small benefit to the allowable pressure calculated using elastic compensation,
however the analysis is more difficult. Figure 7.2.1-4 shows the Mises equivalent stress at the limit
load. Analysis time to calculate limit load using non-linear FE - analysis was 290 seconds.
6. Additional Comments
Additional analysis was completed to ascertain any effect on the results for different mesh density
and for lower order elements (4-node). Limit loads and shakedown were calculated using Mises'
condition, the results are summarised in Table 7.2.1-1.
Number of
Elements
Element Type Number of
Iterations
Lower Bound
Limit Load
Shakedown
Load
Processor
Time
119 4 node 8 56.9 98.5 40.7
339 4 node 8 58.8 101.9 107
119 8 node 8 57 98.75 70.2
339 8 node 8 58.95 102.1 190.3
Table 7.2.1-1. Limit and shakedown analysis summary
Both the lower bound limit load and the shakedown load were calculated using the same method as
described above. As can be seen from Table 7.2.1-1, essentially no difference is noted between the
results for the different element types. For this geometry, higher order elements offer no benefit
over the lower order elements. The geometry is simple and at the smaller mesh density the 4-node
elements fit the curvature well, therefore little change in the results would be expected. A
difference in the results can be noted between the two mesh densities. A slightly larger shakedown
and limit load result is obtained from the analysis using the higher mesh density, although small,
approximately 3%. In general, the results for both the limit load and shakedown calculations show
very little sensitivity to the element type and element density for this geometry.
As the geometry in this problem is outside the scope of DBF, the DBA calculations are a quick and
simple alternative for this simple problem.
The two elastic compensation methods used to calculate the lower bound limit load (direct from
Tresca's criterion or via a correction of Mises') show good correlation; the Mises corrected value is
slightly conservative as would be expected. However, carrying out the Mises elastic compensation
will give both the limit load and shakedown load in one analysis.
Good agreement was shown between the elastic compensation results and the non-linear analysis
results. Processing time is longer for the non-linear analysis however, due to the simplicity of the
model this time was also short.
MPa PS
p
GPD
8 . 62
2
3 87
max
= =

Analysis Details
Example 1.1
Page
7.7(S)
DBA
Design by Analysis
Analysis Type: GPD - Check and PD - Check
Direct Route using elasto-plastic FE calculations
FE-Software: ANSYS 5.4
Element Types:
4 node, 2 D axisymmetric solid PLANE42
Boundary Conditions: Zero vertical displacement in the nodes at the undisturbed end of
the shell.
Symmetry boundary conditions in the nodes in the centre of the flat
end.
Model and Mesh:
Member:
A&AB
Whole height of model:
751.2 mm
Number of elements:
1294
Results: Maximun allowable pressure according to the GRD-Check: PS
maxGPD
=60.87 Mpa
Shakesown limit pressure: PS
maxGPD
=101.45 Mpa
Analysis Details
Example 1.1
Page
7.8(S)
DBA
Design by Analysis
1. Elements, mesh fineness, boundary conditions
The model of the structure is shown on the preceeding page, a total number of 1294 4-node
axisymmetric solid elements, PLANE42 in ANSYS 5.4, was used. The linear shape function of
these elements is sufficient, since
the number of elements in the (linear-elastic) high stressed region is large,
the computation time in an analysis using nonlinear material properties is much larger
for elements with midside nodes and quadratic shape functions, although, close to the
limit load, the results are almost identical compared to those using elements with a linear
shape function,
there is no need to compute linear-elastic peak stresses very exactly, because the check
against PD can be carried out using the stress-concentration-free structure (according to
prEN 13445-3 Annex B.9.3.2) and the structures geometry is modelled exactly in the
example considered.
The boundary conditions applied to the model are symmetry ones in the nodes in the centre of the
plate (where the horizontal direction is perpendicular to the plane of symmetry) and constraining the
vertical degree of freedom in the nodes at the undisturbed end of the cylindrical shell to zero.
2. Determination of the maximum admissible pressure according to the GPD-Check
The partial safety factor
R
according to prEN-13445-3 Annex B, Table B.9-3 is 25 . 1 . Therefore,
the analysis using Trescas yield condition (delivered by an ANSYS distributor) was carried out
with a linear-elastic ideal-plastic material law, a design material strength parameter of MPa 204
(corresponding to a material strength parameter of MPa RM 255 = according to EN 10222-2) for
the shell and the plate, associated flow rule, and first order theory.
The elastic modulus used in all calculations is GPa E 212 = .
The analysis was carried out using the arc-length method, which showed faster convergence for the
considered structure than the Newton - Raphson method; since at the limit load the structure is fully
plastified in the shell and in the adjacent part of the plate, a maximum horizontal displacement of 10
mm at the upper face of the shell was used as termination criterion.
To restrict the computation time in a reasonable manner, the analysis was terminated after 17 hours
on a Compaq Professional Workstation 5000 with two Pentium Pro processors and 256 MB
RAM. The last convergent solution showed an internal pressure of MPa 05 . 73 this pressure was
used as limit pressure.
Figure 7.2.2-1 shows the horizontal displacement in the upper end of the shell versus the internal
pressure.
Analysis Details
Example 1.1
Page
7.9(S)
DBA
Design by Analysis
Figure 7.2.2-1
Figure 7.2.2-2 shows the distribution
of Tresca's equivalent stress at this
limit pressure. Because of the almost
full plastification in the shell there is a
small region where, due to numerical
effects, the equivalent stress exceeds
MPa 204 , but this has no effect on
further analyses.
The maximum absolute value of the
principal strains in the structure at this
limit pressure is 1.4 %, smaller than
5%, as required in the standard.
According prEN 13445-3 Annex B,
Table B.9-2, the partial safety factor
for pressure (without natural limit),
P
,
is 1.2.
Figure 7.2.2-2
Therefore, the (internal) allowable pressure according to GPD is given by
. 87 . 60
2 . 1
05 . 73
max
MPa PS
GPD
= =
A less time-consuming method to determine a limit pressure according to GPD is given by usage of
the limit pressure result from the check against PD (see chapter 2 of section 3 - Procedures). With
the partial safety factors 25 . 1 =
R
and 2 . 1 =
P
, the internal limit pressure according to GPD is, in
this approach,
MPa PS
R P
GPD
4 . 59
2
3
25 . 1 2 . 1
89 . 102
2
3 89 . 102
max
=

=

.
Analysis Details
Example 1.1
Page
7.10(S)
DBA
Design by Analysis
The maximum absolute value of the principal strains in the structure, at the limit pressure used here,
and for the design material strength parameter of the check against PD-check, is 4 %, smaller than
5%, as required in the standard.
3. Check against PD
The elasto-plastic FE analysis was carried out as stated in prEN 13445-3 Annex B, Sec. B.9.3.1,
using Mises yield condition and associated flow rule, a linear-elastic ideal-plastic constitutive law
with a material strength parameter of 255 MPa for shell and plate, and first order theory.
By defining and using load cases in
ANSYS, the superposition of stress
fields can be done easily. Therefore the
first load step of the analysis was
defined at a very low load level (5
MPa), so that there was linear-elastic
response of the structure. All other
linear-elastic stress fields can then be
determined easily by multiplication
with a suitable scale-up factor.
Again, the analysis was carried out
using the arc-length method; since at
the limit load the structure is fully
plastified in the shell and in the
adjacent part of the plate, a maximum
horizontal displacement of 10 mm at
the upper face of the shell was used as
termination criterion.
The computation time of the limit load
was 1 hour and 15 minutes on the
Compaq Professional Workstation
5000 with two Pentium Pro
processors and 256 MB RAM.
The termination criterion was fulfilled
for an internal pressure of 102.89 MPa
this pressure was used as limit
pressure. Figure 7.2.2-3 shows the
horizontal displacement in the
undisturbed shell versus the internal
pressure; according to this figure the
limit state is reached.
Figure 7.2.2-4 shows the elasto-plastic
Mises equivalent stress distribution at
the limit pressure of MPa 89 . 102 .
Figure 7.2.2-3
Figure 7.2.2-4
Analysis Details
Example 1.1
Page
7.11(S)
DBA
Design by Analysis
Figure 7.2.2-5 shows the linear-
elastic Mises equivalent stress
distribution for the limit pressure
the stress maximum is located
in the fillet.
Figure 7.2.2-5
Figure 7.2.2-6 shows the Mises
equivalent stress distribution of
the corrected residual stress field,
the scaling factor is given by
96 . 0 (see subsection 3.3.2.5 of
section 3 Procedures). The site
of the stress maximum is now
located at the outer surface of the
cylinder.
Figure 7.2.2-6
Analysis Details
Example 1.1
Page
7.12(S)
DBA
Design by Analysis
Figure 7.2.2-7 shows the Mises
equivalent stress distribution at
the lower bound shakedown
limit. The scaling factor is
given by 986 . 0 (see subsection
3.3.2.5 of section 3 Proce-
dures).
Figure 7.2.2-7
Thus, the shakedown limit pressure is given by
MPa PS
SD
45 . 101 89 . 102 986 . 0
max
= = .
Analysis Details
Example 1.1
Page
7.13(S)
DBA
Design by Analysis
Analysis Type:
Stress Categorization Route
FE-Software: BOSOR
Element Types: Axisymmetric shell elements.
Model and Mesh:
As shown in diagram Shell numbers in calculation model
Remark:
In this example the structure is very thick walled. As BOSOR operates with thin-walled elements
the pressure acting on element 3 and 4 has been reduced with the factor
m i
R R / = 250.4 / 301.2
Results:
Maximum admissible action according to the Stress Categorization Route:
With reduction factor (Calc NO 11E) - Internal pressure PS
max SC
= 69.7 MPa
Member:
TKS
Analysis Details
Example 1.1
Page
7.14(S)
DBA
Design by Analysis
The following figure shows the deformed model, the figures thereafter the distribution of stresses
membrane and membrane and membrane plus bending - in the surfaces of the various parts of
the model.
With the designation list in subsection 3.6 the various plots are self-explaining.
The plots are for a pressure of 50 MPa. The limiting part is the cylindrical shell, the general
membrane stress criterion is the governing one see the membrane stress distribution in shell 4
on the last page of this contribution.
J
O
B

N
O
1
1
E
9
9
-
0
8
-
2
5

1
0
.
5
4
.
2
6

P
O
S
T
B
O
S
O
R

1
.
0
4







*
G
E
O
M
E
T
R
Y

P
L
O
T
A
L
L

S
H
E
L
L
S

D
E
F
O
R
M
.

U
N
D
E
F
O
R
M
.
W
I
N
D
O
W


(
X
,
Y
)
:
M
I
N
:

0
M
A
X
:

4
.
2
6
E
+
0
2
M
I
N
:

-
5
.
2
2
E
+
0
1
M
A
X
:

1
.
2
5
E
+
0
3
Analysis Details
Example 1.1
Page
7.15(S)
DBA
Design by Analysis
Shell No. 1 Stresses


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0
1
0
2

-
.
7
5

-
.
5
0

-
.
2
5


.
0
0


.
2
5


.
5
0


.
7
5

1
.
0
0

1
.
2
5
1
0
2
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
0
9
.
0
5

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
8
.
0
3
E
+
0
1
M
A
X
:

1
.
3
2
E
+
0
2


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0
1
0
2


.
4
0


.
5
0


.
6
0


.
7
0


.
8
0


.
9
0

1
.
0
0

1
.
1
0

1
.
2
0

1
.
3
0
1
0
2
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
1
1
.
1
0

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

3
.
5
4
E
+
0
1
M
A
X
:

1
.
3
2
E
+
0
2


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0
1
0
2
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
0
8
.
1
5

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

1
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

2
.
6
1
E
+
0
1
M
A
X
:

2
.
6
1
E
+
0
1


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0
1
0
2

-
.
7
5

-
.
5
0

-
.
2
5


.
0
0


.
2
5


.
5
0


.
7
5

1
.
0
0

1
.
2
5
1
0
2
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
0
9
.
4
4

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
8
.
0
3
E
+
0
1
M
A
X
:

1
.
3
2
E
+
0
2
Analysis Details
Example 1.1
Page
7.16(S)
DBA
Design by Analysis
Shell No. 2 Stresses

1
.
5
0

1
.
7
5

2
.
0
0

2
.
2
5

2
.
5
0

2
.
7
5

3
.
0
0

3
.
2
5

3
.
5
0
1
0
2
-
1
.
0
0

-
.
5
0


.
0
0


.
5
0

1
.
0
0

1
.
5
0
1
0
2
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
1
3
.
4
0

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
1
.
2
3
E
+
0
2
M
A
X
:

1
.
7
5
E
+
0
2

1
.
5
0

1
.
7
5

2
.
0
0

2
.
2
5

2
.
5
0

2
.
7
5

3
.
0
0

3
.
2
5

3
.
5
0
1
0
2


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0

1
.
6
0
1
0
2
J
O
B

N
O
1
1
E
9
9
-
0
8
-
2
5

1
0
.
5
6
.
0
2

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

2
.
0
5
E
+
0
0
M
A
X
:

1
.
7
5
E
+
0
2

1
.
5
0

1
.
7
5

2
.
0
0

2
.
2
5

2
.
5
0

2
.
7
5

3
.
0
0

3
.
2
5

3
.
5
0
1
0
2
1
2
.
0
0
1
4
.
0
0
1
6
.
0
0
1
8
.
0
0
2
0
.
0
0
2
2
.
0
0
2
4
.
0
0
2
6
.
0
0
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
1
3
.
0
2

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

2
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

1
.
1
2
E
+
0
1
M
A
X
:

2
.
6
1
E
+
0
1

1
.
5
0

1
.
7
5

2
.
0
0

2
.
2
5

2
.
5
0

2
.
7
5

3
.
0
0

3
.
2
5

3
.
5
0
1
0
2
-
2
.
0
0


.
0
0

2
.
0
0

4
.
0
0

6
.
0
0

8
.
0
0
1
0
1
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
1
4
.
3
0

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
3
.
0
8
E
+
0
1
M
A
X
:

8
.
2
9
E
+
0
1
Analysis Details
Example 1.1
Page
7.17(S)
DBA
Design by Analysis
Shell No. 3 Stresses

3
.
6
0

3
.
8
0

4
.
0
0

4
.
2
0

4
.
4
0

4
.
6
0
1
0
2
-
1
.
0
0


.
0
0

1
.
0
0

2
.
0
0

3
.
0
0

4
.
0
0

5
.
0
0

6
.
0
0
1
0
1
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
1
7
.
0
5

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
1
.
3
6
E
+
0
1
M
A
X
:

6
.
9
4
E
+
0
1

3
.
6
0

3
.
8
0

4
.
0
0

4
.
2
0

4
.
4
0

4
.
6
0
1
0
2
2
0
.
0
0
3
0
.
0
0
4
0
.
0
0
5
0
.
0
0
6
0
.
0
0
7
0
.
0
0
8
0
.
0
0
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
1
8
.
3
2

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

1
.
9
8
E
+
0
1
M
A
X
:

8
.
1
8
E
+
0
1

3
.
6
0

3
.
8
0

4
.
0
0

4
.
2
0

4
.
4
0

4
.
6
0
1
0
2
4
0
.
0
0
4
5
.
0
0
5
0
.
0
0
5
5
.
0
0
6
0
.
0
0
6
5
.
0
0
7
0
.
0
0
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
1
5
.
2
1

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

3
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

3
.
7
2
E
+
0
1
M
A
X
:

7
.
0
6
E
+
0
1

3
.
6
0

3
.
8
0

4
.
0
0

4
.
2
0

4
.
4
0

4
.
6
0
1
0
2
2
0
.
0
0
3
0
.
0
0
4
0
.
0
0
5
0
.
0
0
6
0
.
0
0
7
0
.
0
0
8
0
.
0
0
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
1
7
.
5
7

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

1
.
9
8
E
+
0
1
M
A
X
:

8
.
1
8
E
+
0
1
Analysis Details
Example 1.1
Page
7.18(S)
DBA
Design by Analysis
Shell No. 4 Stresses


.
6
0


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0
1
0
3
1
0
.
0
0
2
0
.
0
0
3
0
.
0
0
4
0
.
0
0
5
0
.
0
0
6
0
.
0
0
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
2
0
.
3
0

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

4

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

9
.
0
8
E
+
0
0
M
A
X
:

6
.
5
2
E
+
0
1


.
6
0


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0
1
0
3


.
7
0


.
8
0


.
9
0

1
.
0
0

1
.
1
0

1
.
2
0
1
0
2
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
2
1
.
4
1

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

4

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

6
.
6
6
E
+
0
1
M
A
X
:

1
.
2
9
E
+
0
2


.
6
0


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0
1
0
3


.
7
0


.
8
0


.
9
0

1
.
0
0

1
.
1
0

1
.
2
0
1
0
2
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
1
9
.
4
7

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

4
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

6
.
9
9
E
+
0
1
M
A
X
:

1
.
2
7
E
+
0
2
M
e
m
.

1
2
2
M
P
a
(

e
q
)
P
m

f


.
6
0


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0
1
0
3


.
7
0


.
8
0


.
9
0

1
.
0
0

1
.
1
0

1
.
2
0
1
0
2
J
O
B

N
O
1
1
E
9
9
-
0
8
-
1
8

1
4
.
2
1
.
1
1

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

4

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

6
.
6
6
E
+
0
1
M
A
X
:

1
.
2
9
E
+
0
2
DBA
Design by Analysis
Analysis Summary
Example 1.1 / F - Check
Page
7.19 (A)
Analysis Type:
F - Check
Member:
A&AB
FE Software: ANSYS 5.4
Results:
Fatigue life = infinity
Height of the model : 1000 mm
Element Types:
4 node, 2 D axisymmetric solid PLANE42
Boundary Conditions: Zero vertical displacement in the nodes at the undisturbed
end of the shell.
Symmetry boundary conditions in the nodes in the centre of
the flat end.
Model and Mesh:
Whole height of model:
751.2 mm
Number of elements:
1294
1
,
Data
t
max
= 20 C
t
min
= 20C
t* = 0,75 t
max
+ 0,25 t
min
= 20 C
R
m
= 460 MPa
R
p0,2/t*
= 255 MPa
Effective stress concentration factorK
eff
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors. K
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then tensors are added and the new stress range is calculated.
R
z
= 50 m (table 18-8)
e
n
= 101,6 mm

D
= 310.8 MPa (table 18-10 for N 2
.
10
6
cycles)
N = 2.10
6
(for the first iteration)
Stresses Critical point: Point of maximum equivalent stress (Tresca)

eq,t
(total or notch equivalent stress range) = 75.55 MPa for p=15.3 MPa

struc
(structural equivalent stress range obtained by quadratic extrapolation from the shell side into the critical point) = 116.28 MPa
0 . 1
D

struc

t
K 0,5 1
1)
t
(K 1,5
1
eff
K
+

+
1
1
]
1

Theoretical elastic stress concentration factor K


t
K
t
=
eq,t
/
struc
= 1.0
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..

total
= k
e
.
eq,t
= .. MPa
Else
total
=
eq,t
= 75.55 MPa
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
l eq,

0,4
0,5
0,7
k

k

= ..

total
= k

.
eq,t
= .. MPa
Else
total
=
eq,t
= .. MPa
) stress equivalent notch (mean MPa 37.78
eq

18-10-6-2 Temperature correction factor f
t*
For t* > 100 C
f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
18-11-1-1 Surface finish correction factor f
s
f
s
= F
s
[0,1ln(N)-0,465]
if N < 2.10
6
, f
s
= F
s
if N 2.10
6
,
with F
s
= 1- 0,056 [ln (R
z
)]
0,64
[ln(R
m
)] +0,289 [ln (R
z
)]
0,53
F
s
= 0,7735
f
s
= 0,7735

eq,struc
=
t
K
total
= 75.55 MPa (for usage in 18-11-3)

eqmax
= 75.55 MPa (maximum notch equivalent stress)
Analysis Details
Example 1.2 / Rev.0
Page
7.20 (A)
Analysis Details
Example 1.1 / F - Check
DBA
Design by Analysis

struc
=
eq, t
if
struc
>
eq, t,

struc
= 75.55 MPa
2
18-11-2 Thickness correction factor f
e
e
n
25 mm 25 mm < e
n
150 mm e
n
> 150 mm
F
e
= 1 Fe = (25/e
n
)
0.182
= 0.7748 F
e
= 0.7217
For N 2.10
6
: f
e
= F
e
For N < 2.10
6
:
f
e
= 1 f
e
= F
e
[0,1ln(N)-0,465]
f
e
= 0,7217
[0,1ln(N)-0,465]
f
e
=

0.7748 f
e
= ..
18-11-1-3 Mean stress correction factor f
m
If
struc
< 2 R
p0,2/t*
and
eqmax
< R
p0,2/t*
If
eq
> 0 then
r eq,
= R
p0,2/t*
-
2
t eq,
If
eq
< 0 then
r eq,
=
2
t eq,
- R
p0,2/t*
and
eq
=
r eq,
= ..
For N 2 10
6
cycles
See figure 18-14
f
m
= 1
If
struc
>2 R
p0,2/t*
18-11-2-1 Overall correction factor f
u
f
u
=
eff
K
* t
.f
m
.f
e
.f
s
f
= 0.5993
equ, struc
/ f
u
= 126.1 MPa
18-11-3 Allowable number of cycles N
N =
2
5 , 11
m
0,63R
u
f
struc eq,

4
10 4.6
1
1
1
1
1
]
1

if N 2
.
10
6
, N = if
eq, struc
/ f
u

D
if R
p0,2/t*

eq

( ) M 1 2
R

+
then
f
m
=
( )
5 , 0
eq
2
M 1
M 2 M
- 1
R
1
1
]
1

,
_

+
+
= ..
N =
If
struc
< 2
.
R
p0,2/t*
and
eqmax
> R
p0,2/t*
if
( ) M 1 2
R

+

eq
R
p0,2/t*
then
f
m
=

,
_

+
+
R

eq
2
3
M
M 1
3 M 1
= .
For N 2 10
6
cycles
M = 0,00035 R
m
0,1= 0,061
f
m
= 1
Page
7.21 (A)
Analysis Details
Example 1.1 / F - Check
DBA
Design by Analysis
Analysis Details
Example 1.2
Page
7.22 (S)
DBA
Design by Analysis
Analysis Type GPD-Check and PD-Check
Direct Route using Elastic Compensation
FE-Software: ANSYS 5.4
Element Types: 8 node, 2-D structural axisymmetric solid.
Boundary Conditions: No vertical displacement in the undisturbed end of the shell remote
from the flat end.
Axisymmetry from elements
Model and Mesh:
Results:
Maximum internal pressure according to the GPD-Check: PS
max GPD
= 4.5 MPa
Check against PD: Shakedown limit pressure PS
max SD
= 7.25 MPa
Member:
Strathclyde
Number of elements 384
Height of model 500 mm
Analysis Details
Example 1.2
Page
7.23 (S)
DBA
Design by Analysis
Figure 7.3.1-1 Limit Stress Field (Tresca)
1 Finite Element Mesh
The geometry model was constructed according to the problem specification for example 1.2.
Element divisions for the finite element mesh were defined parametrically to allow modification of
the element density. Analysis was carried out for various mesh densities to note any alterations this
may have on the results. Finite element models were created using linear 4-node 2-D axisymmetric
solids, the analysis was repeated with higher order 8-node structural 2-D axisymmetric solids, any
alterations this may have on the results could also be noted. The allowable pressure according to
GPD and the shakedown pressure according to PD were calculated from the model giving the
greatest lower bounds. Here, the fine density mesh with 4-node higher order elements gave the
highest lower bounds.
Boundary conditions applied to the model are axisymmetry, applied via a key option when defining
the element type in the FE-software (axisymmetry around the vertical axis Y). The nodes at the
undisturbed end of the cylindrical shell have their vertical degree of freedom constrained to ensure
that plane sections remain plane.
2 Material properties
Material strength parameter RM = 255 MPa, modulus of elasticity E = 212 GPa.
3 Determination of the maximum admissible pressure according to the GPD-Check
Using the application rule in prEN-13445-3 Annex B.9.2.2, to check against GPD the principle is
fulfilled when for any load case the combination of the design actions do not exceed the design
resistance. This may be shown by calculating the limit load. The limit load has to be determined
using Trescas yield condition and associated flow rule. As elastic compensation is based upon a
series of elastic equilibrium stress fields, it is a relatively simple procedure to derive a lower bound
limit load direct from the Tresca yield model.
From prEN-13445-3 Annex B, Table B.9-3 the partial safety factor
R
on the resistance is 1.25.
Therefore, the design or material strength parameter is given by RM/
R
= 204 MPa. The analysis
was carried out using the elastic compensation method conforming to the direct route rules for GPD
in Annex B: - linear elastic ideal
plastic material law, Trescas yield
condition and associated flow rule
and first order theory.
In each elastic compensation
iteration an equilibrium stress field
is produced where the elastic
modulus of each subsequent
iteration is defined by the previous
elastic solution. In this way
regions of the FE-model may be
systematically stiffened or
weakened depending upon the
stress magnitudes in the previous
solution. The result is that every
equilibrium stress field is a lower
bound of the limit load, the stress
Analysis Details
Example 1.2
Page
7.24 (S)
DBA
Design by Analysis
Figure 7.3.1-2 Residual Stress Field (Mises)
field giving the greatest limit load is taken as the closest to the actual limit state and defines the
limit load in the analysis. Convergence of the equilibrium stress field will occur typically between
8 and 15 iterations, i.e. no further benefit in the limit load will be noted with subsequent iterations.
The total computing time to run the analysis on a 300 MHz Pentium two processor with 128 Mb
RAM Windows NT workstation was 106 seconds. The stress field was shown to converge after
twelve iterations giving a lower bound on the limit pressure of 5.4 MPa. Figure 7.3.1-1 shows the
limit stress (intensity) field according to the Tresca condition. The limit pressure is given by
scaling the limit stress field so that the stress anywhere in the model does not exceed the design
material strength, 204 MPa, i.e. the applied pressure (10 MPa) is scaled by the factor (204/378.7)
0.54. According to prEN-13445-3 Annex B, Table B.9-2 for pressure loads (without natural limit)
the partial safety factor
p
is 1.2. Thus, the internal pressure limit according to failure by GPD is
It is also possible to determine a limit pressure from the check against PD. In this case, the elastic
compensation is based on Von Mises yield criterion. The partial safety factor on the resistance
R
is not applied for the PD-check. However, as the analysis is wholly elastic it is possible to scale the
stress fields at any time (similarly as was done above). The maximum ratio of Mises' equivalent
stress to Tresca's equivalent stress for the same load is 2/3. Therefore applying a factor of 3/2 to
the yield stress in the Mises analysis (or to the limit load, as the analysis is elastic) will always lead
to a conservative result. From the Mises analysis the limit load was found to be 7.5 MPa and with
the partial safety factors
R
= 1.25 and
p
= 1.2, the internal pressure limit according to GPD can be
found as
4 Check against PD
In the check for progressive plastic
deformation, the principle in prEN-
13445-3, B.9.3.1 is fulfilled if the
structure can be shown to shake
down. When a structure has been
shown to shake down, the failure
modes of progressive plastic
deformation and alternating plasticity
can not occur.
In elastic compensation the load at
which the structure will shakedown
can be calculated simply. Based on
Melans shakedown theorem, the
self-equilibrating residual stress field
that would result after a loading cycle
can be calculated by subtracting the
linear-elastic stress field at the limit
pressure from the limit stress field.
The residual stress field is in effect
MPa PS
GPD
5 . 4
2 . 1
4 . 5
max
= =
MPa PS
R p
GPD
33 . 4
2
3 5 . 7
max
=

=

Analysis Details
Example 1.2
Page
7.25 (S)
DBA
Design by Analysis
Figure 7.3.1-3 Limit Stress Field (Mises)
Figure 7.3.1-4 Limit Stress Field (non-linear analysis)
the resulting stress from an elastic unloading from the limit state back to zero pressure. If no stress
in the residual field violates the yield condition, i.e. if there is no stress above the design material
(yield) strength, then the shakedown load is equal to the limit load. Where the residual stress field
of a structure does exceed the yield condition, the shakedown limit can be calculated easily from the
stress plots because of the linearity.
The residual stress field from the elastic
compensation analysis using Mises'
criterion is shown in Figure 7.3.1-2.
Because the applied load is arbitrary
and the resulting stress fields are scaled
to the yield condition, the maximum
residual stress is compared with the
maximum stress in the limit stress field,
Figure 7.3.1-3. The maximum residual
stress of 249.5 MPa is lower than that
of 339.5 MPa for the limit stress field.
The shakedown limit is therefore the
same as the calculated limit load from
the Mises condition, given by scaling
up the load by a factor of material yield
parameter to maximum stress in the
limit field (255/339.5) 0.75. With an
applied load of 10 MPa the shakedown
limit is 7.5 MPa.
5 Check on GPD Using Non-linear Analysis
A check against GPD was also
performed for the same FE-model
using conventional non-linear
analysis. In this way, a direct
comparison may be made between the
two limit approaches. The FE-
geometry, mesh and boundary
conditions are the same as those used
in the elastic compensation analysis.
Material non-linearities were applied
corresponding to the design material
strength parameter, 204 MPa, and
perfect plasticity. A ramped load is
applied and the analysis runs until the
applied load is such that convergence
can no longer occur due to
unrestrained displacement Gross
Plastic Deformation. It is assumed
that the last converged solution is the
limit load. Here, the last converged
Analysis Details
Example 1.2
Page
7.26 (S)
DBA
Design by Analysis
solution was at a load of 7.9 MPa, using the same method as above the allowable pressure
according to GPD using Mises' criterion is
The result offers a considerable benefit to the allowable pressure calculated using elastic
compensation. Figure 7.3.1-4 shows the Mises equivalent stress at the limit. Analysis time to
calculate limit load using non-linear FE-analysis was 360 seconds.
6 Additional Comments
Additional analysis was completed to ascertain any effect on the results for different mesh density
and for higher order elements (8-node). Limit loads and shakedown were calculated using Mises
condition, the results are summarised in Table 7.3.1-1.
Number of
Elements
Element Type Number of
Iterations
Lower Bound
Limit Load
Shakedown
Load
Processor
Time
151 4 node 8 4.7 6.8 50
151 8 node 8 4.4 6.4 87.3
267 4 node 8 5.0 7.25 85.5
267 8 node 8 4.8 6.9 152.1
383 4 node 12 5.2 7.5 184
383 8 node 12 5.1 7.4 204
Table 7.3.1-1. Limit and shakedown analysis summary
Both the lower bound limit load and the shakedown load were calculated using the same method as
described above. As can be seen from Table 7.3.1-1, the choice of element type and mesh density
can make a considerable difference to the results (maximum difference of 14%). The greatest
difference is noted over the various mesh densities, with the highest density mesh giving the
greatest lower bound on the limit pressure and shakedown pressure. It can be noted that the higher
order elements give a lower limit pressure than the lower order elements, although the difference is
less significant than that due to mesh density. As the element density increases the difference in
limit pressure and shakedown pressure for the different element type is reduced.
The utilisation of elastic compensation in this DBA calculation has given a small increase in the
allowable pressure over that given by the DBF calculation.
The two methods used to calculate the lower bound limit load (direct from Tresca's criterion or
correction of Mises') show good correlation; the Mises corrected value is slightly conservative as
would be expected. However, carrying out the Mises elastic compensation will give both the limit
load and shakedown load in one analysis.
Non-linear calculations offer a considerable benefit in terms of higher allowable pressure than the
pressures calculated by DBF and elastic compensation in this example.
MPa PS
p
GPD
7 . 5
2
3 9 . 7
max
= =

Analysis Details
Example 1.2
Page
7.27 (A)
DBA
Design by Analysis
Analysis Type:
Direct Route using elasto-plastic FE calculations
FE-Software: ANSYS 5.4
Element Types:
4 node, 2 D axisymmetric solid PLANE42
Boundary Conditions: Zero vertical displacement in the nodes at the unsdisturbed end of
the shell;
Symmetry boundary conditions in the nodes in the centre of the flat
end.
Model and Mesh:
Results: Maximum allowable pressure according to GPD: MPa PS
GPD
7 . 5
max
=
Shakedown limit pressure: MPa PS
SD
9 . 7
max
=
Member:
A&AB
Whole height of model:
476,2 mm
Number of elements
2640
Analysis Details
Example 1.2
Page
7.28 (A)
DBA
Design by Analysis
1. Elements, mesh fineness, boundary conditions
The model of the structure is shown on page 1, a total number of 2640 4-node axisymmetric solid
elements PLANE42 in ANSYS 5.4 - was used. The linear shape function of these elements is
sufficient, since
the number of elements in the (linear-elastic) high stressed region is large,
the computation time in an analysis using nonlinear material properties is much higher for
elements with midside nodes and quadratic shape functions, although, close to the limit load, the
results are almost identical compared to those using elements with a linear shape function,
there is no need to compute linear-elastic peak stresses very exactly, because the check against
PD can be carried out using the stress-concentration-free structure (according to pr EN 13445-3
Annex B.9.3.2) and the structures geometry is modelled exactly in the example considered.
The boundary conditions applied in the model are symmetry ones in the nodes in the centre of the
plate (where the horizontal direction is perpendicular to the plane of symmetry) and constraining the
vertical degree of freedom in the nodes at the undisturbed end of the cylindrical shell to zero.
2. Determination of the maximum admissible pressure according to GPD
Since the subroutine using Trescas yield condition showed bad convergence, the results from the
check against PD (see subsection 3.2 of section 3 Procedures) have been used. The partial safety
factor
R
according to prEN 13445-3 Annex B, Table B.9-3 is 25 . 1 and the partial safety factor for
pressure (without natural limit) according to prEN 13445-3 Annex B, Table B.9-2
P
is 2 . 1 .
Therefore, the (internal) allowable pressure according to the GPD-check is given by
. 72 . 5
2
3
25 . 1 2 . 1
904 . 9
2
3 904 . 9
max
MPa PS
R P
GPD
=

=

The maximum principal strain, at the limit pressure used here and for the design material strength
parameter of the check against PD, is about 3 %, the condition in prEN 13445-3 Annex B.9.2.2 is
fulfilled.
3. Check against PD
The elasto-plastic FE analysis was carried out as stated in prEN 13445-3 Annex B, Sec. B.9.3.1,
using Mises yield condition and associated flow rule, an linear-elastic ideal-plastic law with a
design material strength parameter of 255 MPa for shell and plate (according to EN 10222-2 and for
the structure turned from one forged part), and first order theory. The elastic modulus used in the
calculations is E = 212 GPa.
By defining and using load cases in ANSYS, the superposition of stress fields can be done easily.
Therefore an early load step of the analysis was defined at a very low load level (2 MPa), so that
there was linear-elastic response of the structure. All other linear-elastic stress fields can then be
determined easily by multiplication with a suitable scale-up factor.
Analysis Details
Example 1.2
Page
7.29 (A)
DBA
Design by Analysis
The analysis was carried out using the arc-length method; since at the limit load the structure is
fully plastified in the plate, a maximum vertical displacement at the middle of the plate of 10 mm
was used as termination criterion.
The computation time of the limit load was 1 hour and 15 minutes on the Compaq Professional
Workstation 5000 with two Pentium Pro processors and 256 MB RAM.
The termination criterion was fulfilled
at an internal pressure of 9.904 MPa
this pressure was used as limit
pressure. Figure 7.3.2-1 shows the
vertical displacement at the middle of
the plate versus the internal pressure;
according to this figure the limit state
is reached.
Figure 7.3.2-1
Figure 7.3.2-2 shows the elasto-
plastic Mises equivalent stress
distribution at the limit pressure
of MPa 9 . 9 . Here, because of
the almost full plastification in
the plate there are small regions
near the mid-plane of the plate,
where, due to numerical effects,
the equivalent stress exceeds
255 MPa, but this has no effect
on further analyses.
Figure 7.3.2-2
Analysis Details
Example 1.2
Page
7.30 (A)
DBA
Design by Analysis
Figure 7.3.2-3 shows the linear-
elastic Mises equivalent stress
distribution for the limit pressure
the stress maximum is located in
the fillet.
F
Figure 7.3.2-3
Figure 7.3.2-4 shows the Mises
equivalent stress distribution of the
corrected residual stress field. The
used scaling factor is 66 . 0 (see
subsection 3.3.2.5 of section 3 -
Procedures), the maximum stress is
located again in the fillet.
Figure 7.3.2-4
Analysis Details
Example 1.2
Page
7.31 (A)
DBA
Design by Analysis
Figure 7.3.2-5 shows the Mises
equivalent stress distribution at the
lower bound shakedown limit. The
scaling factor is given by 802 . 0
(see subsection 3.3.2.5 of section 3 -
Procedures). The shakedown limit
pressure is given by
MPa PS
SD
943 . 7 904 . 9 802 . 0
max
= =
F
Figure 7.3.2-5
Analysis Details
Example 1.2
Page
7.32 (T)
DBA
Design by Analysis
Analysis Type:
Stress Categorization Route
FE-Software: BOSOR
Element Types: Axisymmetric shell elements
Model and Mesh:
As shown in diagram Shell numbers in Calculation Model
Results:
Maximum admissible action according to the Stress Categorization Route:
Internal pressure PS
max SC
= 5.6 MPa
Member:
TKS
Analysis Details
Example 1.2
Page
7.33 (T)
DBA
Design by Analysis
The following figure shows the deformed model.
The figure following the next show the distributions of stresses membrane and membrane plus
bending in the surfaces of the various parts of the model.
With the designation list in subsection 3.6 the plots are self-explaining.
The calculation pressure used is 4.2 MPa.
The limiting part is the flat end, and the critical point is the center, where the primary membrane
plus bending stress is governing see the 4
th
figure after the next.
The allowable design stress used is 170 MPa. The membrane plus bending stress at the plate to
cylinder junction has been classified as secondary see the classification table of Annex C of
prEN 13445-3.
J
O
B

N
O
1
2
9
9
-
0
8
-
2
5

1
0
.
5
9
.
0
5

P
O
S
T
B
O
S
O
R

1
.
0
4







*
G
E
O
M
E
T
R
Y

P
L
O
T
A
L
L

S
H
E
L
L
S

D
E
F
O
R
M
.

U
N
D
E
F
O
R
M
.
W
I
N
D
O
W


(
X
,
Y
)
:
M
I
N
:

-
1
.
1
6
E
-
1
9
M
A
X
:

3
.
0
2
E
+
0
2
M
I
N
:

-
4
.
7
3
E
+
0
1
M
A
X
:

6
.
0
0
E
+
0
2
Analysis Details
Example 1.2
Page
7.34 (T)
DBA
Design by Analysis
Shell No. 1 Stresses


.
5
0

1
.
0
0

1
.
5
0

2
.
0
0

2
.
5
0
1
0
2
-
2
.
0
0
-
1
.
5
0
-
1
.
0
0

-
.
5
0


.
0
0


.
5
0

1
.
0
0

1
.
5
0

2
.
0
0
1
0
2
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
3
4
.
5
2

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
2
.
1
7
E
+
0
2
M
A
X
:

2
.
4
7
E
+
0
2


.
5
0

1
.
0
0

1
.
5
0

2
.
0
0

2
.
5
0
1
0
2


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0

1
.
6
0

1
.
8
0

2
.
0
0

2
.
2
0

2
.
4
0
1
0
2
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
3
6
.
5
1

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

6
.
3
7
E
+
0
1
M
A
X
:

2
.
4
7
E
+
0
2
1
9
1
M
P
a

(

e
q
)
P


1
.
5
f


.
5
0

1
.
0
0

1
.
5
0

2
.
0
0

2
.
5
0
1
0
2

8
.
0
0

9
.
0
0
1
0
.
0
0
1
1
.
0
0
1
2
.
0
0
1
3
.
0
0
1
4
.
0
0
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
3
4
.
1
8

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

1
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

7
.
6
0
E
+
0
0
M
A
X
:

1
.
4
8
E
+
0
1


.
5
0

1
.
0
0

1
.
5
0

2
.
0
0

2
.
5
0
1
0
2
-
1
.
5
0
-
1
.
0
0

-
.
5
0


.
0
0


.
5
0

1
.
0
0

1
.
5
0
1
0
2
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
3
5
.
4
5

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
1
.
6
2
E
+
0
2
M
A
X
:

1
.
9
1
E
+
0
2
Analysis Details
Example 1.2
Page
7.35 (T)
DBA
Design by Analysis
Shell No. 2 Stresses

2
.
6
0

2
.
7
0

2
.
8
0

2
.
9
0

3
.
0
0
1
0
2
-
2
.
0
0
-
1
.
0
0


.
0
0

1
.
0
0

2
.
0
0

3
.
0
0
1
0
1
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
3
8
.
3
0

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
2
.
5
2
E
+
0
1
M
A
X
:

3
.
3
8
E
+
0
1

2
.
6
0

2
.
7
0

2
.
8
0

2
.
9
0

3
.
0
0
1
0
2
1
0
.
0
0
1
5
.
0
0
2
0
.
0
0
2
5
.
0
0
3
0
.
0
0
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
3
9
.
5
3

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

5
.
9
3
E
+
0
0
M
A
X
:

3
.
3
8
E
+
0
1

2
.
6
0

2
.
7
0

2
.
8
0

2
.
9
0

3
.
0
0
1
0
2

4
.
5
0

5
.
0
0

5
.
5
0

6
.
0
0

6
.
5
0

7
.
0
0
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
3
7
.
5
6

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

2
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

4
.
3
6
E
+
0
0
M
A
X
:

7
.
0
7
E
+
0
0

2
.
6
0

2
.
7
0

2
.
8
0

2
.
9
0

3
.
0
0
1
0
2

-
.
2
0


.
0
0


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0
1
0
1
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
3
9
.
2
4

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
2
.
3
5
E
+
0
0
M
A
X
:

1
.
5
7
E
+
0
1
Analysis Details
Example 1.2
Page
7.36 (T)
DBA
Design by Analysis
Shell No. 3 Stresses

4
.
0
0

5
.
0
0

6
.
0
0

7
.
0
0

8
.
0
0
1
0
2
-
1
.
5
0
-
1
.
0
0

-
.
5
0


.
0
0


.
5
0

1
.
0
0

1
.
5
0

2
.
0
0

2
.
5
0
1
0
1
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
4
1
.
1
1

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
1
.
7
3
E
+
0
1
M
A
X
:

2
.
5
9
E
+
0
1

4
.
0
0

5
.
0
0

6
.
0
0

7
.
0
0

8
.
0
0
1
0
2

5
.
0
0

7
.
5
0
1
0
.
0
0
1
2
.
5
0
1
5
.
0
0
1
7
.
5
0
2
0
.
0
0
2
2
.
5
0
2
5
.
0
0
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
4
4
.
2
9

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

3
.
1
4
E
+
0
0
M
A
X
:

2
.
5
9
E
+
0
1

4
.
0
0

5
.
0
0

6
.
0
0

7
.
0
0

8
.
0
0
1
0
2

5
.
0
0

6
.
0
0

7
.
0
0

8
.
0
0

9
.
0
0
1
0
.
0
0
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
4
0
.
2
9

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

3
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

4
.
3
0
E
+
0
0
M
A
X
:

1
.
0
9
E
+
0
1

4
.
0
0

5
.
0
0

6
.
0
0

7
.
0
0

8
.
0
0
1
0
2

-
.
2
0


.
0
0


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0
1
0
1
J
O
B

N
O
1
2
9
9
-
0
8
-
2
0

1
4
.
4
4
.
0
2

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
2
.
1
6
E
+
0
0
M
A
X
:

1
.
0
9
E
+
0
1
DBA
Design by Analysis
Analysis Details
Example 1.2 / F-Check
Page
7.37 (C)
Analysis Type:
F-Check
Member:
CETIM
FE Software: ABAQUS/Standard version 5.8.1
Element Types:
Quadratic axisymmetric 8 nodes elements (CAX8).
1161 nodes and 344 elements
Boundary Conditions:
The nodes at the high of the cylindrical part are locked in the vertical
direction.
Model and Mesh:
Results:
Fatigue life = 375200 cycles
Height of the model : 1000 mm
2
Data
t
max
= 20 C
t
min
= 20C
t* = 0,75 t
max
+ 0,25 t
min
= 20 C
R
m
= 460 MPa
R
p0,2/t*
= 255 MPa
Effective stress concentration factorK
eff
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors. K
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
R
z
= 50 m (table 18-8)
e
n
= 25,8 mm

D
= 310.8 MPa (table 18-10 for N 2 10
6
cycles)
N = 2
.
10
6
(for the first iteration)

R
= 353.4 MPa (allowable stress range for N cycles at the 10
th
iteration)
Stresses

eq,t
(total or notch equivalent stress range) = 271.6 MPa

struc
(structural equivalent stress range) =
eq,l
(linearised equivalent stress range) = 230.2 MPa
1877 . 1
D

struc

t
K 0,5 1
1)
t
(K 1,5
1
eff
K
+

+
1
1
]
1

Theoretical elastic stress concentration factor K


t
K
t
=
eq,t
/
struc
= 1.1798
18.8 Plasticity correction factor k
e
mechanical loading
If
eq,l
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
l eq,

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..

total
= k
e
.
eq,t
= .. MPa
Else
total
=
eq,t
= 271.66 MPa
Thermal loading
If
eq,l
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
l eq,

0,4
0,5
0,7
k

k

= ..

total
= k

.
eq,t
= .. MPa
Else
total
=
eq,t
= .. MPa
) stress equivalent notch (mean MPa 135.8
eq

18-10-6-2 Temperature correction factor f
t*
For t* > 100 C
f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
18-11-1-1 Surface finish correction factor f
s
f
s
= F
s
[0,1ln(N)-0,465]
with F
s
= 1- 0,056 [ln (R
z
)]
0,64
[ln(R
m
)] +0,289 [ln (R
z
)]
0,53
F
s
= 0.7735
f
s
= 0.8104

eq,struc
=
t
K
total
= 228.7 MPa (for using in 18-11-3)

eqmax
= 271.6 MPa (maximum notch equivalent stress)
Analysis Details
Example 1.2 / Rev.0
Page
7.38 (C)
Analysis Details
Example 1.2 /F-Check
DBA
Design by Analysis
3
18-11-1-2 Thickness correction factor f
e
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
e
= F
e
[0,1ln(N)-0,465]
f
e
= 0,7217
[0,1ln(N)-0,465]
with Fe = (25/en)
0.182
= 0.9943
f
e
= 1 f
e
=

0.9953 f
e
= ..
18-11-1-3 Mean stress correction factor f
m
If
struc
< 2 R
p0,2/t*
and
eqmax
< R
p0,2/t*
If
eq
> 0 then
r eq,
= R
p0,2/t*
-
2
t eq,
If
eq
< 0 then
r eq,
=
2
t eq,
- R
p0,2/t*
and
eq
=
r eq,
= 119.2 MPa
For N 2
.
10
6
cycles
See figure 18-14
f
m
= ..
If
struc
>2 R
p0,2/t*
18-11-2-1 Overall correction factor f
u
f
u
=
eff
K
* t
.f
m
.f
e
.f
s
f
= 0.6514
18-11-3 Allowable number of cycles N
N =
2
5 , 11
m
0,63R
u
f
struc eq,

4
10 4.6
1
1
1
1
1
]
1

if N 2
.
10
6
cycles
if R
p0,2/t*

eq

( ) M 1 2
R

+
then
f
m
=
( )
5 , 0
eq
2
M 1
M 2 M
- 1
R
1
1
]
1

,
_

+
+
= 0.9592
N =
10
u
f
struc eq,

92
m
R 2,7
1
1
1
1
1
]
1

+
if 2
.
10
6
N 10
8
cycles
N = 375200
N is obtained by iterations. If the value of N oscillates, a difference equal to 1 % between two iterations is
acceptable. If the values decrease monotonously, the difference must be less than 0,001 % between two
iterations.
If
struc
< 2 R
p0,2/t*
and
eqmax
> R
p0,2/t*
if
( ) M 1 2
R

+

eq
R
p0,2/t*
then
f
m
=

,
_

+
+
R

eq
2
3
M
M 1
3 M 1
= .
For N 2
.
10
6
cycles
M = 0,00035 R
m
0,1= 0,061
f
m
= 1
Page
7.39(C)
Analysis Details
Example 1.2 / F-Check
DBA
Design by Analysis
Analysis Details
Example 1.3
Page
7.40 (S)
DBA
Design by Analysis
Analysis Type: GPD- and PD-Check
Direct Route using Elastic Compensation
FE-Software: ANSYS 5.4
Element Types: 8 node, 2-D structural axisymmetric solid.
Boundary Conditions: No vertical displacement in the undisturbed end of the shell remote
from the flat end.
Axisymmetry from elements
Model and Mesh:
Results:
Maximum internal pressure according to the GPD-Check: PS
max GPD
= 10.8 MPa
Check against PD: Shakedown limit pressure PS
max SD
= 13.3 Mpa
Member:
Strathclyde
Number of elements 1039
Total height 500 mm
Analysis Details
Example 1.3
Page
7.41 (S)
DBA
Design by Analysis
Figure 7.4.1-1: Limit Stress Field (Tresca)
1. Finite Element Mesh
Finite element models were created using 1040 linear 4-node 2-D axisymmetric solids. The
allowable pressure according to GPD and the shakedown pressure according to PD were calculated
using elastic compensation procedures and the rules prEN-13445-3 Annex B.
Boundary conditions applied in the model are axisymmetry, applied via a key option when defining
the element type in the FE-software (axisymmetry around the vertical axis Y). The nodes at the
undisturbed end of the cylindrical shell had their vertical degree of freedom constrained to zero to
ensure that plane sections remain plane.
2. Material properties
Material strength parameter MPa RM 255 = (cylindrical shell), MPa RM 245 = (flat end),
modulus of elasticity GPa E 212 = (shell and flat end).
3. Determination of the maximum admissible pressure according to the GPD-Check
Using the application rule in prEN-13445-3 Annex B.9.2.2, to check against GPD the principle is
fulfilled when for any load case the combination of the design actions does not exceed the design
resistance. This may be shown by calculating the limit load. The limit load has to be determined
using Trescas yield condition and associated flow rule. As elastic compensation is based upon a
series of elastic equilibrium stress fields, it is a relatively simple procedure to derive a lower bound
limit load direct from the Tresca yield model.
From prEN-13445-3 Annex B, Table B.9-3 the partial safety factor,
R
on the resistance is 1.25.
Therefore, the design material strength parameters given by RM/
R
are 204 MPa for the shell and
196 MPa for the plate. The analysis was carried out using the elastic compensation method
conforming to the direct route rules
for GPD in Annex B: linear elastic
ideal plastic material law, Trescas
yield condition and associated flow
rule and first order theory.
In each elastic compensation
iteration an equilibrium stress field
is produced where the elastic
modulus of each subsequent
iteration is defined by the previous
elastic solution. In this way
regions of the FE-model may be
systematically stiffened or
weakened depending upon the
stress magnitudes in the previous
solution. The result is that every
equilibrium stress field is a lower
bound of the limit load, the stress field giving the greatest limit load is taken as the closest to the
actual limit state and defines the limit load in the analysis. In problems such as this, where there are
materials with different properties the modulus modification has a modified procedure that takes
account of the different material properties. This modified method calculates the limit pressure for
Analysis Details
Example 1.3
Page
7.42 (S)
DBA
Design by Analysis
each component with a different material, allowing the component giving the lowest limit load to
define the limit for the whole model. Convergence of the equilibrium stress field will occur
typically between 8 and 15 iterations, i.e. no further benefit in the limit load will be noted with
subsequent iterations. The total computing time to run the analysis on a 300 MHz Pentium two
processor with 128 Mb RAM Windows NT workstation was 180 seconds. Figure 7.4.1-1 shows
the limit stress (intensity) field according to the Tresca condition. The maximum stresses in the
plate and shell are 152.0 MPa and 152.3 MPa respectively, and the limit pressure for each
component is given by using the linear proportionality:
P
ap
is the applied load, R
d
the design material strength, and
max
the maximum redistributed stress
for the component with design resistance R
d
. The limit loads for the plate and shell are as follows:
plate
shell
Therefore, for this stress field the limiting component is the plate and the limit load for the structure
is 13.0 MPa. According to prEN-13445-3 Annex B, Table B.9-2 for pressure loads (without natural
limit) the partial safety factor
p
is 1.2. Thus, the internal pressure limit according to the check
against GPD is
It is also possible to determine a limit pressure from the check against PD. In this case, the elastic
compensation is based on Mises yield condition. The partial safety factor on the resistance
R
is
not applied for the PD-check. From the Mises analysis in the PD-check, the limit load for the
structure was found as 17.55 MPa using the same method as above. The maximum ratio of Mises'
equivalent stress to Tresca's equivalent stress for the same load is 2/3. Therefore, applying a
factor of 3/2 to the yield stress in the Mises analysis (or to the limit load, as the analysis is elastic)
will always lead to a conservative result. Applying partial safety factors
R
= 1.25 and
p
= 1.2, the
internal pressure limit according to the GPD-check can be found to be
4. Check against PD
In the check for progressive plastic deformation, the principle in prEN-13445-3 B.9.3.1 is fulfilled
if the structure can be show to shake down. When a structure has been shown to shake down, the
failure modes of progressive plastic deformation and alternating plasticity cannot occur.
MPa PS
P
GPD
8 . 10
2 . 1
0 . 13 0 . 13
max
= = =

MPa PS
R p
GPD
13 . 10
2
3
25 . 1 2 . 1
55 . 17
2
3 55 . 17
max
=

=

max

d
ap L
R
P P =
MPa
R
P P
d
ap L
0 . 13
1 . 151
196
10
max
= = =

MPa
R
P P
d
ap L
4 . 13
3 . 152
204
10
max
= = =

Analysis Details
Example 1.3
Page
7.43 (S)
DBA
Design by Analysis
Figure 7.4.1-2: Equivalent Residual Stress Field
Figure 7.4.1-3: Equivalent Limit Stress Field.
In elastic compensation, the limit on the load at which the structure will undergo shakedown can be
calculated simply. Based on Melans shakedown theorem, the self-equilibrating residual stress field
that would result after a loading cycle can be calculated by subtracting the linear-elastic stress field
at the limit pressure from the limit stress field. The residual stress field is in effect the resulting
stress from an elastic unloading from the limit state back to zero pressure. If no stress in the
residual field violates the yield condition, i.e. if there is no stress above the design material strength
parameter, then the shakedown load is
equal to the limit load. Where the residual
stress field of a structure does exceed the
yield condition, the shakedown limit can
be calculated easily from the stress plots
because of the linearity.
The residual stress field from the elastic
compensation analysis using
Mises'condition is shown in Figure7.4.1-2.
In the check against PD there are no
partial safety factors applied to the
material strength or actions. Therefore,
the design material strength is 245 MPa
for the plate and 255 MPa for the shell
respectively. The maximum stress in the
residual field is greater than the stress
defining the limit load in the PD analysis.
The shakedown limit will be less than the
limit load given by the PD-check. By
invoking the elastic proportionality the shakedown load is given as
Figure 7.4.1-3 shows the limit stress field
for the PD-check. This can be used to
calculate the limit load for the structure
based on Mises' condition criterion and
associated flow rule. The maximum stress
in the plate is 139.64 MPa and by invoking
the proportionality of the elastic solution,
the limit load is 17.55 MPa. This can be
used in a check against GPD as shown
above.
5. Check against GPD Using Non-linear
Analysis
MPa
R
P PS
d
ap SD
3 . 13
7 . 184
245
10
max
max
= = =

Analysis Details
Example 1.3
Page
7.44 (S)
DBA
Design by Analysis
A check against GPD was also performed for the same FE-model using conventional non-linear
analysis. In this way, a direct comparison may be made between the two limit approaches. The FE-
geometry, mesh and boundary conditions are the same as those used in the elastic compensation
analysis. Material non-linearities
were applied corresponding to the
design material strength parameters,
204 MPa for the shell, 196 MPa for
the plate and perfect plasticity. A
ramped load is applied and the
analysis runs until the applied load is
such that convergence can no longer
occur due to unrestrained
displacement Gross Plastic
Deformation. It is assumed that the
last converged solution is the limit
load. Here, the last converged
solution was at a load of 17.29 MPa.
Using the same method as above the
allowable pressure according to the
GPD-check using Mises' condition is
The result offers a considerable benefit to the allowable pressure calculated using elastic
compensation. Figure 7.4.1-4 shows the equivalent stress distribution at the limitpressure.
Considerable plasticity can be noted in the plate with some additional plasticity spreading into the
stronger material in the shell. Analysis time to calculate limit load using non-linear FE-analysis
was 360 seconds.
MPa PS
p
GPD
48 . 12
2
3 29 . 17
max
= =

Figure 7.4.1-4 Limit Stress Field (non-linear analysis)


Analysis Details
Example 1.3
Page
7.45 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using elasto-plastic FE calculations
FE-Software: ANSYS 5.4
Element Types:
4 node, 2 D axisymmetric solid PLANE42
Boundary Conditions: Zero vertical displacement in the nodes of the undisturbed end of
the shell;
Symmetry boundary conditions in the nodes in the centre of the flat
end.
Model and Mesh:
Results: Maximum allowable pressure according to GPD: MPa PS
GPD
6 . 12
max
=
Shakedown limit pressure: MPa PS
SD
3 . 13
max
=
Member:
A&AB
Whole height of model:
443 mm
Number of elements:
2690
Analysis Details
Example 1.3
Page
7.46 (S)
DBA
Design by Analysis
1. Elements, mesh fineness, boundary conditions
The model of the structure is shown on page 1, a total number of 2690 4-node axisymmetric solid
elements PLANE42 in ANSYS 5.4 - was used. The linear shape function of these elements is
sufficient, since
the number of elements in the (linear-elastic) high stressed region is large,
the computation time in an analysis using nonlinear material properties is much higher
for elements with midside nodes and quadratic shape functions, although, close to the
limit load, the results are almost identical compared to those using elements with a linear
shape function,
there is no need to compute linear-elastic peak stresses very exactly, because the check
against PD can be carried out using the stress-concentration-free structure (according to
prEN 13445-3 Annex B.9.3.2) and the structures geometry is modelled exactly in the
example considered.
The boundary conditions applied in the model are symmetry ones in the nodes in the centre of the
plate (where the horizontal direction is perpendicular to the plane of symmetry) and constraining the
vertical degree of freedom in the nodes at the undisturbed end of the cylindrical shell to zero.
2. Determination of the maximum admissible pressure according to GPD
According to prEN 13445-3 Annex B, Table B.9-3, the partial safety factor
R
is 25 . 1 . Therefore,
the analysis using Trescas yield condition (delivered by an ANSYS distributor) was carried out
with a linear-elastic ideal-plastic material law, design material strength parameters of MPa 204 for
the shell and MPa 196 for the plate (corresponding to material strength parameters according to EN
10028-2 of MPa 255 for the shell and MPa 245 for the plate), associated flow rule, and first order
theory. For simplification, the boundary between the two materials was assumed to be in the plane
of the upper surface of the plate.
The elastic modulus of the cylindrical shell and the flat end used in all calculations was E=212 GPa.
The analysis was carried out using the arc-length method, which showed faster convergence for the
considered structure than the Newton - Raphson method. Since at the limit load the structure is fully
plastified in the plate and in the shell adjacent to the plate, a maximum vertical displacement in the
middle of the plate of 10 mm was used as termination criterion.
To restrict computation time in an appropriate manner, the analysis was terminated after 15 hours
on a Compaq Professional Workstation 5000 with two Pentium Pro processors and 256 MB
RAM.
Analysis Details
Example 1.3
Page
7.47 (S)
DBA
Design by Analysis
At termination time the last
convergent solution showed an
internal pressure of MPa 12 . 15 ,
and this pressure was used as
limit pressure. Figure 7.4.2-1
shows the vertical displacement
in the middle of the plate versus
the internal pressure.
Figure 7.4.2-1
Figure 7.4.2-2 shows Tresca's
equivalent stress for this limit
pressure. Because of the almost
full plastification in the plate,
there are small regions near the
mid-plane of the plate, where,
due to numerical effects, the
equivalent stress slightly exceeds
MPa 196 , but this has no effect
on further analyses.
F
Figure 7.4.2-2
Analysis Details
Example 1.3
Page
7.48 (S)
DBA
Design by Analysis
As shown in Figure 7.4.2-3 the
maximum absolute value of the
principal strains in the structure is
less than 5 % - as required in the
standard.
According to prEN 13445-3 Annex
B, Table B.9-2, the partial safety
factor for pressure (without a
natural limit)
P
is given by 1.2.
Therefore, the allowable (internal)
pressure according to the GPD-
check is
. 6 . 12
2 . 1
12 . 15
max
MPa PS
GPD
= =
Figure 7.4.2-3
A less time-consuming method to determine a limit action according to the GPD-check is given by
usage of the limit pressure result form the check against PD (see subsection 3.2 of section 3 -
Procedures). With the partial safety factors 25 . 1 =
R
and 2 . 1 =
P
, the allowable (internal)
pressure according to the GPD-check is, in this approach,
MPa PS
R P
GPD
34 . 12
2
3
25 . 1 2 . 1
4 . 21
2
3 37 . 21
max
=

=

,
where MPa 37 . 21 is the pressure calculated in the check against PD for which the maximum
absolute value of the principal strains does not exceed 5%.
3. Check against PD
The elasto-plastic FE analysis was carried out as stated in prEN 13445-3 Annex B, Sec. B.9.3.1
using Mises yield condition and associated flow rule, a linear-elastic ideal-plastic constitutive law
with design material strength parameters of 255 MPa for the shell and 245 MPa for the plate (for
simplification the boundary between the materials was assumed again to be in the plane of the upper
surface of the plate), and first order theory.
By defining and using load cases in ANSYS, the superposition of stress fields can be done easily.
Therefore the first load step of the analysis was defined at a very low load level, so that there was
Analysis Details
Example 1.3
Page
7.49 (S)
DBA
Design by Analysis
linear-elastic response of the structure. All other linear-elastic stress fields can then be determined
easily by multiplication with a suitable scale-up factor.
The analysis was carried out using the arc-length method. Since at the limit load the structure is
fully plastified in the plate and in the shell adjacent to the plate, a maximum vertical displacement
in the middle of the plate of 10 mm was used as termination criterion.
The computation time of the limit
load was 1 hour and 10 minutes on
the Compaq Professional
Workstation 5000 with two
Pentium Pro processors and 256
MB RAM.
The termination criterion was
fulfilled at an internal pressure of
21.43 MPa this pressure was
used as limit pressure. Figure
7.4.2-4 shows the vertical
displacement in the middle of the
plate versus the internal pressure;
according to this figure the limit
state is reached.
Figure 7.4.2-4
Figure 7.4.2-5 shows the elasto-
plastic Mises equivalent stress
distribution at the limit pressure
of MPa 43 . 21 . Because of the
almost full plastification in the
plate, there are small regions near
the mid-plane of the plate, where,
due to numerical effects, the
equivalent stress exceeds
MPa 245 , but this has no effect
on further analyses.
Figure 7.4.2-5
Analysis Details
Example 1.3
Page
7.50 (S)
DBA
Design by Analysis
Figure 7.4.2-6 shows the
distribution of the linear-
elastic Mises equivalent
stress for the limit pressure
the stress maximum is
located in the groove.
Figure 7.4.2-6
Figure 7.4.2-7 shows the
Mises equivalent stress of
the corrected residual stress
field, with a used scaling
factor given by 444 . 0
(see subsection 3.3.2.5 of
section 3 - Procedures), the
stress maximum is again
located in the groove.
1
.
Figure 7.4.2-7
Analysis Details
Example 1.3
Page
7.51 (S)
DBA
Design by Analysis
Figure 7.4.2-8 shows the Mises
equivalent stress distribution for
the lower bound shakedown limit.
The scaling factor is given by
621 . 0 (see subsection 3.3.2.5 of
section 3 - Procedures), and the
shakedown limit pressure by
MPa PS
SD
3 . 13 43 . 21 621 . 0
max
= =
The greatest equivalent stress is
again located in the groove, the
scaling factor is limited by the
yield stress of the plate.
Figure 7.4.2-8
Figure 7.4.2-9 shows a plot of the stress states in the grooves surface ( = 0 is adjacent to the
shell, = 180 is adjacent to the plate) in the deviatoric map, i.e. given in isometric coordinates by
the principal stresses
1
(perpendicular to the groove's surface),
2
(tangential to the groove's
surface), and hoop stress
3
.
It can easily be seen, that
the sum (thick green curve)
of the linear-elastic stress
(thin green curve, which
corresponds to an internal
pressure of MPa 133 ) and
of the corrected self-
equilibrating stress (yellow
curve) does not violate the
yield condition, which is
given by the circle.
F
i
F
Figure 7.4.2-9
2

= 0
= 180
Analysis Details
Example 1.3
Page
7.52 (S)
DBA
Design by Analysis
Note: Use of the application rule for the check against PD for constant principal stress directions,
see prEN 13445-3 Annex B.9.3.2, leads to a limit pressure of
MPa
RM
PS 4 . 13 43 . 21
8 . 783
245 2
43 . 21
8 . 783
2
=

= ,
where MPa 8 . 783 is the maximum Mises equivalent stress of the linear-elastic solution for an
internal pressure of MPa 43 . 21 .
Since the requirement given in the application rule is only a necessary condition for the fulfilment
of the principle, i.e. it is a necessary but not a sufficient condition for shakedown, the result using
the application rule usually differs from that using Melans theorem, but in this example the
difference of MPa 1 . 0 is negligible.
4. Comments
As shown by this DBA calculation, the DBF result is in this case optimal, since the maximum
allowable internal pressure according to DBF is practically the same as the one according to DBA.
Analysis Details
Example 1.3
Page
7.53 (S)
DBA
Design by Analysis
Analysis Type:
Stress Categorisation Route
FE-Software: BOSOR
Element Types: Axisymmetric shell elements
Model and Mesh:
As shown in diagram Shell Numbers in Calculation Model
Results:
Maximum admissible action according to the Stress Categorization Route:
Internal pressure PS
max SC
= 12.3 MPa
Member:
TKS
Analysis Details
Example 1.3
Page
7.54 (S)
DBA
Design by Analysis
The radius R 18 at the periphery of the flat end has been simulated by two strait lines, see the
figure of the model. This is acceptable, as BOSOR calculates the structural stresses, and this is a
simplification, which has negligible influence of the structural stresses.
The following figure shows the model's displacements.
The plots thereafter the distributions of stresses membrane and membrane plus bending in the
various parts of the model.
The plots are for a pressure of 15 MPa.
The critical points is the connection between the cylinder and the flat end. At this point the
secondary stresses in the cylinder are limited by the shakedown criteria.
The used allowable design stress is f = 255 MPa.
J
O
B

N
O
1
3
2
9
9
-
0
8
-
2
5

1
1
.
0
2
.
1
4

P
O
S
T
B
O
S
O
R

1
.
0
4







*
G
E
O
M
E
T
R
Y

P
L
O
T
A
L
L

S
H
E
L
L
S

D
E
F
O
R
M
.

U
N
D
E
F
O
R
M
.
W
I
N
D
O
W


(
X
,
Y
)
:
M
I
N
:

-
2
.
2
0
E
-
1
9
M
A
X
:

2
.
2
3
E
+
0
2
M
I
N
:

-
3
.
0
0
E
+
0
1
M
A
X
:

3
.
0
0
E
+
0
2
Analysis Details
Example 1.3
Page
7.55 (S)
DBA
Design by Analysis
Shell No. 1 Stresses


.
2
5


.
5
0


.
7
5

1
.
0
0

1
.
2
5

1
.
5
0

1
.
7
5

2
.
0
0
1
0
2
-
2
.
0
0
-
1
.
0
0


.
0
0

1
.
0
0

2
.
0
0

3
.
0
0
1
0
2
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
1
9
.
2
1

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
2
.
5
3
E
+
0
2
M
A
X
:

3
.
2
3
E
+
0
2


.
2
5


.
5
0


.
7
5

1
.
0
0

1
.
2
5

1
.
5
0

1
.
7
5

2
.
0
0
1
0
2


.
5
0

1
.
0
0

1
.
5
0

2
.
0
0

2
.
5
0

3
.
0
0
1
0
2
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
2
0
.
4
3

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

2
.
0
2
E
+
0
1
M
A
X
:

3
.
2
3
E
+
0
2


.
2
5


.
5
0


.
7
5

1
.
0
0

1
.
2
5

1
.
5
0

1
.
7
5

2
.
0
0
1
0
2
2
0
.
0
0
2
5
.
0
0
3
0
.
0
0
3
5
.
0
0
4
0
.
0
0
4
5
.
0
0
5
0
.
0
0
5
5
.
0
0
6
0
.
0
0
6
5
.
0
0
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
1
7
.
2
3

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

1
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

1
.
9
4
E
+
0
1
M
A
X
:

6
.
5
3
E
+
0
1


.
2
5


.
5
0


.
7
5

1
.
0
0

1
.
2
5

1
.
5
0

1
.
7
5

2
.
0
0
1
0
2
-
2
.
0
0
-
1
.
0
0


.
0
0

1
.
0
0

2
.
0
0
1
0
2
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
2
0
.
1
2

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
2
.
5
2
E
+
0
2
M
A
X
:

2
.
9
1
E
+
0
2
Analysis Details
Example 1.3
Page
7.56 (S)
DBA
Design by Analysis
Shell No. 2 Stresses

2
.
2
2

2
.
2
4

2
.
2
6

2
.
2
8

2
.
3
0

2
.
3
2

2
.
3
4

2
.
3
6

2
.
3
8
1
0
2
-
4
.
0
0
-
2
.
0
0


.
0
0

2
.
0
0

4
.
0
0

6
.
0
0
1
0
2
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
2
2
.
3
7

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
4
.
5
2
E
+
0
2
M
A
X
:

6
.
2
4
E
+
0
2

2
.
2
2

2
.
2
4

2
.
2
6

2
.
2
8

2
.
3
0

2
.
3
2

2
.
3
4

2
.
3
6

2
.
3
8
1
0
2

1
.
5
0

2
.
0
0

2
.
5
0

3
.
0
0

3
.
5
0

4
.
0
0

4
.
5
0

5
.
0
0

5
.
5
0

6
.
0
0
1
0
2
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
2
4
.
0
3

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

1
.
3
9
E
+
0
2
M
A
X
:

6
.
2
4
E
+
0
2
6
2
4
M
P
a

e
q
)
P

+
Q

3
f

2
.
2
2

2
.
2
4

2
.
2
6

2
.
2
8

2
.
3
0

2
.
3
2

2
.
3
4

2
.
3
6

2
.
3
8
1
0
2
8
6
.
0
0
8
6
.
5
0
8
7
.
0
0
8
7
.
5
0
8
8
.
0
0
8
8
.
5
0
8
9
.
0
0
8
9
.
5
0
9
0
.
0
0
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
2
1
.
2
7

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

2
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

8
.
5
5
E
+
0
1
M
A
X
:

9
.
0
1
E
+
0
1

2
.
2
2

2
.
2
4

2
.
2
6

2
.
2
8

2
.
3
0

2
.
3
2

2
.
3
4

2
.
3
6

2
.
3
8
1
0
2
-
1
.
5
0
-
1
.
0
0

-
.
5
0


.
0
0


.
5
0

1
.
0
0

1
.
5
0
1
0
2
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
2
3
.
1
8

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
1
.
5
9
E
+
0
2
M
A
X
:

1
.
6
6
E
+
0
2
Analysis Details
Example 1.3
Page
7.57 (S)
DBA
Design by Analysis
Shell No. 3 Stresses

2
.
5
0

3
.
0
0

3
.
5
0

4
.
0
0

4
.
5
0

5
.
0
0
1
0
2
-
1
.
0
0

-
.
5
0


.
0
0


.
5
0

1
.
0
0

1
.
5
0

2
.
0
0

2
.
5
0
1
0
2
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
2
5
.
2
9

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
1
.
3
7
E
+
0
2
M
A
X
:

2
.
9
8
E
+
0
2

2
.
5
0

3
.
0
0

3
.
5
0

4
.
0
0

4
.
5
0

5
.
0
0
1
0
2


.
5
0

1
.
0
0

1
.
5
0

2
.
0
0

2
.
5
0
1
0
2
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
2
6
.
3
8

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

9
.
3
1
E
+
0
0
M
A
X
:

2
.
9
8
E
+
0
2

2
.
5
0

3
.
0
0

3
.
5
0

4
.
0
0

4
.
5
0

5
.
0
0
1
0
2


.
9
0

1
.
0
0

1
.
1
0

1
.
2
0

1
.
3
0

1
.
4
0

1
.
5
0
1
0
2
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
2
4
.
4
0

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

3
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

8
.
0
5
E
+
0
1
M
A
X
:

1
.
5
4
E
+
0
2

2
.
5
0

3
.
0
0

3
.
5
0

4
.
0
0

4
.
5
0

5
.
0
0
1
0
2

-
.
5
0

-
.
2
5


.
0
0


.
2
5


.
5
0


.
7
5

1
.
0
0

1
.
2
5

1
.
5
0
1
0
2
J
O
B

N
O
1
3
2
9
9
-
0
8
-
1
8

1
6
.
2
6
.
0
9

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
7
.
2
4
E
+
0
1
M
A
X
:

1
.
6
0
E
+
0
2
DBA
Design by Analysis
Analysis Details
Example 1.3 / F-Check
Page
7.58(A)
Analysis Type:
F-Check
Member:
A&AB
FE Software: ANSYS 5.4
Element Types:
4 node, 2 D axisymmetric solid PLANE42
Boundary Conditions:
Zero vertical displacement in the nodes of the undisturbed end of the shell;
Symmetry boundary conditions in the nodes in the centre of the flat end.
Model and Mesh:
Results: Fatigue life N = 6373 cycles (welded part)
t
max
= 20 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
= 20 C
R
m
= 410 MPa
R
p0,2/t*
= 255 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . K
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 21,5 mm

D (5
.
10
6
cycles)
= 46 MPa (class 63)
equivalent stresses or principal stresses
m = 3 C = .. m = 3 C

= 5.10
11
C
//
= ..
m = 5 C = .. m = 5 C

= ..
C
//
= ..
Stresses

struc
= 416.7 MPa (structural equivalent stress range, determined by extrapolation)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..
= k
e

eq,l
= .. MPa
Else

=
struc
= 416.7 MPa
Thermal loading
If
eq,l
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

eq,l
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0,639
Page
7.59 (A)
Analysis Details
Example 1.3 / F-Check
DBA
Design by Analysis
DataCritical point: Weld end to shell, inside; Principal stress range approach
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 416.7 MPa
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
If
w
f

>
5
.
10
6
cycles
then
M = 3 and C (C

or C
//
) = 5
.
10
11
N =
m
w
f
C

,
_


= 6910 cycles
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
M = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
Page
7.60 (A)
Analysis Details
Example 1.3 / F-Check
DBA
Design by Analysis
Note: All unwelded regions are less critical
Note: In this example both approaches the equivalent stress range approach and the principal stress range
approach have been used. The maximum principal stress is the tangential stress component and it is
positive, the minimum one pressure is negative and small. Therefore, the difference in the results
is small.
Page
7.61(A)
Analysis Details
Example 1.3 / F-Check
DBA
Design by Analysis
t
max
= 20 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
= 20 C
R
m
= 410 MPa
R
p0,2/t*
= 255 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . K
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the thermal
stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 21,5 mm

D (5 10
6
cycles)
= 46 MPa (class 63)
equivalent stresses or principal stresses
m = 3 C = 5
.
10
11
m = 3 C

=
C
//
= ..
m = 5 C = 1.08
.
10
15
m = 5 C

= ..
C
//
= ..
Stresses

struc
= 428.1 MPa (structural equivalent stress range,determined by extrapolation)
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

eq,l
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0,639
Data
Critical point: Weld end to shell, inside; Equivalent stress range approach
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..
= k
e

eq,l
= .. MPa
Else

=
struc
= 428.1 MPa
Page
7.62(A)
Analysis Details
Example 1.3 / F-Check
DBA
Design by Analysis
18-11-3 Allowable number of cycles N 18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
If
w
f

>
5
.
10
6
cycles
then
M = 3 and C (C

or C
//
) = 5.10
11
N =
m
w
f
C

,
_


= 6373 cycles
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
M = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
18-10-7 Allowable number of cycles N
w
f

= 416.7 MPa
Note: All unwelded regions are less critical
Note: In this example both approaches the equivalent stress range approach and the principal stress range
approach have been used. The maximum principal stress is the tangential stress component and it is
positive, the minimum one pressure is negative and small. Therefore, the difference in the results
is small.
Analysis Details
Example 1.4
Page
7. 63 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using Elastic Compensation
FE-Software: ANSYS 5.4
Element Types: 8 node, 2-D structural axisymmetric solid.
Boundary Conditions: No vertical displacement in the undisturbed end of the shell remote
from the flat end.
Axisymmetry from elements
Model and Mesh:
Results:
Maximum internal pressure according to the GPD-Check PS
max GPD
= 8.42 MPa
Check against PD: Shakedown limit pressure PS
max SD
= 12.8 MPa
Member:
Strathclyde
Analysis Details
Example 1.4
Page
7. 64 (S)
DBA
Design by Analysis
Figure 7.5.1-1: Limit Stress Field (Tresca)
1. Finite Element Mesh
Element divisions for the finite element mesh were defined parametrically to allow modification of
the element density. The analysis was carried out on two models; one with fillets at the
discontinuities at the nozzle the other without. Mesh densities for the two models remained similar:
655 nodes for the fillet model 627 nodes for the model with no fillets at the nozzle. Finite element
models were created using linear 4-node 2-D axisymmetric solids, the analysis was repeated with
higher order 8-node structural 2-D axisymmetric solids, any alterations this may have on the results
could also be noted. The allowable pressure according to the GPD-Check and the shakedown
pressure according to the PD-check were calculated from the model giving the greatest lower
bounds. The analysis discussed is for the model containing fillets at the nozzle with 4-node
elements.
Boundary conditions applied to the model are axisymmetry, applied via a key option when defining
the element type in the FE-software (axisymmetry around the vertical axis Y). The nodes at the
undisturbed end of the cylindrical shell have their vertical degree of freedom constrained to ensure
that plane sections remain plane.
2. Material properties
Material strength parameter MPa RM 255 = (cylindrical shell), MPa RM 245 = (flat end),
MPa RM 265 = (nozzle), modulus of elasticity GPa E 212 = (all parts of the structure).
3. Determination of the maximum admissible pressure according to the GPD-Check
Using the application rule in prEN-13445-3 Annex B.9.2.2 to check against GPD the principle is
fulfilled when for any load case the combination of the design actions do not exceed the design
resistance. This may be shown by calculating the limit load. The limit load has to be determined
using Trescas yield condition and associated flow rule. As elastic compensation is based upon a
series of elastic equilibrium stress fields, it is a relatively simple procedure to derive a lower bound
limit load direct from the Tresca yield model.
From prEN-13445-3 Annex B, Table B.9-3 the partial safety factor,
R
on the resistance is 1.25.
Therefore, the design material strength parameters given by RM/
R
were: 204 MPa for the shell,
196 MPa for the plate and 112MPa
for the nozzle. The analysis was
carried out using the elastic
compensation method conforming
to the direct route rules for GPD in
Annex B: linear elastic ideal plastic
material law, Trescas yield
condition and associated flow rule
and first order theory.
In each elastic compensation
iteration an equilibrium stress field
is produced where the elastic
modulus of each subsequent
iteration is defined by the previous
elastic solution. In this way
regions of the FE- model may be
Analysis Details
Example 1.4
Page
7. 65 (S)
DBA
Design by Analysis
systematically stiffened or weakened depending upon the stress magnitudes in the previous
solution. The result is that every equilibrium stress field is a lower bound of the limit load, the
stress field giving the greatest limit load is taken as the closest to the actual limit state and defines
the limit load in the analysis. In problems such as this, where there are materials with different
properties the modulus modification has a modified procedure that takes account of the different
material properties. This modified method calculates the limit pressure for each component with a
different material, allowing the component giving the lowest limit load to define the limit for the
whole model. Convergence of the equilibrium stress field will occur typically between 8 and 15
iterations, i.e. no further benefit in the limit load will be noted with subsequent iterations. The total
computing time to run the analysis on a 300 MHz Pentium two processor with 128 Mb RAM
Windows NT workstation was 180 seconds. The stress field was shown to converge after eight
iterations giving a lower bound on the limit pressure of 10.1 MPa. Figure 7.5.1-1 shows the limit
stress (intensity) field according to the Tresca condition. The limit pressure is given by scaling the
limit stress field so that the stresses anywhere in the model do not exceed the material resistance for
that component. The analysis showed that the plate was the first component to fail. The limit load
can be calculated by scaling the applied pressure (10 MPa) by the factor 1.01 (design material
strength/max stress: 196/195). According to prEN-13445-3 Annex B, Table B.9-2 for pressure
loads (without natural limit) the partial safety factor
p
is 1.2. Thus, the internal pressure limit
according to the GPD-check is
It is also possible to determine a limit pressure from the check against PD. In this case, the elastic
compensation is based on Mises yield condition. The partial safety factor on the resistance
R
is
not applied for the PD-check. However, as the analysis is wholly elastic it is possible to scale the
stress fields at any time (similarly as was done above). The maximum ratio of Mises' equivalent
stress to Tresca's equivalent stress for the same load is 2/3. Therefore, applying a factor of 3/2 to
the yield stress in the Mises analysis (or to the limit load, as the analysis is elastic) will always lead
to a conservative result. From the Mises analysis the limit load was found to be 13.9 MPa and with
the partial safety factors
R
= 1.25 and
p
= 1.2, the internal pressure limit according to the GPD-
check can be found as
4. Check against PD
In the check for progressive plastic deformation, the principle in prEN-13445-3 B.9.3.1 is fulfilled
if the structure can be show to shake down. When a structure has been shown to shake down, the
failure modes of progressive plastic deformation and alternating plasticity can not occur.
In elastic compensation the load at which the structure will shakedown can be calculated simply.
Based on Melans shakedown theorem, the self-equilibrating residual stress field that would result
after a loading cycle can be calculated by subtracting the linear-elastic stress field at the limit
pressure from the limit stress field. The residual stress field is in effect the resulting stress from an
MPa PS
GPD
42 . 8
2 . 1
1 . 10
max
= =
MPa PS
R p
GPD
03 . 8
2
3 9 . 13
max
=

=

Analysis Details
Example 1.4
Page
7. 66 (S)
DBA
Design by Analysis
Figure 7.5.1-3: Limit Stress Field (Mises)
Figure 7.5.1-2: Residual Stress Field (Mises)
elastic unloading from the limit state back to zero pressure. If no stress in the residual field violates
the yield condition, i.e. if there is no
stress above the design material
strength, then the shakedown load is
equal to the limit load. Where the
residual stress field of a structure
does exceed the yield condition, the
shakedown limit can be calculated
easily from the stress plots because
of the linearity.
The residual stress field from the
elastic compensation Mises analysis
is shown in Figure 7.5.1-2. Because
the applied load is arbitrary and the
resulting stress fields are scaled to
the yield condition, the maximum
residual stress is compared to the
maximum stress in the limit stress
field, Figure 7.5.1-3. As stated
above, the presence of different
materials requires here some
modification. A shakedown
pressure is calculated for each
component with a different material
and the component giving the lowest
shakedown limit defines the limit for
the structure. The analysis showed
that the plate may be the first
component to suffer from
progressive plastic deformation. In
many of the iterations the maximum
residual stress was greater than the
maximum from the limit stress field,
thus the shakedown limit is less than
the limit load. As all the iterations
are lower bounds, the stress field
yielding the greatest value of
shakedown load is used to define the
limit. With an applied load of 10
MPa the shakedown limit is 12.8 MPa.
5. Check against GPD Using Non-linear Analysis
A check against GPD was also performed for the same FE-model using conventional non-linear
analysis. In this way, a direct comparison may be made between the two limit approaches. The FE-
geometry, mesh and boundary conditions are the same as those used in the elastic compensation
analysis. Material non-linearities were applied corresponding to the design material parameters,
Analysis Details
Example 1.4
Page
7. 67 (S)
DBA
Design by Analysis
Figure 7.5.1-4: Limit Stress Field (non-linear analysis)
Figure 7.5.1-5: Maximum total principal strain
204 MPa for the shell, 196 MPa for
the plate, 112 MPa for the nozzle
and perfect plasticity. A ramped
load is applied and the analysis runs
until the applied load is such that
convergence can no longer occur
due to unrestrained displacement
Gross Plastic Deformation. It is
assumed that the last converged
solution is the limit load. Figure
7.5.1-4 shows the Mises equivalent
stress distribution at the limit load.
Here, the last converged solution
was at a load of 13.75 MPa, using
the same method as above the
allowable pressure according to the
GPD-check using Mises condition is
At this load, there is a considerable
amount of plasticity and a check on
the maximum total principal strain is
required to be below +/- 5%. Figure
7.5.1-5 shows that the maximum
absolute value of the total principal
strain is 4.7% and is within the limit:
Therefore,
The result offers a considerable
benefit to the allowable pressure
calculated using elastic
compensation. Analysis time to
calculate limit load using non-linear
FE-analysis was 360 seconds.
MPa PS
p
GPD
92 . 9
2
3 75 . 13
max
= =

Analysis Details
Example 1.4
Page
7. 68 (S)
DBA
Design by Analysis
6. Additional Comments
Additional analysis was completed to ascertain any effect on the results for higher order elements
(8-node) and omission of the fillets at the nozzle discontinuities . Limit loads and shakedown values
were calculated using Mises' condition, the results are summarised in the following table.
Use of Fillets Element Type Number of
Iterations
Lower Bound
Limit Load
Shakedown
Load
Processor
Time
No 4 node 8 8.08 12.8 273
No 8 node 8 7.96 12.3 284
Yes 4 node 8 8.42 12.8 278
Yes 8 node 8 7.9 12.3 289.7
Table 7.5.1-1: Limit and shakedown analysis summary
Both the lower bound limit load and shakedown load are calculated using the same method as
described above. As can be seen from the table above, the choice of element type and application
of fillets is of negligible difference to the results.
The utilisation of elastic compensation in this DBA calculation has given an increase in the
allowable pressure over that given by the DBF calculation.
The two methods used to calculate the lower bound limit load (direct from Tresca or correction of
Mises) show good correlation; the Mises corrected value is slightly conservative as would be
expected. However, carrying out the Mises elastic compensation will give both the limit load and
shakedown load in one analysis.
Non-linear DBA calculations offer a further benefit than elastic compensation in terms of higher
allowable pressure than those pressures calculated by DBF in this example.
Analysis Details
Example 1.4
Page
7. 69 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using elasto-plastic FE calculations
FE-Software: ANSYS 5.4
Element Types:
4 node, 2 D axisymmetric solid PLANE42
Boundary Conditions: Zero vertical displacement in the nodes of the unsdisturbed end of
the shell; Longitudinal stress corresponding to pressure force on
the nozzle's closed end.
Model and Mesh:
Results: Maximum allowable pressure according to GPD: MPa PS
GPD
9 . 9
max
=
Shakedown limit pressure: MPa PS
SD
2 . 12
max
=
Member:
A&AB
Length of shell: 400 mm
Length of nozzle (weld incl.):150 mm
Total length of model: 593 mm
Number of elements: 3693
Analysis Details
Example 1.4
Page
7. 70 (S)
DBA
Design by Analysis
1. Elements, mesh fineness, boundary conditions
The model of the structure is shown on page 1, a total
number of 3693 4-node axisymmetric solid elements
PLANE42 in ANSYS 5.4 - was used. To avoid stress
singularities in the linear-elastic calculation, the boundary
between the weld and the nozzle and the weld and the
plate, respectively, is modelled with fillets corresponding
to a thickness of the weld influence zone of 2 mm.
Therefore, the corresponding fillet radii are given by
mm R 7
1
= and mm R 5 . 10
2
= - see Figure 7.5.2-1.
Linear-elastic calculations with different mesh finenesses
in the high stress regions showed, that the used mesh
fineness is sufficient (doubling the number of elements in
this regions leads to a difference in the stress results of
approximately 1 %). Furthermore, the used linear shape
function of the elements is here sufficient, since the
computation time in an analysis using non-linear material
properties is much higher for elements with mid-side
nodes and quadratic shape functions, and, close to the
limit load, the results are almost identical.
The boundary condition applied in the model is
constraining the vertical degree of freedom in the nodes at
the undisturbed end of the cylindrical shell to zero. At the
end of the nozzle a longitudinal stress corresponding to a
closed end is applied.
Figure 7.5.2-1
2. Determination of the maximum allowable pressure according to GPD
Since the subroutine using Trescas yield condition showed bad convergence, the results from the
check against PD, see chapter 2 of subsection 3 -Procedures, have been used. The partial safety
factor
R
according to prEN 13445-3 Annex B, Table B.9-3 is 25 . 1 , and the partial safety factor for
pressure (without a natural limit) according to prEN 13445-3 Annex B, Table B.9-2
P
is 2 . 1 .
Therefore, the (internal) allowable pressure according to the GPD-check is by
. 92 . 9
2
3
25 . 1 2 . 1
18 . 17
2
3 18 . 17
max
MPa PS
R P
GPD
=

=

The maximum principal strain for this state is about 5 %, see Figure 7.5.2-2, the condition in prEN
13445-3 Annex B.9.2.2 is fulfilled.
Analysis Details
Example 1.4
Page
7. 71 (S)
DBA
Design by Analysis
Figure 7.5.2-2
3. Check against PD
The elasto-plastic FE analysis was carried out as stated in prEN 13445-3 Annex B, Sec. B.9.3.1,
using Mises yield condition and associated flow rule, a linear-elastic ideal-plastic constitutive law
with design material strength parameters of 255 MPa for the shell (according to EN 10028-2), 245
MPa for the plate (according to EN 10028-2), and 265 MPa for the nozzle (according to prEN
10216-2), and first order theory. For simplicity, the boundaries between the materials were assumed
to be in the planes of the lower and upper surface of the plate. The elastic modulus used in the
calculations was E=212 GPa for all parts of the structure.
Defining and using load cases in ANSYS,
the superposition of stress fields can be done
easily; therefore an early load step of the
analysis was defined at a very low load level
(1 MPa), so that there was linear-elastic
response of the structure. All other linear-
elastic stress fields can then be determined
by multiplication with a suitable scale-up
factor.
The analysis was carried out using the arc-
length method. Since at the limit load the
structure is fully plastified in the plate and
in the nozzle adjacent to the plate, a
Figur 7.5.2-3
Analysis Details
Example 1.4
Page
7. 72 (S)
DBA
Design by Analysis
maximum vertical displacement at the inner edge of the plate of 5 mm was used as termination
criterion.
The computation time of the limit
load was 54 minutes on the
Compaq Professional Workstation
5000 with two Pentium Pro
processors and 256 MB RAM.
The termination criterion was
fulfilled at an internal pressure of
17.18 MPa this pressure was used
as limit pressure. Figure 7.5.2-3
shows the vertical displacement at
the inner edge of the plate versus the
internal pressure. According to this
figure the limit state is reached.
Figure 7.5.2-4 shows the elasto-
plastic Mises' equivalent stress
distribution at the limit pressure of
MPa 18 . 17 .
Figure 7.5.2-4
Figure 7.5.2-5 shows the linear-
elastic Mises equivalent stress
distribution at the limit pressure the
stress maximum is located at the
junction of nozzle and weld.
Figure 7.5.2-5
Analysis Details
Example 1.4
Page
7. 73 (S)
DBA
Design by Analysis
Figure 7.5.2-6 shows the Mises
equivalent stress distribution of
the corrected residual stress
field, the scaling factor is
given by 538 . 0 (see subsection
3.3.2.5 of section 3 -
Procedures). The maximum
(allowable) stress, which is used
for the determination of , is
now located in the stress relief
groove of the plate.
Figure 7.5.2-6
At the lower bound shakedown
limit the scaling factor is
now determined by the
maximum stress at the junction
between the nozzle and the weld
see Figures 7.2.5-7 and
7.2.5-8 and given by 710 . 0
(see subsection 3.3.2.5 of
section 3 - Procedures).
Figure 7.5.2-7
Analysis Details
Example 1.4
Page
7. 74 (S)
DBA
Design by Analysis
Figure 7.5.2-8
Thus, the shakedown limit pressure is given by
MPa PS
SD
2 . 12 18 . 17 710 . 0
max
= = .
Analysis Details
Example 1.4
Page
7. 75 (S)
DBA
Design by Analysis
Analysis Type:
Stress Categorization Route
FE-Software: BOSOR
Element Types: Axisymmetric shell elements.
Model and Mesh:
As shown in diagram Shell Numbers in Calculation Model
Results:
Maximum admissible action according to Stress Categories:
Internal pressure PS
max SC
= 10.0 MPa
Member:
TKS
Analysis Details
Example 1.4
Page
7. 76 (S)
DBA
Design by Analysis
The following figure shows the deformed model.
The figures thereafter show the distribution of stresses membrane and membrane plus bending
in the surfaces of the various parts of the model.
With the designation list in subsection 3.6 the various plots are self-explaining.
The plots are for a pressure of 7.9 MPa. The limiting part is the nozzle close to the flat end.
There the limiting stress category is the membrane stresses at the (spatial) local stress see the
membrane stress plot on the page after this one.
The allowable design stress used is MPa f 170 = , the allowable value for the local membrane
stress 187 MPa.
J
O
B

N
O
1
4
9
9
-
0
8
-
2
5

1
1
.
0
5
.
5
4

P
O
S
T
B
O
S
O
R

1
.
0
4







*
G
E
O
M
E
T
R
Y

P
L
O
T
A
L
L

S
H
E
L
L
S

D
E
F
O
R
M
.

U
N
D
E
F
O
R
M
.
W
I
N
D
O
W


(
X
,
Y
)
:
M
I
N
:

1
.
0
0
E
+
0
2
M
A
X
:

2
.
2
3
E
+
0
2
M
I
N
:

-
4
.
3
3
E
+
0
1
M
A
X
:

4
.
5
0
E
+
0
2
Analysis Details
Example 1.4
Page
7. 77 (S)
DBA
Design by Analysis
Shell No. 1 Stresses


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0

1
.
2
0
1
0
2
-
2
.
0
0
-
1
.
0
0


.
0
0

1
.
0
0

2
.
0
0

3
.
0
0
1
0
2
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
2
.
5
5
.
1
4

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
2
.
2
6
E
+
0
2
M
A
X
:

3
.
0
6
E
+
0
2


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0

1
.
2
0
1
0
2

1
.
0
0

1
.
5
0

2
.
0
0

2
.
5
0

3
.
0
0
1
0
2
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
2
.
5
7
.
3
8

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

7
.
1
7
E
+
0
1
M
A
X
:

3
.
2
8
E
+
0
2


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0

1
.
2
0
1
0
2


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0

1
.
6
0

1
.
8
0
1
0
2
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
2
.
5
4
.
3
1

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

1
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

7
.
5
2
E
+
0
1
M
A
X
:

1
.
8
4
E
+
0
2
L
im
it

f
o
r

lo
c
a
l m
e
m
b
r
a
n
1
.
1
f

=

1
8
8
M
P
a
1
.
0
f
1
4
8
M
P
a


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0

1
.
2
0
1
0
2


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0

1
.
6
0

1
.
8
0

2
.
0
0

2
.
2
0

2
.
4
0

2
.
6
0
1
0
2
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
2
.
5
6
.
5
8

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

1

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

7
.
1
7
E
+
0
1
M
A
X
:

2
.
6
2
E
+
0
2
Analysis Details
Example 1.4
Page
7. 78 (S)
DBA
Design by Analysis
Shell No. 2 Stresses

1
.
4
0

1
.
6
0

1
.
8
0

2
.
0
0

2
.
2
0

2
.
4
0
1
0
2
-
1
.
5
0
-
1
.
0
0

-
.
5
0


.
0
0


.
5
0

1
.
0
0

1
.
5
0
1
0
2
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
3
.
0
3
.
0
9

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
1
.
5
9
E
+
0
2
M
A
X
:

1
.
9
9
E
+
0
2

1
.
4
0

1
.
6
0

1
.
8
0

2
.
0
0

2
.
2
0

2
.
4
0
1
0
2


.
2
0


.
4
0


.
6
0


.
8
0

1
.
0
0

1
.
2
0

1
.
4
0

1
.
6
0

1
.
8
0
1
0
2
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
3
.
0
7
.
4
8

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

1
.
3
9
E
+
0
1
M
A
X
:

1
.
9
9
E
+
0
2

1
.
4
0

1
.
6
0

1
.
8
0

2
.
0
0

2
.
2
0

2
.
4
0
1
0
2
1
0
.
0
0
1
5
.
0
0
2
0
.
0
0
2
5
.
0
0
3
0
.
0
0
3
5
.
0
0
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
3
.
0
1
.
5
1

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

2
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

6
.
4
3
E
+
0
0
M
A
X
:

3
.
8
1
E
+
0
1

1
.
4
0

1
.
6
0

1
.
8
0

2
.
0
0

2
.
2
0

2
.
4
0
1
0
2
-
1
.
0
0

-
.
5
0


.
0
0


.
5
0

1
.
0
0
1
0
2
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
3
.
0
7
.
0
5

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

2

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
1
.
1
7
E
+
0
2
M
A
X
:

1
.
3
6
E
+
0
2
Analysis Details
Example 1.4
Page
7. 79 (S)
DBA
Design by Analysis
Shell No. 3 Stresses

2
.
5
0

3
.
0
0

3
.
5
0

4
.
0
0

4
.
5
0

5
.
0
0
1
0
2
-
2
.
0
0
-
1
.
0
0


.
0
0

1
.
0
0

2
.
0
0

3
.
0
0
1
0
2
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
3
.
1
2
.
5
6

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


G
E
N
.
D
I
R


S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
2
.
6
5
E
+
0
2
M
A
X
:

3
.
6
1
E
+
0
2

2
.
5
0

3
.
0
0

3
.
5
0

4
.
0
0

4
.
5
0

5
.
0
0
1
0
2


.
5
0

1
.
0
0

1
.
5
0

2
.
0
0

2
.
5
0

3
.
0
0

3
.
5
0
1
0
2
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
3
.
1
5
.
2
7

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


T
R
E
S
C
A



S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

9
.
4
8
E
+
0
0
M
A
X
:

3
.
6
1
E
+
0
2

2
.
5
0

3
.
0
0

3
.
5
0

4
.
0
0

4
.
5
0

5
.
0
0
1
0
2
4
5
.
0
0
5
0
.
0
0
5
5
.
0
0
6
0
.
0
0
6
5
.
0
0
7
0
.
0
0
7
5
.
0
0
8
0
.
0
0
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
3
.
1
1
.
3
0

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
O
M
P
.
-
S
T

S
H
E
L
L

3
"
M
E
M
B
R
"


F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

4
.
3
4
E
+
0
1
M
A
X
:

8
.
1
3
E
+
0
1

2
.
5
0

3
.
0
0

3
.
5
0

4
.
0
0

4
.
5
0

5
.
0
0
1
0
2
-
8
.
0
0
-
6
.
0
0
-
4
.
0
0
-
2
.
0
0


.
0
0

2
.
0
0

4
.
0
0

6
.
0
0

8
.
0
0
1
0
1
J
O
B

N
O
1
4
9
9
-
0
8
-
1
9

1
3
.
1
4
.
5
3

P
O
S
T
B
O
S
O
R

1
.
0
4









*
S
T
R
E
S
S
E
S
*


C
I
R
C
U
M
F


S
H
E
L
L

3

"
I
N
S
I
D
E
"
"
O
U
T
S
I
D
E
"
F
U
N
C
.
V
A
L
U
E
S
:
M
I
N
:

-
9
.
4
6
E
+
0
1
M
A
X
:

9
.
3
0
E
+
0
1
DBA
Design by Analysis
Analysis Summary
Example 1.4 / F-Check
Page
7.80 (A)
Analysis Type:
F-Check
Member:
A&AB
FE Software: ANSYS 5.4
Element Types:
4 node, 2 D axisymmetric solid PLANE42
Boundary Conditions:
Zero vertical displacement in the nodes of the undisturbed end of
the shell;
Model and Mesh:
Results:
Fatigue life N = 21010 cycles (welded part shell-plate junction)
Length of shell: 400 mm
Length of nozzle (weld incl.):150 mm
Total length of model: 593 mm
Number of elements: 3693
Three locations, where local stress maxima occur, are of interest with regard to the check
against fatigue:
the weld toe in the nozzle (inside) at the nozzle-plate junction (node 3052 see Figure
7.5.4-1)
the weld toe in the nozzle (outside) at the nozzle-plate junction (node 3038 see Figure
7.5.4-2)
the weld toe at the inside of the shell at the shell-plate junction (node 1438 see Figure
7.5.4-3).
At these points the structural (equivalent and principal) stresses were determined by quadratic
extrapolation. The figures 7.5.4-1, 7.5.4-2 and 7.5.4-3 show the corresponding nodes, which
are used as pivot points for the extrapolation. The distances between these points are
approximately the ones recommended in Figure 18-3 of prEN 13445-3.
Figure 7.5.4-1
Page
7.81 (A)
Analysis Details
Example 1.4 / F-Check
DBA
Design by Analysis
Figure 7.5.4-2
Figure 7.5.4-3
DBA
Design by Analysis
Analysis Details
Example 1.4 / F-Check
Page
7.82 (A)
Data
t
max
= 20 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
= 20 C
R
m
= 410 MPa
R
p0,2/t*
= 265/245 MPa
Used: R
p0,2/t*
= 245 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 13,7 mm

D (5 10
6
cycles)
= 52 MPa (class 71)
equivalent stresses or principal stresses
m = 3 C = m = 3 C

= 7.16
.
10
11
C
//
= ..
m = 5 C =

m = 5 C

= ..
C
//
= ..
Stresses
Critical point: Weld toe in nozzle (inside): Node 3052

struc
= 225,6 MPa (structural equivalent stress range,determined by extrapolation)
(obtained by quadratic extrapolation on inside of nozzle)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..
= k
e

struc
= .. MPa
Else

=
struc
= 225.6 MPa
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0.639
Page
7. 83 (A)
Analysis Details
Example 1.4 / F-Check
DBA
Design by Analysis
Welded region / Principal stress range approach / Critical point: Node 3052
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 225.6 MPa
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = 7,16
.
10
11
N =
m
w
f
C

,
_


= 62380 cycles
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
Page
7.84 (A)
Analysis Details
Example 1.4 / F-Check
DBA
Design by Analysis
Since large differences between the principal stress range approach and the structural equivalent stress one
are expected, both approaches are used here. The details of the structural equivalent stress range approach
are given on the next two pages.
Data
t
max
= 20 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
= 20 C
R
m
= 410 MPa
R
p0,2/t*
= 265/245 MPa
Used: R
p0,2/t*
= 245 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 13,7 mm

D (5
.
10
6
cycles)
= 52 MPa (class 71)
equivalent stresses or principal stresses
m = 3 C = 7.16
.
10
11
m = 3 C

=
C
//
= ..
m = 5 C = 1.96
.
10
15
m = 5 C

= ..
C
//
= ..
Stresses

struc
= 350,7 MPa (structural equivalent stress range)
(total value used, since quadratic extrapolation on inside of nozzle gives larger value)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..
= k
e

struc
= .. MPa
Else

=
struc
= 350.7 MPa
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0.639
Page
7. 85 (A)
Analysis Details
Example 1.4 / F-Check
DBA
Design by Analysis
Critical point: Weld toe in nozzle (inside): Node 3052
Welded region/ Structural equivalent stress range approach Critical point: Node 3052
For N 2 10
6
cycles
See figure 18-14
f
m
= 1
if
( ) M 1 2
R

+

eq
R
p0,2/t*
then
f
m
=

,
_

+
+
R

eq
2
3
M
M 1
3 M 1
= .
f
m
= 1
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 350.7 MPa
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
Page
7.86 (A)
Analysis Details
Example 1.4 / F-Check
DBA
Design by Analysis
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = 7.16
.
10
11
N =
m
w
f
C

,
_


= 16600 cycles
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
Data
t
max
= 20 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
= 20 C
R
m
= 410 MPa
R
p0,2/t*
= 265/245 MPa
Used: R
p0,2/t*
= 245 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 13,7 mm

D (5 10
6
cycles)
= 52 MPa (class 71)
equivalent stresses or principal stresses
m = 3 C = m = 3 C

= 7.16
.
10
11
C
//
= ..
m = 5 C = m = 5 C

= ..
C
//
= ..
Stresses
Critical point: Weld toe in nozzle (outside): Node 3038

struc
= 318,0 MPa (structural equivalent stress range)
(maximum value used, since quadratic extrapolation on outside of nozzle gives larger value)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..
= k
e

struc
= .. MPa
Else

=
struc
= 318.0 MPa
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0.639
Page
7.87 (A)
Analysis Details
Example 1.4 / F-Check
DBA
Design by Analysis
Welded region / Principal stress range approach Critical point: Node 3038
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 318,0 MPa
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
Page
7.88 (A)
Analysis Details
Example 1.4 / F-Check
DBA
Design by Analysis
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = 7.16
.
10
11
N =
m
w
f
C

,
_


= 22270 cycles
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
Since large differences between the principal stress range approach and the structural equivalent stress one
are expected, both approaches are used here. The details of the structural equivalent stress range approach
are given on the next two pages
Data
t
max
= 20 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
= 20 C
R
m
= 410 MPa
R
p0,2/t*
= 265/245 MPa
Used: R
p0,2/t*
= 245 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of
the stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and
thermal equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied
to the thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 10,25 mm

D (5 10
6
cycles)
= 52 MPa (class 71)
equivalent stresses or principal stresses
m = 3 C = 7.16
.
10
11
m = 3 C

= ..
C
//
= ..
m = 5 C = 1,96 10
15
m = 5 C

= ..
C
//
= ..
Stresses
Critical point: Weld toe in nozzle (outside): Node 3038

struc
= 225,6 MPa (structural equivalent stress range)
(obtained by quadratic extrapolation on outside of nozzle)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..
= k
e

struc
= .. MPa
Else

=
struc
= 289.1 MPa
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0.639
Page
7.89 (A)
Analysis Details
Example 1.4 / F-Check
DBA
Design by Analysis
Welded region / Structural equivalent stress range approach / critical point: node 3038
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 289.1 MPa
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
Page
7.90 (A)
Analysis Details
Example 1.4 / F-Check
DBA
Design by Analysis
If
w
f

>
5 10
6
cycles
then
m = 3 and C (C

or C
//
) = 7.16
.
10
11
N =
m
w
f
C

,
_


= 29630 cycles
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
Due to extrapolation differences this value is larger than the one via the principal stress range approach.
Data
t
max
= 20 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
= 20 C
R
m
= 410 MPa
R
p0,2/t*
= 265/245 MPa
Used: R
p0,2/t*
= 245 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 21,5mm

D (5 10
6
cycles)
= 52 MPa (class 71)
equivalent stresses or principal stresses
m = 3 C = m = 3 C

= 7.16
.
10
11
C
//
= ..
m = 5 C =

m = 5 C

= ..
C
//
= ..
Stresses
Critical point: Weld toe at inside of shell: Node 1438

struc
= 234,2 MPa (structural equivalent stress range)
(obtained by quadratic extrapolation on inside of shell)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..
= k
e

struc
= .. MPa
Else

=
struc
= 324.2 MPa
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0.639
Page
7.91 (A)
Analysis Details
Example 1.4 / F-Check
DBA
Design by Analysis
Welded region / Principal stress range approach / critical point: node 1438
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 324.2 MPa
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
Page
7.92 (A)
Analysis Details
Example 1.4 / F-Check
DBA
Design by Analysis
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = 7.16
.
10
11
N =
m
w
f
C

,
_


= 21010 cycles
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
Analysis Details
Example 2
Page
7.93 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using Elastic Compensation
FE-Software: ANSYS 5.4
Element Types: 4 node, 3-D isoparametric shells.
Boundary Conditions:
Symmetry boundary conditions along the vertical cutting plane of
the model.
All degrees of freedom at open end of small cylinder have
displacements constrained to 0.
Model and Mesh:
Results:
All load conditions admissible according to checks against GPD and PD.
Member:
Strathclyde
Number of elements - 3570
Analysis Details
Example 2
Page
7.94 (S)
DBA
Design by Analysis
Figure 7.6.1-1: Equivalent Stress for Wind Loading.
1. Finite Element Mesh
The geometry model was constructed according to the problem specification. Half symmetry was
used with finite element models created using 3570 low order 4-node 3-D shell elements.
Calculations to check the admissibility of the defined load cases were carried out according to the
GPD- and PD-check rules in the code.
As half symmetry was utilised, symmetry boundary conditions were applied to the model along the
vertical cutting plane. The nodes at the open end of the narrow cylinder (bottom of storage tank)
have all their degrees of freedom constrained to zero displacement (fully fixed). When internal
pressure is applied, a corresponding thrust is applied to the top open edge of the elements of the
thick cylinder to model the closed end condition. The mass of the top cover and insulation of the
top cover are also applied as a
pressure over this edge.
Hydrostatic pressure is
modelled by applying a
pressure gradient over the
internal surface of the cylinder
corresponding to gh. The
wind loading is calculated as a
total wind force for each
section of the tank and
distributed evenly over the
nodes in that section of the
tank (in the direction of the
wind). Figure 7.6.1-1 shows
the resulting Mises equivalent
stress distribution resulting
from the wind load. Dead
weight is applied by defining a
density that includes the
insulation and density of the
material. Acceleration equal
to gravity is then applied.
2. Material properties:
Shell (X6Cr Ni Ti 18-10): material strength parameter MPa RM 224 = , modus of elasticity
, 193GPa E = coefficient of linear thermal expansion . / 1 10 4 . 16
6
K

=
ring (P235 GH): material strength parameter , 202MPa RM = modulus of elasticity
, 209GPa E = coefficient of linear thermal expansion . / 1 10 2 . 12
6
K

=
3. Admissiblity check against to GPD.
Using the application rule in prEN-13445-3 Annex B.9.2.2 to check against GPD, the principle is
fulfilled when for any load case the combination of the design actions do not exceed the design
resistance. The analysis was carried out using the elastic compensation method conforming to the
direct route rules for GPD in Annex B: linear elastic ideal - plastic material law, Trescas yield
condition and associated flow rule and first order theory. For shell elements it is not currently
Analysis Details
Example 2
Page
7.95 (S)
DBA
Design by Analysis
Figure 7.6.1-2: Equivalent stress for dead weight
possible to calculate limit loads based directly on Trescas condition from elastic compensation.
Models utilising shells have only one element through thickness. Instead of carrying out the
analysis using a Tresca or Mises model directly, a generalised yield model is used which considers
the element's thickness. In elastic compensation, the Ilyushin generalised yield model is used in the
calculation of limit stress fields. Ilyushin's model is based Mises' condition, the limit load will
require correction to meet the code rules on the use of the Tresca condition. The maximum ratio of
Mises' equivalent stress to Tresca's equivalent stress for the same load is 2/3. Therefore, applying a
factor of 3/2 to the yield stress in the Mises analysis will always lead to a conservative result for
the Tresca condition.
Admissibility checks are required for the following three load cases:
(1) Hydrostatic pressure at maximum medium level, dead weight, wind load.
(2) Hydrostatic pressure at minimum medium level, draining pressure, dead weight and wind load.
(3) Draining pressure, dead weight and wind load
The full model is used in the check against GPD for the above three load cases. Materials defined
for the analysis have proof strengths of 224 MPa and 202 MPa for the shell and ring respectively.
From prEN-13445-3 Annex B, Table B.9-3, the partial safety factor
R
on the resistance is 1.18 and
1.25 for the shell and ring respectively. The design material strength parameter used in the
calculations are given by applying the partial safety factors and Mises' correction, i.e. the design
material strength parameter for the shell is 164.4 MPa and for the ring it is 140 MPa.
For the loading, according to prEN-13445-3 Table 5.B.9.2 the partial safety factors are
Hydrostatic pressure (pressure with natural limit)
p
= 1.0
Draining pressure (pressure without natural limit)
p
= 1.2
Dead weight (action with unfavourable effect)
G
= 1.35
Wind load (variable action)
G
= 1.0
In the specification, the weight of
the insulation is 220 N/m
2
(weight
per unit surface area) or 110 kg/m
2
.
To apply this as a density to the
model, the value has to be divided
by the thickness of the shell. The
resulting densities for the
insulation are then added to those
of the steel, 7930 kg/m
3
for the
shell and 7850 kg/m
3
for the ring.
A pressure equivalent to the weight
of the roof is applied over the top
edge of the shell. The partial
safety factor for the dead weight is
applied to the densities and roof
weight for use in the model. The
equivalent stress distribution for
the dead weight only is shown in
Figure 7.6.1-2.
Analysis Details
Example 2
Page
7.96 (S)
DBA
Design by Analysis
Figure 7.6.1-3: Limit Ilyushin field (Load case 1)
Figure 7.6.1-4: Limit Ilyushin field (Load case 2)
Load case 1
For load case 1, hydrostatic pressure at maximum medium level of 19.68 m, dead weight and wind
load corresponding to that defined in the specification are checked against GPD. In checking for
admissibility of the applied loads using
elastic compensation a lower bound
limit stress field must have an Ilyushin
function less than 1 where the Ilyushian
function f(IL) is
Where
e
is the element stress and R
d
the design material strength parameter.
Figure 7.6.1-3 shows the limit Ilyushin
stress field (where the scale is given by
the square of the Ilyushin function) for
the applied loading. The maximum
square of the Ilyushin function of 0.33
is less than 1, and therefore, the loading
is below the limit load and the load case
is admissible according to the GPD-
check.
Load case 2
For load case 2 the actions are:
hydrostatic pressure at minimum
medium level, draining pressure, dead
weight and wind load. Draining
pressure is 0.06 MPa with a partial
safety factor of 1.2, giving a design
value for the pressure action of A
pd
=
0.072. Figure 7.6.1-4 shows the limit
Ilyushin stress field for load case 2.
The maximum square of the Ilyushin
function of 0.73 is less than the 1
therefore the action is below the limit.
Admissibility for load case 2 is shown.
d
e
R
IL f

= ) (
Analysis Details
Example 2
Page
7.97 (S)
DBA
Design by Analysis
Figure 7.6.1-5: Limit Ilyushin field (Load case 3)
Figure 7.6.1-6: Equivalent (thermal) stress distribution.
Load case 3
For load case 3 the actions are:
draining pressure, dead weight and
wind load. Figure 7.6.1-5 shows
the limit Ilyushin stress field for
load case 3. The maximum square
of the Ilyushin function of 0.73 is
less than the 1, and therefore, the
action is below the limit. The
loading is very similar to that in
load case 2, as can be seen from
the resulting stress plot. Thus, the
admissibility for load case 3 is
shown.
4. Admissibility check against PD
In the check for progressive plastic deformation, the principle in prEN-13445-3 B.9.3.1 is fulfilled
if the structure can be shown to shake down.
In the check against progressive plastic deformation the thermal stresses have to be considered.
Thermal stresses arise in this
problem due to the different
thermal expansion coefficients of
the shell and support ring. Figure
7.6.1-6 shows the Mises
equivalent (thermal) stress
distribution resulting from the
different thermal expansion at a
temperature of 60
o
C.
For the PD-check there are no
partial safety factors on the
material strength, giving design
strengths of 224 MPa and 202
MPa for the shell and reinforcing
ring respectively.
Checks against PD are for the
load cases defined above in the
GPD-check. Since there are no
partial safety factors on the
actions in the check against PD, the results for each of the load cases is found by superposition of
the thermal result on the load cases defined above.
Analysis Details
Example 2
Page
7.98 (S)
DBA
Design by Analysis
Figure 7.6.1-7: Elastic Equivalent Stress for Load Case 1
Figure 7.6.1-8: Elastic Equivalent Stress for Load Case 2
Figures 7.6.1-7 through 7.6.1-9 show
the elastic equivalent stress
distribution for the three load cases as
follows.
Figure 7.6.1-7 Load case 1. The
maximum equivalent stress in the
shell is 225 MPa, approximately the
upper limit of the elastic range for the
material. As the model does not pass
into the plastic range, admissibility
against PD of load case 1 is shown.
Figure 6.7.1-8 Load case 2. The
maximum equivalent stress in the
model is 196.3 MPa. The
maximum stress is within the
elastic range, proving the
admissibility of load case 2 against
PD.
Analysis Details
Example 2
Page
7.99 (S)
DBA
Design by Analysis
Figure 7.6.1-9: Elastic Equivalent Stress for Load Case 3
Figure 7.6.1-9 Load case 3. As
with the check against GPD, the
resulting stress distribution is
similar to that of load case 2. The
maximum stress in the model is
196.3 MPa, within the elastic
range of the materials. Therefore,
the admissibility in the check
against PD for load case 3 is
shown.
Analysis Details
Example 2
Page
7.100 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using elasto-plastic FE calculations
FE-Software: ANSYS 5.4
Element Types: 4-node, 2-D axisymmetric solid PLANE42 for axisymmetric model
and 8-node, 3-D shell elements for 3-D model used for the wind
load calculation.
Boundary Conditions: Axisymmetric model: No vertical displacement in the nodes at
the lower end of the model;
3-D model: Symmetry boundary conditions in the symmetry
plane of the structure; no vertical displacement in the nodes at the
lower end of the model; rigid region concerning vertical
displacements at the upper end of the model to apply the moment
caused by the wind action; vertical displacement set to 0 for
dummy end of beam.
Model and Mesh:
Results: Actions and action cycles, as given in the specification, are admissible according
to the GPD- and SD-checks.
Member:
A&AB
Axialsymmetric model: Height of upper and lower cylinder 500 mm
Total number of elements: 3965;
3-D model: Height of upper cylinder 1500 mm, height of lower cylinder 500 mm
Total number of elements: 5361.
Analysis Details
Example 2
Page
7.101 (S)
DBA
Design by Analysis
1. Elements, mesh fineness, boundary conditions
Since the wind action, a variable action according to EN 13445-3 Annex B.7.1, had to be taken into
account when performing the checks against GPD and PD, a 3-D model representing the interesting
part of the structure, i.e. the cone-cylinder intersections, has been used. Under usage of this model,
the linear-elastic stress corresponding to the moment caused by the wind action was determined and
further used in all other calculations, which were carried out under usage of an axisymmetrical
model. In elasto-plastic calculations, the yield strengths of the linear-elastic ideal-plastic material
law used in the calculations were the (design) material strength parameters decreased by the value
of the maximum equivalent linear-elastic stress caused by the wind action. Since this latter value is
small, this conservative approach could be used.
The 3D-model consisted of a total number of 3560 8-node shell elements SHELL93 and one elastic
beam element BEAM4 (which is necessary to create a rigid region). The two cone-cylinder
intersections with the stiffener ring at the narrow end of the cone and the adjacent lower cylindrical
parts with a length of 500 mm and the adjacent upper cylindrical part with a length of 1500 mm
have been modelled. Because of the symmetry of the structure and the action, only one half of the
structure was considered. To investigate the influence of the length of the cylindrical part at the
upper end on the stress results, calculations with different lengths were performed (always using the
same moment value). The (stress) results were the same for shell lengths of 1500 and 2500 mm, and
about 10 % smaller for a length of 500 mm. Therefore, the model with an upper cylindrical shell
length of 1500 mm was used.
The vertical displacements in the nodes in the lower end of the model were constrained to zero
(corresponding to an undisturbed membrane stress state), and a symmetry boundary condition was
applied to all nodes in the symmetry plane of the structure. To apply the moment, a beam element
(3D elastic beam BEAM4 in ANSYS 5.4) was attached to the structure at the centre of the
structures upper end. To transfer the moment from the beam to the structures model , the nodes in
the upper cross-section of the structure (slave nodes) have been rigidly linked to the lower node of
the beam as master node (under usage of the rigid region command in ANSYS, where the
vertical displacement was associated to the corresponding constraint equations). To obtain a stable
model, the vertical displacement of the upper (dummy) end of the beam element was constrained to
zero.
For the axisymmetric model of the structure a total number of 3965 4-node axisymmetric elements,
PLANE42 in ANSYS 5.4, were used. As boundary condition the vertical displacement in the
nodes at the lower end of the model were constrained to zero. According to the specification,
different material properties were used for the stiffener ring (and the corresponding welds) and the
shell. The welds between stiffener ring and shell were modelled with fillets according to weld
influence zone of about 2 mm (a corresponding fillet radius of 5 mm was used). To avoid stress
singularities, the shell welds at the cone cylinder junctions were modelled with appropriate fillet
radii.
Analysis Details
Example 2
Page
7.102 (S)
DBA
Design by Analysis
2. Admissibility check against GPD
The admissibility checks have to be carried out for the following 3 load cases:
LC1: Wind, dead weight and hydrostatic pressure according to maximum medium level
LC2: Wind, dead weight, hydrostatic pressure according to minimum medium level and
draining pressure
LC3: Wind, dead weight and draining pressure
Note: The admissibility of the load case wind, dead weight and hydrostatic pressure corresponding
to minimum medium level is shown by the admissibility of LC1.
To avoid the usage of a computation time intensive 3-D model in the elasto-plastic calculations, the
maximum linear-elastic equivalent stresses in the shell and the stiffener ring due to the wind action
were calculated using the 3-D shell model. Furthermore, the GPD checks were performed with the
axisymmetrical model without application of the wind action (which would not be possible within
this model) and with decreased design material strength parameters given by the difference of the
original design material strength parameters minus the maximum linear-elastic equivalent stresses
in the shell and the stiffener ring due to the wind action. This procedure is admissible due to the
positive definiteness of the equivalent stress: The sum of the maximum equivalent stresses of two
stress tensors is greater or equal to the maximum equivalent stress of the sum of the two stress
tensors.
The elastic moduli used in the models are given by 193 GPa for the shell and 209.5 GPa for the
stiffener ring, respectively.
The moment, which has to be applied to the upper cross-section of the 3-D model and which
corresponds to the wind action on the vessel above the wide end of the cone, was calculated with a
partial safety factor of 0 . 1 =
Q
, since the wind action is specified in the specification is as a limit
value (see the specification of the example):

= = =
i i i W ges i i i W i i W
a h q d c a A q c a W M
, ,
,
where
i W
q
,
is the wind stagnation pressure depending on height h (specified)
: 6 0 m h m
2
/ 81 . 0 m kN q
W
=
: 10 6 m h m <
2
/ 88 . 0 m kN q
W
=
: 15 10 m h m <
2
/ 94 . 0 m kN q
W
=
: 25 15 m h m <
2
/ 02 . 1 m kN q
W
=
i
W the wind force on section i:
i i W i
A q c W =
,
c the drag coefficient 44 . 0 = c (specified)
i
A the projection of the surface of the vessel in wind direction:
i ges i
h d A =
i
a the vertical distance of the resultant wind force of
i
W from the wide end of the cone,
and
ges
d the vessel diameter including insulation, mm d
ges
6900 = .
Analysis Details
Example 2
Page
7.103 (S)
DBA
Design by Analysis
Therefore, the moment corresponding to the wind and with respect to the centre point of the cross
section at the cones wide end is given by MNmm M
W
4 . 356 = . Since only half of the structure was
modelled, half of this moment was applied in the FE-model.
Figure 7.6.2-1 shows the
Mises equivalent stress
distribution in the shell
model according to the
wind action. The maxima
are given by MPa 9 . 26 in
the shell at the wide end
of the cone and by about
MPa 9 in the ring.
Figure 7.6.2-1
Since the subroutine using Trecass yield condition showed bad convergence, all elasto-plastic
calculations were carried out using Mises yield condition and associated flow rule, a linear-elastic
ideal-plastic material law, first order theory; under usage of the Newton-Raphson method, and
scaled down material strength parameters with the factor 2 / 3 (see subsection 3.2 of section 3 -
Procedure). The partial safety factors according prEN 13445-3 Annex B, Table B.9.3, are given by
18 . 1 =
R
for the shell material and 25 . 1 =
R
for the ring material. Therefore, the design material
strength parameters used in the calculations are given by
MPa R
d
4 . 137 9 . 26
2
3
18 . 1
224
= = for the shell, and
MPa R
d
5 . 131 9
2
3
25 . 1
8 . 202
= = for the ring.
where RM = 224 MPa is the material strength parameter of the shell according to prEN 10028-7 and
RM = 202.8 MPa the material strength parameter of the ring according to EN 10028-2, both
determined for the calculation temperature of the vessel.
Analysis Details
Example 2
Page
7.104 (S)
DBA
Design by Analysis
According to prEN 13445-3 Annex B, Table B.9.2, the partial safety factor for the hydrostatic
pressure is given by 0 . 1
,
=
hydro P
, the partial safety factor for the draining pressure by 2 . 1
,
=
drain P
,
and the partial safety factor for the dead weight by 35 . 1 =
G
.
For the application of the dead weight, densities of
3
/ 7930 m kg for the shell material,
3
/ 7850 m kg
for the ring material, and
3
/ 110 m kg for the insulation were used (the densities used in the input file
were the sum of the densities of the corresponding material and the insulation multiplied by the
factor 35 . 1 =
G
). The dead weight of the part of the structure above the modelled part was
calculated to be MN 239 ; a
corresponding pressure of MPa 96 . 1
multiplied by the factor 35 . 1 =
G
was
applied in the FE- model.
The internal (draining) pressure applied
in the FE-model is given by the product
MPa 072 . 0 2 . 1 06 . 0 = ; a
corresponding longitudinal stress was
applied at the upper end of the model.
Figure 7.6.2-2 shows the Mises
equivalent stress distribution for LC1.
The corresponding maximum absolute
value of the principal strains is 0.09 %,
and, therefore, the admissibility of this
load case according to the GPD-check
is shown.
Figure 7.6.2-2
Figure 7.6.2-3 shows the Mises
equivalent stress distribution for LC2.
The corresponding maximum absolute
value of the principal strains is 0.32 %,
and, therefore, the admissibility of this
load case according to the GPD-check
is shown.
Figure 7.6.2-3
Analysis Details
Example 2
Page
7.105 (S)
DBA
Design by Analysis
Figure 7.6.2-4 shows the
Mises equivalent stress
distribution for load
case LC3. The
corresponding
maximum of the
absolute values of the
principal strains is 0.32
%, and, therefore, the
admissibility of this load
case according to the
GPD-check is shown.
The computation time of
the elasto-plastic
solutions were about 7
minutes on the
Compaq Professional
Workstation 5000 with
two Pentium Pro
processors and 256 MB
RAM for every state.
Figure 7.6.2-4
3. Admissibility check against PD
In the following, it is shown that Melans shakedown theorem is fulfilled for all load cases under
consideration, i.e. the structure shakes down, and, therefore, progressive plastic deformation cannot
occur.
The following load cases have to be considered, including the thermal stress according to the
different thermal expansion coefficients of the shell and the ring material.
LC1: Wind, dead weight and hydrostatic pressure according to maximum medium level
LC2: Wind, dead weight, hydrostatic pressure according to minimum medium level and
draining pressure
LC3: Wind, dead weight and draining pressure.
Again, to avoid the usage of a computation time intensive 3-D model in the calculations, the
maximum linear-elastic equivalent stresses in the shell and the stiffener ring due to the windaction,
calculated using the 3-D shell model, were used. Therefore, the calculations for the shakedown
check were carried out without the wind action, using the axisymmetrical model, and with
decreased material strength parameters, given by the difference of the original material strength
parameters and the value of the maximum linear-elastic equivalent stresses in the shell and the
stiffener ring due to the wind action. Thus, the reduced (design) material strength parameters used
in the shakedown check are given by
Analysis Details
Example 2
Page
7.106 (S)
DBA
Design by Analysis
MPa RM 1 . 197 9 . 26 224 = = for the shell, and
MPa RM 8 . 193 9 8 . 202 = = for the ring.
The thermal expansion
coefficients to be used
in the calculation, are
K 1 10 4 . 16
6
= for
the shell material, and
K 1 10 2 . 12
6
= for
the ring material.
Figure 7.6.2-5 shows
the linear-elastic
(thermal) Mises
equivalent stress
distribution according
to the different thermal
expansion coefficients
at the temperature
60C.
Figure 7.6.2-5
The following figures show the linear-elastic Mises equivalent stress distributions of the
considered load cases: Figure 7.6.2-6 shows results for LC1, and Figure 7.6.2-7 for LC3. The
results of LC2 are not shown since they are practically the same as for LC3. The results for all these
load cases were calculated by superposition of the load cases for the different loads.
Analysis Details
Example 2
Page
7.107 (S)
DBA
Design by Analysis
Figure 7.6.2-6
Figure 7.6.2-7
Analysis Details
Example 2
Page
7.108 (S)
DBA
Design by Analysis
As can be seen from these figures, the behaviour of the structure is completely elastic for LC1, but
in LC2 and in LC3 the maximum Mises equivalent stress is larger than the used material strength
parameter, and, therefore, some plastification will occur.
From the behaviour of the structure it was concluded, that the stresses due to the draining pressure
were the reason for the plastification. Therefore, a self-equilibrating stress field based on internal
pressure solutions was used to show that Melan s shakedown theorem is fulfilled.
An elasto-plastic FE
calculation was
carried out for a
value of the internal
pressure of
MPa 1 . 0 . And the
self-equilibrating
stress field, shown
in Figure 7.6.2-8
was found by the
difference of the
elasto-plastic stress
field minus the
corresponding
linear-elastic stress
field. The
computation time of
the elasto-plastic
solution was 10
minutes on the
Compaq
Professional
Workstation 5000
with two Pentium
Pro processors and 256 MB RAM.
Figure 7.6.2-8
Using this self-equilibrating stress field, Melans theorem could not be fulfilled for LC1, and,
therefore, the self-equilibrating stress field was multiplied with a suitable factor, which was found
to be 0.4. Using this corrected self-equilibrating stress field, Melans theorem was fulfilled for all
load cases under consideration. Figure 7.6.2-9 shows the sum of the corrected self-equilibrating
stress field and the linear-elastic stress field of LC1, and Figure 7.6.2-10 shows the sum
corresponding LC2. The sum of the corrected self-equilibrating stress field and the linear-elastic
stress field of LC3 is not shown since it is practically the same as the sum corresponding to LC2.
Analysis Details
Example 2
Page
7.109 (S)
DBA
Design by Analysis
Figure 7.6.2-9
Figure 7.6.2-10
DBA
Design by Analysis
Analysis Details
Example 2 / F-Check
Page
7.110 (C)
Analysis Type:
F-Check
Member:
CETIM
FE Software: ABAQUS / Standard version 5.8.1
Element Types:
Quadratic axisymmetric 8-node elements .
4743 nodes and 1126 elements
Boundary Conditions:
The base ring of the vessel is fixed in the vertical direction.
Model and Mesh:
Results:
Fatigue life N = 1984 cycles (junction shell/stiffener)
DBA
Design by Analysis
Analysis Details
Example 2 / F-Check
Page
7.111 (C)
1. Introduction:
The calculations performed were linear-elastic ones. In the welded zones, the mesh takes into
account the nodes where the results are extrapolated. A quadratic extrapolation was used, with
the first node at a distance equal to 0,4 e
n
(e
n
= thickness of the shell) from the weld, the second
at 0,9 e
n
of the weld and the third node at 1,4 e
n
of the weld.
2. Material parameters:
X6CrNiTi 18-10 EN 10028-7 for the shell, MPa R MPa R
t p m
224 , 520
/ 0 . 1
.
P235 GH EN 10028-2 for reinforcing, . 8 . 202 , 360
/ 2 . 0
MPa R MPa R
t p m

3. Operating Cycles:
The load cases used in determining the cycles are.
LC 1 : media level = h
max
; internal pressure = 0 ; temperature = 60 C
LC 2 : media level = h
min
; internal pressure = 0 ; temperature = 60 C
LC 3 : media level = h
max
; internal pressure = 0 ; temperature = 60 C
LC 4 : media level = h
min
; internal pressure = 0 ; temperature = 60 C
LC 5 : media level = 0 ; internal pressure = 0,6 bar ; temperature = 60 C
LC 6 : media level = 0 ; internal pressure = 0 ; temperature = 20 C
LC 7 : media level = h
max
; internal pressure = 0 ; temperature = 60 C
Stress range :
The are five (welded) zones of possible interest.
weld 1
weld 2
weld 3
weld 4
weld 5
DBA
Design by Analysis
Analysis Details
Example 2 / F-Check
Page
7.112 (C)
The selection of welds to be investigated in detail was based on the principal stress range in the
steps and the class of the weld. Since only the wide end and the narrow and of the cone are
specified as to be investigated, there remains:
Weld 3 : = 224 MPa class 63
Weld 5 : = 253,1 MPa class 80
With the list of load cases given above the full cycle, relevant at the two welded regions, can be
written as
Weld 3, weld 5: 1 x (LC 6 - LC 3), 99 x (LC 3 LC 2) and 1 x (LC 6 - LC 5)
The structural stress components and the equivalent structural stresses for the load cases and the
load case differences at the two welds were obtained via quadratic extrapolation, and are
Weld 3 (structural stresses obtained with quadratic extrapolation)
steps
11
(MPa)

22
(MPa)

33
(MPa)

Tresca
(MPa)
2 -5,2 -35,8 -34,6 -
3 -37 -203,3 -117,8 -
5 -4,3 -31,6 -33,5 -
6 0 0 0 -
6 - 3 37 203,3 117,8 166,3
3 - 2 -31,8 -167,5 -83,2 157,7
6 - 5 4,3 31,6 33,5 29,2
Weld 5 (structural stresses obtained with quadratic extrapolation)
steps
11
(MPa)

22
(MPa)

33
(MPa)

Tresca
(MPa)
2 0 0 0 -
3 -0,2 -0,4 83,3 -
4 -20,1 276,9 -1,1 -
4 - 3 -19,9 277,3 -84,4 361,7
3 - 2 -0,2 -0,4 83,3 83,5
Since thermal stresses are negligeable in the weld region 5, the load cases used for the two
regions are different.
It is still not obvious which of the two welded regions is the more critical one, both need detailed
investigation, see the following-sheets.
DBA
Design by Analysis
Analysis Details
Example 2 / F-Check
Page
7.113 (C)
Data
t
max
= 60 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
= 50 C
R
m
= 520 MPa
R
p1,0/t*
= 224 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 8 mm

D (5 10
6
cycles)
= 46 MPa (class 63)
equivalent stresses or principal stresses
m = 3 C = 5
.
10
11
m = 3 C

= ..
C
//
= ..
m = 5 C = 1,08
.
10
15
m = 5 C

= ..
C
//
= ..
Stresses

struc
= 116,3 / 135,7 / 29,2 MPa (structural equivalent stress range, determined by extrapolation)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p1,0/t*

1
1
]
1

+ 1
* p1,0/t
R 2
struc

4 , 0 1
e
k
k
e
= ..
= k
e

struc
= .. MPa
Else

=
struc
= 166,3 / 135,7 / 29,2 MPa
Thermal loading
If
struc
> 2 R
p1,0/t*
1
1
]
1

* p1,0/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
=(25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0,639
Weld 3 / Equivalent stress range approach
DBA
Design by Analysis
Analysis Details
Example 2 / F-Check
Page
7.114 (C)
If
w
f

>
5
.
10
6
cycles
then
M = 3 and C (C

or C
//
) = 5
.
10
11
N =
m
w
f
C

,
_


= 108720 / 200090 cycles
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
M = 5 and C (C

or C
//
) =
N =
m
w
f
C

,
_


=
N = since 29.2 <
10
8
cycles
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 166,3 / 135,7 / 29,2 MPa


w
f

=29.2 N =
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,043 4,3 10
-4
t* = ..
Else f
t*
= 1
The complete operating cycle corresponds to 1 cycle for the stress ranges 166,3 MPa and 29,2 MPa and 99
cycles for the stress range 135,7 MPa. The global allowable number of cycles is equal to N with :

+ +
N
200090
N 99
108720
N
= 1
N = 1984 cycles (each cycle is equal to 100 variations between h
max
and h
min
and 1 complete draining
and filling)
DBA
Design by Analysis
Analysis Details
Example 2 / F-Check
Page
7.115 (C)
Data
t
max
=60 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
= 50 C
R
m
= 520 MPa
R
p1,0/t*
= 224 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made in each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 6 mm

D (5 10
6
cycles)
= 59 MPa (class 80)
equivalent stresses or principal stresses
m = 3 C = 1,02
.
10
12
m = 3 C

= ..
C
//
= ..
m = 5 C = 3,56
.
10
15
m = 5 C

= ..
C
//
= ..
Stresses

struc
= 361,7 / 83,5 MPa (structural equivalent stress range, determined by extrapolation)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p1,0/t*

1
1
]
1

+ 1
* p1,0/t
R 2
struc

4 , 0 1
e
k
k
e
= ..
= k
e

struc
= .. MPa
Else =


struc
= 361,7 / 83,5 MPa
Thermal loading
If
struc
> 2 R
p1,0/t*
1
1
]
1

* p1,0/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
=(25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0,639
Weld 5 / Equivalent stress range approach
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 361,7 / 83,5 MPa


18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,043 4,3 10
-4
t* = ..
Else f
t*
= 1
If
w
f

>
5
.
10
6
cycles
then
M = 3 and C (C

or C
//
) = 1,02
.
10
12
N =
m
w
f
C

,
_


= 21555 / 1752000 cycles
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
M = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= .. cycles
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
The complete operating cycle corresponds to 1 cycle for the stress ranges 361,9 MPa and 99 cycles for the
stress range 83,7 MPa. The global allowable number of cycles is equal to N with :
175200
N 99
21555
N
+ = 1
N = 9718 cycles (each cycle is equal to 100 variations between h
max
and h
min
and 1 complete draining
and filling)
DBA
Design by Analysis
Analysis Details
Example 2 / F-Check
Page
7.116 (C)
Analysis Details
Example 2
Page
7.117(A)
DBA
Design by Analysis
Stability check of ring at small end of cone:
Load case: Self weight of vessel, vessel fully filled, wind:
Maximum radial force resultant at inner ring surface: (diameter 4000 mm), determined
from FE-analyses (Direct Route using elasto-plastic FE, corresponding linear-elastic
calculation):
mm N q
r
/ 99 =
This value corresponds to the characteristic value of the actions.
The partial safety factors for the actions are
35 . 1 =
G
for self weight
0 . 1
,
=
hydro p
for the hydrostatic pressure due to filling
0 . 1 =
a
for the wind action,
see pages 7.104.
Since the force (per unit length) due to the filling weight is much larger than the forces
due to wind moment and self weight, see page 7.120, the characteristic value of the
effective radial force given above is used also as design value, i. e. 0 . 1 =
a
is used for
all actions in the following. If a more accurate result is required, the contribution of self
weight, given by
sw
n , could be multiplied by 35 . 1 =
G
.
This results in a design value of the effective radial force of
mm N q q
n n n
n n n
q
re re
sw w f
sw G w f
d re
/ 6 . 97 023 . 1
,
= =
+ +
+ +
=

with values for
sw w f
n n n , , of page 7.120.
Effective radial force at center of gravity of ring (diameter 4150)
mm N q q
r re
/ 4 . 95 4150 / 4000 = =
Allowable radial force (per unit length) acc. to Section 8 of prEN 13445-3, where

sH H r
L p q to s correspond :
Eq. (8.4.3-48):
3
/ 3 R I E L p
eH SH H
= .
For simplicity the ring only is considered:

4 6
10 . 4375 . 8 , 2000 mm I mm R
s
= = :
Analysis Details
Example 2
Page
7.118(A)
DBA
Design by Analysis
mm N L p
SH H
/ 66 . 610 =
Safety factor according to 8.2.3 : k = 1.5
Coefficient for hot bent stiffener according to 8.4.3.5a : 2 . 1 =
f
k
Allowable radial force:
mm N k k L p q
f sH H ra
/ 3 . 339 ) /( = = :
Conclusion:
o.k. check 95
,
= >
d re ra
q q
Stability check of small cylindrical shell:
Maximum axial force resultant (per unit length):
Load case: Vessel fully filled, self weight of vessel and wind:
, / 0 . 328
max
mm N n n n n
f sw w
= + + =
see SE-check, page 7.120.
Allowable axial stress resultant (per unit length):
Roarks Formulas for Stress and Strain, 6
th
edition, Table 35 Case 15:
Theoretical value:
mm N v r Ee n
ath
/ 3745 )) 1 ( 3 ( / ) / (
5 . 0 2 2
= =
Practical value:
mm N n
pr
/ 1498 =
with reduction (knock-down) factor 0.4 (see Roark)
With a safety factor 5 . 1 = k , assuming that the pressure test cannot be performed with the
required test pressure, the allowable axial stress resultant
mm N n
a
/ 8 . 998 =
results. There follows:
Allowable value > Maximum value:
> 0 . 328 8 . 998 check o. k.
The civil engineering standard DIN 18800 gives, for the boundary conditions
shell clamped at lower end
shell radial displacement restrained (to zero) at upper end,
a theoretical value of
Analysis Details
Example 2
Page
7.119(A)
DBA
Design by Analysis
mm N n
ath
/ 3763 =
and a reduction factor of 0.544, and, thus, a reduced (practical) value of
mm N n
pr a
/ 2048
,
=
With a partial safety factor of 5 . 1 =
I
, assuming that the pressure test cannot be
performed with the required test pressure, an allowable axial force per unit length of
mm N n
dr a
/ 1366
,
=
results; thus, the conclusion is, that the allowable (design resistance) value is larger than
the (design value) of the maximum axial force (per unit length),
o.k check 328 1366
max ,
= > = n n
dr a
.
Analysis Details
Example 2
Page
7.120(A)
DBA
Design by Analysis
Moment due to wind with respect to centre of base:
MNmm M
W
95 . 582 =
Corresponding maximum axial force resultant (per unit length, in shell mm d
m
3992 = ):
mm N d M n
m w w
/ 58 . 46 ) /( 4
2
= =
Total weight of vessel (exclusive base plate):
kN F
sw
44 . 269 =
Corresponding axial force resultant:
mm N d F n
m sw sw
/ 29 . 21 ) /( = =
Weight of filling above conical shell:
MN F
f
263 . 3 =
Corresponding axial force resultant:
mm N n
f
/ 17 . 260 =
Check against overturning:
Load case: Vessel empty, self weight of vessel, and wind:
> = + 0 / 73 . 24 mm N n n
sw w
Bolting required
Bolt force for 8 bolts:
kN d n n F
m sw w B
8 . 38 8 / ) ( = + =
Check of bolt load:
Load case: Vessel empty, self weight of vessel, and wind:
N F
B
8 . 38 =
Load case: Vessel empty, self weight of vessel, internal pressure and wind:
Axial force resultant for internal pressure:
mm N d p n
i p
/ 76 . 59 4 / = =
Maximum bolt force for 8 bolts:
kN d n n n F
m p sw w B
3 . 133 8 / ) ( = + + =
Analysis Details
Example 2
Page
7.121(A)
DBA
Design by Analysis
Check of maximal pressure under base ring:
Load case: Self weight of vessel, vessel fully filled and wind:
Maximal pressure under base ring (minimum pretensioning of bolts, stiff foundation,
triangular pressure distribution):
MPa n n n p
f sw w b
86 . 5 ) 3864 4120 ( / ) 2 ( 4 = + + =
Check against displacement:
Load case: Vessel empty, self weight of vessel and wind
Horizontal wind force (at base):
kN F
w
68 . 52 =
Friction force:
kN F F
sw F
8 . 80 = =
with friction coefficient 3 . 0 =
o.k. check >
w F
F F ; bolting required only because of wind moment.
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.122 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using Elastic Compensation
FE-Software: ANSYS 5.4
Element Types: 8 node, 3-D isoparametric shells.
Boundary Conditions: Symmetry boundary conditions in the plane along the longitudinal
direction of the shell.
Hoop displacements in nodes at both ends of the shell constrained
to zero.
Longitudinal displacements in the nodes at one end of the shell
constrained to zero, longitudinal displacements coupled in the
nodes at the other end of the shell (plane sections remain plane).
A rigid region is set up on the top surface nodes of the nozzle to
apply the bending moment.
Model and Mesh:
Results:
Maximum admissible action according to the GPD-Check: Internal pressure PS
max GPD
= 0.467 MPa
(for the constant moment of 15644.4 Nm)
Check against PD: Shakedown limit pressure PS
max SD
= 1.34 MPa
Member:
Strathclyde
Number of elements - 576
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.123 (S)
DBA
Design by Analysis
1. Finite Element Mesh
Element divisions for the finite element mesh were defined parametrically to allow modification of
the element density. Half symmetry was used with finite element models created using high order
8-node 3-D shell elements. The allowable pressure according to GPD and the shakedown pressure
according to PD were calculated for this model.
Boundary conditions applied to the model are symmetry about the xz plane therefore the applied
moment is half the specified value. The nodes at both ends of the main cylinder are constrained in
the hoop direction but are allowed to move radially. The nodes at one end of the main cylinder
were constrained fully in the longitudinal direction with the nodes at the other end coupled in the
longitudinal direction, thus maintaining plane sections remaining plane. Pressure was applied over
the whole internal structure with the equivalent thrust that would be produced from closed ends
applied at the longitudinally coupled degree of freedom end of the cylinder. The moment applied to
the intersecting nozzle was modelled by creating a rigid region over the top edge of the nozzle via
constraint equations. A mass element was created in the centre at the top of the nozzle (to act as the
master node to which the moment is applied) with the nodes on the top edge of the nozzle (slave
nodes) connected to it by constraint equations. The moment applied to the master node would then
be transmitted by the rigid constraint region evenly to the slave nodes on the nozzle. This results in
the application of a pure moment to the top surface of the nozzle with no warping, the cross-section
is forced to remain plane.
2. Material properties
Material strength parameter MPa RM 272 = , modulus of elasticity GPa E 125 . 210 =
3. Determination of the maximum admissible pressure according to the GPD-Check
Using the application rule in prEN 13445-3, Annex B.9.2.2 to check against GPD the principle is
fulfilled when for any load case the combination of the design actions does not exceed the design
resistance. This may be shown by calculating the limit load. The limit load has to be determined
using Trescas yield condition and associated flow rule. However, for shell elements the limit load
from elastic compensation is calculated using a generalised yield criterion based on Mises'.
From prEN-13445-3 Annex B, Table B.9-3 the partial safety factor
R
on the resistance is 1.25.
Therefore, the design material strength parameter is given by RM/
R
= 217.6 MPa. The analysis
was carried out using the elastic compensation method conforming to the direct route rules for GPD
in Annex B: linear elastic - ideal plastic material law, Trescas yield condition and associated flow
rule and first order theory. For shell elements it is currently not possible to calculate limit loads
based directly on Trescas condition from elastic compensation. Models utilising shells have only
one element through thickness. Therefore, it is not possible to carry out an elastic compensation
analysis in the same way as in solid elements. Instead of carrying out the analysis using a Tresca
or Mises model directly, a generalised yield model was used which considers the elements
thickness. In elastic compensation, the Ilyushin generalised yield model is used in the limit load
calculation. Ilyushin's model is based upon Mises' condition, the limit load will require correction
to meet the code rules on the use of the Tresca condition. The maximum ratio of Mises' equivalent
stress to Tresca's equivalent stress for the same load is 2/3. Therefore, applying a factor of 3/2 to
the yield stress in the Mises analysis (or to the limit load, as the analysis is elastic) will always lead
to a conservative result on the Tresca condition. In effect the Mises yield locus is reduced to fit
within the Tresca yield locus.
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.124 (S)
DBA
Design by Analysis
Pressure-Moment Limit Locus
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 5000 10000 15000 20000 25000 30000
Moment (Nm)
P
r
e
s
s
u
r
e

(
M
P
a
)
A
Pd
=0.56 MPa
A
Md
=21120 Nm
Figure 7.7.1-1: Pressure Moment Limit Locus
In each elastic compensation iteration an equilibrium stress field is produced where the elastic
modulus of each subsequent iteration is defined by the previous elastic solution based upon the
Ilyushin generalised yield model. In this way regions of the FE-model may be systematically
stiffened or weakened depending upon the stress magnitudes in the previous solution. The result is
that every equilibrium stress field is a lower bound of the limit load, the stress field giving the
greatest limit load is taken as the closest to the actual limit state and defines the limit load in the
analysis. Convergence of the equilibrium stress field will occur typically between 8 and 15
iterations, i.e. no further benefit in the limit load will be noted with subsequent iterations.
In the check against GPD, the loading considered is a constant moment and increasing pressure, i.e.
the limit pressure is to be calculated. As elastic compensation is based on linear-elastic analysis, it
is not possible to calculate the limit load directly where there are two separate actions with one
constant. The elastic compensation procedure would scale the combined load stress field and would
therefore scale both sets of loads. However, by carrying out a series of elastic compensation
analyses where the ratio of the load sets are altered for each analysis a limit locus can be
constructed. The limit locus will then describe the limit loads for any ratio of the load sets.
Correction of the analysis based on Mises' condition was applied to the design material strength
parameter to give a corrected value of = ) 2 / 3 ( ) 6 . 217 ( 188.5 MPa.
The total computing time to run each analysis on a 300 MHz Pentium two processor with 128 Mb
RAM Windows NT workstation was 750 seconds. The stress fields were shown to converge after
fourteen iterations, eight separate analyses were carried out to describe the limit locus. Figure
7.7.1-1 shows the pressure moment limit locus from which the design pressure is found. According
to prEN-13445-3 Annex B,
Table B.9-2 for permanent
actions with an unfavourable
effect, e. g. the moment, the
partial safety factor
G
is
1.35. Therefore, the design
value for the moment action,
A
Md
, is the allowable
moment (defined in the
specification) multiplied by

G
giving A
Md
=21120 Nm.
From Figure 7.7.1-1, for a
design moment action of
21120 Nm a design value for
the pressure action, A
Pd
, of
0.56 MPa results. For a
pressure without natural limit
the partial safety factor
p,
given by prEN 13445-3, Table B.9-2, is 1.2. Thus the allowable pressure
is given as
By carrying out a final elastic compensation analysis using the calculated value of design pressure,
a check of the result can be made. Figure 7.7.1-2 shows the limit stress field based on the Ilyushin
function at the design actions. The contour units are dimensionless and are termed Ilyushin
MPa PS
p
GPD
467 . 0
2 . 1
56 . 0 56 . 0
max
= = =

Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.125 (S)
DBA
Design by Analysis
Figure 7.7.1-2: Ilyushin Function Limit Field
Figure 7.7.1-3: Residual Ilyushin Field
function. The Ilyusin function
represents the ratio of actual
equivalent stress to the yield
stress of the material, i.e.
Where f(IL) is the Ilyushin
function,
e
is the element
equivalent stress, and
Y
the
yield stress of the design
material (or in this analysis the
design material strength
parameter). Therefore, for the
applied loading to be a lower
bound on the limit load the
Ilyushin function anywhere in
the redistributed limit field
cannot exceed 1. In Figure
7.7.1-2 it can be seen, that the
maximum square of the Ilyushin function is 1.001. Therefore, for the given moment, the allowable
pressure loading of 0.467 MPa fulfils the check against GPD. The plot also shows the extent of
plasticity on and around the nozzle at the limit.
4. Check against PD
In the check against progressive
plastic deformation, the principle
in prEN-13445-3 B.9.3.1 is
fulfilled if the structure can be
shown to shake down.
In elastic compensation the load
at which the structure will shake
down is simple to calculate.
Based on Melans shakedown
theorem, the self-equilibrating
residual stress field that would
result after a loading cycle can be
calculated by subtracting the
linear-elastic stress field at the
limit pressure from the limit
stress field. The residual stress
field is in effect the resulting
stress from an elastic unloading
from the limit state back to zero
pressure. If no stress in the
Y
e
IL f

= ) (
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.126 (S)
DBA
Design by Analysis
Pressure-Moment Shakedown Locus
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0 10000 20000 30000 40000 50000
Moment (Nm)
P
r
e
s
s
u
r
e

(
M
P
a
)
M=15644.4 Nm
P =1.34 MPa
SH
Figure 7.7.1-4: Pressure-Moment Shakedown Locus
residual field violates the yield condition, i.e. if there is no stress above the design material strength
parameter, then the shakedown load is equal to the limit load. Where the residual stress field of a
structure does exceed the yield condition, the shakedown limit can be calculated easily from the
stress plots because of the linearity.
The residual stress field
from the elastic
compensation analysis
using the Ilyushin
function is shown in
Figure 7.7.1-3. It can be
seen that the maximum
residual Ilyushin function
is 0.388 which is less than
the yield condition of 1.
Therefore, there is no
equivalent stress greater
than the design material
strength parameter and the
shakedown load is equal
to the limit load. As with
the check against GPD, a
shakedown locus can be created covering all of the pressure-moment combinations. Due to the
linearity, the limit locus in Figure 7.7.1-1 can be scaled up to meet the PD rules: Mises' condition
and no partial safety factor on the material strength. Figure 7.7.1-4 shows the shakedown pressure-
moment locus. As there are no partial safety factors applied to the actions for the PD-check, the
applied moment is 15644.4 Nm, giving a maximum shakedown pressure of
PS
maxSD
= 1.34 MPa.
This load gives a maximum stress at the discontinuity of 1124 MPa, which is greater than the 2 RM
placed on the shakedown load as defined in the code. The equivalent stress near to (but not on) the
discontinuity is 628 MPa, which is also greater than the 2 RM-limit. The value of the shakedown
load calculated by elastic compensation using the generalised yield criterion is non-conservative
and may not be used to give a lower bound shakedown load. Therefore, the use of elastic
compensation in this shakedown check is inconclusive.
5. Check on GPD Using Non-linear Analysis
A check against GPD was also performed for the same FE-model using conventional non-linear
analysis. In this way, a direct comparison may be made between the two limit approaches. The FE-
geometry, mesh and boundary conditions are the same as those used in the elastic compensation
analysis. Material non-linearities were applied corresponding to the material strength parameter,
188.45 MPa and pseudo-perfect plasticity. Limit analysis using shell elements can have problems
with convergence, so a small value of plastic modulus is applied. The pressure is ramped and the
analysis continues to converge to a pressure of 0.55 MPa. The maximum absolute value of the
principal strain may not exceed 5%. The maximum absolute value of the principal strain occurs at
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.127 (S)
DBA
Design by Analysis
Figure 7.7.1-6: Equivalent Stress At Limit
Figure 7.7.1-5: Maximum Principal Strain at 5% limit
the intersection of the two cylinders.
Due to the simplification in
modelling with shells, the
intersection of the two cylinders is
not used for the strain evaluation.
An evaluation cross-section is taken
at the location of the weld toe for
assessment of the strain (see
subsection 3.3.1). At this position
the load of 0.55 MPa exceeds the
5% limit. At a load of 0.455 MPa
the value of maximum principal
strain is within the 5% limit, Figure
7.7.1-5. With a constant moment of
21120 Nm, the maximum allowable
load according to the GPD-check
using non-linear analysis is given as
Figure 7.7.1-6 shows the
equivalent stress at the limit
defined by the maximum allowable
principal strain. The extent of the
plasticity region covers the area
immediately round the intersection
of the two cylinders. The
allowable load calculated using the
non-linear analysis is considerably
lower than that calculated using
elastic compensation. The
structure can still support further
increase in load above the
maximum principal strain limit of
+/- 5%. Elastic compensation has
no such limit applied to the
analysis and gives a limit load
based upon the point at which
deformations become unconstrained.
Running on the same equipment as the elastic compensation analysis, the non-linear analysis
required a CPU time of 4874 seconds.
6. Additional Comments
As the geometry in this problem is outside the scope of DBF, the DBA calculations are a quick and
simple alternative for this simple problem.
MPa PS
p
GPD
38 . 0
455 . 0
max
= =

Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.128 (S)
DBA
Design by Analysis
0
0.1
0.2
0.3
0.4
0.5
0.6
0 10 20 30 40 50
Displacement (mm)
P
r
e
s
s
u
r
e

(
M
P
a
)
Figure 7.7.1-7: Pressure-Displacement Graph (no principal strain limit)
Figure 7.7.1-8: Non-linear equivalent stress at P=0.51 MPa
The load given by the non-linear analysis is considerably lower than that given by elastic
compensation. The reason is the limit on maximum principal strain. Elastic compensation has no
limit on the maximum
principal strain and the
limit load is not restricted
by this. If the tangent
intersection method is
considered again and the
limit load calculated with
no restriction on the
maximum principal strain
(Figure 7.7.1-7) a limit
pressure of 0.51 MPa is
found. Giving a maximum
allowable pressure
according to the GPD-
Check of 0.425 MPa, a
value closer to that
calculated by elastic
compensation.
Usage of the generalised yield function in elastic compensation for shells works successfully in
performing a conventional limit analysis. When the code rules are considered and the restriction on
plastic strain is applied, the limit is considerably reduced. The structure is still relatively stiff.
Elastic compensation does not
take account of this and will give
a limit load based upon the true
limit state of the structure when
deformations become unbounded.
This may be seen by considering
the limit defining stress fields of
each method. The elastic
compensation limit field (Figure
7.7.1-2) shows a large plastic
region covering most of the
nozzle and extending far into the
main cylinder. The non-linear
limit field shows a considerably
smaller extent of plasticity, an
area restricted to the region
immediate to the cylinder
intersection. If the stress field for
the non-linear analysis is
examined at the limit load given
by elastic compensation, Figure 7.7.1-8, it can be seen that the extent of the plasticity region is
similar, for the same load, to that for elastic compensation, Figure 7.7.1-2.
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.129 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using elasto-plastic FE calculations
FE-Software: ANSYS 5.4
Element Types: Model: 8 node structural shell SHELL93
Submodell: 8 node, 3 - D structural solid SOLID45
Boundary Conditions: Model: Symmetry b.c. in the symmetry plane in longitudinal
direction of the shell; vertical displacement in the middle-nodes
of the flat ends of the shell constrained to 0; longitudinal
displacement in the middle node of the left flat end of the shell
constrained to 0; coupling of the rotational DOF about x-axis (see
Figure) at the end of the nozzle. Submodel: Symmetry b.c. in
the symmetry plane in longitudinal direction of the shell; b.c. at
the cut-boundaries according to submodelling.
Model and Mesh:
Results: Max. allowable pressure acc. to GPD, for constant moment, MPa PS
GPD
39 . 0
max
=
Shakedown pressure, for constant moment, MPa PS
SD
51 . 0
max
=
Member:
A&AB
Model:
Number of shell elements: 1324
Geometry: as in example specification
Submodel:
Number of solid elements: 13600
(4 in thickness direction)
Geometry: see Analysis Details
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.130 (S)
DBA
Design by Analysis
1. Elements, mesh fineness, boundary conditions
Principally, the model size and the boundary conditions used for modelling a shell-nozzle
intersection strongly influence the calculation results; the influence increases with increasing ratio
of nozzle diameter to shell diameter and with decreasing wall thickness. The example under
consideration was used to investigate these influences. Some results are given briefly in the
following.
For a shell-nozzle intersection with an inner shell diameter of 496 mm, an inner nozzle diameter
of 367 mm, and wall thicknesses of 4 mm, the linear-elastic behaviour of FE models with
different shell lengths and with boundary conditions as applied in Example 3.2 of this project,
was studied, with internal pressure action only. The maximum of Mises equivalent stress
showed a strong dependence of the length of the shell the difference was about 12 % for
models with 1500 mm and 2000 mm total length of shell, respectively. Therefore, a restriction
of the model size to minimise computation time is not admissible. The location of the maximum
of Mises equivalent stress was the lowest point of the intersection curve (in the symmetry plane
normal to the main shell axis).
Since the example was specified initially to compare results with fatigue tests, performed at
WTCM-CRIF (a member of the project group), a model with a geometry exactly the same as
used in the test was studied too. This model had a greater shell thickness near the ends, and flat
ends at the shell and the nozzle (see specification of examples). Again, the linear-elastic
behaviour of this model was studied under internal pressure action only: The maximum Mises
equivalent stress was about half of the one of the model stated above (with the same total length
of the shell). Moreover, the site of the maximum of the Mises equivalent stress was completely
different it was the saddle point of the intersection curve.
The limit pressure of this latter model, corresponding to the fatigue tests, was almost twice the
one of the other model (the limit pressures were determined with the very same material
strength parameter, Mises yield condition and constant nozzle moment).
As a straightforward conclusion: The knowledge of the real boundary conditions of a real shell-
nozzle intersection with such extreme diameter ratios and small wall thicknesses is a basic
requirement for FE-calculations of practical significance. Therefore, the FE-model used for
Example 3.1 had exactly the same geometric features as the one tested (which are the ones stated in
the specification of the examples).
Due to the large ratio of diameter to wall thickness of the structure, 8-node shell elements
SHELL93 have been used in the check against GPD and for the coarse model of the shakedown
check. To model the whole structure with solid elements using an appropriate number of elements
in wall thickness direction would result in an FE-model with too many elements and nodes and,
therefore, in an unacceptable long computation time. To evaluate the 5% principal strain limit in the
check against GPD in the so called evaluation cross section (see chapter 1 of the Procedure
Guide) exactly, 8 rows of smaller elements corresponding to the weld geometry were located at the
intersection curve see Figure 7.7.2-1.
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.131 (S)
DBA
Design by Analysis
Because of the symmetry of the structure
and the actions, only the half structure was
modelled. Therefore, the moment applied to
the model was half of the specified one. In
the plane of the symmetry a symmetry
boundary condition was applied to all nodes.
To apply the moment on the top of the
nozzle properly, the rotational degrees of
freedom about the axis normal to the plane
of symmetry of all nodes of the flat end of
the nozzle were coupled.
Furthermore, the middle node of the left flat
end of the shell was supported in vertical
and horizontal direction and the middle
node of the right flat end of the shell was
supported in vertical direction.
Figure 7.7.2-1
The shakedown check was performed by submodel technique, and, for this purpose, the coarse
model was identical with the shell model described above. The submodel consisted of 13600 8-node
brick solid elements SOLID45. To obtain proper stress results, 4 elements were arranged over the
wall thickness. The submodel had a total length of 440 mm, the highest cross-section of the nozzle
had a distance of 280 mm to the centreline of the shell, and the angle between the lower cut-
boundary of the shell and the x-axis was 30. To avoid stress concentrations, the weld surface was
modelled with a fillet of 4 mm radius and the inner edge of the intersection was modelled with a
fillet of 2mm radius. In the plane of symmetry a symmetry boundary condition was applied to all
nodes, the boundary conditions of the cut-boundaries were interpolated by the software from the
corresponding displacements of the coarse model.
2. Determination of the maximum allowable pressure according to GPD
Since the subroutine for Tresca's yield condition showed bad convergence, the check against GPD
was carried out using Mises' yield condition (see subsection 3.2 of section 3 - Procedure) only.
Because of the different partial safety factors for the constant moment in the GPD- and the PD-
checks, the results from the PD-check cannot be used, and a separate calculation had to be carried
out for that particular purpose.
Because of the fact, that the moment is specified as being constant, the limit pressure was
determined for constant moment and increase of pressure only, and not for proportional increase of
all actions as required in prEN 13445-3 Annex B.9.2.1. The partial safety factor
G
for permanent
actions (with an unfavourable effect) is 1.35. Therefore, the moment to be applied in the FE-model
is given by
. 10560
2
4 . 15644
35 . 1 Nm M = =
The analysis was carried out with a linear-elastic ideal-plastic material law, Mises yield condition,
a design material strength parameter of MPa 45 . 188 for the shell and the nozzle (corresponding to a
material strength parameter of MPa 272 according to EN 10028-2) , associated flow rule, and first
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.132 (S)
DBA
Design by Analysis
order theory. The analysis was performed in two parts: In the first, the moment load was applied; in
the second, internal pressure was applied additionally, and increased until a state near the limit state
was reached.
The elastic modulus used in the calculations was GPa E 125 . 210 = .
To restrict computation time to
reasonable values, the analysis
was terminated after 7 hours on
the Compaq Professional
Workstation 5000 with two
Pentium Pro processors and 256
MB RAM.
At termination time the last
convergent solution showed an
internal pressure of MPa 645 . 0 .
Since the maximum absolute
value of the principal strain
located in the evaluation cross-
section exceeded 5% at the last
load level the absolute principal
strain value was 17% - a lower
load value with appropriate strains
had to be used as limit value. For
a value of the internal pressure of
MPa 472 . 0 the maximum
absolute value of the principal
strains was less than 5%. Figure
7.7.2-2 shows the corresponding
distribution of the Mises
equivalent stress, and Figure
7.7.2-3 the corresponding
distribution of the maximum
absolute principal strain.
According to prEN 13445-3
Annex B, Table B.9-2, the partial
safety factor for pressure action
(without natural limit)
P
is 1.2.
Thus, the maximum allowable
pressure according to the GPD-
check is given by
. 393 . 0
2 . 1
472 . 0
max
MPa PS
GPD
= =
Figure 7.7.2-2
Figure 7.7.2-3
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.133 (S)
DBA
Design by Analysis
3. Check against PD
The following procedure corresponds to the one given in subsection 3.3.2.6 of section 3
Procedures.
The load case with a constant moment Nm M 2 . 7822 2 / 4 . 15644 = = , as in the specification, and an
internal pressure of MPa p 2 . 1 = was found to be relatively close to the limit state of the structure,
and, therefore, will be called the limit state in the following.
Since there is no possibility within the submodel technique to calculate different load steps during
one analysis, separate calculations for the coarse model and the submodel were necessary to obtain
the required stress states: Internal pressure and linear-elastic calculation; moment only and linear-
elastic calculation; moment only and elasto-plastic calculation; limit state and elasto-plastic
calculation.
The computation times were
2 minutes for the coarse model and 5 minutes for the submodel for the load case internal
pressure and linear-elastic calculation,
2 minutes for the coarse model and 5 minutes for the submodel for the load case moment
only and linear-elastic calculation,
20 minutes for the coarse model and 55 minutes for the submodel for the load case
moment only and elastic-plastic calculation,
1hour and 3 minutes for the coarse model and 3hours and 12 minutes for the submodel
for the load case limit state and elastic-plastic calculation,
on the Compaq Professional Workstation 5000 with two Pentium Pro processors and 256 MB
RAM. The elasto-plastic FE-analyses were carried out as stated in prEN 13445-3 Annex B, Sec.
B.9.3.1 - using Mises yield condition and associated flow rule, a linear-elastic ideal-plastic law
with a design material strength parameter of 272 MPa for shell and nozzle, and first order theory.
A linear combination of self equilibrating stress fields the one according to the limit state
( ) p M res ij + ,
) ( and the one according to moment action only
M res ij ,
) ( was used to fulfil Melans
theorem.
For the determination of the factors
2 1
, and , the deviatoric mappings of the stress states, i.e.
the coordinates of a stress point given by its principal stresses, at the critical locations of the
structure were used, since, due to the increased number of factors and the different critical locations,
load case operations using the FE-software are not feasible.
The two critical locations of the structure are: The inner surface of the shell-nozzle junction (path I
in Figure 7.7.2-4, the fillet is excluded), where the maximum linear-elastic calculated stress arises in
the case of internal pressure only (location on the symmetry plane of the structure -see Figure
7.7.2-4), and the intersection curve of the shell and the nozzle on the outer surface of the structure
(path A in Figure 7.7.2-5), where the maximum linear-elastic calculated stress arises in the case of
moment load only (see Figure 7.7.2-5). As can be seen from the following deviatoric mappings, it is
necessary to consider the whole intersection curve from point X to point Y, a consideration only of
the point where the maximum stress arises would result in non-conservative results.
The self-equilibrating stress field according to the moment
M res ij ,
) ( is equal to 0 at the inner
surface (paths I). Therefore, the necessary conditions (see also subsection 3.3.2.6 of section 3
Procedures) are given by
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.134 (S)
DBA
Design by Analysis
0 ] ) ( ) ( ) ( [
, , ) ( , 1
+ +
+ le p ij le M ij p M res ij

0 ] ) ( ) ( [
, ) ( , 1
+
+ le M ij p M res ij
.
The optimal scaling factors
1
and
can be determined from these two
conditions for the equality sign:
0
1
= ,
858 . 0 = ;
The first result 0
1
= shows that the
self-equilibrating stress field according to
the limit state
( ) p M res ij + ,
) ( cannot be
used, because of the distribution of the
linear-elastic calculated stresses due to
the moment action only at path I.
Therefore the maximum admissible
internal pressure (at path I) is given by
the condition that the sum of the two
linear-elastic stress distributions are
completely inside of the limit circle in the
deviatoric map, which renders the value of for path I.
This behaviour can be visualised directly
in Figure 7.7.2-6, which shows the stress
distribution in the deviatoric map for the
part of path I located at the shell, where
the maximum stresses arise:
linear-elastic stress according to
moment and pressure action
) ( ,
) (
p M le ij +
thin green line,
elasto-plastic stress according to
moment and pressure action
) ( ,
) (
p M ep ij +
thick green line,
linear-elastic stress according to
moment action only
M le ij ,
) (
orange line,
self-equilibrating stress field
according to moment and pressure
action
) ( ,
) (
p M res ij +
violet line.
Figure 7.7.2-4
Figure 7.7.2-5
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.135 (S)
DBA
Design by Analysis
Figure 7.7.2-6
For the points at the outer surface (path A) the necessary conditions are now given by
0 ] ) ( ) ( ) ( [
, , , 2
+ +
le p ij le M ij M res ij
,
0 ] ) ( ) ( [
, , 2
+
le M ij M res ij
.
Figure 7.7.2-7 shows the stress distribution in the deviatoric map for path A:
linear-elastic stress according to pressure load
p le ij ,
) ( 858 . 0 thin green line,
linear-elastic stress according to moment load only
M le ij ,
) ( thin orange line,
elasto-plastic stress according to moment load
M ep ij ,
) ( thick orange line,
self-equilibrating stress according to moment load only
M res ij ,
) ( thin violet line.
It can easily be seen, that the critical point at the outer surface is point X, because of its large value
of linear-elastic stress due to the pressure load, and not the point with the maximum linear-elastic
stress due to the moment load only. Furthermore, since in X the self-equilibrating stress according
to moment load only is 0, the admissible pressure to fulfil Melans theorem is, for the chosen stress
distribution, very low i.e., the admissible value of for path A is (independently from
2
) close
to 0.
Path I, shell
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.136 (S)
DBA
Design by Analysis
The conclusion is, that the used self-equilibrating stress field, according to moment only, is not an
optimal one, and, therefore, another self-equilibrating stress field, according to a moment only, was
used the one which corresponds to a moment of Nm M 14632 = . This moment is scaled up from
the moment in the specification with the factors
R
,
G
and 3 / 2 , and is , therefore, smaller than
the limit moment of the structure. The reason for the usage of this moment is the plastification in
point X in this loading case, and, therefore, a resulting non-zero self-equilibrating stress according
to this moment in point X.
Figure 7.7.2-7
Since elasto-plastic results cannot be scaled up, an additional FE-calculation with this larger
moment had to be carried out. The corresponding computation times were 1 hour 12 minutes for the
coarse model and 4 hours and 7 minutes for the submodel.
The new self-equilibrating stress field according to this larger moment M is again practically 0 at
path I, and, therefore, does not change the admissibility conditions at this path.
At path A now, a value of 1
2
= for the self-equilibrating stress field is used, and a new
corresponding value of for the linear-elastic stress due to pressure (which must be smaller than
0.858) is determined. This value was found to be 0.425 and, therefore, a lower bound shakedown
pressure is given by
MPa MPa p PS
SD
51 . 0 2 . 1 425 . 0
max
= = = .
X
X
X=Y
Y
Y
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.137 (S)
DBA
Design by Analysis
Figure 7.7.2-8 shows some corresponding stress distributions in the deviatoric map for path A:
linear-elastic stress according to the large moment
M le
ij
,
) ( thin orange line,
elasto-plastic stress according to large moment
M ep
ij
,
) ( thick orange line,
self-equilibrating stress according to the large moment
M res
ij
,
) ( thin violet line,
sum of the self-equilibrating stress according to the large moment
M res
ij
,
) ( and the
linear-elastic stress according to the specified moment
M le ij ,
) ( thin yellow line,
sum of the self-equilibrating stress according to the large moment
M res
ij
,
) ( and the
linear-elastic stress according to the specified moment
M le ij ,
) ( and the linear-elastic
stress field according to the pressure
p le ij ,
) ( 51 . 0 - thin red line.
Figure 7.7.2-8
Note: It is recommended to check always the validity of used ( or obtained) self-equilibrating stress
fields, determined by the procedure described above, in the postprocessor of the FE-software - by
superposition with the linear-elastic stress fields according to the load cases under consideration.
X X
X
Y
Y
Y
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.138 (S)
DBA
Design by Analysis
Analysis type: Stress Categorization Route
Materials and Properties: Material strength parameter R
p0.2/t
= 272 MPa
Modulus of elasticity E = 210125 MPa.
FE- Software: ALGOR.
Element types: 19116 8-node brick elements.
Boundary conditions: yz-plane (longitudinal plane of symmetry) fixed in x (T
x
),
the shell ends fixed in z, one end in y (T
y
and T
z
),
rotations free.
Model and mesh:
Maximum admissible internal pressure according to the Stress Categorization Route:
Internal pressure = 0.45 MPa for a constant moment M
c
= 15644.4 Nm.
Member:
WTCM
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.139 (S)
DBA
Design by Analysis
Three classification lines (CL) are to be considered:
- CL 1: through point 1 with coordinates (200.218 , 20.7107 , 146.301) and point 2 with
coordinates (202.649 , 22.5705 , 149.674). The maximum Tresca stress occurs in point 1 and is
309.87 MPa for a combination of longitudinal moment (15644.4 Nm) and internal pressure
(0.28 MPa).
- CL 2: through point 3 with coordinates (198.258 , 0 , 148.788) and point 4 with coordinates
(200.694 , 0 , 152.153). The maximum Tresca stress occurs in point 3 and is 270.81 MPa for a
longitudinal moment (15644.4 Nm) only.
- CL 3: through point 5 with coordinates (16.7046 , 191.002 , 247.433) and point 6 with
coordinates (15.8173 , 190.6 , 251.379). The maximum Tresca stress occurs in point 5 and is
603.52 MPa for an internal pressure of (1.28 MPa) only.
In the figure above the locations of the classification lines are shown. The CLs are situated at the
nozzle-cylinder intersection. Three CLs are considered : CL2 at the location of highest stress for a
longitudinal moment loading only, which is situated in the transversal plane of symmetry. CL3 at
the location of highest stress for internal pressure loading only, situated near the longitudinal plane
of symmetry (2 or 3 degrees from this longitudinal plane) and CL1, at the location of the highest
stress, for a combination of longitudinal moment and internal pressure
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.140 (S)
DBA
Design by Analysis
For the stress classification, the following procedure was followed:
- for each load acting on the vessel, calculate the elementary stresses
ij
(i,j = 1,2,3) in the
different points on the different CLs.
- for each load acting on the vessel and along each CL, calculate the membrane stress components

ij,m
and the bending stress components
ij,b
.
- Classify the membrane stress components
ij,m
in (
ij
)
Pm
, (
ij
)
PL
or (
ij
)
Qm
and the bending
stress components
ij,b
in (
ij
)
Pb
or (
ij
)
Qb
.
- Calculate the sum of the stresses classified in this way for the set of loads acting simultaneously
on the vessel. The stresses resulting from this summation are designated (
ij
)
Pm
, (
ij
)
PL
, (
ij
)
Pb
,
(
ij
)
Qm
, (
ij
)
Qb
.
- From this deduce: (
ij
)
Pm
, (
ij
)
PL
, (
ij
)
P
, (
ij
)
P+Q
.
- Calculate the following equivalent stresses:
- (
eq
)
Pm
or (
eq
)
PL
,
- (
eq
)
P,
- (
eq
)
P+Q
.
According to table 19C-2, the following classification must be used:
- for internal pressure: P
L
and Q
b
,
- for a local load: P
L
and P
b
.
- Verify the admissibility of the equivalent stresses.
The results of this procedure are given on the next pages. Two loads are considered, which act
simultaneously: an internal pressure (initial value = 1.28 MPa) and a longitudinal moment of
15644.4 Nm. The stresses are calculated for those two individual actions, and a stress classification
along the CLs 1, 2 and 3 is applied. The results of the calculations are shown in the next tables.
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.141 (S)
DBA
Design by Analysis
The FE-calculations for internal pressure only are performed with a pressure of 1.28 MPa. Since
this is a linear elastic analysis, one can calculate the stresses for all other pressures (e.g. 0.45 MPa)
from this FE-calculation.
1. Pressure (1.28 MPa)
a. Along CL3
5 6 5 6 (
ij
)
m
(
ij
)
b
(
ij
)
m
(
ij
)
b
S
11
382,1 477,4 134,3 167,8 429,75 -47,65 151,1 -16,8
S
22
-203,9 227,7 -71,7 80,1 11,9 -215,8 4,2 -75,9
S
33
27,2 59,3 9,6 20,8 43,25 -16,05 15,2 -5,6
S
12
-53,8 -24 -18,9 -8,4 -38,9 -14,9 -13,7 -5,2
S
23
-7,1 -45,1 -2,5 -15,9 -26,1 19 -9,2 6,7
S
31
-25,5 -29,3 -9,0 -10,3 -27,4 1,9 -9,6 0,7
b.Along CL2
3 4 5 6 (
ij
)
m
(
ij
)
b
(
ij
)
m
(
ij
)
b
S
11
-23,2 20,7 -8,2 7,3 -1,25 -21,95 -0,4 -7,7
S
22
-139 12,5 -48,9 4,4 -63,25 -75,75 -22,2 -26,6
S
33
-48,3 22,4 -17,0 7,9 -12,95 -35,35 -4,6 -12,4
S
12
-3,1 -3,4 -1,1 -1,2 -3,25 0,15 -1,1 0,1
S
23
4,3 3,8 1,5 1,3 4,05 0,25 1,4 0,1
S
31
33,8 -22,1 11,9 -7,8 5,85 27,95 2,1 9,8
a. Along CL1
1 2 5 6 (
ij
)
m
(
ij
)
b
(
ij
)
m
(
ij
)
b
S
11
-23,5 35,1 -8,3 12,3 5,8 -29,3 2,0 -10,3
S
22
-140,1 53,6 -49,3 18,8 -43,25 -96,85 -15,2 -34,0
S
33
-44,1 64,5 -15,5 22,7 10,2 -54,3 3,6 -19,1
S
12
-28,6 -31,3 -10,1 -11,0 -29,95 1,35 -10,5 0,5
S
23
43,7 32,7 15,4 11,5 38,2 5,5 13,4 1,9
S
31
25 -39,6 8,8 -13,9 -7,3 32,3 -2,6 11,4
pressure = 0.45 MPa pressure = 1.28 MPa
pressure = 0.45 MPa
pressure = 0.45 MPa
pressure = 0.45 MPa
pressure = 1.28 MPa
pressure = 1.28 MPa
pressure = 0.45 MPa
pressure = 0.45 MPa
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.142 (S)
DBA
Design by Analysis
The FE-calculations for longitudinal moment only are performed for the given moment of 15644.4
Nm.
2. Longitudinal moment (15644.4 Nm)
a. Along CL3
5 6 (
ij
)
m
(
ij
)
b
S
11
61.4 142.4 101.9 -40.5
S
22
-127.5 111.2 -8.15 -119.35
S
33
11.8 29 20.4 -8.6
S
12
-18.6 -1.4 -10 -8.6
S
23
-1 -19.6 -10.3 9.3
S
31
-6 -9 -7.5 1.5
b. Along CL2
3 4 (
ij
)
m
(
ij
)
b
S
11
0 0 0 0
S
22
0 0 0 0
S
33
0 0 0 0
S
12
-85.8 -23.1 -54.45 -31.35
S
23
104.7 31.1 67.9 36.8
S
31
0 0 0 0
a. Along CL1
1 2 (
ij
)
m
(
ij
)
b
S
11
25.5 -12.3 6.6 18.9
S
22
23.1 -27.2 -2.05 25.15
S
33
63.2 -4.1 29.55 33.65
S
12
-79.2 -22.5 -50.85 -28.35
S
23
96.7 26.6 61.65 35.05
S
31
-30.6 15.8 -7.4 -23.2
Analysis Details
Example 3.1 / GPD- & PD-Check
Page
7.143 (S)
DBA
Design by Analysis
The calculated equivalent stresses must meet the assessment criteria:
The highest equivalent stresses are reached on CL3. These stresses are used to meet the assessment
criteria.
With f = R
p0.2/t
/ 1.5 = 272 / 1.5 MPa = 181.3 MPa, the assessment criteria are met for an internal
pressure of 0.45 MPa and a longitudinal moment of 15664.4 Nm.
3. Pressure (0.45 MPa) + longitudinal moment (15644.4 Nm)
a. Along CL3
5 6 (
ij
)
PL
(
ij
)
Qb
(
ij
)
P+Q
(
ij,PL
)
1
256.23 (
ij,P+Q
)
1
314.006
S
11
195.7 310.2 253.0 -57.3 310.2 (
ij,PL
)
2
43.382 (
ij,P+Q
)
2
54
S
22
-199.2 191.3 -4.0 -195.2 -199.2 (
ij,PL
)
3
-15.011 (
ij,P+Q
)
3
-207.23
S
33
21.4 49.8 35.6 -14.2 49.8
S
12
-37.5 -9.8 -23.7 -13.8 -37.5 (
eq
)
Pl
271.241 MPa
S
23
-3.5 -35.5 -19.5 16.0 -35.5 (
eq
)
P+Q
521.236 MPa
S
31
-15.0 -19.3 -17.1 2.2 -19.3
b.Along CL2
3 4 (
ij
)
PL
(
ij
)
Qb
(
ij
)
P+Q
(
ij,PL
)
1
75.678 (
ij,P+Q
)
1
100.739
S
11
-8.2 7.3 -0.4 -7.7 -8.2 (
ij,PL
)
2
0 (
ij,P+Q
)
2
-0.075
S
22
-48.9 4.4 -22.2 -26.6 -48.9 (
ij,PL
)
3
-102.88 (
ij,P+Q
)
3
-174.76
S
33
-17.0 7.9 -4.6 -12.4 -17.0
S
12
-86.9 -24.3 -55.6 -31.3 -86.9 (
eq
)
Pl
178.554 MPa
S
23
106.2 32.4 69.3 36.9 106.2 (
eq
)
P+Q
275.503 MPa
S
31
11.9 -7.8 2.1 9.8 11.9
a. Along CL1
1 2 (
ij
)
PL
(
ij
)
Qb
(
ij
)
P+Q
(
ij,PL
)
1
108.744 (
ij,P+Q
)
1
165.111
S
11
17.2 0.0 8.6 8.6 17.2 (
ij,PL
)
2
8.359 (
ij,P+Q
)
2
7.641
S
22
-26.2 -8.4 -17.3 -8.9 -26.2 (
ij,PL
)
3
-92.702 (
ij,P+Q
)
3
-134.05
S
33
47.7 18.6 33.1 14.6 47.7
S
12
-89.3 -33.5 -61.4 -27.9 -89.3 (
eq
)
Pl
201.446 MPa
S
23
112.1 38.1 75.1 37.0 112.1 (
eq
)
P+Q
299.163 MPa
S
31
-21.8 1.9 -10.0 -11.8 -21.8
(
eq
)
Pl
< 1.5f
(
eq
)
Pl+Qb
< 3f
DBA
Design by Analysis
Analysis Details
Example 3.1 / F-Check
Page
7.144 (C)
Analysis Type:
F-Check (Equivalent structural stress approach)
Member:
CETIM
FE Software: ABAQUS / Standard version 5.8.1
Element Types:
Quadratic shell rectangular 8-node and triangular 6-node elements .
13247 nodes and 3362 elements
Boundary Conditions:
The model takes into account a plane of symmetry in the axial direction. In
this plane, the perpendicular displacement to this plane and the rotations in
this plane are locked. One end of the horizontal shell is locked in the
vertical and the axial direction. The other end is locked in the vertical
direction.
Model and Mesh:
Results:
Fatigue life N = 122770 cycles (for cyclic pressure)
1042 cycles (for cyclic moment)
Data
t
max
= 50 C
t
min
= 50 C
t* = 0,75 t
max
+ 0,25 t
min
= 50 C
R
m
= 460 MPa
R
p0,2/t*
= 272 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 4 mm

D (5
.
10
6
cycles)
= 52 MPa (class 71)
equivalent stresses or principal stresses
m = 3 C = 7,16
.
10
11
m = 3 C

= ..
C
//
= ..
m = 5 C = 1,96
.
10
15
m = 5 C

= ..
C
//
= ..
Stresses

struc
= 180 MPa (structural equivalent stress range, determined by extrapolation)
Note: As the maximum stresses are obtained in the weld, it is not necessary to combine the constant loading with the varying loading.
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..
= k
e

struc
= .. MPa
Else

=
struc
= 180 MPa
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0,639
DBA
Design by Analysis
Analysis Details
Example 3.1 / F-Check
Page
7.145 (C)
Cyclic pressure
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = 7,16 10
11
N =
m
w
f
C

,
_


= 122770 cycles
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 180 MPa N = if
w
f

<
10
8
, else:
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
DBA
Design by Analysis
Analysis Details
Example 3.1 / F-Check
Page
7.146 (C)
Data
t
max
= 50 C
t
min
= 50 C
t* = 0,75 t
max
+ 0,25 t
min
= 50 C
R
m
= 460 MPa
R
p0,2/t*
= 272 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 4 mm

D (5
.
10
6
cycles)
= 52 MPa (class 71)
equivalent stresses or principal stresses
m = 3 C = 7,16
.
10
11
m = 3 C

= ..
C
//
= ..
m = 5 C = 1,96
.
10
15
m = 5 C

= ..
C
//
= ..
Stresses

struc
= 761 MPa (structural equivalent stress range, determined by extrapolation)
Note: As the maximum stresses are obtained in the weld, it is not necessary to combine the constant loading with the varying loading.
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= 0,4
k
e
= 1,1596
= k
e

struc
= 882,4 MPa
Else =
struc
= ..
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= 1,0887


= k

struc
=
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0,639
DBA
Design by Analysis
Analysis Details
Example 3.1 / F-Check
Page
7.147 (C)
Cyclic moment
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 882,4 MPa N = if
w
f

<
10
8
, else:
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= = 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
If
w
f

>
5
.
10
6
cycles
then
M = 3 and C (C

or C
//
) = 7,16
.
10
11
N =
m
w
f
C

,
_


= 1042 cycles
If
w
f

<
5
.
10
6
cycles
with other
cycles where
w
f

>
5
.
10
6
cycles
then
M = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= .. cycles
If
w
f

<
5
.
10
6
cycles
with all other
cycles where
w
f

<
5
.
10
6
cycles
then
N =
DBA
Design by Analysis
Analysis Details
Example 3.1 / F- Check
Page
7.148 (C)
Analysis Details
Example 3.2
Page
7.149 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using Elastic Compensation
FE-Software: ANSYS 5.4
Element Types: 8 node, 3-D isoparametric solids.
Boundary Conditions: Symmetry boundary conditions in the plane along the longitudinal
direction of the shell.
Hoop displacements in nodes at both ends of the shell constrained
to zero.
Longitudinal displacements in the nodes at one end of the shell
constrained to zero, longitudinal displacements coupled in the
nodes at the other end of the shell (plane sections remain plane).
A rigid region is set up on the top surface nodes of the nozzle to
apply the bending moment.
Model and Mesh:
Results:
Maximum admissible action according to the GPD-Check: Internal pressure PS
max GPD
= 11.5 MPa
(for the constant moment of Nm M
c
1 . 711 = )
Check against PD: Shakedown limit pressure PS
max SD
= 13.7 MPa
Member:
Strathclyde
Length of shell 500 mm
Length of nozzle 80mm
Number of elements - 1307
Analysis Details
Example 3.2
Page
7.150 (S)
DBA
Design by Analysis
1. Finite Element Mesh
The geometry model was constructed according to the problem specification for example 3.2.
Element divisions for the finite element mesh were defined parametrically to allow modification of
the element density. Half symmetry was used with finite element models created using low order 8-
node 3-D isoparametric solid elements. The maximum allowable pressure according to GPD and
the shakedown pressure according to PD were calculated for this model.
Boundary conditions applied to the model are symmetry about the xz plane, therefore the applied
moment is half the specified value. The nodes at both ends of the main cylinder are constrained in
the hoop direction but are allowed to move radially. The nodes at one end of the main cylinder
were constrained fully in the longitudinal direction with the nodes at the other end coupled in the
longitudinal direction, thus maintaining plane sections remaining plane. Pressure was applied over
the whole internal structure with the equivalent thrust that would be produced from closed ends
applied at the longitudinally coupled degree of freedom end of the cylinder. The moment applied to
the intersecting nozzle was modelled by creating a rigid region over the top edge of the nozzle via
constraint equations. A mass element was created in the centre at the top of the nozzle (to act as the
master node to which the moment is applied) with the nodes on the top edge of the nozzle (slave
nodes) connected to it by constraint equations. The moment applied to the master node would then
be transmitted by the rigid constraint region evenly to the slave nodes on the nozzle. Resulting in
the application of a pure moment to the top surface of the nozzle with no warping, the cross-section
remains plane.
2. Material properties:
- Shell (P265GH): material strength parameter MPa RM 234 = , modulus of elasticity
. 125 , 210 GPa E =
- Nozzle (11CrMo9-10): material strength parameter , 343MPa RM = modulus of elasticity
. 125 , 210 GPa E =
3. Determination of the maximum admissible pressure according to the GPD-Check
Using the application rule in prEN-13445-3 Annex B.9.2.2 to check against GPD, the principle is
fulfilled when for any load case the combination of the design actions do not exceed the design
resistance. This may be shown by calculating the limit load. The limit load has to be determined
using Trescas yield condition and associated flow rule. As elastic compensation is based upon a
series of elastic equilibrium stress fields, it is a relatively simple procedure to derive a lower bound
limit load direct from the Tresca yield model.
From prEN 13445-3 Annex B, Table B.9-3 the partial safety factor,
R
on the resistance is 1.25.
Therefore, the design material strength parameter given by RM/
R
for the shell and nozzle is 187.2
MPa and 274.4 MPa, respectively. The analysis was carried out using the elastic compensation
method conforming to the direct route rules for GPD in Annex B: linear elastic - ideal plastic
material law, Trescas yield condition and associated flow rule and first order theory. In each elastic
compensation iteration an equilibrium stress field is produced where the elastic modulus of each
subsequent iteration is defined by the previous elastic solution. In this way regions of the FE-model
may be systematically stiffened or weakened depending upon the stress magnitudes in the previous
solution. The result is that every equilibrium stress field is a lower bound of the limit load, the
stress field giving the greatest limit load is taken as the closest to the actual limit state and defines
Analysis Details
Example 3.2
Page
7.151 (S)
DBA
Design by Analysis
Pressure-Moment Limit Locus
0
2
4
6
8
10
12
14
16
0 200 400 600 800 1000 1200 1400
Moment (Nm)
P
r
e
s
s
u
r
e

(
M
p
a
)
A = 960 Nm
Md
A = 12 MPa
Pd
Figure 7.8.1-1: Pressure Moment Limit Locus
Figure 7.8.1-2: Tresca Limit Stress Field
the limit load in the analysis. In problems such as this, where there are materials with different
properties, the modulus modification has a modified procedure that takes account of the different
material properties. This modified method calculates the limit pressure for each component with a
different material, allowing the component giving the lowest limit load to define the limit for the
whole model. Convergence of the equilibrium stress field will occur typically between 8 and 15
iterations, i.e. no further
benefit in the limit load will
be noted with subsequent
iterations.
In the check against GPD,
the loading considered is
constant moment and
increasing pressure, i.e. the
limit pressure is to be
calculated. As elastic
compensation is based upon
linear-elastic analysis it is
not possible to calculate the
limit load directly, where the
load set contains two
separate loads with one
constant. The elastic
compensation procedure will scale the combined load stress field therefore both sets of loads are
scaled. However, by carrying out a series of elastic compensation analyses, where the ratio of the
load sets are altered for each analysis, a limit locus can be constructed. The limit locus will then
describe the limit load for any ratio of the loads within the load set.
The total computing time to run each analysis on a 300 MHz Pentium two processor with 128 Mb
RAM Windows NT workstation was
approximately 458 seconds. The
stress fields were shown to converge
after eight iterations, fourteen
separate analyses were carried out to
describe the limit locus. Figure
7.8.1-1 shows the pressure moment
limit locus from which the design
pressure is found. The limit locus
follows an unusual pan, this may be
a result of the mixed material model.
When the analysis results are
studied, it can be shown that the
failing component moves from the
shell when the pressure is
predominant to the nozzle when the
moment is predominant. Resulting
in a transition in the limit defining
locus from that defining the limit of
the shell to the locus defining the limit of the nozzle.
Analysis Details
Example 3.2
Page
7.152 (S)
DBA
Design by Analysis
Mises Limit Locus
0
5
10
15
20
25
0 200 400 600 800 1000 1200 1400 1600 1800
Moment (Nm)
P
r
e
s
s
u
r
e

(
M
P
a
) P = 19.9 MPa
L
A = 960 Nm
Md
Figure 7.8.1-3: Mises Limit Locus
According to prEN-13445-3 Annex B, Table B.9-2 for permanent actions with an unfavourable
effect (this covers the moment) the partial safety factor
G
is 1.35. Therefore, the design value for
the moment action, A
Md
, is the allowable moment (defined in the specification) multiplied by
G,
giving A
Md
=960 Nm. From Figure 7.8.1-1, for a design moment action of 960 Nm the limit locus
gives a design value for the pressure action, A
Pd
, of 12 MPa. For a pressure without natural limit
the partial safety factor,
p,
given by Table B.9-2 is 1.2. The allowable pressure is given as
By carrying out a final elastic compensation analysis using the calculated value of design pressure,
a check on the result can be made. Figure 7.8.1-2 shows the limit stress intensity field based on
Tresca's criterion at the design loads. The maximum stress intensity in the shell is 188 MPa, which
is approximately equal to the material strength parameter for the shell. The maximum stress in the
nozzle is 228 MPa, which is lower than the design strength parameter for the nozzle. Clearly, the
shell is the limit defining component with the stress in the nozzle still some way below its material
strength parameter. Therefore,
for the given moment, the
allowable pressure loading of 10
MPa fulfils the check against
GPD.
The maximum pressure according
to GPD can also be calculated
from the limit load results of the
PD check using Mises' condition,
where a lower bound on the
Tresca limit load is given by
applying a factor of 3/2 to the
yield strength. Elastic
compensation based on Mises'
condition was carried out as
described in the PD-check, the
resulting limit locus is as shown in Figure7.8.1-3. The calculated limit pressure for a constant
design moment action of 960 Nm is P
L
= 19.9 MPa. The maximum allowable pressure according to
the GPD-check is obtained by applying the partial safety factors for the material and pressure (as
before) along with the Mises correction:
The value for maximum allowable pressure will be taken from the Mises analysis, analysis based on
Mises' and Tresca's condition both give lower bounds on the limit load. Therefore, the maximum
will be used in this analysis.
MPa PS
p
GPD
10
12
max
= =

MPa PS
R p
GPD
5 . 11
2
3
25 . 1 2 . 1
9 . 19
2
3 9 . 19
max
=

=

Analysis Details
Example 3.2
Page
7.153 (S)
DBA
Design by Analysis
Mises Shakedown Locus
0
2
4
6
8
10
12
14
16
18
0 200 400 600 800 1000 1200 1400 1600 1800
Moment (Nm)
P
r
e
s
s
u
r
e

(
M
P
a
)
P = 13.7 MPa
sh
M = 711.1 Nm
Figure 7.8.1-4: Mises Shakedown Locus
Similarly as with the Tresca locus the Mises limit locus shows the transition from failure occurring
in the shell to failure occurring in the nozzle at large moments, hence the unusual shape of the
Mises limit locus.
4. Check against PD
In the check for progressive plastic
deformation, the principle in prEN-
13445-3 Annex B.9.3.1 is fulfilled
if the structure can be shown to
shakedown.
In elastic compensation the load at
which the structure will shakedown
is simple to calculate. Based on
Melans shakedown theorem, the
self-equilibrating residual stress
field that would result after a
loading cycle, can be calculated by
subtracting the linear-elastic stress field at the limit pressure from the limit stress field. The residual
stress field is in effect the resulting stress from an elastic unloading from the limit state back to zero
pressure. If no stress in the residual field violates the yield condition, i.e. if there is no equivalent
stress above the material (yield) parameter, then the shakedown load is equal to the limit load.
Where the residual stress field of a structure does exceed the yield condition, the shakedown limit
can be calculated easily from the stress plots because of the linearity of elastic compensation.
As with the check against GPD, a shakedown locus can be created covering all pressure-moment
combinations. A second series of elastic compensation calculations are made based on Mises'
condition. Design material strength parameters of 234 MPa and 343 MPa were used for shell and
nozzle respectively. The resulting pressure-moment shakedown locus is shown in Figure 7.8.1-4.
As there are no partial safety factors applied to the actions for the PD-check then the applied
moment is 711.1 Nm, giving a maximum shakedown pressure
PS
maxSD
= 13.7 MPa.
As with the limit loci the shakedown locus follows an unusual path. Here the shakedown limit was
always defined by the shell, however the graph shows the effect the different material yield
parameters have on the result. It may also be noted that when we consider the moment only the
limit load and the shakedown load are similar, and as the pressure becomes a higher percentage of
the combined loading the shakedown load becomes considerably less than the limit load.
5. Additional Comments
Application of the DBA rules to this problem gives little benefit over the result obtained from DBF,
in-fact the results given by elastic compensation are less than that obtained from DBF.
Analysis Details
Example 3.2
Page
7.154 (S)
DBA
Design by Analysis
Elastic compensation in this problem shows some unusual results. Maximum allowable pressure is
greater when calculated from the corrected Mises PD limit load than that given by the Tresca
calculation. By definition, this should not be possible, as the application of the correction should
make the Mises calculated value a lower bound on the Tresca value. However, both calculations
are lower bounds on the limit load and the pressure given from the Mises model defines the limit.
An interesting observation in the limit loci of elastic compensation results is the transition in failure
from one component to another.
The non-linear analysis calculates a limit load slightly higher than that given by DBF and
considerably higher than those given by elastic compensation. Elastic compensation in this problem
is too conservative.
Analysis Details
Example 3.2
Page
7.155 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using elasto-plastic FE calculations
FE-Software: ANSYS 5.4
Element Types: 8 node (brick element), 3 D structural solid SOLID45;
3 D elastic beam BEAM4 (to apply the moment load).
Boundary Conditions: Symmetry boundary conditions in nodes in the symmetry plane in
longitudinal direction of the shell; hoop displacements in nodes at
both ends of the shell constrained to 0; longitudinal displacement
in one node of one end of the shell constrained to 0; rigid region
concerning vertical displacements at the end of the nozzle to apply
the moment; vertical displacement set to 0 for dummy end of
beam element.
Model and Mesh:
Results: Max. allowable pressure acc. to GPD, for constant moment, MPa PS
GPD
79 . 14
max
=
Shakedown pressure, for constant moment, MPa PS
SD
65 . 17
max
=
Member:
A&AB
Length of shell: 200 mm Length of nozzle: 100 mm
Number of elements: 3281 (1 beam element)
Analysis Details
Example 3.2
Page
7.156 (S)
DBA
Design by Analysis
1. Elements, mesh fineness, boundary conditions
To model the structure, 3-D 8-node brick elements SOLID45 in ANSYS 5.4 have been used.
For comparison, a higher order version of this elements - the 20-node brick element SOLID95 with
midside nodes was used additionally. To show the differences for these two types of elements,
linear-elastic and elasto-plastic calculations have been carried out, leading to the following results:
the linear-elastic peak stresses differ by about 5 to 8 % (for pressure as well as for moment
action), the larger values were obtained by the model with the 20-node elements;
the elasto-plastic results near the limit action of the structure are almost identical, i.e. nearly
independent of the kind of elements used;
the time for computing the results, using the Compaq Professional Workstation 5000 with
two Pentium Pro processors and 256 MB RAM, applying an internal pressure of MPa 26
at constant moment (which is very close to the limit state of the structure) was 1 hour 15
minutes using the 8-node elements, and 63 hours 15 minutes using the 20-node elements.
Because of the major difference in computation time and the fact that there is no need to compute
linear-elastic peak stresses very exactly, because the check against PD can be carried out using the
stress-concentration-free structure (according to prEN 13445-3 Annex B.9.3.2), the 8-node elements
were used for all calculations.
Because of the possibility of using a stress-concentration-free structure in the check against PD, the
weld was modelled with a fillet (radius 6.5 mm) located entirely inside the weld. At the inner edge
of the cylinder-cylinder intersection a fillet with 1mm radius was used.
Note: Modelling details of a structure to avoid stress singularities or peak stresses influence linear-
elastic calculations, but do not affect results near the limit load of the structure.
The nozzle of the considered structure is set-on. Therefore, using the material characteristics of the
shell also for the welding zone leads to conservative results - the material of the nozzle is stronger
than the material of the shell.
Calculations with the moment action only showed that the region of the shell which is influenced by
the moment is rather small see Figure 7.8.2-1, which shows the Mises equivalent stress of the
elasto-plastic calculation with moment load only. Therefore, FE-models with different lengths of
the shell render almost identical results in the intersection region. Since the opening in the shell is
small and the wall thicknesses are considerable, the results obtained for internal pressure are
independent of the kind of boundary condition at the end of the nozzle (rigid region or free) and
also of the length of the shell and the nozzle. Therefore, to save computation time, the total length
of the model was restricted to 200 mm. The length of the set-on nozzle in the model was 100 mm
(including the weld).
Analysis Details
Example 3.2
Page
7.157 (S)
DBA
Design by Analysis
To apply the moment load, the 3-D
elastic beam BEAM4 of ANSYS
5.4 was attached to the structure at
the centre of the nozzles upper
end. To transfer the moment from
the beam to the nozzle, the nodes
in the upper cross-section of the
nozzle (slave nodes) have been
linked to the lower node of the
beam (master node) rigidly. The
constraining equations of this
rigid region couple the vertical
displacements of the nodes in the
upper cross section of the nozzle
with the rotation in the node of the
beam caused by the moment load.
As a consequence of this boundary
condition the upper cross-section
of the nozzle remains plane, the
cross-section can ovalize, but it
cannot warp.
Figure 7.8.2-1
Because of the symmetry of the structure and the actions, only half of the structure was modelled,
and, therefore, the applied moment was half of the specified value. In the plane of symmetry a
symmetry boundary condition was applied to all nodes.
As boundary condition in the end cross-sections of the shell, the hoop displacements were set to
zero. This allows for warping and ovalization, although ovalization is constrained to some extent,
see also Example 4. In this example this boundary condition is considered to be appropriate, the
main shell being considered to be a part of a piping. Setting the hoop displacements to zero and
creating additionally rigid regions in the end cross-sections of the shell leads to plane end cross-
sections and avoids warping. Since linear-elastic calculations showed that the maximum stresses in
the model with the boundary condition for rigid ends are lower than the ones for boundary condition
without, the one without rigid ends was used, for reasons of conservativity.
To obtain a stable model, the horizontal displacement in one node in one end cross-section of the
main shell has been constrained to zero. The vertical displacement in the node of the upper
(dummy) end of the elastic beam has been constrained to zero too.
2. Determination of the maximum allowable pressure according to GPD
The partial safety factors
R
according to EN-UFPV Annex B, Table B.9-3 are 25 . 1 . Therefore, the
analysis was carried out with a linear-elastic ideal-plastic material law, Trescas yield condition
(delivered as a subroutine by an ANSYS distributor) with design material strength parameters of
MPa 2 . 187 for the shell and MPa 4 . 274 for the nozzle (corresponding to material strength
parameters of MPa 2 . 187 for the shell, according to EN 10028-2 and MPa 343 according to prEN
10216-2 for the nozzle), associated flow rule, and first order theory. For simplicity, the boundary
Analysis Details
Example 3.2
Page
7.158 (S)
DBA
Design by Analysis
between the two materials was assumed to be at the upper end of the weld. The elastic modulus
used for the shell and the nozzle was E = 210,125 GPa.
Because of the fact, that the moment is constant, the limit pressure was determined for constant
moment and increase of pressure only, and not for proportional increase of all actions as required in
prEN 13445-3 Annex B.9.2.1. The partial safety factor
G
for permanent actions (with an
unfavourable effect) is 1.35. Therefore, the moment to be applied to the FE-model is given by
. 480
2
1 . 711
35 . 1 Nm M = =
The analysis was carried out with the Newton-Raphson method using the initial stiffness matrix in
every equilibrium iteration, which showed the best convergence using Trescas yield condition for
the structure considered.
The analysis was carried out in two parts: in the first, the moment load was applied, and in the
second, internal pressure and corresponding longitudinal stresses at the ends of the shell and at the
end of the nozzle were applied additionally, and increased until a state near the limit state was
reached.
To restrict computation time to
reasonable values, the analysis was
terminated after 6 hours on the
Compaq Professional Workstation
5000 with two Pentium Pro
processors and 256 MB RAM.
At termination time the last
convergent solution showed an
internal pressure of MPa 66 . 17
this pressure was used as limit
pressure. This value is close to the
theoretical limit pressure of the
undisturbed cylindrical shell, which
is given by MPa 34 . 18 for Trescas
yield condition.
The Figure 7.8.2-2 shows the
distribution of Tresca's equivalent
stress for a pressure of MPa 66 . 17 .
Figure 7.8.2-2
As shown in Figure 7.8.2-3, the maximum absolute value of the principal strains in the structure is
less than 5 %, as required in the standard.
According to EN-UFPV Annex B, Table B.9-2, the partial safety factor for pressure actions
(without natural limit)
P
is 1,2. Therefore, the maximum allowable pressure according to GPD is
. 72 . 14
2 . 1
66 . 17
max
MPa PS
GPD
= =
Analysis Details
Example 3.2
Page
7.159 (S)
DBA
Design by Analysis
If a subroutine for Trescas yield
condition is not available, Mises
yield condition has to be used for the
check against GPD (see subsection 3.2
of section 3 - Procedures). Since the
partial safety factor for the constant
moment is different from 1, the results
from the check against PD cannot be
used, and an additional calculation has
to be carried out for that particular
purpose.
In this case, the analysis was carried
out with a linear-elastic ideal-plastic
material law, Mises yield condition
with design material strength
parameters of MPa 1 . 162 for the shell
and MPa 6 . 237 for the nozzle (again,
the boundary between the two
materials was assumed to be at the
upper end of the weld), associated
flow rule, and first order theory.
Figure 7.8.2-3
The analysis was carried out in the same way as the one using Trescas yield condition, except that
the kind of stiffness matrix used in the Newton-Raphson method was program-chosen.
After two hours computation time, the analysis on the Compaq Professional Workstation 5000
with two Pentium Pro processors and 256 MB RAM was terminated, because at this time the last
convergent solution showed an
internal pressure of MPa 91 . 17 ,
which is close to the theoretical
limit pressure of the undisturbed
cylindrical shell, which is given by
MPa 73 . 18 for Mises' yield
condition. Since the maximum
absolute value of the principal
strains exceeds 5 % at the last load
level the maximum absolute
principal strain value is 14 % a
lower load level with appropriate
strains had to be used as limit value.
For an internal pressure of
MPa 75 . 17 the maximum absolute
value of the principal strain in the
structure is less than 5 % - as
required in the standard see Figure
7.8.2-4.
Figure 7.8.2-4
Analysis Details
Example 3.2
Page
7.160 (S)
DBA
Design by Analysis
Figure 7.8.2-5 shows the
distribution of Mises
equivalent stress at this
pressure of MPa 75 . 17 .
According to prEN 13445-3
Annex B, Table B.9-2 the
partial safety factor for
pressure (without natural
limit)
P
is given by 1.2, and,
therefore, the maximum
allowable pressure according
to GPD by
MPa PS
GPD
79 . 14
2 . 1
75 . 17
max
= =
Figure 7.8.2-5
3. Check against PD
The following procedure corresponds to the one given in subsection 3.6 of section 3 Procedures.
The used loading state with constant moment Nm M 5 . 355 2 / 1 . 711 = = as in the specification and
internal pressure of MPa p 26 = is close to the limit state of the structure, since the theoretical
limit pressure of the undisturbed shell is MPa 04 . 27 for Mises yield condition and a material
strength parameter of MPa 234 .
To obtain the required stress states and the elasto-plastic stress state corresponding to the moment
only during one analysis which is advantageous to realise load case operations in ANSYS5.4 -
the following load path was used in the FE analysis:
state 1 Nm M MPa p 5 . 35 , 0 = = (linear-elastic path)
state 2 Nm M MPa p 5 . 35 , 6 . 2 = = (linear-elastic path)
state 3 Nm M MPa p 0 , 0 = = (linear-elastic path)
state 4 Nm M MPa p 5 . 355 , 0 = = (elasto-plastic path)
state 5 Nm M MPa p 5 . 355 , 26 = = (elasto-plastic path).
The required linear-elastic stress field corresponding to the limit state is obtained by scaling the
stress field of state 2. The one corresponding to an internal pressure of MPa p 26 = is obtained by
scaling the stress field of state 1, and the one corresponding to the moment of Nm M 5 . 35 = by
scaling the stress field of the difference of the stress fields of the states 2 and 1.
Analysis Details
Example 3.2
Page
7.161 (S)
DBA
Design by Analysis
The computation time was 3 hours and 10 minutes on the Compaq Professional Workstation 5000
with two Pentium Pro processors and 256 MB RAM. The elasto-plastic FE analysis was carried
out as stated in prEN 13445-3 Annex B, Sec. B.9.3.1 - using Mises yield condition and associated
flow rule, a linear-elastic ideal-plastic law with (design) material strength parameters of 234 MPa
for the shell and 343 MPa for the nozzle (again, the boundary between the two materials was
assumed to be at the upper end of the weld), and first order theory.
In the case under consideration, no factors and (see subsection 3.3.2.6 of section 3
Procedures) were found, so that Melans theorem was fulfilled.
Therefore, a linear combination of self equilibrating stress fields the one according to the limit
state
( ) p M res ij + ,
) ( and the one according to moment load only
M res ij ,
) ( was used to fulfil Melans
theorem.
For the determination of the factors
2 1
, and (see subsection 3.3.2.6 of section 3
Procedures) the deviatoric maps of the stress states, i.e. the coordinates of a stress point given by its
principal stresses, at the critical locations of the structure are used, since due to the increased
number of factors and the different critical locations load case operations using the FE software are
not feasible.
The two critical
locations are: The
outer surface of the
weld fillet (path A in
Figure 7.8.2-6) with
the adjacent part of
the nozzle (path N in
Figure 7.8.2-6), and
the inner surface of
the shell of the
junction and the
nozzle (path I in
Figure 7.8.2-6) with
the adjacent part of
the nozzle (path Z in
Figure 7.8.2-6). To
ensure, that the
critical part of the
structure is not in a
part of the
undisturbed shell, a
corresponding point
at the inner surface
was also taken into account. Figure 7.8.2-6
Since the self-equilibrating stress field according to the moment load
M res ij ,
) ( is equal to 0 in the
critical point at the inner surface (paths I and Z), the point of the linear-elastic stress distribution for
moment and pressure load with greatest distance to the centre in deviatoric map, see Figure7.8.2-7,
the necessary conditions in this point are now given by
Analysis Details
Example 3.2
Page
7.162 (S)
DBA
Design by Analysis
0 ] ) ( ) ( ) ( [
, , ) ( , 1
+ +
+ le p ij le M ij p M res ij

0 ] ) ( ) ( [
, ) ( , 1
+
+ le M ij p M res ij
.
Thus, the optimal scaling factors
1
and can be determined from these two conditions
567 . 0
1
= ,
679 . 0 = .
Figure 7.8.2-7 shows the stress distribution in the deviatoric map for the inner surface:
linear-elastic stress according to moment and pressure
) ( ,
) (
p M le ij +
thin green line,
elasto-plastic stress according to moment and pressure
) ( ,
) (
p M ep ij +
thick green line,
linear-elastic stress according to moment only
M le ij ,
) ( yellow line,
self-equilibrating stress according to moment and pressure
) ( ,
) (
p M res ij +
black line.
connection line (no physical meaning)
fillet
nozzle
Figure 7.8.2-7
Analysis Details
Example 3.2
Page
7.163 (S)
DBA
Design by Analysis
For the critical point on the outer surface (paths A and N), the point of the linear-elastic stress
distribution for moment and pressure with greatest distance to the centre in deviatoric map see
Figure 7.8.2.-8 -, the necessary conditions are given by
0 ] ) ( 679 . 0 ) ( ) ( ) ( 569 . 0 [
, , , 2 ) ( ,
+ + +
+ le p ij le M ij M res ij p M res ij
,
0 ] ) ( ) ( ) ( 569 . 0 [
, , 2 ) ( ,
+ +
+ le M ij M res ij p M res ij
.
It can be shown, that the first condition is fulfilled for all values of 0
2
, and, therefore, the
second condition renders for the equality sign:
4 . 2
2
= .
Figure 7.8.2-8 shows the stress distribution in the deviatoric map for the outer surface:
linear-elastic stress according to moment and pressure
) ( ,
) (
p M le ij +
thin green line,
elasto-plastic stress according to moment and pressure
) ( ,
) (
p M ep ij +
thick green line,
linear-elastic stress according to moment only
M le ij ,
) ( thin yellow line,
self-equilibrating stress according to moment and pressure
) ( ,
) (
p M res ij +
thin black line
self-equilibrating stress according to moment only
M res ij ,
) ( thin red line.
Figure 7.8.2-8
nozzle
connection line (no
physical meaning)
fillet
Analysis Details
Example 3.2
Page
7.164 (S)
DBA
Design by Analysis
Thus, the limit pressure according to the shakedown condition is not smaller than
MPa MPa p PS
SD
65 . 17 26 679 . 0
max
= = = .
4. Comments
If the mechanical properties as specified in EN 10028-2 at calculation temperature (50C) are used
instead of the ones at ambient temperature, there is a remarkable difference in the corresponding
DBF results.
Generally, the influence of the total model length (length of the shell) of shell-nozzle intersections
on the calculation results increases with the ratio of the nozzle to shell diameter. In the example
considered the influence is still negligible.
Analysis Details
Example 3.2
Page
7.165 (S)
DBA
Design by Analysis
Analysis Type:
Stress Categorization Route
FE-Software: ANSYS 5.4
Element Types: 8 node, 3-D isoparametric solids.
Boundary Conditions: Symmetry boundary conditions in the plane along the longitudinal
direction of the shell.
Hoop displacements in nodes at both ends of the shell constrained
to zero.
Longitudinal displacements in the nodes at one end of the shell
constrained to zero, longitudinal displacements coupled in the
nodes at the other end of the shell (plane sections remain plane).
A rigid region is set up on the top surface nodes of the nozzle to
apply the bending moment.
Model and Mesh:
Results:
Maximum admissible action according to the Stress Categorisation Route
Internal pressure PS
max SC
= 14.25 MPa
Member:
Strathclyde
Length of shell 500 mm
Length of nozzle 80mm
Number of elements - 2560
Analysis Details
Example 3.2
Page
7.166 (S)
DBA
Design by Analysis
1. Finite Element Mesh
Boundary conditions applied to the model are symmetry about the xz plane. Therefore the applied
moment is half the specified value. The nodes at both ends of the main cylinder are constrained in
the hoop direction but are allowed to move radially. The nodes at one end of the main cylinder
were constrained fully in the longitudinal direction with the nodes at the other end coupled in the
longitudinal direction, thus maintaining plane sections remaining plane. Pressure was applied over
the whole internal structure with the equivalent thrust that would be produced from closed ends
applied at the longitudinally coupled degree of freedom end of the cylinder. The moment applied to
the intersecting nozzle was modelled by creating a rigid region over the top edge of the nozzle via
constraint equations. A mass element was created in the centre at the top of the nozzle (to act as the
master node to which the moment is applied) with the nodes on the top edge of the nozzle (slave
nodes) connected to it by constraint equations. The moment applied to the master node is then
transmitted by the rigid constraint region evenly to the slave nodes on the nozzle, resulting in the
application of a pure moment to the top surface of the nozzle with no warping, the cross-section
remains plane.
2. Material properties
Shell: Material strength parameter MPa RM 234 = , modulus of elasticity GPa E 125 . 210 =
Nozzle: Material strength parameter MPa RM 343 = , modulus of elasticity GPa E 125 . 210 =
3. Determination of the Maximum Allowable Pressure
Annex C of prEN 13445-3 defines the rules and methods for the interpretation of stresses calculated
on an elastic basis and the verification of their admissibility by means of appropriate assessment
criteria. Elastic stresses are linearised along stress classification lines into membrane and bending
components. The membrane and bending components are then categorised depending upon the
location of the stress classification line. Figure 7.8.3-1 shows the selected classification lines.
Classification lines 1-3 are within the region of local primary membrane stress occurring as a result
of the discontinuity, prEN 13445-3 Annex C. Classification line 4 is at a considerable distance
from the discontinuity where no local effects occur.
Two separate elastic analyses are carried out one for each action; a bending moment applied to the
nozzle of 711.1 Nm and for an applied pressure of 10 MPa. Superposition may be used to combine
the moment and pressure load case. As the analysis is elastic, the pressure load case may be scaled
to the limit of admissibility to find the maximum allowable pressure according to the stress category
approach. Figure 7.8.3-2 shows the stress intensity in the structure for moment load of 711.1 Nm
only.
Analysis Details
Example 3.2
Page
7.167 (S)
DBA
Design by Analysis
CL 1
CL 4
CL 2
CL 3
Figure 7.8.3-1: Position of Classification Lines
Figure 7.8.3-2: Stress Intensity for Moment Load (711.1Nm)
Analysis Details
Example 3.2
Page
7.168 (S)
DBA
Design by Analysis
Figure 7.8.3-3 shows the stress intensity of the structure for an applied pressure of 10 MPa.
Figure 7.8.3-4 Shows the stress intensity in the structure when the above load cases are
superimposed with a scale factor of 1.425 applied to the pressure load case i.e. the internal pressure
is 14.25 MPa.
Figure 7.8.3-3: Stress Intensity for Internal Pressure (10MPa)
Figure 7.8.3-4: Stress intensity for Combined Load (Moment 711.1 Nm, Pressure 14.25 MPa)
Analysis Details
Example 3.2
Page
7.169 (S)
DBA
Design by Analysis
The assessment criteria of Annex C states that the following limits be placed on the stress
categories:

eq
denotes Tresca's equivalent stress intensity, P
m
denotes general primary membrane stress, P
L
is,
local primary membrane stress, Pb primary bending stress, P+Q primary plus secondary stress and f
is the value of the nominal design stress. Clause 6 of prEN 13445-3 gives details on the value to be
taken as the nominal design stress f; 6.2.1 states that the value of f is taken as 0.2% proof strength
of the material at calculation temperature with a safety factor of 1.5. Therefore, for the shell f =
234/1.5 = 156 MPa and for the nozzle f = 343/1.5 = 228.7 MPa.
For the classification lines in Figure 7.8.3-1, the linearised stress components fall into the stress
categories as shown in the following Table (given from Table C-2 in prEN 13445-3 Annex C)
Stress Categories for Linearised Stresses
Classification
line
Type of Stress Pressure Load Moment Load

ij,m
P
L
P
L
1

ij,b
Q
b
Q
b

ij,m
P
L
P
L
2

ij,b
Q
b
Q
b

ij,m
P
L
P
L
3

ij,b
Q
b
Q
b

ij,m
P
m
P
m
4

ij,b
Q
b
P
b
The following Table summarises the stress categorisation results for a moment load of 711.1 Nm
and pressure of 14.25 MPa. Stress linearisation along the chosen classification lines was carried
using the post processing linearisation routines within the ANSYS program.
f
f
f
f
Q P eq
b P eq
P eq
P eq
L
m
3 ) (
5 . 1 ) (
5 . 1 ) (
) (

Analysis Details
Example 3.2
Page
7.170 (S)
DBA
Design by Analysis
Results for Stress Categories on Linearised Stresses
Classification
line
Type of stress Stress category Linearised Stress
Intensity (MPa)
Limit on stress
categories (MPa)
membrane (
eq
)
PL
211.3 1.5f = 234
bending (
eq
)
Qb
24.17 -
1
mem+bend (
eq
)
P+Q
234.7 3f = 468
membrane (
eq
)
PL
91.67 1.5f = 343
bending (
eq
)
Qb
33.46 -
2
mem+bend (
eq
)
P+Q
120.3 3f = 686.1
membrane (
eq
)
PL
227.9 1.5f = 234
bending (
eq
)
Qb
26.25 -
3
mem+bend (
eq
)
P+Q
253.5 3f = 468
membrane (
eq
)
Pm
156.0 f = 156
bending (
eq
)
Qb
3.82 -
4
mem+bend (
eq
)
P+Q
158.8 3f = 468
This Table shows that the limiting stress category is the primary membrane stress in the shell
resulting from the internal pressure ((
eq
)
Pm
on classification line 4). General primary membrane
stress is limited to f (156 MPa in the shell) and at an internal pressure of 14.25 MPa this limit is
reached, therefore
MPa PS
SC
25 . 14
max
=
DBA
Design by Analysis
Analysis Summary
Example 3.2 / F-Check
Page
7.171 (A)
Analysis Type:
Check against Fatigue
Member:
A&AB
FE Software: ANSYS 5.4
Element Types: 8 node (brick element), 3 D structural solid SOLID45;
3 D elastic beam BEAM4 (to apply the moment load).
Boundary Conditions:
Symmetry boundary conditions in nodes in the symmetry plane in longitudinal direction
of the shell; hoop displacements in nodes at both ends of the shell constrained to 0;
longitudinal displacement in one node of one end of the shell constrained to 0; rigid
region concerning vertical displacements at the end of the nozzle to apply the moment;
vertical displacement set to 0 for dummy end of beam element.
Model and Mesh:
Results:
Fatigue life N = 70865 cycles (for cycling pressure)
1160 cycles (for cycling moment)
Length of shell: 200 mm Length of nozzle: 100 mm
Number of elements: 3281 (1 beam element)
1
Data
t
max
= 50 C
t
min
= 50 C
t* = 0,75 t
max
+ 0,25 t
min
=50 C
R
m
= 410 MPa
R
p0,2/t*
= 234 MPa
Effective stress concentration factorK
eff
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors. k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
R
z
= 50 m (table 18-8)
e
n
= 21.2 mm

D
= 279,3 MPa (table 18-10 for N 2
.
10
6
cycles)
N = 200000 (for the first iteration)

R
= 293.4 MPa (allowable stress range for N cycles) at the last iteration
Stresses

eq,t
(total or notch equivalent stress range) = 361.7 MPa

struc
(structural equivalent stress range) = 293.4 MPa
(obtained by quadratic extrapolation along the main shell inside surface)
2120 . 1
D

struc

t
K 0,5 1
1)
t
(K 1,5
1
eff
K
+

+
1
1
]
1

Theoretical elastic stress concentration factor K


t
K
t
=
eq,t
/
struc
= 1.2329
18.8 Plasticity correction factor k
e
mechanical loading
If
eq,l
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..

total
= k
e
.
struc
= .. MPa
Else
total
=
struc
= 361.7 MPa
Thermal loading
If
eq,l
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..

total
= k

.
sruc
= .. MPa
Else
total
=
struc
= .. MPa
) stress equivalent notch (mean MPa 402.0
eq

18-10-6-2 Temperature correction factor f
t*
For t* > 100 C
f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= ..
Else f
t*
= 1
18-11-1-1 Surface finish correction factor f
s
f
s
= F
s
[0,1ln(N)-0,465]
with F
s
=1- 0,056 [ln (R
z
)]
0.64
[ln(R
m
)] +0,289 [ln (R
z
)]
0,53
= 0.7889
f
s
= 0.8568

eqmax
= 442.2 MPa (maximum notch equivalent stress)
DBA
Design by Analysis
Analysis Details
Example 3.2 / F-Check
Page
7.172 (A)
Cyclic pressure
The maximum total stress range occurs in the crotch corner an unwelded region
DBA
Design by Analysis
Analysis Details
Example 3.2 / F-Check
Page
7.173 (A)
18-11-1-2 Thickness correction factor f
e
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
e
= F
e
[0,1ln(N)-0,465]
f
e
= 0,7217
[0,1ln(N)-0,465]
with Fe = (25/en)
0.182
=
f
e
= 1 f
e
=

.. f
e
= ..
18-11-1-3 Mean stress correction factor f
m
If
struc
< 2 R
p0,2/t*
and
eqmax
< R
p0,2/t*
If
eq
> 0 then
r eq,
= R
p0,2/t*
-
2
t eq,
If
eq
< 0 then
r eq,
=
2
t eq,

- R
p0,2/t*

eq
=
r eq,
= 53.14 MPa
For N 2
.
10
6
cycles
See figure 18-14
f
m
= ..
If
struc
>2 R
p0,2/t*
18-11-2-1 Overall correction factor f
u
f
u
=
eff
K
* t
.f
m
.f
e
.f
s
f
= 0.6993
18-11-3 Allowable number of cycles N
N =
2
5 , 11
m
0.63R
u
f
struc eq,

4
10 4.6
1
1
1
1
1
]
1

if N 2
.
10
6
cycles
if R
p0,2/t*

eq

( ) M 1 2
R

+
then
f
m
=
( )
5 , 0
eq
2
M 1
M 2 M
- 1
R
1
1
]
1

,
_

+
+
= 0.9892
N =
10
u
f
struc eq,

92
m
R 2.7
1
1
1
1
1
]
1

+
if 2
.
10
6
N 10
8
cycles
N = 70865 cycles
N is obtained by iterations. If the value of N oscillates, a difference equal to 1 % between two iterations is
acceptable. If the values decrease monotonously, the difference must be less than 0,001 % between two
iterations.
If
struc
< 2 R
p0,2/t*
and
eqmax
> R
p0,2/t*
if
( ) M 1 2
R

+

eq
R
p0,2/t*
then
f
m
=

,
_

+
+
R

eq
2
3
M
M 1
3 M 1
= .
For N 2
.
10
6
cycles
M = 0.00035 R
m
0,1 = 0.0435
f
m
= 1
Data
t
max
= 50 C
t
min
= 50 C
t* = 0,75 t
max
+ 0,25 t
min
= 50 C
R
m
= 540 MPa
R
p0,2/t*
= 343 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of
the stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and
thermal equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied
to the thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 7,5 mm

D (5
.
10
6
cycles)
= 52 MPa (class 71)
equivalent stresses or principal stresses
m = 3 C = m = 3 C

= 7.16
.
10
11
C
//
= ..
m = 5 C = m = 5 C

= ..
C
//
= ..
Stresses

struc
= 797.8 MPa (structural equivalent stress range)
(obtained by quadratic extrapolation on outside of nozzle)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= 0.4133
k
e
= 1.0674
= k
e

struc
= 851.5 MPa
Else

=
struc
= ..
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

=


= k

struc
=
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
e
= (25/e
n
)
0.25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0.639
DBA
Design by Analysis
Analysis Details
Example 3.2 / F-Check
Page
7.174 (A)
Cyclic moment / Principal stress range approach
Critical point: Weld nozzle to shell, weld toe in branch longitudinal symmetry plane
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 851.5 MPa
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= = 1.03 1.5
.
10
-4
t* -1.5
.
10
-6
t*
2
= ..
Else f
t*
= 1
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = 7.16
.
10
11
N =
m
w
f
C

,
_


= 1160 cycles
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= .. cycles
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
DBA
Design by Analysis
Analysis Details
Example 3.2 / F-Check
Page
7.175 (A)
Analysis Details
Example 4
Page
7.176 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using Elastic Compensation
FE-Software: ANSYS 5.4
Element Types: 8 node, 3-D isoparametric shell elements
Boundary Conditions: Symmetry boundary conditions along the vertical cutting plane of
the model
Hoop and longitudinal displacements in nodes at the open end of
the cylindrical shell constrained to zero
Hoop displacement in nodes at open end of nozzle constrained to
zero
Longitudinal displacements coupled at nodes on open end of
nozzle (plane sections remain plane)
Model and Mesh:
Results:
Maximum admissible action according to the GPD-Check:Internal pressure PS
max GPD
= 0.41 MPa
Check against PD: Shakedown limit pressure PS
max SD
= 0.27 MPa
Member:
Strathclyde
Number of elements - 1002
Analysis Details
Example 4
Page
7.177 (S)
DBA
Design by Analysis
1. Finite Element Mesh
Element divisions for the finite element mesh were defined parametrically to allow modification of
the element density. Half symmetry was used with finite element models created using high order
8-node 3-D shell elements. The allowable pressure according to GPD and the shakedown pressure
according to PD were calculated for this model.
Boundary conditions applied to the model are symmetry about the vertical plane as both geometry
and loading are symmetrical about this plane. The nodes at the open end of the cylinder are
constrained in the hoop direction and the longitudinal direction. The nodes on the top surface of
the nozzle are constrained in the hoop direction and are coupled in the longitudinal direction (plane
sections remain plane). Internal pressure is applied over the entire inside surface of the model. A
thrust is applied to the top edge of the elements on the nozzle to model the thrust on the closed end
resulting from the internal pressure. It is important to model a large shell length as the tangential
nozzle has a large effect down the length of the cylinder.
2. Material properties
Material strength parameter MPa RM 202 = ; modulus of elasticity GPa E 6 . 183 =
3. Determination of the maximum admissible pressure according to the GPD-Check
Using the application rule in prEN-13445-3 Annex B.9.2.2 to check against GPD the principle is
fulfilled when for any load case the combination of the design actions do not exceed the design
resistance. This may be shown by calculating the limit load. The limit load has to be determined
using Trescas yield condition and associated flow rule. However, for shell elements the limit load
from elastic compensation is calculated using a generalised yield condition based on Mises'.
From prEN 13445-3 Annex B, Table B.9-3 the partial safety factor,
R
for the resistance is 1.25.
With a proof strength of 202 MPa the design material parameter is 161.6 MPa. The analysis was
carried out using the elastic compensation method conforming to the direct route rules for GPD in
Annex B: linear elastic - ideal plastic material law, Trescas yield condition and associated flow rule
and first order theory. For shell elements it is not currently possible to calculate limit loads based
directly on Trescas condition from elastic compensation. Models utilising shells have only one
element through thickness. Therefore, it is not possible to carry out an elastic compensation
analysis in the same way as in solid elements. Instead of carrying out the analysis using a Tresca
or Mises model directly a generalised yield model is used which considers the elements thickness.
In elastic compensation, the Ilyushin generalised yield model is used in the limit load calculation.
Ilyushin's model is based upon Mises' condition the limit load will require correction to meet the
code rules on the use of the Tresca condition. The maximum ratio of Mises' equivalent stress to
Tresca's equivalent stress for the same load is 2/3. Therefore, applying a factor of 3/2 to the yield
stress in the Mises analysis (or to the limit load, as the analysis is elastic) will always lead to a
conservative result on the Tresca condition. In effect, the Mises yield locus is being reduced to fit
within the Tresca yield locus. The applied design material parameter, considering Mises' correction
is 140 MPa.
In each elastic compensation iteration an equilibrium stress field is produced where the elastic
modulus of each subsequent iteration is defined by the previous elastic solution based upon the
Ilyushin generalised yield model. In this way regions of the FE-model may be systematically
stiffened or weakened depending upon the stress magnitudes in the previous solution. The result is
that every equilibrium stress field is a lower bound of the limit load, the stress field giving the
Analysis Details
Example 4
Page
7.178 (S)
DBA
Design by Analysis
Figure 7.9.1-1: Limit Ilyushian Stress Field
greatest limit load is taken as the closest to the actual limit state and defines the limit load in the
analysis. Convergence of the equilibrium stress field will occur typically between 8 and 15
iterations, i.e. no further benefit in the limit load will be noted with subsequent iterations.
In the check against GPD, the loading considered is increasing pressure, i.e. the limit pressure is to
be calculated. The total computing time to run each analysis on a 300 MHz Pentium two processor
with 128 Mb RAM Windows NT workstation was 820 seconds. The stress fields were shown to
converge after ten iterations. Iteration number five gives the highest lower bound on the limit load
and the square of the squared Ilyusin function distribution is shown in Figure 7.9.1-1. The contour
units are dimensionless and are
termed the square of the Ilyushin
function, where the Ilyushin
function represents the ratio of
actual stress to the yield stress of
the material, i.e.
Where f(IL) is the Ilyushin
function,
e
is the element stress
and
Y
the yield stress of the
material (or in this analysis the
design material strength
parameter). Therefore, for the
applied loading to be a lower
bound on the limit load the
Ilyushian function anywhere in
the redistributed limit field
cannot exceed 1. Therefore, the
limit multiplier on the applied load is given as
Where P
L
is the limit load, P
ap
the applied load and IL
max
the maximum of the Ilyushin function in
the limit field. For an applied load of 1 MPa the maximum squared Ilyushin function in the limit
field in Figure 7.9.1-1 is 4.142 giving a limit pressure of
Application of the partial safety factor given in prEN-13445-3 Annex B for a pressure without
natural limit to the value of the design action is
p
=1.2. The allowable load according to GPD is
thus given by
Y
e
IL f

= ) (
max
1
IL
P P
ap L
=
MPa P
L
491 . 0
142 . 4
1
1 = =
MPa PS
p
GPD
41 . 0
2 . 1
491 . 0 491 . 0
max
= = =

Analysis Details
Example 4
Page
7.179 (S)
DBA
Design by Analysis
Figure 7.9.1-2: Residual Stress Field (Mises)
4. Check against PD
In the check against progressive plastic deformation, the principle in prEN-13445-3 Annex B.9.3.1
is fulfilled if the structure can be shown to shake down.
In elastic compensation the load at which the structure will shake down is simple to calculate. Using
Melans lower bound shakedown
theorem, as self-equilibrating stress
field the residual stress field that
would result after a loading cycle can
be used. It can usually be calculated
by subtracting the linear-elastic stress
field at the limit pressure from the
limit stress field. The residual stress
field is in effect the resulting stress
from an elastic unloading from the
limit state back to zero pressure. If no
stress in the residual field violates the
yield condition, i.e. if there is no
equivalent stress above the material
parameter, then the shakedown load is
equal to the limit load. Where the
residual stress field of a structure does
exceed the yield condition, the
shakedown limit can be calculated by evoking the proportionality of the linear elastic solution. It is
not practical to calculate the shakedown load using a generalised yield model as used for calculating
the limit pressure. In calculating the shakedown load the limit equilibrium stress fields are created
based on the generalised yield model, with the residual stress fields calculated using Mises'
equivalent stress. For the material strength parameter of 202 MPa and an applied load of 0.345
MPa, the residual stress field is shown in Figure 7.9.1-2. The maximum equivalent residual stress
is 256.8 MPa and the equilibrium limit stress field for this iteration has a maximum of 237.8 MPa.
Therefore, the residual stress field violates the yield condition. Invoking the proportionality of the
elastic solution gives an allowable load according to PD of
For the same applied load of 0.345 MPa the maximum elastic Mises equivalent stress is 474.8 MPa,
giving a load to first yield of 0.147 MPa. The upper limit on the pressure according to PD given in
the application rule in Annex B.9.3.2 is the pressure where the elastic stress is limited to twice RM.
For case 4 the upper limit on the pressure according to this application rule is 0.294 MPa.
5. Check against GPD Using Non-linear Analysis
A check against GPD was also performed for the same FE-model using conventional non-linear
analysis. In this way, a direct comparison may be made between the two limit approaches. The FE
geometry, mesh and boundary conditions are the same as those used in the elastic compensation
analysis. Material non-linearities were applied corresponding to the design material parameter,
MPA PS
SD
271 . 0
8 . 256
202
345 . 0
max
= =
Analysis Details
Example 4
Page
7.180 (S)
DBA
Design by Analysis
Figure 7.9.1-4: Principal Strain Field Prior to Loss of Equilibrium (shell top)
Figure 7.9.1-3: Equivalent Stress Field Prior to Loss of Equilibrium (shell top)
140 MPa and again perfect
plasticity. The pressure is
ramped until the analysis
terminates due to non-
convergence, i.e. equilibrium
can no longer be maintained
between the external applied
forces and the internal forces.
The limit load can be defined
as the last converged solution
in the analysis. Figure
7.9.1-3 shows the equivalent
stress, at the top surface of
the shell, for the last
converged solution (prior to
loss of equilibrium). The
corresponding load is 0.572
MPa.
According to the rules for
the GPD-check in prEN-
13445-3 Annex B, the
maximum principal strain
must not exceed +/- 5%.
Figure 7.9.1-4 shows the
absolute maximum total
principal strain on the top
shell surface, as can be seen
the maximum principal
strain is 13.85%. According
to the code, the load at the
loss of equilibrium is too
high. At a load of 0.502
MPa the maximum absolute
value of principal strain is
within the 5% limit. Figure
7.9.1-5 shows the Mises
equivalent stress field at the
design value of the pressure
0.502 MPa.
Analysis Details
Example 4
Page
7.181 (S)
DBA
Design by Analysis
Applying the rules for the GPD-check in prEN-13445-3 Annex B, the allowable load is found by
applying the partial safety factors on the design value of the action. The limit analysis was carried
out using Mises' condition and correction was applied to the material strength parameter as with the
elastic compensation solution above. The allowable load according to GPD using non-linear
analysis is given as
Running on the same equipment as the elastic compensation analysis, the non-linear analysis
required a CPU time of 3810 seconds.
MPa PS
p
GPD
418 . 0
2 . 1
502 . 0 502 . 0
max
= = =

Figure 7.9.1-5: Equivalent Stress (shell top) at Design Value of Pressure (0.502MPa)
Analysis Details
Example 4
Page
7.182 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using elasto-plastic FE calculations
FE-Software: ANSYS 5.4
Element Types: Model: 8-node structural shell SHELL93
Submodel:10-node, 3-D tetrahedral structural solid SOLID87
Boundary Conditions: Model: Symmetry b.c. in the symmetry plane of the structure
and in the horizontal plane at the lower end of the cylinder;
horizontal displacement in one node at the lower end of the
cylinder constrained to 0.
Submodel: Symmetry b.c. in the symmetry plane of the
structure; b.c. at the cut boundaries according to submodelling.
Model and Mesh:
Results: Max. allowable pressure, according to GPD, MPa PS
GPD
375 . 0
max
=
Shakedown pressure MPa PS
SD
289 . 0
max
= ,
Number of allowable action cycles with a maximum pressure of
GPD
PS
max
:
1440 =
PD
N
Member:
A&AB
Model: Number of shell elements: 1002
Submodel: Number of solid elements: 19526
Geometry: see Analysis Details
Analysis Details
Example 4
Page
7.183 (S)
DBA
Design by Analysis
1. Model geometry, boundary conditions, elements
The structure is modelled with the nozzle closed by a flat end, as specified. Without the flat end
relatively large warping and ovalization of the nozzles upper end occur.
Due to the large diameter-to-wall thickness ratio of the structure 8-node shell elements SHELL93
have been used for the GPD check and for the shakedown check. Additionally, a submodel was
used to ensure that the maxima of the self-equilibrating stress state used in the shakedown check
were located in the surface of the structure (see subsection 3.3.3.2 of section 3 - Procedures). To
model the whole structure with solid elements and using an appropriate number of elements in wall
thickness direction would result in an FE-model with too many elements and nodes, and, therefore,
in an unacceptable long computation time.
Because of the non-rotational symmetry of the structure, model details and boundary conditions can
influence the results of the calculations strongly. Generally, whenever a cylindrical shell structure is
prone to ovalization and/or warping model details and boundary conditions require special
attention, and even investigation.
The influence of three different kinds of boundary conditions and four different lengths of the
cylindrical shell adjacent to the dished end on linear-elastic stress results and on the internal limit
pressure was investigated, the results are given briefly in the following. Because of the problems
symmetry, for all calculations a model comprising half of the structure was used. Following prEN
13445-3 Annex B.9.2.1 a linear-elastic ideal-plastic constitutive law with a design material strength
parameter of 138.5 MPa (see the determination of the maximum allowable pressure according to
GPD) was used. The modulus of elasticity used in the calculations was 183.6 GPa.
In all cases the 5 % strain limit criterion - see prEN 13445-3 Annex B.9.2.2 was governing the
calculation of the (limit) pressure.
The different shell lengths used were: 500mm, 1000mm, 2000mm and 3000mm.
The boundary conditions considered were:
(a) Symmetry boundary condition in the nodes in the symmetry plane of the structure; vertical
(longitudinal) and hoop displacements in the nodes in the end of the cylindrical shell
constrained to zero, i.e. no warping and practically no ovalization of the cylinder is possible.
(b) Symmetry boundary condition in the nodes in the symmetry plane of the structure; vertical
(longitudinal) displacements in the nodes in the end of the cylindrical shell constrained to zero,
i.e. no warping but ovalization of the cylinder is possible.
(c) Symmetry boundary condition in the nodes in the symmetry plane of the structure; symmetry
boundary condition (vertical displacement, longitudinal and hoop rotations constrained to zero)
in the nodes in the end of the cylindrical shell constrained to zero, i.e. no warping but
ovalization of the cylinder is possible.
Table 7.9.2-1 shows the calculation results corresponding to the different shell lengths and
boundary conditions, i.e. the (limit) pressure where the maximum absolute principal strain does not
exceed 5 %, and the maximum Mises linear-elastic equivalent stress according to an internal
pressure of 0.1 MPa.
Analysis Details
Example 4
Page
7.184 (S)
DBA
Design by Analysis
Shell length boundary condition (a) boundary condition (b) boundary condition (c)
(limit) pressure lin.-el. 0.1MPa (limit) pressure lin.-el. 0.1MPa (limit) pressure lin.-el. 0.1MPa
[mm] [MPa] [MPa] [MPa] [MPa] [MPa] [MPa]
500 5.8 129.1 4.5 137.4 4.5 137.4
1000 5.7 132.5 4.5 138.4 4.5 138.4
2000 5.0 137.7 4.5 138.6 4.5 138.6
3000 4.5 138.6 4.5 139.0 4.5 139.0
Table 7.9.2-1
As can be seen from this table, the results are for shell lengths greater 500 mm independent from
the length of the cylindrical shell in the case of boundary condition (b) or (c), and the results are
practically the same for both boundary conditions. If boundary condition (a) is used, the results
depend on the length of the cylindrical shell, the structures stiffness increases with decreasing
cylinder length, and the results are different from those obtained by using boundary condition (b) or
(c), with the exception of the longest model. Therefore, boundary condition (c) and a cylindrical
shell length of 1000 mm was used for all further calculations.
Additional remark: Generally, if a 90 - model is used instead of a 180 - model, the structure
becomes slightly stiffer, the (limit) pressure of the 90 - model is about 3 % to 10% larger than the
corresponding one of the 180 - model.
The submodel used for the check of the self-equilibrating stress field in the shakedown check
consisted of 19526 elements SOLID87. To obtain proper stress results, elements with midside-
nodes and a maximum global element size of 5 mm were used in the (free) meshing of the structure.
The submodel is bounded by a cylinder with radius 186,5 mm and by a horizontal plane, located 85
mm below the flat end. In the plane of symmetry a symmetry boundary condition was applied to all
nodes, the boundary conditions of the cut-boundaries were interpolated by the software from the
corresponding displacements of the coarse (shell) model.
2. Determination of the maximum allowable pressure according to the GPD-check
Since the subroutine for Trescas yield condition showed bad convergence, the check against GPD
was carried out using Mises yield condition only. To evaluate the 5% limit of the absolute principal
strain more accurately than with the results obtained from the shakedown check, a separate
calculation was carried out (see subsection 3.3.1 of section 3 - Procedures). The analysis was
carried out with a linear-elastic ideal-plastic material law, Mises' yield condition with a design
material strength parameter of MPa 5 . 138 for the shell, the nozzle and the flat end, associated flow
rule and first order theory. The value for the design material strength parameter is equal to the 1.0 %
proof strength of X6CrNiMoTi 17-12-2 according to prEN 10028-7 at 180C, MPa 200 , divided
by the partial safety factor for the material 25 . 1 =
R
and multiplied with the factor 2 3 (since
Mises yield condition was used instead of Trescas).
Analysis Details
Example 4
Page
7.185 (S)
DBA
Design by Analysis
The last convergent
solution showed an
internal pressure of
MPa 457 . 0 after a
computation time of
about 3 hours on the
Compaq Professional
Work-station 5000 with
two Pentium Pro
processors and 256 MB
RAM. Since the
maximum absolute
value of the principal
strain exceeded 5% at
the last load level the
absolute principal strain
value was 5.8 % (but
only in a very small part
of the structure) - a
lower load value with
appropriate strains had
to be used as limit
value.
Figure 7.9.2-1
At an internal pressure
of MPa 451 . 0 the
maximum absolute
value of the principal
strain was about 5%.
Figure 7.9.2-1 shows
the distribution of the
corresponding Mises
equivalent stress, and
Figure 7.9.2-2 shows
the distribution of the
corresponding maxi-
mum absolute principal
strain the maximum
value is located in the
symmetry plane of the
structure, on the inside
of the nozzle, slightly
above the intersection
curve.
Figure 7.9.2-2
Analysis Details
Example 4
Page
7.186 (S)
DBA
Design by Analysis
According to prEN 13445-3 Annex B, Table B.9-2, the partial safety factor for pressure loads
(without natural limit)
P
is 1.2. Therefore, the maximum allowable pressure according to GPD is
. 375 . 0
2 . 1
451 . 0
max
MPa PS
GPD
= =
3. Check against PD
The elasto-plastic FE-analysis was carried out as stated in prEN 13445-3 Annex B.9.3.1 - using
Mises yield condition and associated flow rule, a linear-elastic ideal-plastic constitutive law with
design material strength parameters of 200 MPa for the shell , the flat end and the nozzle, and first
order theory.
To ensure that the maxima of the used self-equilibrating stress field at the critical parts are located
in the surface of the structure (see subsection 3.3.3.2 of section 3 Procedures), a submodel was
used. According to this proof, the shell model can be used for the shakedown check.
By defining and using load cases in ANSYS, the superposition of stress fields can be done easily.
Therefore, the first load step of the analysis was defined at a very low load level ( MPa 1 . 0 ), so that
there was linear-elastic response of the structure. All other linear-elastic stress fields due to pressure
can then be found easily by scaling.
The analysis was carried
out using the Newton-
Raphson method. A
pressure close to the
(unknown) limit pressure
was found to be
MPa 651 . 0 after a
computation time of 3
hours on the Compaq
Professional Workstation
5000 with two Pentium
Pro processors and 256
MB RAM.
Figure 7.9.2-3 shows the
elasto-plastic Mises
equivalent stress field for
the pressure of
MPa 651 . 0 .
Figure 7.9.2-3
Analysis Details
Example 4
Page
7.187 (S)
DBA
Design by Analysis
Figure 7.9.2-4 shows the
linear-elastic Mises
equivalent stress field for
the pressure of
MPa 651 . 0 the stress
maximum is located in the
symmetry plane of the
structure, on the inside of
the nozzle, slightly above
the intersection curve. The
second region with large
linear-elastic stresses is a
part of the intersection
curve.
Figure 7.9.2-4
As a self-equilibrating stress field of the structure, the difference of the elasto-plastic and the linear-
elastic stress fields corresponding to the pressure of MPa 651 . 0 was used. A check with the
submodel verified, that the corresponding stress maxima are located in the surface of the structure.
The self-equilibrating stress field of the submodel was generated by performing two calculations:
An elastic-plastic one and a corresponding linear-elastic one. Creating two load cases, copying the
linear-elastic load case into the working directory of the elastic-plastic load case and superposition
of the load cases renders the stress field. The corresponding computation times were 7 minutes for
the linear-elastic and 4 hours and 5 minutes for the elastic-plastic calculation on the Compaq
Professional Workstation 5000 with two Pentium Pro processors and 256 MB RAM. Since the
stress maxima are located in the surface of the structure, the self-equilibrating stress field obtained
by the shell model was scaled down the scaling-down factor, for which the yield condition is not
violated by the stress field, is given by 281 . 0 = (see subsection 3.3.2.5 of section 3 - Procedures).
Analysis Details
Example 4
Page
7.188 (S)
DBA
Design by Analysis
Figure 7.9.2-5 shows the Mises
equivalent stress distribution of this
scaled self-equilibrating stress
field.
Figure 7.9.2-5
Figure 7.9.2-6 shows the Mises
equivalent stress distribution at the
lower bound shakedown limit, the
scaling factor is given by 443 . 0
(see subsection 3.3.2.5 of section 3 -
Procedures).
Thus, the shakedown limit pressure
is given by
MPa PS
SD
289 . 0 651 . 0 443 . 0
max
= =
Figure 7.9.2-6
Analysis Details
Example 4
Page
7.189 (S)
DBA
Design by Analysis
The determination of the allowable pressure according to the PD - check using the application rule
given in prEN 13445-3, Annex B.9.3.2, leads to a limit pressure of 0.289 MPa the very same
value as obtained by performing the SD - check. Since the limit pressure according to the GPD -
check is larger than the one resulting from the SD-check and the application rule in Annex B.9.3.2,
further investigations of the cyclic behaviour of the structure were performed.
FE - calculations with an internal pressure cycling between a maximum value of 0.375 MPa and a
minimum value of 0 MPa were performed. The behaviour of the structure was considered for 4 full
action cycles (the numbers 0, 2, 4, and 8 in the load history correspond to an internal pressure of 0
MPa, and the numbers 1, 3, 5 and 7 to an internal pressure of 0.375 MPa; these numbers also
correspond to the calculated half-cycles).
Plastification occured only in the final part of each half cycle, close to the maximum and minimum
pressures, as can be seen from the MxPl parameter from the solution history information file
(jobname.mntr) written by the FE software ANSYS. To investigate whether alternating
plasticity and/or progressive plastic deformation occurs, the cyclic stress and strain behaviour of the
structure has to be considered. A necessary (but not sufficient) condition for alternating plasticity is
that the stress states for the upper (or lower) extreme values of the actions remain constant, a
necessary and sufficient condition for alternating plasticity is that the sum of the plastic strains (for
every component) during a full cycle is zero.
To identify the critical location of the structure, i.e. the locations where detailed investigations have
to be performed, it is useful to create a difference - load case, i.e. a fictitious load case where the
results are given by the differences of the results of two real load cases for the same action. By
consideration of the stresses and the strains of this difference - load case the critical location can
usually be identified, but quite often only the stresses can be used for this purpose, since the
differences in the (plastic) strains are too small.
Figure 7.9.2-7 shows the
Mises equivalent stress of
the difference - load case
created with the load states
5 and 7 (maxiumum
pressure). As can be seen
from this figure, there are
two locations where the
stress differences are
noticeable: around node
2503, which is located in the
symmetry plane of the
structure, and around node
1161, which is located in the
intersection seam of the
shell and the nozzle.
Figure 7.9.2-7
ANSYS 5.5.3
OCT 4 1999
09:09:50
NODAL SOLUTION
STEP=3
SUB =1
SEQV (AVG)
PowerGraphics
EFACET=1
AVRES=Mat
DMX =.226E-03
SMN =.506E-06
SMX =5.205
1
MX
2
MX
.506E-06
.578302
1.157
1.735
2.313
2.892
3.47
4.048
4.626
5.205
.506E-06
.578302
1.157
1.735
4.048
4.626
5.205

Analysis Details
Example 4
Page
7.190 (S)
DBA
Design by Analysis
For plotting and listing the plastic strains which occur in these locations during the cycles, the time
history post processor of the FE - software was used. To obtain accurate results without averaging,
the element solution values, i.e. the strains calculated in the integration points and copied to the
adjacent node of the element, are used. Since one node belongs to more than one element the
corresponding maximum values shall be used in this procedure. Figure 7.9.2-8 shows the principal
plastic strains on the inner and outer surfaces of the shell model in the critical nodes 2503 and 1161
versus the load history.
Figure 7.9.2-8
Figure 7.9.2-8 shows clearly, that, in the considered history interval, alternating plasticity and progressive plastic
deformation occur on the inner surface in node 2503 and on the outer surface in node 1161, and that progressive plastic
deformation only occurs on the outer surface in node 2503, and that the plastic strains remains constant on the inner
surface in node 1161.
The strain differences of the full cycles decrease with increasing cycle number, and, therefore, it is
possible that the structure shakes down to constant cyclic behaviour, i.e. alternating plasticity, after
a certain number of action cycles, but this behaviour can only be verified by a large number of
simulation cycles, and, therefore, is unsuitable for a practical procedure. Thus, the maximum strain
difference can be used to calculate an allowable number of cycles, using the following proposal
1
EPPL1
EPPL2
EPPL3

1
EPPL1
EPPL2
EPPL3

1
EPPL1
EPPL2
EPPL3

1
EPPL1
EPPL2
EPPL3

Analysis Details
Example 4
Page
7.191 (S)
DBA
Design by Analysis
concerning the PD - check in prEN 13445-3, Annex B: If it can be shown via cyclic simulations,
that the maximum/minimum principle strain after the specified number of cycles is less than +-5%
PD does technically not occur.
Thus, the allowable number of cycles
i PD
N for a specific location of the structure and for the i
th
principal strain (i = 1, 2, 3) can be calculated according to the following formulae:
n n pl i
n pl i c el i
i PD
N
) 2 (
) (
) ( ) ( 5


,
where n number of half-cycles in the simulation,
c el i
) ( maximum linear-elastic i
th
principal strain occuring during the
full cycle in [%],
n pl i
) ( plastic i
th
principal strain strain after the n
th
half-cycle in [%],
n n pl i ) 2 (
) (

difference of the i
th
plastic principal strain for the last full cycle
in [%].
According to the number of calculated cycles for the structure under consideration, the above
formulae is now given by
8 6
8
) (
) ( ) ( 5
pl i
pl i c el i
i PD
N


= .
Table 7.9.2-2 lists the strain values and the corresponding number of allowable action cycles for the
two critical nodes and the inner and outer surface of the structure, respectively.
The most critical location is given by node 1161 on the outer surface, where the difference of the 1
st
plastic principal strain for the last action cycle is calculated to be 2.92
.
10
-3
%. Thus, the number of
allowable action cycles for an maximum internal pressure of 0.375 MPa is given by
1440 =
PD
N .
Note: If this number of action cycles is too small, further cyclic simulations with probably
decreasing plastic strain differences would be necessary.
Analysis Details
Example 4
Page
7.192 (S)
DBA
Design by Analysis
node 1161, element 945
outer surface
node 1161, element 945
inner surface
node2503, element 959
outer surface
node2503, element 959
inner surface
(
1 el
)
c
0,11302 not relevant 0,11273 0,11311
(
1 pl
)
24
5,71
.
10
-3
0 0,34
.
10
-3
7,92
.
10
-3
(
1 pl
)
46
3,93
.
10
-3
0 0,32
.
10
-3
4,14
.
10
-3
(
1 pl
)
68
2,92
.
10
-3
0 0,23
.
10
-3
2,25
.
10
-2
(
1 pl
)
8
0,68170 not relevant 0,04036 0,34479
N
PD1
1440 21059 2018
(
2 el
)
c
-0,02212 not relevant -0,0184 -0,0273
(
2 pl
)
24
-4,34
.
10
-3
0 -0,12
.
10
-3
-7,41
.
10
-3
(
2 pl
)
46
-3,03
.
10
-3
0 -0,11
.
10
-3
-3,93
.
10
-3
(
2 pl
)
68
-2,24
.
10
-3
0 -0,08
.
10
-3
-2,14
.
10
-3
(
2 pl
)
8
0,02476 not relevant -0,01405 0,01051
N
PD2
2211 62100 2319
(
3 el
)
c
-0,11612 not relevant -0,0421 -0,11242
(
3 pl
)
24
-1,42
.
10
-3
0 -0,22
.
10
-3
-0,52
.
10
-3
(
3 pl
)
46
-0,93
.
10
-3
0 -0,21
.
10
-3
-0,16
.
10
-3
(
3 pl
)
68
-0,72
.
10
-3
0 -0,16
.
10
-3
-0,10
.
10
-3
(
3 pl
)
8
-0,70646 not relevant -0,02631 -0,33427
N
PD3
5802 30825 45583
Table 7.9.2-2
Analysis Details
Example 4
Page
7.193 (S)
DBA
Design by Analysis
Materials and Properties: Material strength parameter R
p1.0/t
= 200 MPa,
modulus of elasticity E = 183600 MPa.
FE- Software: ALGOR
Element types: 7088 shell elements
Boundary conditions: yz-plane fixed in x (T
x
),
the shell ends fixed in z (T
z
),
xz-plane fixed in y (T
y
) and fixed rotations (R
xyz
)
Model and mesh:
Maximum admissible internal pressure according to the Stress Categorisation Route:
Internal pressure = 0.263 MPa.
Analysis Type:
Stress Categorisation Route
Member:
WTCM
Analysis Details
Example 4
Page
7.194 (S)
DBA
Design by Analysis
Six classification lines (CL) are considered :
- CL 1: through point 1 with coordinates (995;0;40).
- CL 2: through point 2 with coordinates (983.053;60.062;68.961).
- CL 3: through point 3 with coordinates (984.062;77.4474;56.3427).
- CL 4: through point 4 with coordinates (909.304;139.843;166.708).
- CL 5: through point 5 with coordinates (985.185;54.281;62.2765).
- CL 6: through point 6 with coordinates (886.55;149.268;180.84).
In the figure above the locations of the six classification lines are shown. The CLs are situated at
the nozzle-cylinder intersection. They are drawn through one point, perpendicular to the shell
surface. The stresses are calculated at the inner and outer diameter. These values are used to
calculate the membrane and bending components of the stresses.
Analysis Details
Example 4
Page
7.195 (S)
DBA
Design by Analysis
For the stress classification, the following procedure was followed :
- for each load acting on the vessel, calculate the elementary stresses
ij
(i,j = 1,2,3) in the
different points on the different CLs.
- for each load acting on the vessel and each CL, calculate the membrane stress components
ij,m
and the bending stress components
ij,b
.
- Classify the membrane stress components
ij,m
in (
ij
)
Pm
, (
ij
)
PL
or (
ij
)
Qm
and the bending
stress components
ij,b
in (
ij
)
Pb
or (
ij
)
Qb
.
- Calculate the sum of the stresses classified in this way for the set of loads acting simultaneously
on the vessel. The stresses resulting from this summation are designated (
ij
)
Pm
, (
ij
)
PL
, (
ij
)
Pb
,
(
ij
)
Qm
, (
ij
)
Qb
.
- From this deduce: (
ij
)
Pm
, (
ij
)
PL
, (
ij
)
P
, (
ij
)
P+Q
.
- Calculate the following equivalent stresses:
- (
eq
)
Pm
or (
eq
)
PL
,
- (
eq
)
P,
- (
eq
)
P+Q
.
According to table C-2, the following classification must be used for internal pressure: P
L
and
Q
b
.
- Verify the admissibility of the equivalent stresses.
In the case under consideration only one load is considered: internal pressure with an initial value =
0.583 MPa. The stresses were calculated for this load and a stress classification along the CLs 1 to
6 is applied. The results of the calculations are shown in the following tables.
CL 1 through point 1 (995;0;40) - Pressure = 0.583 MPa.
1 1' (
ij
)
PL
(
ij
)
Qb
(
ij
)
P+Q
S
11
148,4 -456 -153,8 302,2 -456
S
22
180 -538,5 -179,25 359,25 -538,5
S
33
0 0 0 0 0
S
12
-161,2 289,9 64,35 -225,55 289,9
S
23
0 0 0 0 0
S
31
0 0 0 0 0
(
ij,PL
)
1
0 (
ij,P+Q
)
1
0
(
ij,PL
)
2
-100,929 (
ij,P+Q
)
2
-204,43
(
ij,PL
)
3
-232,121 (
ij,P+Q
)
3
-790,07
(
eq
)
Pl
232,121 MPa
(
eq
)
P+Q
790,07 MPa
Analysis Details
Example 4
Page
7.196 (S)
DBA
Design by Analysis
CL 2 through point 2 (983.053;60.062;68.961) -
Pressure = 0.583 MPa.
2 2' (
ij
)
PL
(
ij
)
Qb
(
ij
)
P+Q
S
11
-345,5 280,6 -32,45 -313,05 -345,5
S
22
-444,1 304,2 -69,95 -374,15 -444,1
S
33
0 0 0 0 0
S
12
3,8 28,4 16,1 -12,3 28,4
S
23
0 0 0 0 0
S
31
0 0 0 0 0
(
ij,PL
)
1
0 (
ij,P+Q
)
1
0
(
ij,PL
)
2
-26,486 (
ij,P+Q
)
2
-337,905
(
ij,PL
)
3
-75,914 (
ij,P+Q
)
3
-451,695
(
eq
)
Pl
75,914 MPa
(
eq
)
P+Q
451,695 MPa
CL 3 trough point 3 (984.062;77.4474;56.3427) - Pressure = 0.583 MPa.
3 3' (
ij
)
PL
(
ij
)
Qb
(
ij
)
P+Q
S
11
-485,2 47,1 -219,05 -266,15 -485,2
S
22
-216,6 378,5 80,95 -297,55 378,5
S
33
0 0 0 0 0
S
12
98,5 16,3 57,4 41,1 98,5
S
23
0 0 0 0 0
S
31
0 0 0 0 0
(
ij,PL
)
1
83,615 (
ij,P+Q
)
1
389,591
(
ij,PL
)
2
0 (
ij,P+Q
)
2
0
(
ij,PL
)
3
-221,715 (
ij,P+Q
)
3
-496,291
(
eq
)
Pl
305,33 MPa
(
eq
)
P+Q
885,882 MPa
Analysis Details
Example 4
Page
7.197 (S)
DBA
Design by Analysis
CL 4 through point 4 (909.304;139.843;166.708) - Pressure = 0.583 MPa.
3 3' (
ij
)
PL
(
ij
)
Qb
(
ij
)
P+Q
S
11
324,8 293,9 309,35 15,45 324,8
S
22
215 -28,3 93,35 121,65 215
S
33
0 0 0 0 0
S
12
3,2 -70,8 -33,8 37 -70,8
S
23
0 0 0 0 0
S
31
0 0 0 0 0
(
ij,PL
)
1
347,522 (
ij,P+Q
)
1
359,492
(
ij,PL
)
2
55,178 (
ij,P+Q
)
2
180,308
(
ij,PL
)
3
0 (
ij,P+Q
)
3
0
(
eq
)
Pl
347,522 MPa
(
eq
)
P+Q
359,492 MPa
CL 5 through point 5 (985.185;54.281;62.2765) - Pressure = 0.583 MPa.
3 3' (
ij
)
PL
(
ij
)
Qb
(
ij
)
P+Q
S
11
-290,4 205,2 -42,6 -247,8 -290,4
S
22
-524,2 222,6 -150,8 -373,4 -524,2
S
33
0 0 0 0 0
S
12
8,7 83,3 46 -37,3 83,3
S
23
0 0 0 0 0
S
31
0 0 0 0 0
(
ij,PL
)
1
0 (
ij,P+Q
)
1
0
(
ij,PL
)
2
-7,596 (
ij,P+Q
)
2
-281,675
(
ij,PL
)
3
-185,804 (
ij,P+Q
)
3
-532,925
(
eq
)
Pl
185,804 MPa
(
eq
)
P+Q
532,925 MPa
Analysis Details
Example 4
Page
7.198 (S)
DBA
Design by Analysis
CL 6 through point 6 (886.55;149.268;180.84) - Pressure = 0.583 MPa.
3 3' (
ij
)
PL
(
ij
)
Qb
(
ij
)
P+Q
S
11
298,8 233,2 266 32,8 298,8
S
22
272,2 -51,9 110,15 162,05 272,2
S
33
0 0 0 0 0
S
12
-4,9 -96 -50,45 45,55 -96
S
23
0 0 0 0 0
S
31
0 0 0 0 0
(
ij,PL
)
1
302,142 (
ij,P+Q
)
1
337,674
(
ij,PL
)
2
74,008 (
ij,P+Q
)
2
233,326
(
ij,PL
)
3
0 (
ij,P+Q
)
3
0
(
eq
)
Pl
302,142 MPa
(
eq
)
P+Q
337,674 MPa
The calculated equivalent stresses must meet the assessment criteria:
With f = R
p1.0/t
/ 1.5 = 200 / 1.5 MPa = 133.3 MPa, the assessment criteria are met if the internal
pressure is reduced to 0.263 MPa (the most severe CL is CL3).
(
eq
)
Pl
< 1.5f
(
eq
)
Pl+Qb
< 3f
DBA
Design by Analysis
Analysis Details
Example 4 / F-Check
Page
7.199 (A)
Analysis Type:
Check against Fatigue
Member:
A&AB
FE Software: ANSYS 5.4
Element Types:
Model: 8-node structural shell SHELL93
Boundary Conditions:
Symmetry b.c. in the symmetry plane of the structure and in the
horizontal plane at the lower end of the cylinder;
horizontal displacement in one node at the lower end of the cylinder
constrained to 0.
Model and Mesh:
Results:
Fatigue life N = 4072 cycles
Model: Number of shell elements: 1002
Data
t
max
= 180 C
t
min
= 180 C
t* = 0,75 t
max
+ 0,25 t
min
= 180 C
R
m
= 540 MPa
R
p1,0/t*
= 202 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 8,5 mm

D (5
.
10
6
cycles)
= 52 MPa (class 71)
equivalent stresses or principal stresses
m = 3 C = 7.16
.
10
11
m = 3 C

= ..
C
//
= ..
m = 5 C = 1.96
.
10
15
m = 5 C

= ..
C
//
= ..
Stresses

struc
= 491.1 MPa (structural equivalent stress range)
(obtained by quadratic extrapolation along the outside nozzle generatrix)
Note: In this example the extrapolation pivot distances, as specifies in the draft standard, are too small to be used compared to the
used, mesh-size; the three closest nodal points have been used for extrapolation.
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p1,0/t*

1
1
]
1

+ 1
* p1,0/t
R 2
struc

4 , 0 1
e
k
k
e
= 1,0862
= k
e

struc
= 533.5 MPa
Else

=
struc
= ..
Thermal loading
If
struc
> 2 R
p1,0/t*
1
1
]
1

* p1,0/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= ..
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0.25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0.639
DBA
Design by Analysis
Analysis Details
Example 4 / F-Check
Page
7.200 (A)
Welded region / Structural equivalent stress range approach
Critical point: Weld toe in nozzle, maximum stress range point at outside nozzle
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 0.9656
18-10-7 Allowable number of cycles N
w
f

= 552.5 MPa
If
w
f

>
5 10
6
cycles then
M = 3 and C (C or C//
) = 7.16
.
10
11
N =
m
w
f
C

,
_


= 4246 cycles
If
w
f

<
5
.
10
6
cycles and other
cycles with
w
f

>
5
.
10
6
cycles
then
M = 5 and C (C or C//) = ..
N =
m
w
f
C

,
_


= .. cycles
If
w
f

<
5
.
10
6
cycles and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,043 4,3 10
-4
t* = 0.9656
Else f
t*
= 1
DBA
Design by Analysis
Analysis Details
Example 4 / F-Check
Page
7.201 (A)
In this example a large difference between the structural equivalent stress range approach and the principal
stress range approach is to be expected. Therefore both approaches have been applied. Details on the
principal stress range approach are given on the next two pages.
The minimum total stress range is in the unwelded region, at the outside of the nozzle, in the symmetric
plane of the model, and slightly above the weld. The details of the corresponding fatigue calculation is
given on the two pages after the next two.
Data
t
max
= 180 C
t
min
= 180 C
t* = 0,75 t
max
+ 0,25 t
min
= 180 C
R
m
= 540 MPa
R
p1,0/t*
= 202 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 8.5 mm

D (5
.
10
6
cycles)
= 52 MPa (class 71)
principal stresses or equivalent stresses
m = 3 C

= 7.16 10
11
m = 3 C = 7,16 10
11
C
//
= ..
m = 5 C

= m = 5 C = 1,96 10
15
..
C
//
= ..
Stresses

struc
= 432.9 MPa (structural equivalent stress range)
(obtained by quadratic extrapolation along the outside nozzle generatrix)
Note: In this example the extrapolation pivot distances, as specifies in the draft standard, are too small to be used, compared to
the used mesh-size; the three closest nodal points have been used for extrapolation.
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p1,0/t*

1
1
]
1

+ 1
* p1,0/t
R 2
struc

4 , 0 1
e
k
k
e
= 1,0286
= k
e

struc
= 445.3 MPa
Else

=
struc
= ..
Thermal loading
If
struc
> 2 R
p1,0/t*
1
1
]
1

* p1,0/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= ..
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0.25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0.639
Welded region / Principal stress range approach
Critical point: Weld toe in nozzle, maximum stress range point at outside nozzle
DBA
Design by Analysis
Analysis Details
Example 4 / F-Check
Page
7.202 (A)
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 0.9656
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,043 4,3 10
-4
t* = 0.9656
Else f
t*
= 1
DBA
Design by Analysis
Analysis Details
Example 4 / F-Check
Page
7.203 (A)
This value is noticeably larger than the result for the structural equivalent stress range approach.
The details of the calculation for the maximum total stress range in the unwelded region is given on the
next two pages.
18-10-7 Allowable number of cycles N
w
f

= 461.1 MPa
If
w
f

>
5
.
10
6
cycles then
M = 3 and C (C or C//
) = 7.16
.
10
11
N =
m
w
f
C

,
_


= 7303 cycles
If
w
f

<
5
.
10
6
cycles and other
cycles with
w
f

>
5
.
10
6
cycles
then
M = 5 and C (C or C//) = ..
N =
m
w
f
C

,
_


= .. cycles
If
w
f

<
5
.
10
6
cycles and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
Data
t
max
= 180 C
t
min
= 180 C
t* = 0,75 t
max
+ 0,25 t
min
=180 C
R
m
= 540 MPa
R
p0,2/t*
= 202 MPa
Effective stress concentration factorK
eff
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component
of the stress tensors. k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical
and thermal equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is
applied to the thermal

stress tensors. Then both tensors are added and the new stress range is calculated.

eq.struc
=
total
/ K
t
= 877.6 MPa (for age in 18-11-3)
R
z
= 200 m (table 18-8)
e
n
= 8.5 mm

D
= 361.2 MPa (table 18-10 for N 2 10
6
cycles)
N = 7000 (for the first iteration)

R
= 1049.6 MPa (allowable stress range for N cycles) at the last iteration
Stresses

eq,t
(total or notch equivalent stress range) = 808.0 MPa

struc
(structural equivalent stress range) = 808.0 MPa
(the maximum occurs in a region which is stress-concentration-free)
1
D

struc

t
K 0,5 1
1)
t
(K 1,5
1
eff
K
+

+
1
1
]
1

Theoretical elastic stress concentration factor K


t
K
t
=
eq,t
/
struc
= 1
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= 1.0862

total
= k
e
.
struc
= 877.6 MPa
Else
total
= .....
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..

total
= k

.
sruc
= .. MPa
Else
total
=
struc
= .. MPa
) stress equivalent notch (mean MPa 404.0
eq

18-10-6-2 Temperature correction factor f
t*
For t* > 100 C
f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= 0.9656
Else f
t*
= 1
18-11-1-1 Surface finish correction factor f
s
f
s
= F
s
[0,1ln(N)-0,465]
with F
s
=1- 0,056 [ln (R
z
)]
0,64
[ln(R
m
)] +0,289 [ln (R
z
)]
0,53
F
s
= 0.6751
f
s
= 0.8568

eqmax
= 808.0 MPa (maximum notch equivalent stress)
DBA
Design by Analysis
Analysis Details
Example 4 / F-Check
Page
7.204 (A)
18-11-1-3 Mean stress correction factor f
m
DBA
Design by Analysis
Analysis Details
Example 4 / F-Check
Page
7.205 (A)
18-11-1-2 Thickness correction factor f
e
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
e
= F
e
[0,1ln(N)-0,465]
f
e
= 0,7217
[0,1ln(N)-0,465]
with Fe = (25/en)
0.182
=
f
e
= 1 f
e
=

.. f
e
= ..
If
struc
< 2 R
p0,2/t*
and
eqmax
< R
p0,2/t*
If
eq
> 0 then
r eq,
= R
p0,2/t*
-
2
t eq,
If
eq
< 0 then
r eq,
=
2
t eq,

- R
p0,2/t*

eq
=
r eq,
=
For N 2 10
6
cycles
See figure 18-14
f
m
= 1
If
struc
>2 R
p0,2/t*
18-11-2-1 Overall correction factor f
u
f
u
=
eff
K
* t
.f
m
.f
e
.f
s
f
= 0.8362
eq, struc
/ f
u
= 1049.5 MPa
18-11-3 Allowable number of cycles N
N =
2
5 . 11
m
0,63R
u
f
struc eq,

4
10 4.6
1
1
1
1
1
]
1

if N 2
.
10
6
cycles
if R
p0,2/t*

eq

( ) M 1 2
R

+
then
f
m
=
( )
5 , 0
eq
2
M 1
M 2 M
- 1
R
1
1
]
1

,
_

+
+
=
N = if
eq, struc
<
2
.
10
6
N = 4072 cycles
N is obtained by iterations. If the value of N oscillates, a difference equal to 1 % between two iterations is
acceptable. If the values decrease monotonously, the difference must be less than 0,001 % between two
iterations.
If
struc
< 2 R
p0,2/t*
and
eqmax
> R
p0,2/t*
if
( ) M 1 2
R

+

eq
R
p0,2/t*
then
f
m
=

,
_

+
+
R

eq
2
3
M
M 1
3 M 1
= .
For N 2
.
10
6
cycles
M = 0.00035 R
m
0.1 = 0.0435
f
m
= 1
Analysis Details
Example 4 / GPF- & PD-Check
Page
7.176 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using Elastic Compensation
FE-Software: ANSYS 5.4
Element Types: 8 node, 3-D isoparametric shell elements
Boundary Conditions: Symmetry boundary conditions along the vertical cutting plane of
the model
Hoop and longitudinal displacements in nodes at the open end of
the cylindrical shell constrained to zero
Hoop displacement in nodes at open end of nozzle constrained to
zero
Longitudinal displacements coupled at nodes on open end of
nozzle (plane sections remain plane)
Model and Mesh:
Results:
Maximum admissible action according to the GPD-Check:Internal pressure PS
max GPD
= 0.41 MPa
Check against PD: Shakedown limit pressure PS
max SD
= 0.27 MPa
Member:
Strathclyde
Number of elements - 1002
Analysis Details
Example 4 / GPF- & PD-Check
Page
7.176 (S)
DBA
Design by Analysis
1. Finite Element Mesh
Element divisions for the finite element mesh were defined parametrically to allow modification of
the element density. Half symmetry was used with finite element models created using high order
8-node 3-D shell elements. The allowable pressure according to GPD and the shakedown pressure
according to PD were calculated for this model.
Boundary conditions applied to the model are symmetry about the vertical plane as both geometry
and loading are symmetrical about this plane. The nodes at the open end of the cylinder are
constrained in the hoop direction and the longitudinal direction. The nodes on the top surface of
the nozzle are constrained in the hoop direction and are coupled in the longitudinal direction (plane
sections remain plane). Internal pressure is applied over the entire inside surface of the model. A
thrust is applied to the top edge of the elements on the nozzle to model the thrust on the closed end
resulting from the internal pressure. It is important to model a large shell length as the tangential
nozzle has a large effect down the length of the cylinder.
2. Material properties
Material strength parameter MPa RM 202 = ; modulus of elasticity GPa E 6 . 183 =
3. Determination of the maximum admissible pressure according to the GPD-Check
Using the application rule in prEN-13445-3 Annex B.9.2.2 to check against GPD the principle is
fulfilled when for any load case the combination of the design actions do not exceed the design
resistance. This may be shown by calculating the limit load. The limit load has to be determined
using Trescas yield condition and associated flow rule. However, for shell elements the limit load
from elastic compensation is calculated using a generalised yield condition based on Mises'.
From prEN 13445-3 Annex B, Table B.9-3 the partial safety factor,
R
for the resistance is 1.25.
With a proof strength of 202 MPa the design material parameter is 161.6 MPa. The analysis was
carried out using the elastic compensation method conforming to the direct route rules for GPD in
Annex B: linear elastic - ideal plastic material law, Trescas yield condition and associated flow rule
and first order theory. For shell elements it is not currently possible to calculate limit loads based
directly on Trescas condition from elastic compensation. Models utilising shells have only one
element through thickness. Therefore, it is not possible to carry out an elastic compensation
analysis in the same way as in solid elements. Instead of carrying out the analysis using a Tresca
or Mises model directly a generalised yield model is used which considers the elements thickness.
In elastic compensation, the Ilyushin generalised yield model is used in the limit load calculation.
Ilyushin's model is based upon Mises' condition the limit load will require correction to meet the
code rules on the use of the Tresca condition. The maximum ratio of Mises' equivalent stress to
Tresca's equivalent stress for the same load is 2/3. Therefore, applying a factor of 3/2 to the yield
stress in the Mises analysis (or to the limit load, as the analysis is elastic) will always lead to a
conservative result on the Tresca condition. In effect, the Mises yield locus is being reduced to fit
within the Tresca yield locus. The applied design material parameter, considering Mises' correction
is 140 MPa.
In each elastic compensation iteration an equilibrium stress field is produced where the elastic
modulus of each subsequent iteration is defined by the previous elastic solution based upon the
Ilyushin generalised yield model. In this way regions of the FE-model may be systematically
stiffened or weakened depending upon the stress magnitudes in the previous solution. The result is
that every equilibrium stress field is a lower bound of the limit load, the stress field giving the
Analysis Details
Example 4 / GPF- & PD-Check
Page
7.176 (S)
DBA
Design by Analysis
Figure 7.9.1-1: Limit Ilyushian Stress Field
greatest limit load is taken as the closest to the actual limit state and defines the limit load in the
analysis. Convergence of the equilibrium stress field will occur typically between 8 and 15
iterations, i.e. no further benefit in the limit load will be noted with subsequent iterations.
In the check against GPD, the loading considered is increasing pressure, i.e. the limit pressure is to
be calculated. The total computing time to run each analysis on a 300 MHz Pentium two processor
with 128 Mb RAM Windows NT workstation was 820 seconds. The stress fields were shown to
converge after ten iterations. Iteration number five gives the highest lower bound on the limit load
and the square of the squared Ilyusin function distribution is shown in Figure 7.9.1-1. The contour
units are dimensionless and are
termed the square of the Ilyushin
function, where the Ilyushin
function represents the ratio of
actual stress to the yield stress of
the material, i.e.
Where f(IL) is the Ilyushin
function,
e
is the element stress
and
Y
the yield stress of the
material (or in this analysis the
design material strength
parameter). Therefore, for the
applied loading to be a lower
bound on the limit load the
Ilyushian function anywhere in
the redistributed limit field
cannot exceed 1. Therefore, the limit multiplier on the applied load is given as
Where P
L
is the limit load, P
ap
the applied load and IL
max
the maximum of the Ilyushin function in
the limit field. For an applied load of 1 MPa the maximum squared Ilyushin function in the limit
field in Figure 7.9.1-1 is 4.142 giving a limit pressure of
Application of the partial safety factor given in prEN-13445-3 Annex B for a pressure without
natural limit to the value of the design action is
p
=1.2. The allowable load according to GPD is
thus given by
Y
e
IL f

= ) (
max
1
IL
P P
ap L
=
MPa P
L
491 . 0
142 . 4
1
1 = =
MPa PS
p
GPD
41 . 0
2 . 1
491 . 0 491 . 0
max
= = =

Analysis Details
Example 4 / GPF- & PD-Check
Page
7.176 (S)
DBA
Design by Analysis
Figure 7.9.1-2: Residual Stress Field (Mises)
4. Check against PD
In the check against progressive plastic deformation, the principle in prEN-13445-3 Annex B.9.3.1
is fulfilled if the structure can be shown to shake down.
In elastic compensation the load at which the structure will shake down is simple to calculate. Using
Melans lower bound shakedown
theorem, as self-equilibrating stress
field the residual stress field that
would result after a loading cycle can
be used. It can usually be calculated
by subtracting the linear-elastic stress
field at the limit pressure from the
limit stress field. The residual stress
field is in effect the resulting stress
from an elastic unloading from the
limit state back to zero pressure. If no
stress in the residual field violates the
yield condition, i.e. if there is no
equivalent stress above the material
parameter, then the shakedown load is
equal to the limit load. Where the
residual stress field of a structure does
exceed the yield condition, the
shakedown limit can be calculated by evoking the proportionality of the linear elastic solution. It is
not practical to calculate the shakedown load using a generalised yield model as used for calculating
the limit pressure. In calculating the shakedown load the limit equilibrium stress fields are created
based on the generalised yield model, with the residual stress fields calculated using Mises'
equivalent stress. For the material strength parameter of 202 MPa and an applied load of 0.345
MPa, the residual stress field is shown in Figure 7.9.1-2. The maximum equivalent residual stress
is 256.8 MPa and the equilibrium limit stress field for this iteration has a maximum of 237.8 MPa.
Therefore, the residual stress field violates the yield condition. Invoking the proportionality of the
elastic solution gives an allowable load according to PD of
For the same applied load of 0.345 MPa the maximum elastic Mises equivalent stress is 474.8 MPa,
giving a load to first yield of 0.147 MPa. The upper limit on the pressure according to PD given in
the application rule in Annex B.9.3.2 is the pressure where the elastic stress is limited to twice RM.
For case 4 the upper limit on the pressure according to this application rule is 0.294 MPa.
5. Check against GPD Using Non-linear Analysis
A check against GPD was also performed for the same FE-model using conventional non-linear
analysis. In this way, a direct comparison may be made between the two limit approaches. The FE
geometry, mesh and boundary conditions are the same as those used in the elastic compensation
analysis. Material non-linearities were applied corresponding to the design material parameter,
MPA PS
SD
271 . 0
8 . 256
202
345 . 0
max
= =
Analysis Details
Example 4 / GPF- & PD-Check
Page
7.176 (S)
DBA
Design by Analysis
Figure 7.9.1-4: Principal Strain Field Prior to Loss of Equilibrium (shell top)
Figure 7.9.1-3: Equivalent Stress Field Prior to Loss of Equilibrium (shell top)
140 MPa and again perfect
plasticity. The pressure is
ramped until the analysis
terminates due to non-
convergence, i.e. equilibrium
can no longer be maintained
between the external applied
forces and the internal forces.
The limit load can be defined
as the last converged solution
in the analysis. Figure
7.9.1-3 shows the equivalent
stress, at the top surface of
the shell, for the last
converged solution (prior to
loss of equilibrium). The
corresponding load is 0.572
MPa.
According to the rules for
the GPD-check in prEN-
13445-3 Annex B, the
maximum principal strain
must not exceed +/- 5%.
Figure 7.9.1-4 shows the
absolute maximum total
principal strain on the top
shell surface, as can be seen
the maximum principal
strain is 13.85%. According
to the code, the load at the
loss of equilibrium is too
high. At a load of 0.502
MPa the maximum absolute
value of principal strain is
within the 5% limit. Figure
7.9.1-5 shows the Mises
equivalent stress field at the
design value of the pressure
0.502 MPa.
Analysis Details
Example 4 / GPF- & PD-Check
Page
7.176 (S)
DBA
Design by Analysis
Applying the rules for the GPD-check in prEN-13445-3 Annex B, the allowable load is found by
applying the partial safety factors on the design value of the action. The limit analysis was carried
out using Mises' condition and correction was applied to the material strength parameter as with the
elastic compensation solution above. The allowable load according to GPD using non-linear
analysis is given as
Running on the same equipment as the elastic compensation analysis, the non-linear analysis
required a CPU time of 3810 seconds.
MPa PS
p
GPD
418 . 0
2 . 1
502 . 0 502 . 0
max
= = =

Figure 7.9.1-5: Equivalent Stress (shell top) at Design Value of Pressure (0.502MPa)
Analysis Details
Example 4 / GPF- & PD-Check
Page
7.176 (S)
DBA
Design by Analysis
Analysis Details
Example 5
Page
7.206 (W)
DBA
Design by Analysis
Analysis Type:
Direct Route using Elastic Compensation
FE-Software: ANSYS 5.4
Element Types: 8 node, axisymmetric 2-D isoparametric solids.
Boundary Conditions: Constraint on the hoop displacement at the cut plane.
Axisymmetry
Model and Mesh:
Results:
Maximum admissible action according to the GPD-Check: Internal pressure PS
max GPD
= 11.1 MPa
Check against PD: Outside the scope of elastic compensation
Member:
Strathclyde
Number of elements - 611
Analysis Details
Example 5
Page
7.207 (W)
DBA
Design by Analysis
1. Finite Element Mesh
Finite element models were created using 8-node 2-D axisymmetric solids. The model used in the
analysis contained 620 elements. The allowable pressure according to the GPD-check and
admissibility of the DBF pressure and thermal stresses according to the PD-check required
assessment.
Boundary conditions applied to the model are axisymmetry, applied via a key option when defining
the element type in the FE- software (axisymmetry around the vertical axis, Y). The nodes at the
undisturbed end of the spherical shell have their hoop degree of freedom constrained to ensure that
plane sections remain plane.
2. Material properties
Shell: Material strength parameter MPa RM 230 = ; modulus of elasticity GPa E 200 =
Nozzle (reinforced part): Material strength parameter MPa RM 284 = ; modulus of elasticity
GPa E 200 =
Nozzle (unreinforced part): Material strength parameter MPa RM 5 . 147 = ; modulus of elasticity
GPa E 200 =
3. Determination of the maximum admissible pressure according to GPD
Using the application rule in prEN-13445-3 Annex B.9.2.2 to check against GPD, the principle is
fulfilled when for any load case the combination of the design actions does not exceed the design
resistance. This may be shown by calculating the limit load. The limit load has to be determined
using Trescas yield condition and associated flow rule. As elastic compensation is based upon a
series of elastic equilibrium stress fields, it is a relatively simple procedure to derive a lower bound
limit load direct from the Tresca yield model.
In the check against GPD only the pressure load has to be considered in this problem. The thermal
stresses that arise due to the cold medium injection are self-equilibrating and, therefore, do not
affect the pressure limit of the vessel.
From prEN-13445-3 Annex B, Table B.9-3 the partial safety factor,
R
on the material resistance is
1.25. At a temperature of 180
o
C the proof strengths of the materials for the shell, nozzle
reinforcement and nozzle are 230 MPa, 284 MPa and 147.5 MPa, respectively. The analysis was
carried out using the elastic compensation method conforming to the direct route rules in the GPD-
Check in Annex B: linear elastic - ideal plastic material law, Trescas yield condition and associated
flow rule and first order theory.
In each elastic compensation iteration an equilibrium stress field is produced where the elastic
modulus of each subsequent iteration is defined by the previous elastic solution. In this way regions
of the FE-model may be systematically stiffened or weakened depending upon the stress
magnitudes in the previous solution. The result is that every equilibrium stress field is a lower
bound of the limit load, the stress field giving the greatest limit load is taken as the closest to the
actual limit state and defines the limit load in the analysis. In problems such as this, where there are
materials with different properties, the modulus modification requires a modified procedure that
takes account of the different material properties. This modified method calculates the limit
pressure for each component with a different material, allowing the component giving the lowest
limit load to define the limit for the whole model. Convergence of the equilibrium stress field will
occur typically between 8 and 15 iterations, i.e. no further benefit in the limit load will be noted
Analysis Details
Example 5
Page
7.208 (W)
DBA
Design by Analysis
Figure 7.10.1-1: Limit Stress Field (Tresca)
with subsequent iterations. The total computing time to run the analysis on a 300 MHz Pentium II
processor with 128 Mb RAM Windows NT workstation was 197 seconds. The stress field was
shown to converge after ten iterations. The highest limit pressure supplied from iteration number 3.
Figure 7.10.1-1 shows the limit stress (intensity) field according to the Tresca condition. The limit
pressure is given by scaling the limit stress field so that the equivalent stress anywhere in the model
does not exceed the material strength parameter for that component:
Where P
L
is the limit load, P
ap
the
applied load in the elastic
compensation analysis, RM is the
yield strength of the component and

max
is the maximum stress intensity
in that component. In this problem
the shell gives the lowest limit
pressure and defines the limit for the
whole structure. For a pressure of
10 MPa the maximum stress
intensity in the shell from the
redistributed stress field is 138.3
MPa giving a limit pressure of
Application of the partial safety
factors given in prEN 13445-3
Annex B to the limit pressure is necessary to calculate the allowable pressure according to GPD.
The partial safety factor on the material yield strength is
R
= 1.25, for a pressure without natural
limit the partial safety factor on the action is
p
=1.2. The allowable pressure according to the GPD-
check is thus given by
It is also possible to determine a limit pressure using Mises' condition and associated flow rule.
However, due to the fact that the Tresca yield envelope lies within the Mises yield envelope,
correction is required. The maximum ratio of Mises' equivalent stress to Tresca's equivalent stress
for the same load is 2/3. Therefore, applying a factor of 3/2 to the yield stress in the Mises
analysis (or to the limit load, as the analysis is elastic) will always lead to a conservative result.
From the Mises analysis the limit load was found to be 17.57 MPa and with the partial safety
factors
R
= 1.25 and
p
= 1.2, the internal pressure limit according to the GPD-check can be found
as,
,
max

RM
P P
ap L
=
MPa
RM
P P
ap L
63 . 16
3 . 138
230
10
max
= = =

MPa PS
R p
GPD
1 . 11
25 . 1 2 . 1
734 . 0 63 . 16
max
=

=

MPa PS
R p
GPD
14 . 10
2
3
25 . 1 2 . 1
57 . 17
2
3 57 . 17
max
=

=

Analysis Details
Example 5
Page
7.209 (W)
DBA
Design by Analysis
Figure 7.10.1-2: Limit Equivalent Stress Field (shell top)
Figure 7.10.1-3: Limit Equivalent Strain Field (shell top)
As would be expected, using a corrected Mises analysis gives a lower bound on the Tresca limit
pressure . The value calculated using the Tresca based elastic compensation is used for this
problem.
4. Check against PD
In the check against progressive plastic deformation, thermal stresses created by the cold medium
injection will affect the residual stress and shakedown. Elastic compensation deals solely with
structural loads and cannot be used in any assessment involving thermal transients. Therefore, this
check is outside the scope of the direct route of DBA using elastic compensation.
5. Check on GPD Using Non-linear Analysis
A check against GPD was also performed for the same FE-model using conventional non-linear
analysis. In this way, a direct
comparison may be made between the
two limit approaches. The FE-geometry,
mesh and boundary conditions are the
same as those used in the elastic
compensation analysis. Material non-
linearities were applied corresponding to
the material yield strengths, 230 MPa
(shell), 284 MPa (nozzle reinforcement),
147.5 MPa (nozzle) and perfect
plasticity. Mises' yield condition and
associated flow rule was used in the non-
linear analysis. Therefore, correction of
Mises' yield surface is required, as the
check against GPD requires the use of
Trescas yield condition. As stated
above the maximum ratio of Mises'
equivalent stress to Tresca's equivalent
stress for the same load is 2/3.
Therefore, applying a factor of 3/2 to
the yield stress in the Mises analysis (or
to the limit load) will always lead to a
conservative result.
Pressure is ramped until the analysis
terminates due to non-convergence, i.e.
equilibrium can no longer be maintained
between the external applied forces and
the internal forces. The limit load can be
defined as the last converged solution in
the analysis. Figure 7.10.1-2 shows the
equivalent stress at the top surface for the
last converged solution. The
corresponding pressure is the limit one,
Analysis Details
Example 5
Page
7.210 (W)
DBA
Design by Analysis
P
L
= 19.8 MPa. According to the rules for the GPD check in prEN-13445-3 Annex B, the
maximum principal strain must not exceed +/- 5%. Figure 7.10.1-3 shows the maximum total
principal strain on the top shell surface, as can be seen the maximum principal strain is 4.66%. The
limit pressure for the structure is therefore 0.948 MPa.
Applying the rules for the GPD-check in prEN 13445-3 Annex B, the allowable pressure is found
by applying the partial safety factors on the action and material strength parameter as with the
elastic compensation solution above. The limit analysis was carried out using Mises' condition and
correction is required to bring the result within the bounds of the Tresca condition as stated above.
The allowable load according to the GPD-Check using non-linear analysis is given as
Running on the same equipment as the elastic compensation analysis, the non-linear analysis
required a CPU time of 155 seconds.
6. Additional Comments
The utilisation of elastic compensation in this DBA-calculation gives a lower allowable pressure
according to the GPD-check than that given by the DBF-calculation.
Close correlation was shown between the various methods used to calculate the maximum
allowable pressure according to GPD. The lowest value was calculated by elastic compensation
based on Mises' yield condition, and the highest by the non-linear analysis.
Elastic compensation could not be used in the admissibility check against progressive plastic
deformation. The loading cycles involved thermal transients that could not be dealt with within the
elastic compensation routines.
MPa PS
R p
GPD
43 . 11
2
3
25 . 1 2 . 1
8 . 19
2
3 8 . 19
max
=

=

Analysis Type:
Direct Route using elasto-plastic FE calculations
Member:
A&AB
Analysis Details
Example 5
Page
7.211 (W)
DBA
Design by Analysis
1. Elements, mesh fineness, boundary conditions
FE-Software: ANSYS 5.4
Element Types: 4 node, 2 D axisymmetric solid PLANE42 (structural analysis)
and PLANE55 (thermal analysis), respectively.
Boundary Conditions (structural analysis):
Zero meridional displacement in the nodes at the undisturbed end
of the hemispherical shell;
Longitudinal stress in the end of the nozzle according to a closed
end.
Model and Mesh:
Results: Action cycles as given in the specification not admissible according to the
checks against PD.
Total number of elements
1889
Analysis Details
Example 5
Page
7.212 (W)
DBA
Design by Analysis
To model the structure a total number of 1889 4-node axisymmetric elements PLANE42 in
ANSYS 5.4 for the structural analysis and PLANE55 for the steady-state thermal analysis was
used. The hemispherical part of the structure was modelled up to an angle of 30, measured from
the vertical axis of the structure. The boundary condition applied in the model for the structural
analysis is given by restraining the meridional displacement in the nodes at the undisturbed end of
the shell to 0. Furthermore, a longitudinal stress according to a closed end was applied at the end of
the nozzle. The boundary conditions used for the transient thermal analysis are given in the next
chapter.
The welds between the shell and the nozzle reinforcement, and the nozzle reinforcement and the
nozzle, respectively, were modelled with fillets corresponding to weld influence zone thicknesses of
3 mm between shell and nozzle reinforcement, and 2 mm between nozzle reinforcement and nozzle.
2. Transient thermal analysis
The transient thermal analysis according to the cold media injection (bulk temperature 80C) in the
vessel with an initially uniform temperature of 325 was carried out considering the specified 10
minutes of injection and the following 10 minutes. For the injection phase the heat transfer
coefficient of the medium to the nozzle wall was specified as 10.8 kW/m
2
K. The heat transfer
coefficient of the medium to the vessel wall, and to the nozzle wall if there is no injection, was
specified as 1.16 kW/m
2
K. The outer surface of the vessel was assumed to be insulated perfectly.
The density of the materials was specified to be 7.85 kg/dm
3
. For the input of the temperature
dependent specific heat, the following interpolation knots were used for the linear interpolation in
the FE software: 20C: 461 J/(kg
.
K), 100C: 479 J/(kg
.
K), 200C: 499 J/(kg
.
K), 300C: 517
J/(kg
.
K), 400C: 531 J/(kg
.
K). Furthermore, temperature dependent thermal conductivities were
used, and the following interpolation knots were used for the material P265GH: 20C: 51 W/(m
.
K),
100C: 50.8 W/(m
.
K), 200C: 48.7 W/(m
.
K), 300C: 45.8 W/(m
.
K), 400C: 42.5 W/(m
.
K), and for
the material 11CrMo9-10: 20C 34.9 W/(m
.
K), 100C: 37.3 W/(m
.
K), 200C: 38.2 W/(m
.
K),
300C: 37.8 W/(m
.
K), 400C: 36.6 W/(m
.
K).
Note: Since the structural and thermal calculations were carried out with the consistent mm-t-s unit
system (1 t corresponds to 1000 kg) as usual in structural calculations if the stress output shall be
in MPa -, the input unit is given for the thermal conductivity by mW/(mm
.
K), for the heat transfer
coefficient by mW/(mm
2.
K), for the specific heat by mJ/(t
.
K), and for the density by t/mm
3
.
In the non-linear transient analysis the transient load is step-changed, i. e. the load is step-changed
at the first substep of the corresponding load steps (at the beginning and the end of the injection) to
the value of this load step. To obtain proper results, a minimum of 10 substeps was used each time
for the following time intervals, given in [s], where 0 s corresponds to the beginning of the injection
and 600 s to the end of the injection: [0, 0.1]; [0.1, 1]; [1, 10]; [10, 100]; [100, 600]; [600, 600.1];
[600.1, 601]; [601, 610]; [610, 700]; [700, 1200].
Analysis Details
Example 5
Page
7.213 (W)
DBA
Design by Analysis
Figure 7.10.2-1 shows
the temperature field at
the time 0.1 s, Figure
7.10.2-2 the temperature
field at 1 s, and Figure
7.10.2-3 the one at 10 s.
Figure 7.10.2-4 shows
the temperature field at
100 s, and Figure 7.10.2-
5 the one at 600 s, and
Figure 7.10.2-6 the one
at 1200 s.
Figure 7.10.2-1
Figure 7.10.2-2
Analysis Details
Example 5
Page
7.214 (W)
DBA
Design by Analysis
Figure 7.10.2-3
Figure 7.10.2-4
Analysis Details
Example 5
Page
7.215 (W)
DBA
Design by Analysis
Figure 7.10.2-5
Figure 7.10.2-6
3. Admissibility check against GPD
Since the subroutine for Tresca's yield condition showed bad convergence, the check against GPD
was carried out using Mises' yield condition (see subsection 3.3.1 of section 3 Procedures) only.
Thus, the elastic-plastic calculation was carried out using Mises' yield condition and associated flow
rule, a linear-elastic ideal-plastic constitutive law, first order theory, and under usage of the
Newton-Raphson method.
According to prEN 13445-3 Annex B, Table B.9.3 the partial safety factor for the materials is given
by 25 . 1 =
R
, and, therefore, the design material strength parameters (corresponding to the relevant
material strength parameters for a temperature of 325C the maximum calculation temperature of
the structure) are given by
MPa
RM
R
35 . 159
2
3
25 . 1
230
2
3
= =

for the shell,


MPa
RM
R
11 . 197
2
3
25 . 1
284
2
3
= =

for the nozzle reinforcement,


MPa
RM
R
19 . 102
2
3
25 . 1
5 . 147
2
3
= =

for the nozzle.


For the welds the lower value of the welded parts was used. The modulus of elasticity used for all
parts of the structure was E = 192 GPa.
Analysis Details
Example 5
Page
7.216 (W)
DBA
Design by Analysis
The last convergent solution showed an internal pressure value of 15.65 MPa, which is close to the
theoretical limit pressure of the corresponding undisturbed hemispherical shell of 15.95 MPa. Thus,
15.65 MPa was used as limit pressure in the check against GPD, since the maximum absolute value
of the principal strains was 3.3%,
and, therefore, the requirement of
the standard was fulfilled. The
computation time of this (internal
pressure) limit load was 10 minutes
on the Compaq Professional
workstation 5000 with two Pentium
Pro processors and 256 MB RAM.
Figure 7.10.2-7 shows the
distribution of the Mises' equivalent
stress at the limit pressure.
According to prEN 13445-3 Annex
B, Table B.9.2 the partial safety
factor for the pressure is 2 . 1 =
P
,
and, therefore, the admissible
(internal) pressure according to the
check against GPD is given by
MPa PS
GPD
04 . 13
2 . 1
65 . 15
max
= =
4. Admissibility check against PD
The linear-elastic stress states due to an internal pressure MPa PS
DBF
71 . 11 01 . 13 9 . 0 9 . 0
max
= =
and due to the thermal stresses according to the temperature distributions of the transient thermal
analysis have been calculated, using temperature dependent elastic moduli and coefficients of
thermal expansion.
As input the following interpolation
knots were used for the material P265:
Elastic modulus: 20C: 212 GPa,
100C: 207 GPa, 200C: 199 GPa,
300C: 192 GPa, 400C: 184 GPa;
Coefficient of thermal expansion:
20C: 11.9e-6 1/K, 100C: 12.5e-6 1/K,
200C: 13e-6 1/K, 300C: 13.6e-6 1/K,
400C: 14.1e-6 1/K.
Figure 7.10.2-7
Analysis Details
Example 5
Page
7.217 (W)
DBA
Design by Analysis
Figure 7.10.2-8
As input the following interpolation
knots were used for the other material:
11CrMo9-10:
Elastic modulus: same values as for
P265; coefficient of thermal expansion:
20C: 11.5e-6 1/K, 100C: 12.1e-6 1/K,
200C: 12.7e-6 1/K, 300C: 13.2e-6 1/K,
400C: 13.6e-6 1/K.
Figure 7.10.2-9
Figure 7.10.2-8 shows the Mises'
equivalent stress distribution for
time = 0 s; figure 7.10.2-9 the Mises'
equivalent stress distribution for
0.1 s; figure 7.10.2-10 the Mises'
equivalent stress distribution for
1 s; figure 7.10.2-11 the Mises'
equivalent stress distribution for
10 s; figure 7.10.2-12 the Mises'
equivalent stress distribution for
100 s; figure 7.10.2-13 the Mises'
equivalent stress distribution for
600 s; and figure 7.10.2-14 the
Mises' equivalent stress distribution
for 1200 s.
Figure 7.10.2-10
Analysis Details
Example 5
Page
7.218 (W)
DBA
Design by Analysis
Figure 7.10.2-11
Analysis Details
Example 5
Page
7.219 (W)
DBA
Design by Analysis
Figure 7.10.2-12
Figure 7.10.2-13
Analysis Details
Example 5
Page
7.220 (W)
DBA
Design by Analysis
Figure 7.10.2-14
As can be seen easily from these figures, the stresses are very large, and the variation of the stresses
(the difference between a stress state under consideration and the zero-stress-state, which
corresponds to shutdown) larger than twice the relevant design material strength parameters, which
are given according to prEN 13445-3 Annex B.9.3.4 by the arithmetic mean of the yield or 1%
proof strength for the highest and lowest calculation temperatures at the position under
consideration during the whole action cycle, in more than one point of the structure for more than
one instant of time. The corresponding design material strength parameters are given by 270 MPa
for the shell, 319.75 MPa for the nozzle reinforcement, and 206.25 MPa for the nozzle. At the
locations where the large stresses arise on the inner surface of the nozzle reinforcement no local
structural discontinuity, exists.
Thus, the application rule in prEN 13445-3 Annex B.9.3.2 cannot be fulfilled, and, therefore, the
cycle cannot be shown by this procedure to be admissible according to the check against PD.
Since the stress range is larger than twice the design material strength parameter, there follows that
the structure cannot shake down under the given action cycle, and, therefore, the behaviour of the
structure was investigated by performing cyclic elastic-plastic FE-calculations as follows.
Analysis Details
Example 5
Page
7.221 (W)
DBA
Design by Analysis
For this calculation, an elasto-plastic constitutive law with the following (guaranteed) yield strength
values according to the relevant material standards was used:
P265 nozzle according to prEN 10216-2:
20C: 265 MPa, 100C: 226 MPa, 200C: 192 MPa, 300C: 154 MPa, 325C: 147.5 Mpa.
11CrMo9-10 shell - according to EN 10028-2:
20C: 310 MPa, 250C: 255 MPa, 300C: 235 MPa, 325C: 230 MPa.
11CrMo9-10 reinforcement - according to prEN 10216-2:
20C: 355 MPa, 100C: 323 MPa, 200C: 304 MPa, 300C: 289 MPa, 325C: 284 MPa.
Figure 7.10.2-15 shows the simulated cycles. At first a start-up from 0 MPa and 20C to 11.71 MPa
and 325C, one thermal cycle (injection cycle) and one shutdown cycle (unloading and reloading)
were applied. Afterwards three thermal cycles and one shutdown cycle were appended and then
these cycles were repeated once again.
The calculation of the whole cycling needed about 2 hours on a Pentium Pro with 200 MHz and 128
MB memory.
Figure 7.10.2-15
The accumulated plastic strain after this load history is shown in Figure 7.10.2-16. The maximum
strain after the complete action history is already 4%.
Figure 7.10.2-16
Analysis Details
Example 5
Page
7.222 (W)
DBA
Design by Analysis
To check if there is progressive plastic deformation in the next step, only the last 3 thermal cycles
and the shutdown cycle afterwards were considered. Therefore, the results of point 1 were
subtracted from the results of point 4 (Figure 7.10.2-15).
The difference in the absolute plastic strain (Figure 7.10.2-17) is about 0.004% (Mises' equivalent
strain). But, by considering the absolute values of the total strain (Figure 7.10.2-16), it can be seen
that there may be a numerical problem. The difference is too small in comparison to the absolute
value to achieve good numerical results.
Figure 7.10.2-17 (values to be divided by 1.5)
Analysis Details
Example 5
Page
7.223 (W)
DBA
Design by Analysis
To obtain better insight into this problem, the difference in the displacements for these three thermal
and one shutdown cycle was plotted (Figure 7.10.2-18). If the large displacement at the corner is
neglected, and only the largest mean displacement over the wall thickness considered, the value is
about 0.0002 mm. This value is mainly due to the growing of the pipe diameter near the
reinforcement. But there may be also numerical problems as well.
Figure 7.10.2-18
Analysis Details
Example 5
Page
7.224 (W)
DBA
Design by Analysis
For progressive plastic deformation to occur, there must be a failure mechanism, and, therefore, at
least one cross section must fail (plastify as a whole). To check this behaviour, the accumulated
plastic strains (Mises) for these three thermal and one shutdown cycles were plotted (Figure
7.10.2-19). It can be seen that there is one cross section (connection of pipe to the reinforcement
piece) with plastification during this cycling of almost the whole cross section. In this cross section
progressive plastic deformation is likely. In the region above, where the large deformation were
seen (Figure 7.10.2-18), the radial growing of the displacements may stop after further cycling.
Figure 7.10.2-19
To obtain information whether the structure develops to stable cycling, the time history behaviour
of a point at the inside of the pipe, slightly above the corner with the maximal plastic strain
increment, was investigated. The plot of the plastic equivalent strain (Mises) vs. time (Figure
7.10.2-20) shows that there are relevant differences between the plastic strains before and after a
thermal cycle at the three thermal cycles after the first shutdown. For the considered three thermal
cycles after the second shutdown, this amount of plastic strain is very small in comparison with the
plastic strain accumulated at the cycle.
Analysis Details
Example 5
Page
7.225 (W)
DBA
Design by Analysis
Figure 7.10.2-20
The history of accumulated equivalent plastic strain (Figure 7.10.2-21) shows also that the amount
of plastic strain due to the closed cycles is much larger than the amounts of strain due to progressive
plastic deformation (if there were progressive plastic deformation).
Figure 7.10.2-21
Analysis Details
Example 5
Page
7.226 (W)
DBA
Design by Analysis
To determine whether the problem with progressive plastic deformation is more relevant at the
thermal cycle or at the shutdown cycle, the accumulated plastic strains at the last thermal cycle
(Figure 7.10.2-22) and at the last shutdown cycle (Figure 7.10.2-23) were plotted. It can be seen
that the problem results from the thermal cycle with the same picture as above. At the shutdown
cycle only a little plastification in a small part of the cross section can be discovered, i. e. t, the
problem is due to the thermal cycling.
Figure 7.10.2-22
Figure 7.10.2-23
Analysis Details
Example 5
Page
7.227 (W)
DBA
Design by Analysis
Result:
Due to numerical problems, it is difficult to judge whether progressive plastic deformation occurs.
A large number of cycles may be needed to reach an (almost) steady-state cycle, and because of the
large number of necessary life cycles the very small strain differences (small in comparison to the
absolute strain) of one cycle may be important.
Therefore, in this example an exact answer cannot be given.
But it can be seen that the strain accumulated during the thermal cycle is much larger than the strain
due to possible progressive plastic deformation. Therefore, the fatigue calculation, which must
consider the plastic strain accumulated within the cycle, should give appropriate numbers of cycles,
e.g. Fatigue and not PD is likely to be the relevant failure mechanism.
On the other hand, it does not seem to be appropriate to approve of a design for which in an
operating cycle only a very small part of a cross-section remains elastic. A fatigue calculation
starting from results of a linear elastic analysis may in such a case give results which are not
relevant for the fatigue life.
To avoid this type of numerical difficulties, appropriate application rules are required.
Analysis type: Stress Categorization Route
Member:
WTCM
Analysis Details
Example 5
Page
7.228 (W)
DBA
Design by Analysis
Materials and Properties:
For shell and nozzle reinforcement: R
p0.2/t
= 215 MPa and E = 190000
MPa at 325 C.
For nozzle: R
p0.2/t
= 147.5 MPa and E = 190000 MPa at 325 C.
FE- Software: ALGOR.
Element types: 2-D axisymmetric elements
Boundary conditions: No x- and z-displacement (T
x
and T
z
).
No rotations (R
x
, R
y
and R
z
).
Model and mesh:
Maximum admissible actions according to the Stress Categorisation Route:
Superposition of internal pressure and cold injection not admissible
Analysis Details
Example 5
Page
7.229 (W)
DBA
Design by Analysis
The following two figures show the temperature distribution after timestep 1 (5 sec) and 59 time
steps (295 sec), respectively.
Fig. 7.10.3-1: Temperature distribution after 5 sec of cold injection
Fig. 7.10.3-2: Temperature distribution after 295 sec of cold injection
Analysis Details
Example 5
Page
7.230 (W)
DBA
Design by Analysis
Initially, four classification lines were chosen as shown in the figure below. The stress distribution
shown corresponds to an internal pressure of 11.712 MPa.
Fig. 7.10.3-3: Tresca's equivalent stress for internal pressure
For the stress classification, the following procedure was followed:
- for each load acting on the vessel, calculate the elementary stresses
ij
(i,j = 1,2,3) in the
different points on the different CLs.
- for each load acting on the vessel and along each CL, calculate the membrane stress components

ij,m
and the bending stress components
ij,b
.
- Classify the membrane stress components
ij,m
in (
ij
)
Pm
, (
ij
)
PL
or (
ij
)
Qm
and the bending
stress components
ij,b
in (
ij
)
Pb
or (
ij
)
Qb
.
- Calculate the sum of the stresses classified in this way for the set of loads acting simultaneously
on the vessel. The stresses resulting from this summation are designated (
ij
)
Pm
, (
ij
)
PL
, (
ij
)
Pb
,
(
ij
)
Qm
, (
ij
)
Qb
.
- From this deduce: (
ij
)
Pm
, (
ij
)
PL
, (
ij
)
P
, (
ij
)
P+Q
.
- Calculate the following equivalent stresses:
- (
eq
)
Pm
or (
eq
)
PL
,
- (
eq
)
P,
- (
eq
)
P+Q
.
According to Annex C of prEN13445-3, the following classification shall be used:
- for internal pressure: P
L
and Q
b
,
- for a thermal load: Q
m
and Q
b
.
- Verify the admissibility of the equivalent stresses.
Analysis Details
Example 5
Page
7.231 (W)
DBA
Design by Analysis
Two loads are considered, which act simultaneously: an internal pressure (initial value = 11.712
MPa) and a thermal load. This thermal load is divided in timesteps to simulate the cooling effect of
the injected cold medium inside the vessel. A total of 100 timesteps is considered and each timestep
is 5 seconds long. Only timestep 1 (after 5 seconds), timestep 2 (after 10 seconds) and timestep 59
(after 295 seconds) are taken into account because at those timesteps, the Tresca equivalent stresses
are the largest.
The stresses were calculated for those two individual loads and a stress classification along the CLs
1, 2, 3 and 4 was applied. On scrutinising the results of these calculations, it was observed that even
without internal pressure, the assessment criteria can not be met.
The elastic Mises stress equivalent distribution for timestep 1 is shown in the figure below.
Fig. 7.10.3-4: Mises' equivalent thermal stress after 5 sec of cold injection
The classification is most severe along CL 1. The results are shown in the next tables:
Analysis Details
Example 5
Page
7.232 (W)
DBA
Design by Analysis
With f = R
p0.2/t
/ 1.5 = 147.5 / 1.5 MPa = 98.3 MPa (in the nozzle R
p0.2/t
= 147.5 MPa), the
assessment criteria (
eq,P+Q
< 3f = 295 MPa) are not met. The cold medium injection is too severe
for any kind of internal pressure, according to the Stress Categorisation Route.
Point on CL 1 2 3 4 5 6
Stress comp. (MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
S
11
487,8 138,5 9,1 -5,1 -17,6 -99,6
S
22
1,9 101,2 49,3 -74,9 -178,0 -192,5
S
33
349,4 131,4 -22,3 -121,6 -192,4 -229,1
S
12
0,3 3,6 5,7 10,4 24,6 47,4
S
23
0 0 0 0 0 0
S
31
0 0 0 0 0 0
Membrane Timestep 1 Bending Timestep 1
stress comp. (MPa) stress comp. (MPa)

11,m
58,668
11,b
-244,25

22,m
-37,319
22,b
-182,693

33,m
-32,526
33,b
-283,088

12,m
13,161
12,b
21,658

13,m
0
13,b
0

23,m
0
23,b
0
i,j
(
ij
)
Qm
(
ij
)
Qb
(
ij
)
P+Q
11
58,668 -244,25 302,918
22
-37,319 -182,693 -220,012
33
-32,526 -283,088 -315,614
12
13,161 21,658 34,819
13
0 0 0
23
0 0 0

1,P+Q
305,226 MPa

2,P+Q
-222,32 MPa

3,P+Q
-315,614 MPa

eq,P+Q
= 620,84 MPa
Analysis Details
Example 5 / GPD- & PD-Check
Page
7.206 (S)
DBA
Design by Analysis
In the shell, CL 3 is the most severe. The results for thermal stresses only (after 295 sec of cold
injection) are shown in the next tables:
With f = R
p0.2/t
/ 1.5 = 215 / 1.5 MPa = 143.3 MPa (in the shell R
p0.2/t
= 215 MPa), the assessment
criteria (
eq,P+Q
< 3f = 430 MPa) are not met. The cold medium injection is too severe for any value
of internal pressure.
1 2 3 4 5 6 7 8
S
11
74,9 12,6 42,2 58,1 69,7 76,8 80,6 81,8
S
22
47,0 -7,7 -2,1 0,5 0,6 -2,6 -8,8 -17,5
S
33
498,6 361,3 295,4 253,0 214,1 177,1 141,7 108,1
S
12
-21,4 5,9 -28,2 -32,2 -35,8 -37,3 -36,6 -33,7
S
23
0 0 0 0 0 0 0 0
S
31
0 0 0 0 0 0 0 0
9 10 11 12 13 14 15 16
S
11 79,5 77,7 73,1 64,2 44,9 -1,9 -12,9 -18,3
S
22 -20,8 -33,2 -46,4 -58,3 -61,3 -26,5 -25,9 -29,5
S
33 103,7 70,1 38,6 8,9 -18,8 -34,4 -41,8 -52,8
S
12 -34,6 -30,1 -24,0 -15,9 5,1 1,3 -18,7 -47,6
S
23 0 0 0 0 0 0 0 0
S
31 0 0 0 0 0 0 0 0
Timestep 59 Timestep 59
(MPa) (MPa)

11,m
57,668
11,b
-6,84

22,m
-20,525
22,b
-37,305

33,m
150,407
33,b
-231,551

12,m
-25,469
12,b
4,608

13,m
0
13,b
0

23,m
0
23,b
0
i,j
(
ij
)
Qm
(
ij
)
Qb
(
ij
)
P+Q
11
57,668 -6,84 64,508
22
-20,525 -37,305 -57,83
33
150,407 -231,551 381,958
12
-25,469 4,608 -30,077
13
0 0 0
23
0 0 0

1,P+Q
381,958 MPa

2,P+Q
71,503 MPa

3,P+Q
-64,825 MPa

eq,P+Q
= 446,783 MPa
DBA
Design by Analysis
Analysis Details
Example 5 / F-Check
Page
7.234 (C)
Analysis Type:
Check against Fatigue
Member:
CETIM
FE Software: ABAQUS / Standard version 5.8.1
Element Types:
Quadratic axisymmetric 8-node elements .
1241 nodes and 352 elements
Boundary Conditions:
The mid-plane of the spherical part is locked in the vertical direction.
Model and Mesh:
Results:
Fatigue life N = 9,8 full cycles (start up shut down + 500 cold media injection)
One (full) operating cycle consists of 1 start up shut down cycle and 500 cold media
injections.
Two points show themselves as possibly critical:
One at the inside of the nozzle reinforcement, slightly below the crotch corner see the
following figures.
The following drawing shows the Tresca equivalent stress for the thermal loading only (cold
media injection).
This figure shows the position and the value of the maximum total equivalent stress for cold
media injection, i. e. thermal stresses only. The corresponding value for cold media injection
plus (maximum) pressure action is 881.2 MPa. This point is in a region where the theoretical
stress concentration factor 1
t
K .
The fatigue results for this point are shown on the following four pages.
The other possibly critical point, and certainly a point of interest, is at the weld toe of the weld
seam nozzle reinforcement to nozzle. The calculation details for the structural equivalent
stress range approach are given on the two pages after the next four.
DBA
Design by Analysis
Analysis Details
Example 5 / F-Check
Page
7.235 (C)
Notch stress = 681MPa
1
Data
t
max
= 325 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
=248,75 C
R
m
= 540 MPa
R
p0,2/t*
= 296,2 MPa
Effective stress concentration factorK
eff
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of
the stress tensors. K
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and
thermal equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied
to the thermal

stress tensors. Then both tensors are added and the new stress range is calculated. For this loading, we
have both mechanical and thermal loadings. The combination of the two correction factors and the tensors gives the
following result :
total
= 992,6 MPa

struc
=
t
K
total
= 992,6 MPa (for using in 18-11-3)
R
z
= 200 m (table 18-8)
e
n
= 57.5 mm (distance from critical point to outside weld surface)

D
= 361,2 MPa (table 18-10 for N 2
.
10
6
cycles)
N = 2300 (for the first iteration)

R
= 1303,0 MPa (allowable stress range for N cycles) at the 2
nd
iteration
Stresses

eq,t
(total or notch equivalent stress range) = 881,2 MPa

struc
(structural equivalent stress range) = 881,2 MPa
1
D

struc

t
K 0,5 1
1)
t
(K 1,5
1
eff
K
+

+
1
1
]
1

Theoretical elastic stress concentration factor K


t
K
t
=
eq,t
/
struc
= 1
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= 0,4133
k
e
= 1,2015

total
= k
e
.
eq,t
= .. MPa
Else
total
=
eq,t
= ..
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= 1,1033

total
= k

.
eq,t
= .. MPa
Else
total
=
eq,t
= .. MPa
) stress equivalent notch (mean MPa 440,6
eq

18-10-6-2 Temperature correction factor f
t*
For t* > 100 C
f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= 0,8999
Else f
t*
= 1
18-11-1-1 Surface finish correction factor f
s
f
s
= F
s
[0,1ln(N)-0,465]
with F
s
= 1- 0,056 [ln (R
z
)]
0,64
[ln(R
m
)] +0,289 [ln (R
z
)]
0,53
=0,675
f
s
= 0,8868

eqmax
= 881,2 MPa (maximum notch equivalent stress)
DBA
Design by Analysis
Analysis Details
Example 5 / F-Check
Page
7.236 (C)
Critical point: Unwelded region / Inside nozzle reinforcement
Start up shut down + 1 cold media injection
1 DBA
Design by Analysis
Analysis Details
Example 5 / F-Check
Page
7.237 (C)
18-11-1-2 Thickness correction factor f
e
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
e
= F
e
[0,1ln(N)-0,465]
f
e
= 0,7217
[0,1ln(N)-0,465]
few = (25/en)
0.25
= 0.8593
f
e
= 1 f
e
=

0,9547 f
e
= ..
18-11-1-3 Mean stress correction factor f
m
If
struc
< 2 R
p0,2/t*
and
eqmax
< R
p0,2/t*
If
eq
> 0 then
r eq,
= R
p0,2/t*
-
2
t eq,

If
eq
< 0 then
r eq,
=
2
t eq,
- R
p0,2/t*
and
eq
=
r eq,
= .. MPa
For N 2 10
6
cycles
See figure 18-14
f
m
= ..
If
struc
>2 R
p0,2/t*
18-11-2-1 Overall correction factor f
u
f
u
=
eff
K
* t
.f
m
.f
e
.f
s
f
= 0,7619
struc
/ f
u
= 1303 MPa
18-11-3 Allowable number of cycles N
N =
2
5 , 11
m
0,63R
u
f
struc

4
10 4.6
1
1
1
1
1
]
1

if N 2
.
10
6
cycles
if R
p0,2/t*

eq

( ) M 1 2
R

+
then
f
m
=
( )
5 , 0
eq
2
M 1
M 2 M
- 1
R
1
1
]
1

,
_

+
+
= ..
N = if
struc
/ f
u

2
.
10
6
N = 2229 cycles (second iteration with an error of 1%)
N is obtained by iterations. If the value of N oscillates, a difference equal to 1 % between two iterations is
acceptable. If the values decrease monotonously, the difference must be less than 0,001 % between two
iterations.
If
struc
< 2 R
p0,2/t*
and
eqmax
> R
p0,2/t*
if
( ) M 1 2
R

+

eq
R
p0,2/t*
then
f
m
=

,
_

+
+
R

eq
2
3
M
M 1
3 M 1
= .
For N 2
.
10
6
cycles
M = 0,00035 R
m
0,1 = ..
f
m
= 1
Data
t
max
= 325 C
t
min
= 80 C
t* = 0,75 t
max
+ 0,25 t
min
=263,75 C
R
m
= 540 MPa
R
p0,2/t*
= 294,1 MPa
Effective stress concentration factorK
eff
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of
the stress tensors. k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and
thermal equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied
to the thermal

stress tensors. Then both tensors are added and the new stress range is calculated.

struc
=
t
K
total
= 708,6 MPa (for using in 18-11-3)
R
z
= 200 m (table 18-8)
e
n
= 57.5 mm (distance from critical point to outside weld surface)

D
= 361,2 MPa (table 18-10 for N 2
.
10
6
cycles)
N = 200000 (for the first iteration)

R
= 986,0 MPa (allowable stress range for N cycles) at the 7
th
iteration
Stresses

eq,t
(total or notch equivalent stress range) = 681 MPa

struc
(structural equivalent stress range) = 681 MPa
1
D

struc

t
K 0,5 1
1)
t
(K 1,5
1
eff
K
+

+
1
1
]
1

Theoretical elastic stress concentration factor K


t
K
t
=
eq,t
/
struc
= 1
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..

total
= k
e
.
eq,t
= .. MPa
Else
total
=
eq,t
= ..
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= 1,0405

total
= k

.
eq,t
= 708,6 MPa
Else
total
=
eq,t
= .. MPa
) stress equivalent notch (mean MPa 540.7
eq

18-10-6-2 Temperature correction factor f
t*
For t* > 100 C
f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= 0,8861
Else f
t*
= 1
18-11-1-1 Surface finish correction factor f
s
f
s
= F
s
[0,1ln(N)-0,465]
with F
s
= 1- 0,056 [ln (R
z
)]
0,64
[ln(R
m
)] +0,289 [ln (R
z
)]
0,53
= 0,675
f
s
= 0,8597

eqmax
= 881.2 MPa (maximum notch equivalent stress)
DBA
Design by Analysis
Analysis Details
Example 5 / F-Check
Page
7.238 (C)
Critical point: Unwelded region / Inside nozzle reinforcement
Cold media injection
DBA
Design by Analysis
Analysis Details
Example 5 / F-Check
Page
7.239 (C)
18-11-1-2 Thickness correction factor f
e
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
e
= F
e
[0,1ln(N)-0,465]
f
e
= 0,7217
[0,1ln(N)-0,465]
few = (25/en)
0.25
= 0.8593
f
e
= 1 f
e
=

0,9434 f
e
= ..
18-11-1-3 Mean stress correction factor f
m
If
struc
< 2 R
p0,2/t*
and
eqmax
< R
p0,2/t*
If
eq
> 0 then
r eq,
= R
p0,2/t*
-
2
struc

If
eq
< 0 then
r eq,
=
2
struc

- R
p0,2/t*
and
eq
=
r eq,
= .. MPa
For N 2 10
6
cycles
See figure 18-14
f
m
= ..
If
struc
>2 R
p0,2/t*
18-11-2-1 Overall correction factor f
u
f
u
=
eff
K
* t
.f
m
.f
e
.f
s
f
= 0,7187
struc
/ f
u
= 986.0 MPa
18-11-3 Allowable number of cycles N
N =
2
5 , 11
m
0,63R
u
f
struc

4
10 4,6
1
1
1
1
1
]
1

+
if N 2
.
10
6
cycles
if R
p0,2/t*

eq

( ) M 1 2
R

+
then
f
m
=
( )
5 , 0
eq
2
M 1
M 2 M
- 1
R
1
1
]
1

,
_

+
+
= ..
N = if
struc
/ f
u

2
.
10
6
N = 4898 cycles
N is obtained by iterations. If the value of N oscillates, a difference equal to 1 % between two iterations is
acceptable. If the values decrease monotonously, the difference must be less than 0,001 % between two
iterations.
The complete operating cycle corresponds to 1 cycle with start up and shut down and 499 cycles with the
cold media injection. The global allowable number of cycles is equal to N with :
4898
N 499
2229
N
+ = 1
N = 9.8 full cycles
If
struc
< 2 R
p0,2/t*
and
eqmax
> R
p0,2/t*
if
( ) M 1 2
R

+

eq
R
p0,2/t*
then
f
m
=

,
_

+
+
R

eq
2
3
M
M 1
3 M 1
= .
For N 2
.
10
6
cycles
M = 0,00035 R
m
0,1 = ..
f
m
= 1
Data
t
max
= 325 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
= 248,75 C
R
m
= 410 MPa
R
p0,2/t*
= 171.5 MPa
e
n
= 14,3mm

D (5
.
10
6
cycles)
= 46 MPa (class 631)
equivalent stresses or principal stresses
m = 3 C = 5
.
10
11
m = 3 C

= ..
C
//
= ..
m = 5 C = 1.08
.
10
15
m = 5 C

= ..
C
//
= ..
Stresses

struc
=422.9 MPa (structural equivalent stress range, determined by extrapolation)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= 0,4
k
e
= 1,0932
= k
e

struc
= .......MPa
Else

=
struc
= ..MPa
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

=1,0571
= k

struc
= .....
DBA
Design by Analysis
Analysis Details
Example 5 / F-Check
Page
7.240 (C)
Start up shut down + 1 cold media injection
Critical point: Weld seam nozzle to reinforced nozzle
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of
the stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and
thermal equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied
to the thermal

stress tensors. Then both tensors are added and the new stress range is calculated. For this loading,
we have both mechanical and thermal loadings. The combination of the two correction factors and the tensors
given the following result:
struc
449.3 MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
e
= (25/e
n
)
0.25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0,639
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = 5
.
10
11
N =
m
w
f
C

,
_


= 4018 cycles
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 0,8999
18-10-7 Allowable number of cycles N
w
f

= 499.3 MPa
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= 0,8999
Else f
t*
= 1
DBA
Design by Analysis
Analysis Details
Example 5 / F-Check
Page
7. 241 (C)
Data
t
max
= 325 C
t
min
= 80 C
t* = 0,75 t
max
+ 0,25 t
min
= 263,75 C
R
m
= 410 MPa
R
p0,2/t*
= 166,3 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of
the stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and
thermal equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied
to the thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 14,3 mm

D (5
.
10
6
cycles)
= 46 MPa (class 63)
equivalent stresses or principal stresses
m = 3 C = 5
.
10
11
m = 3 C

= ..
C
//
= ..
m = 5 C = 1,08
.
10
15
m = 5 C

= ..
C
//
= ..
Stresses

struc
= 360,7 MPa (structural equivalent stress range, determined by extrapolation)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p0,2/t*

1
1
]
1

+ 1
* p0,2/t
R 2
struc

0
A 1
e
k
0,5 for 800 MPa R
m
1000 MPa
with A
0
= 0,4 for R
m
500 MPa
MPa 800
m
R MPa 500 for
3000
500
m
R
0,4

+
1
]
1

A
0
= ..
k
e
= ..
= k
e

struc
= .. MPa
Else

=
struc
= .....
Thermal loading
If
struc
> 2 R
p0,2/t*
1
1
]
1

* p0,2/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0,25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0,639
DBA
Design by Analysis
Analysis Details
Example 5 / FE-Check
Page
7. 242 (C)
Critical point: Weld seam nozzle to reinforced nozzle
Cold media injection
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = 5
.
10
11
N =
m
w
f
C

,
_


= 6929 cycles
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 0,8861
18-10-7 Allowable number of cycles N
w
f

= 416.3 MPa
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,03 1,5 10
-4
t* -1,5 10
-6
t*
2
= 0,8861
Else f
t*
= 1
DBA
Design by Analysis
Analysis Details
Example 5 / F-Check
Page
7. 243 (C)
The complete operating cycle corresponds to 1 cycle with start up and shut down and 499 cycles with the cold
media injection. The global allowable number of cycles is equal to N with:
1
6929
499
4018
+
N N
N = 13.8 full cycles
This value is larger than the one for the unwelded region the unwelded region governs the fatigue life.
Analysis Details
Example 6
Page
7.244 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using Elastic Compensation
FE-Software: ANSYS 5.4
Element Types: 4 node, axisymmetric 2-D isoparametric solids for model 1
8 node 3-D structural shell selements for model 2
Boundary Conditions:
Model 1:
Axisymmetry about Y axis.
Bottom nodes on shell and jacket constrained longitudinally
(symmetry).
Nodes at centre of head constrained (axisymmetry)
Model and Mesh:
Axisymetric solid mesh (model 1)
Member:
Strathclyde
Number of elements - 660
Results:
Check against GPD: Actions not admissible.
Analysis Details
Example 6
Page
7.245 (S)
DBA
Design by Analysis
Results:
Check against GPD: Actions not admissible.
Shell element mesh
Number of elements - 435
Boundary Conditions:
Model 2:
Symmetry conditions on both longitudinal cut edges.
Nodes on open end of cylinder constrained longitudinally and in the hoop
direction (displacement allowed in the radial direction)
Analysis Details
Example 6
Page
7.246 (S)
DBA
Design by Analysis
1. Finite Element Mesh
Two different finite element models of the geometry were created. The first one a 2-D solid
axisymmetric model, the second a partial 3-D shell model utilising a quarter rotational symmetry
and higher order 8-node elements.
The 2-D model used in the analysis was created using 660 2-D 4-node solid axisymmetric elements.
Boundary conditions for this model involved symmetry on the free end of the cylinder and jacket,
axisymmetry and constraint on the radial displacement of the nodes at the centre of the dished end.
The 3-D model was created using 435 3-D 8-node isoparametric shell elements. The boundary
conditions on this model involved symmetry along the three cutting planes.
The pressure combinations described in the problem specification were applied in accordance with
prEN 13445-3 Annex B and checks are performed against GPD and PD.
2. Admissibility checks against GPD
Using the application rule in EN-UFPV Annex B.9.2.2, to check against GPD, the principle is
fulfilled when for any load case the combination of the design actions do not exceed the design
resistance. Analysis is performed using Trescas yield condition and associated flow rule.
Elastic compensation is generally used to calculate lower bounds on the shakedown and limit loads.
In the case of multiple loads, the process involves constructing yield loci to describe the structural
limits for every load combination. However, if the applied loads are known, it is a simple
procedure to perform an admissibility check against GPD.
In the check against GPD the pressure action only has to be considered in this problem. The
thermal stresses that arise due to the differential temperatures in the two chambers are self-
equilibrating, and do not affect the pressure limit of the vessel.
From prEN 13445-3 Annex B, Table B.9-3 the partial safety factor,
R
for the material resistance is
1.2. At a temperature of 160
o
C the proof strength of the material is 194 MPa. The analysis was
carried out using the elastic compensation method conforming to the direct route rules for the GPD-
check in Annex B: linear elastic - ideal plastic material law, Trescas yield condition and associated
flow rule and first order theory.
In each elastic compensation iteration an equilibrium stress field is produced where the elastic
modulus of each subsequent iteration is defined by the previous elastic solution. In this way regions
of the FE-model may be systematically stiffened or weakened depending upon the stress
magnitudes in the previous solution. The result is, that every equilibrium stress field is a lower
bound of the limit load, the stress field giving the greatest limit load is taken as the closest to the
actual limit state and defines the limit load in the analysis.
For this problem the load cases are defined and the admissibility of the load cases are checked
against GPD. To check against GPD using elastic compensation, the loads are applied according to
the code rules for GPD and the resulting redistributed equilibrium stress fields found. For the load
cases to be admissible, the stress anywhere in the model should have a stress no greater than the
design material strength parameter. Convergence of the equilibrium stress field will occur
typically between 8 and 15 iterations, i.e. no further benefit in the equilibrium stress field will be
noted with subsequent iterations. The total computing time to run an analysis on the 2-D solid
model on a 300 MHz Pentium two processor with 128 Mb RAM Windows NT workstation was 240
seconds. The equilibrium stress field was shown to converge after fourteen iterations.
According to the action combinations laid down in the problem specification, three checks against
GPD have to be performedfor the following load cases:
Analysis Details
Example 6
Page
7.247 (S)
DBA
Design by Analysis
Figure 7.11.1-1: Equilibrium stress (intensity) field at design
action.
LC1: Inner chamber internal pressure PS
i
= 1.3 MPa
Outer chamber internal pressure PS
o
= 0.5 MPa
temperature in both chambers TS
i
= 160
o
C
LC2: Inner chamber internal pressure PS
i
= 1.3 MPa
Outer chamber internal pressure PS
o
= 0.0 MPa
temperature in both chambers TS
i
= 160
o
C
LC3: Inner chamber internal pressure PS
i
= -0.1 MPa
Outer chamber internal pressure PS
o
= 0.5 MPa
temperature in inner chamber TS
i
= 160
o
C
temperature in outer chamber TS
o
= 10
o
C
Check for LC 1
According to the rules for checks against GPD, the partial safety factors applied to this analysis are
as follows. As stated above the partial safety factor used to give the design resistance is
R
= 1.2,
giving a design material strength parameter of 161.67 MPa. According to Table B.9-2 the partial
safety factor for pressure loads without natural limit is
P
= 1.2. Therefore, the pressures applied to
the model for the analysis are
P
.
PS
i
= 1.56 MPa and
P
.
PS
o
= 0.6 MPa.
In the elastic compensation analysis using the 2-D model, the 14
th
iteration gave the lowest
equilibrium stress field, shown in Figure 7.11.1-1. The plot shows that the main area of plasticity
occurs in the knuckle region of the dished end. The maximum equivalent stress in the model is
176.6 MPa, which is greater than the design material strength parameter. Therefore, the defined
loads are inadmissible according to the check against GPD using elastic compensation. It should be
noted that the maximum stress is very
near the design strength and this shows
that the load case is very near the
maximum limit. Lower bounds
calculated using elastic compensation
can be conservative and it may be that
under different analysis types the loads
may be found admissible. As a check,
the same analysis was carried out on a
shell element model as described above.
In elastic compensation applied to shell
elements, a different yield function is
used, as there is only one element
modelling the thickness. A generalised
yield function is used in the analysis; the
Ilyushin function is based on Mises'
condition and correction is required to
make it a lower bound
on the Tresca condition. Therefore, the
design material strength parameter is scaled down by a factor of 3/2 to make the analysis a lower
bound of the Tresca yield condition. The design material strength parameter applied to the shell
analysis is therefore 140 MPa. Running on the same equipment as the solid model above, the
equilibrium stress field converged after fourteen iterations with iteration fourteen giving the lowest
Analysis Details
Example 6
Page
7.248 (S)
DBA
Design by Analysis
Figure 7.11.1-3 Equilibrium stress field at design load
Figure 7.11.1-2: Ilyushin equilibrium stress field
maximum for the Ilyushin function.
Figure 7.11.1-2 shows the Ilyushin
equilibrium stress field, the contours
relate to the squared Ilyushin
function, where
f(IL)
2
is the square of the Ilyushin
function from the contour plot,
e

the maximum equivalent stress in


the equilibrium stress field and R
d
the design material strength
parameter. For the load case to be
admissible according to the check
against GPD, the maximum Ilyushin
function can be no greater than 1. From Figure 7.11.1-2 the maximum Ilyushin function is 1.08
(1.167) - the load case is found inadmissible according to the GPD-check. As with the 2-D solid
model the shell model shows that the load case is very near the pressure limit for the structure.
Check for LC 2
As with the check for load case 1, the partial safety factors are applied according to the check
against GPD laid down in the rules. As the operating temperature is TS
i
= 160
o
C the design
material strength parameter remains 161.67 MPa. Applying the safety factor of
P
= 1.2 on the
actions gives pressures, used in the analysis , of
P
.
PS
i
= 1.56 MPa and
P
.
PS
o
= 0.
Elastic compensation carried out on the 2-D model using the Tresca condition gave the lowest
equilibrium stress field in iteration
fourteen, shown in Figure 7.11.1-3.
As would be expected, with the
inner chamber pressure the same as
in load 1, there is considerable
plasticity shown in the knuckle
region as in LC1. The maximum
equivalent stress in the equilibrium
stress field is 176.79 MPa,
occurring in the knuckle region,
which is approximately the same as
in LC 1. The maximum stress is
above the design material strength
parameter and the elastic
compensation analysis deems the
load case inadmissible.
Invoking the proportionality of the
linear result from elastic
compensation, the maximum
2
2
) (

=
d
e
R
IL f

MPa
R
P PS
d
GPD
19 . 1
79 . 176
67 . 161
3 . 1
max
max
= = =

Analysis Details
Example 6
Page
7.249 (S)
DBA
Design by Analysis
Figure 7.11.1-4: Equilibrium stress (intensity) field at designaction.
allowable load in the dished end can be found. The load multiplier can be given by the ratio to the
design material strength parameter to the maximum equivalent stress found in the model. This
renders
Check for LC 3
In this load case, the temperature of the material will vary as the inner chamber is at 120
o
C and the
outer chamber at 10
o
C. The material strength parameters will, therefore, vary throughout the
structure. However, a conservative approach is to take the lowest material strength parameter and
apply it over the whole structure. The lowest material strength parameter is 194 MPa, given at the
temperature of 160
o
C. As before, the resulting design material strength parameter is 161.67 MPa.
Applying the safety factor to the actions results in design pressures of
P
.
PS
i
= -0.12 MPa and
P
.
PS
o
= 0.6 MPa.
Elastic compensation carried out on the 2-D solid model resulted in the lowest equilibrium stress
field being obtained at iteration fourteen, shown in Figure 7.11.1-4. The maximum equivalent
stress found is 115.2 MPa occurring in the relief groove in the outer chamber. As the maximum
equivalent stress is considerably lower than the design material strength parameter, the load case is
deemed admissible according to GPD.
3. Check against GPD Using Non-linear Analysis
Although the check against GPD via elastic compensation deemed the first two load steps
inadmissible, the maximum equivalent stress was very close to the design strength parameter in
both cases. As a check, a conventional non-linear analysis based on Mises' condition was
performed for the first two load cases, to assess if the elastic compensation results are too
conservative.
Check for LC 1
Partial safety factors, design material strength parameters and action as in the elastic compensation
analysis above. As the analysis is based on Mises' condition, a factor of 3/2 was applied to the
design strength to make it a lower bound of the Tresca condition. The design material strength
parameter is thus 140 MPa. An elastic-perfect plastic material model was used in accordance with
the code. Non-linear analysis was carried out on the 2-D solid model with the pressures ramped
simultaneously. Figures 7.11.1-5 shows the equivalent stress at the applied design actions.
Figure 7.11.1-5: Equivalent stress at design actions
Figure 7.11.1-6: Equivalent Stress at Limit Load
Analysis Details
Example 6
Page
7.250 (S)
DBA
Design by Analysis
The solution continued to converge to the design load indicating that the structure was still within
the limit. As can be seen from Figure 7.11.1-5 the plasticity at the knuckle region of the dished end
is extensive as was shown in the elastic compensation result. For the loading to be admissible, the
maximum absolute value of total principal strain must not be greater than 5%. The maximum
value of principal total strain is 0.35%. Therefore, from the non-linear analysis the load case is
admissible according to the GPD-check.
Figure 7.11.1-6 shows the equivalent stress at the limit pressure of 1.654 MPa. The resulting
maximum allowable pressure for the cylinder head is 1.654/
P
= (1.654/1.2), i.e. PS
maxGPD
=1.378
MPa. The maximum total principal strain at this load is 1.1%.
The same procedure was carried out for load case 2, the non-linear result converged up to the design
loads and the maximum total principal strain was 0.38%. This would be expected as the pressure
on the ellipsoidal head is limiting in both load cases 1 and 2. Therefore, the admissibility of load
case 2 according to GPD using non-linear analysis is proved.
4. Admissibility check against PD.
Presently it is not possible to apply the elastic compensation shakedown procedure to load sets
involving thermal stresses, as a result of differential thermal expansions. Elastic compensation at
present can not deal with thermally induced stress.
Analysis Details
Example 6
Page
7.251 (S)
DBA
Design by Analysis
Analysis Type:
Direct Route using elasto-plastic FE calculations
FE-Software: ANSYS 5.4
Element Types: 4 node, 2 D axisymmetric solid PLANE42 (structural analysis)
and PLANE55 (thermal analysis), respectively.
Boundary Conditions (structural analysis):
Whole model: Symmetry boundary conditions in the nodes in the
centre of the dished end and in the nodes in the horizontal
symmetry plane of the structure.
Partial model dished end: Symmetry boundary conditions in the
nodes in the centre of the dished end, and vertical displacement in
the nodes in the edge of the cylindrical part constrained to 0.
Model and Mesh:
Member:
A&AB
Total number of elements
2833
Analysis Details
Example 6
Page
7.252 (S)
DBA
Design by Analysis
1. Elements, mesh fineness, boundary conditions
To model the structure (taking account of the symmetries), a total number of 2833 4-node
axisymmetric elements PLANE42 in ANSYS 5.4 for the structural analysis and PLANE55 for
the steady-state thermal analysis - was used (see figure on previous page). The boundary conditions
applied to the model are symmetry ones in the nodes in the centre of the dished end (with the
horizontal direction perpendicular to the plane of symmetry), and symmetry ones in the nodes in the
horizontal symmetry plane of the structure.
The welds of the stiffening rings and the weld between the inner vessel and the flat end of the jacket
were modelled with fillets completely inside of the welds.
In cases where thermal stresses need not to be included, e.g. in the GPD-check, the adjacent vessel
does not influence the dished ends stresses and strains significantly. Therefore, partial models of
the dished end and the jacket with the jacketed part of the inner vessel could be used to determine
the admissible pressure according to GPD conservatively. This was done additionally for guideline
purposes for the dished end (see chapter 4), using a model with a total number of 1776 4-node
axisymmetric PLANE42
elements. A thermal analysis
was not necessary, since the
model (i.e. this part of the
whole structure) had a
uniform temperature of
160C during all operating
conditions as can bee seen
from the results of the
thermal analysis of the whole
model. The model of the
dished end is shown in
Figure 7.11.2-1. To use
partial models for the
shakedown check could lead
to non-conservative results,
since the linear-elastic stress
distribution including the
thermal stresses of the whole
structure is different from the
ones using partial models.
Figure 7.11.2-1
2. Steady-state thermal analysis
The steady-state thermal analysis was carried out for medium (bulk) temperature of 10C inside the
jacket and 160C inside the inner vessel. The corresponding heat transfer coefficients were
specified by h
i
= 1.16 kW/(m
2.
K) on the inside of the vessel wall, h
o
= 14.4 kW/(m
2.
K) on the inside
of the jacket on all surfaces, and the outer surface of the jacket and the vessel outside of the jacket
Results: Actions and action cycles, as given in the specification, are admissible according
to the GPD- and SD-checks.
Analysis Details
Example 6
Page
7.253 (S)
DBA
Design by Analysis
were perfectly insulated. A temperature dependent thermal conductivity was used. For the input the
following interpolation knots were used for the linear interpolation in the FE-software: 10C: 14.11
W/(m
.
K); 20C: 14.3 W/(m
.
K); 100C: 15.8 W/(m
.
K); 160C: 16.82 W/(m
.
K).
Note: Since the structural and
thermal calculations were carried
out using the consistent mm-t-s
unit system (1 t corresponds to
1000 kg) as usual in structural
calculations if the stress output is
given in terms of MPa, the input
unit for the thermal conductivity
is given by mW/(mm
.
K), and for
the heat transfer coefficient by
mW/(mm
2.
K).
Figure 7.11.2-2 shows the
temperature distribution in the
structure, the dished end shows a
uniform temperature of 160C
after some distance above the
jacket, and the jacket shows a
uniform temperature of 20C
slightly away from the inner
vessel wall.
Figure 7.11.2-2
3. Admissibility check against GPD of the whole structure
The admissibility checks have to be carried out for the following 4 states:
LC1: Internal pressure in the inner chamber MPa PS
i
3 . 1 = ,
internal pressure in the outer chamber MPa PS
o
5 . 0 = ,
temperature in both chambers C TS TS
o i
= = 160 .
LC2: Internal pressure in the inner chamber MPa PS
i
3 . 1 = ,
internal pressure in the outer chamber MPa PS
o
0 = ,
temperature in both chambers C TS TS
o i
= = 160 .
LC3: Internal pressure in the inner chamber MPa PS
i
1 . 0 = ,
internal pressure in the outer chamber MPa PS
o
5 . 0 = ,
temperature in the inner chamber C TS
i
= 160 ,
temperature in the outer chamber C TS
o
=10 .
LC4: Internal pressure in the inner chamber MPa PS
i
0 = ,
internal pressure in the outer chamber MPa PS
o
5 . 0 = ,
temperature in the inner chamber C TS
i
= 20 .
temperature in the outer chamber C TS
i
=10 .
Analysis Details
Example 6
Page
7.254 (S)
DBA
Design by Analysis
Since the subroutine for Trescas yield condition showed bad convergence, the check against GPD
was carried out using Mises yield condition (see subsection 3.2 of section 3 - Procedures) only.
Thus, the elastic-plastic calculation was carried out using Mises yield condition and associated
flow rule, a linear-elastic ideal-plastic material law, first order theory, and under usage of the
Newton-Raphson method
In the admissibility check of state LC1
the partial safety factor according to
prEN 13445-3 Annex B, Table B.9-3,
is given by 2 . 1 =
R
, and, therefore, a
design material strength parameter
given by
MPa
RM
R
140
2
3
2 . 1
194
2
3
= =

was used (corresponding to a material


strength parameter of 194 MPa for a
temperature of 160C according to
prEN 10028-7).
According to prEN 13445-3 Annex B,
Table B.9-2, the partial safety factor
for pressure without natural limit) is
given by 2 . 1 =
P
. Thus, the check
was carried out with internal pressures
of MPa PS
i P
56 . 1 = for the inner,
and MPa PS
o P
6 . 0 = for the outer
chamber, respectively.
The corresponding distribution of the
Mises equivalent stress is shown in
Figure 7.11.2-3. There is plastification
in the knuckle region of the dished
end, and a very small zone of
plastification in the stress relief
groove of the flat end of the jacket. As
can be seen in Figure 7.11.2-4, the
maximum value of the absolute
principal strains in the structure is
0.38 %, and, therefore, the
admissibility of this state in the GPD
check is shown.
Figure 7.11.2-3
Figure 7.11.2-4
Analysis Details
Example 6
Page
7.255 (S)
DBA
Design by Analysis
Since the temperatures in LC 2 are equal to those of state LC1, and the pressures acting on the
structure are a part of those applied in state LC 1, the admissibility of state LC2 is shown by the
admissibility of LC 1.
In the admissibility check of state LC3 a partial safety factor 2 . 1 =
R
for the material, and
temperature dependent design material strength parameters
2
3

R
RM

,
were used, where RM corresponds to the 1% proof strength values according to prEN 10028-7 at
the considered temperatures. Thus, as input the following interpolation knots were used for the
linear interpolation in the FE software: 20C: 173.2 MPa; 100C: 150.1 MPa; 150C: 141.45 MPa;
160C: 140 MPa. The modulus of elasticity was also specified in temperature dependent form. To
use all of those temperature dependent values in the calculation, the results from the thermal
analysis were used as input for the structural analysis, but since no thermal stresses were considered
in the GPD-check, the coefficient of thermal expansion was set to zero.
The partial safety factor of pressure without a natural limit is given by 2 . 1 =
P
, and with a natural
limit by 0 . 1 . Thus, the check was carried out with pressures given by MPa PS
i
1 . 0 0 . 1 = and
MPa PS
o P
6 . 0 = for inner and outer chamber, respectively.
The resulting distri-
bution of the Mises
equivalent stress is
shown in Figure
7.11.2-5. Only a very
small zone of plasti-
fication in the stress
relief groove and on the
outer surface of the end
plate of the jacket can
be observed. The
corresponding maxi-
mum of the absolute
values of the principal
strains in the jacket and
the jacketed part of the
inner vessel is 0.116 %,
and, therefore, the
admissibility of this
state in the GPD-check
is shown.
Since the temperatures in LC 4 are smaller compared with those of state LC3, thermal stresses have
no influence on the GPD-check, and the pressures acting on the structure are a part of those applied
in state LC 3, the admissibility of state LC4 is shown by the admissibility of LC 3.
Figure 7.11.2-5
Analysis Details
Example 6
Page
7.256 (S)
DBA
Design by Analysis
4. Maximum admissible pressure of the dished end according to GPD
Since the subroutine using Trescas yield condition showed bad convergence, Mises yield
condition and associated flow rule, a linear-elastic ideal-plastic law with a design material strength
parameter of 194 MPa, first order theory, and the Newton-Raphson method was used, and,
therefore, the result was scaled down with the factor 2 / 3 (see subsection 3.2 of section 3 -
Procedures). The materials partial safety factor is given by 2 . 1 =
R
, and the partial safety factor
for pressure (without a natural limit) by 2 . 1 =
P
.
The computation time of
the limit load was 12
minutes on the
Compaq Professional
Workstation 5000 with
two Pentium Pro
processors and 256 MB
RAM, and the internal
limit pressure of the
structure was found to be
2.25 MPa. Figure
7.11.2-6 shows the
corresponding Mises
equivalent stress
distribution. The corres-
ponding maximum
absolute principal strain
was 1.4%, and, thus, the
condition in prEN
13445-3 Annex B.9.2.2
is fulfilled .
Figure 7.11.2-6
Therefore, the allowable (internal) pressure according to the GPD-check is given by
MPa PS
R P
GPD
35 . 1
2
3
2 . 1 2 . 1
18 . 17
2
3 25 . 2
max
=

=

.
Analysis Details
Example 6
Page
7.257 (S)
DBA
Design by Analysis
5. Admissibility check against PD
In the following, it is shown that Melans shakedown theorem is fulfilled for a cycle through all
states under consideration, i.e. the structure shakes down, and, therefore, progressive plastic
deformation cannot occur.
The following states have to be considered, including possible thermal stresses :
LC1: Internal pressure in the inner chamber MPa PS
i
3 . 1 = ,
internal pressure in the outer chamber MPa PS
o
5 . 0 = ,
temperature in both chambers C TS TS
o i
= = 160 .
LC2: Internal pressure in the inner chamber MPa PS
i
3 . 1 = ,
internal pressure in the outer chamber MPa PS
o
0 = ,
temperature in both chambers C TS TS
o i
= = 160 .
LC3: Internal pressure in the inner chamber MPa PS
i
1 . 0 = ,
internal pressure in the outer chamber MPa PS
o
5 . 0 = ,
temperature in the inner chamber C TS
i
= 160 ,
temperature in the outer chamber C TS
o
=10 .
LC4: Internal pressure in the inner chamber MPa PS
i
3 . 1 = ,
internal pressure in the outer chamber MPa PS
o
5 . 0 = ,
temperature in the inner chamber C TS
i
= 160 ,
temperature in the outer chamber C TS
o
=10 .
LC5: Internal pressure in the inner chamber MPa PS
i
0 = ,
internal pressure in the outer chamber MPa PS
o
5 . 0 = ,
temperature in the inner chamber C TS
i
= 20 ,
temperature in the outer chamber C TS
o
=10 .
A linear-elastic calculation was carried out for all these states. Corresponding to the states, either a
uniform temperature or the temperature distribution according to the thermal analysis was used. The
modulus of elasticity and the coefficient of thermal expansion were specified temperature
dependent, with the following interpolation knots used for the linear interpolation in the software:
Modulus of elasticity: 10C: 197 GPa; 20C: 196 GPa; 100C: 190 GPa; 160C: 185.2 GPa.
Coefficient of thermal expansion: 10C: 16.05e-6 1/K; 20C: 16.1e-6 1/K; 100C: 16.7e-6 1/K;
160C: 17e-6 1/K.
The following figures show the linear-elastic Mises equivalent stress distributions for these states:
Figure 7.11.2-7 state LC1, Figure 7.11.2-8 - state LC2, Figure 7.11.2-9 state LC3, Figure
7.11.2-10 state LC4, and Figure 7.11.2-11 state LC5.
The relevant design material strength parameter for the PD-check is given by MPa RM 217 = ,
which corresponds to the arithmetic mean of the 1% proof strength values
t p
R
/ 0 , 1
for the highest
Analysis Details
Example 6
Page
7.258 (S)
DBA
Design by Analysis
(160C) and the lowest (10C and 20C, respectively) calculation temperatures during the whole
action cycle.
Taking this design material strength parameter into account, it can be seen from the Figures 7.11.2-
7 to 7.11.2-11, that the
behaviour of the jacket
and the jacketed part of
the inner vessel is
completely elastic in the
states LC1, LC2 and LC5,
but in the states LC3 and
LC4 the maximum Mises
equivalent stress is larger
than the corresponding
design material strength
parameter, and, therefore,
plastification will occur.
The dished end is
completely elastic in the
state LC3 and LC5, but in
the states LC1, LC2 and
LC4 the maximum Mises
equivalent stress is larger
than the design material
strength parameter, and,
therefore, plastification
will occur.
Figure 7.11.2-7
Figure 7.11.2-8
Analysis Details
Example 6
Page
7.259 (S)
DBA
Design by Analysis
Figure 7.11.2-9
Figure 7.11.2-10
Analysis Details
Example 6
Page
7.260 (S)
DBA
Design by Analysis
Figure 7.11.2-11
Figure 7.11.2-12 shows the Mises equivalent stress distribution of the linear-elastic calculated
thermal stresses only.
Figure 7.11.2-12
Analysis Details
Example 6
Page
7.261 (S)
DBA
Design by Analysis
From the behaviour of the structure, it was concluded that the thermal stresses are the reason for the
plastification in the jacket and in the jacketed part of the inner vessel, whereas the internal pressure
is the reason for the plastification in the dished end. Therefore, a pressure induced self-equilibrating
stress field alone seems to be not sufficient to fulfil Melans theorem with the 5 linear-elastic stress
fields given above, and the same applies for a thermal induced stress field (which is a self-
equilibrating stress field) alone.
To obtain a better understanding of the situation, the stresses in the critical points of the structure
were drawn in a deviatoric map see Figure 7.11.2-14. These critical points are given by the
locations where the maximum (equivalent) stresses of the different load cases occured:
LC1 and LC2: node 93 at the inner side of the knuckle region of the dished end (designation
A in Figure 7.11.2-14),
LC3: node 2694 at the inside of the inner vessel wall (designation D in Figure 7.11.2-14),
LC4: node 441 at the outer surface of the end plate of the jacket (designation B in Figure
7.11.2-14),
LC5: node 751 in the stress relief groove of the annular end plate of the jacket (designation
C in Figure 7.11.2-14).
The locations of these critical points are shown in Figure 7.11.2-13.
Figure 7.11.2-13
Analysis Details
Example 6
Page
7.262 (S)
DBA
Design by Analysis
Figure 7.11.2-14
To obtain a pressure induced self-equilibrating stress field for the knuckle region of the dished end,
a linear-elastic and an elasto-plastic FE-calculation, with an internal pressure of 2 MPa applied to
the main shell of the structure, was carried out. The difference of a elasto-plastic stress field and
linear-elastic one is a self-equilibrating stress field; Figure 7.11.2-15 shows the corresponding
equivalent stress distribution. Point A is the only one of the critical locations where this self-
equilibrating stress field is not zero, the corresponding deviatoric point is designated in Figure
7.11.2-14 by [Res(I)]. The sums of this self-equilibrating stress field and all stress states of point A
are inside the deviatoric limit circle.
B
C
B
B
C
C
D
D
D
D
B
C
A
A
C
A
C
C
C
C
D
B
D
B
D
LC1
LC2
LC5
LC4
LC3
Thermal
Res(O)
Res(I)
0.32 Res(O)
-0.34 Thermal
Analysis Details
Example 6
Page
7.263 (S)
DBA
Design by Analysis
Figure 7.11.2-15
An appropriate self-equilibrating stress field for the application of Melans theorem for the points B
and D is given by the thermal stress field multiplied by 0.34. The corresponding deviatoric
mapping of this stress field for the critical locations is shown in Figure 7.11.2-14. Unfortunately,
the sum of this self-equilibrating stress field and the stress state of point C for LC5 is outside the
limit circle. Therefore, a third self-equilibrating stress field was used: to obtain a self-equilibrating
stress field, which influences only the behavior of point C, a linear-elastic and an elasto-plastic FE
calculation, with an internal pressure
of 1.2 MPa applied only in the jacket
of the structure, was carried out. The
difference of the corresponding
elasto-plastic and linear-elastic stress
fields is a self-equilibrating stress
field; Figure 7.11.2-16 shows the
corresponding equivalent stress
distribution. Point C is the only one
of the critical locations, in which this
self-equilibrating stress field is not
zero. The corresponding deviatoric
point is designated in Figure 7.11.2-
14 by [Res(O)]. The sums of this
self-equilibrating stress field
multiplied by 0.32 - see Figure
7.11.2-14 -, the thermal stress field
multiplied by 0.34 and all stress
states of point C are inside the
deviatoric limit circle.
Figure 7.11.2-16
Analysis Details
Example 6
Page
7.264 (S)
DBA
Design by Analysis
The conclusion is, that a self-
equilibrating stress field, which
is given by the sum of the self-
equilibrating stress fields
[Res(I)], 0.32 [Res(O)] and 0.34
[Thermal] can be used to fulfil
Melans theorem for all critical
locations, and, therefore, the
structure shakes down in these
points for the action cycle under
consideration. Figure 7.11.2-17
shows the equivalent stress
distribution of this self-
equilibrating stress field.
To show that Melans theorem is fulfilled for all points of the structure, this self-equilibrating stress
field is superposed to the linear-elastic stress fields of the load cases. The corresponding stress
distributions are shown in the Figures 7.11.2-18, 7.11.2-19, 7.11.2-20, 7.11.2-21 and 7.11.2-22. As
can be seen, the maximum equivalent stress nowhere exceeds 212 MPa, and, therefore, it is shown
that the structure shakes down in all points for the action cycle under consideration.
Figure 7.11.2-17
Figure 7.11.2-18
Analysis Details
Example 6
Page
7.265 (S)
DBA
Design by Analysis
Figure 7.11.2-19
Figure 7.11.2-20
Analysis Details
Example 6
Page
7.266 (S)
DBA
Design by Analysis
Figure 7.11.2-21
Figure 7.11.2-22
Analysis Details
Example 6
Page
7.267 (S)
DBA
Design by Analysis
DBA
Design by Analysis
Analysis Details
Example 6 / F-Check
Page
7.268 (A)
Analysis Type:
Check against fatigue
Member:
A&AB
FE Software: ANSYS 5.4
Element Types: 4 node, 2 D axisymmetric solid PLANE42 (structural analysis)
and PLANE55 (thermal analysis), respectively.
Boundary Conditions (structural analysis):
Symmetry boundary conditions in the nodes in the centre of the
dished end and in the nodes in the horizontal symmetry plane of the
structure.
.
Model and Mesh:
Results:
Fatigue life N = 20506 cycles
Total number of elements:
2833
In each operating cycle four different action states are taken on:
AS1: P0
i
= 0; T0
i
= 20 C;
P0
o
= 0; T0
o
= 20 C;
AS2: P0
i
= 0; T0
i
= 20 C;
P0
o
= 4.5 bar; T0
i
= 20 C;
AS3: P0
i
= 11 bar; T0
i
= 160 C;
P0
o
= 4.5 bar; T0
o
= 160 C;
AS4: P0
i
= 11 bar; T0
i
= 160 C;
P0
o
= 4.5 bar; T0
o
= 10 C;
Where P0
i
and T0
i
denote operating pressure and temperature in the inner chamber, and P0
o
and T0
o
operating pressure and temperature in the jacket.
A full operating cycle corresponds to the series
AS1 AS2 AS3 AS4 AS3 AS2 - AS1.
The maximum equivalent stress range occurs for AS1 AS4 in the knuckle region of the dished end. A
rough, conservative check shows that the fatigue life in this point is larger than the one for the critical welded
regions the weld seams of the outer jacket shell to the annular end plates and the weld seams of the annular
end plates to the main shell. The fatigue class of the former is 40 the inside can not be visually inspected
and the fatigue class of the latter is 71 welding from both sides.
Again, a rough check shows that the weld seam outer jacket shell to end plates govern the fatigue life, and the
critical point is the weld toe (on inside). Therefore, details for this point are given in the following.
The prinicipal stresses normal and parallel to the weld seam for the four action states are given in the
following table (in the order of their occurence in the operating cycle)
AS1 AS2 AS3 AS4 AS3 AS2 AS1
l

0 138.3 114.2 -12.2 114.2 138.3 0


c

0 -39.0 66.7 228.0 66.7 -39.0 0


There are two stress ranges for the principal stress normal to the weld
MPa
struc
5 . 150 (AS3 AS4)
MPa
struc
3 . 138 (AS1 AS2)
For the principal stress parallel to the weld only one stress range results. Since for this weld fatigue class 80
seems to be justified with proper testing for full penetration and freedom from significant flaws this
principal stress does not govern the fatigue life (N = 86060 cycles).
DBA
Design by Analysis
Analysis Details
Example 6 / F-Check
Page
7.269(A)
Data
t
max
= 160 C
t
min
= 10 C
t* = 0,75 t
max
+ 0,25 t
min
= 122.5C
R
m
= 520 MPa
R
p1,0/t*
= 202.6 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 8 mm

D (5
.
10
6
cycles)
= 46 MPa (class 63)
principal stresses or equivalent stresses
m = 3 C

= 1.28
.
10
11
m = 3 C =
C
//
= ..
m = 5 C

= .. m = 5 C =
C
//
= ..
Stresses

struc
= 150.5 MPa (structural equivalent range)
(obtained by quadratic extrapolation on the jacket shell inside)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p1,0/t*

1
1
]
1

+ 1
* p1,0/t
R 2
struc

4 , 0 1
e
k
k
e
= ..
= k
e

struc
= .. MPa
Else

=
struc
= 150.5 MPa
Thermal loading
If
struc
> 2 R
p1,0/t*
1
1
]
1

* p1,0/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0.25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0,639
DBA
Design by Analysis
Analysis Details
Example 6 / F-Check
Page
7.270 (A)
Welded region / Weld toe of jacket to annular plate: Principal stress range approach
Cycle: AS3 AS4 AS3
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = 1.28
.
10
11
N =
m
w
f
C

,
_


= 36470 cycles
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= ..
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 1
18-10-7 Allowable number of cycles N
w
f

= 152
.
0 MPa
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,043 4,3 10
-4
t* = 0.9903
Else f
t*
= 1
DBA
Design by Analysis
Analysis Details
Example 6 / F-Check
Page
7.271 (A)
Data
t
max
= 160 C
t
min
= 20 C
t* = 0,75 t
max
+ 0,25 t
min
= 125 C
R
m
= 500 MPa
R
p1,0/t*
= 202 MPa
If both mechanical and thermal loadings are to be considered, the correction has to be made on each component of the
stress tensors . k
e
and k

are to be calculated with the above formulas where


eq,l
is the full mechanical and thermal
equivalent stress range. The factor k
e
is applied to the mechanical stress tensors and the factor k

is applied to the
thermal

stress tensors. Then both tensors are added and the new stress range is calculated.
e
n
= 8 mm

D (5 10
6
cycles)
= 46 MPa (class 63)
principal stresses or equivalent stresses
m = 3 C

= 1.28
.
10
11
m = 3 C =
C
//
= ..
m = 5 C

= .. m = 5 C =
C
//
= ..
Stresses

struc
= 138,3MPa (structural equivalent stress range)
(obtained by quadratic extrapolation on the jacket shell inside)
18.8 Plasticity correction factor k
e
mechanical loading
If
struc
> 2 R
p1,0/t*

1
1
]
1

+ 1
* p1,0/t
R 2
struc

4 , 0 1
e
k
k
e
= ..
= k
e

struc
= .. MPa
Else

=
struc
= 138.3 MPa
Thermal loading
If
struc
> 2 R
p1,0/t*
1
1
]
1

* p1,0/t
R
struc

0,4
0,5
0,7
k

k

= ..


= k

struc
= .. MPa
Else =
struc
= .. MPa
18-10-6-1 Thickness correction factor f
ew
e
n
25 mm 25 mm e
n
150 mm e
n
150 mm
f
ew
= (25/e
n
)
0.25
=
f
ew
= 1 f
ew
=

.. f
ew
= 0.639
DBA
Design by Analysis
Analysis Details
Example 6 / F-Check
Page
7.272(A)
Welded region / Weld toe of jacket to annular plate: Principal stress range approach
Cycle: AS1 AS1
18-10-6-3 Overall correction factor f
w
f
w
= f
ew
.f
t*
= 0.9892
18-10-7 Allowable number of cycles N
w
f

= 139.8 MPa
18-10-6-2 Temperature correction factor f
t*
For t* > 100 C f
t*
= 1,043 4,3 10
-4
t* = 0.9892
Else f
t*
= 1
If
w
f

>
5
.
10
6
cycles
then
m = 3 and C (C

or C
//
) = 1.28
.
10
11
N =
m
w
f
C

,
_


= 46845cycles
If
w
f

<
5
.
10
6
cycles
and other
cycles with
w
f

>
5
.
10
6
cycles
then
m = 5 and C (C

or C
//
) = ..
N =
m
w
f
C

,
_


= .. cycles
If
w
f

<
5
.
10
6
cycles
and all other
cycles with
w
f

<
5
.
10
6
cycles
then
N =
DBA
Design by Analysis
Analysis Details
Example 6 / F-Check
Page
7.273 (A)
Combing the two subcycles to the full operating cycle (with allowed number of cycles N) gives, with
, 1
2 1 2
2
1
1
+ +
N
N
N
N
N
n
N
n
finally N = 20506 cycles
Analysis Details
Example 6 / I-Check
Page
7.274 (A)
DBA
Design by Analysis
Analysis Type:
I-Check
FE-Software: ANSYS 5.4
Element Types: 8 node structural shell elements SHELL93
Boundary Conditions: Symmetry b.c. in the nodes in the vertical symmetry plane of the
structure; the displacements in one node at the junction of the inner
shell and its stiffener ring in this symmetry plane was constrained
to zero in horizontal and vertical direction, and in the other one
constrained to zero in vertical direction only.
Model and Mesh:
Results: Actions as given in the specification are admissible according to the I-Check
Member:
A&AB
Total number of elements
4608
Analysis Details
Example 6 / I-Check
Page
7.275 (A)
DBA
Design by Analysis
1. General
The following approach was used here to show the admissibility of the specified actions against
instability: A input to the fully nonlinear analysis of the structure with initial imperfections, an
eigenvalue buckling calculation of the perfect structure (with linear-elastic constitutive law) was
performed, and the so obtained buckling shape of the first buckling mode is scaled with regard to
the specified maximum (admissible) out-of-roundness of the structure and the result used for the
imperfect geometry. The latter analysis was performed using the nonlinear geometry and a linear-
elastic ideal-plastic constitutive law.
2. FE-model
To include the critical non-rotational symmetric buckling mode forms, and the fact that the buckling
forms are not necessarily symmetric to the horizontal symmetry plane of the structure, an FE-model
representing half of the structure was used. Because of the large diameter to wall thickness ratio of
the structure shell elements, i.e. the 8node SHELL93 elements in ANSYS 5.4, were used. In the
nodes in the vertical symmetry plane of the structure symmetry boundary conditions were applied.
Additionally, to avoid rigid body motions, the displacement in one node at the junction of the inner
shell and its stiffener ring in the symmetry plane was constrained to zero in horizontal and vertical
direction, and the other one constrained to zero in vertical direction only. The FE-model used for
the eigenvalue buckling approach was created as usual by the preprocessor of the FE-software. The
model used for the fully nonlinear analysis, which included the shape imperfections corresponding
to the first eigenvalue buckling mode scaled by the allowable out-of-roundness, was created by use
of the macro IMPER (see Annex 5: Model and Solution Input), which determined the geometry -
i.e. the coordinates of the nodes - of the imperfect structure using the perfect geometry and the
results of the eigenvalue buckling calculation. In this procedure the FE-mesh is detached from the
geometric lines and areas, which had been used to create the model of the perfect structure.
Therefore, the different loads which have to be applied on different parts of the structure in the fully
nonlinear analysis, have to be applied directly on the corresponding elements. To make the selection
of these elements easy, the real constant number notation in ANSYS5.4 was used to perform this
selection - see picture on preceeding page, where the different colours correspond to different real
constant numbers.
3. Eigenvalue buckling approach
An eigenvalue buckling analysis predicts the theoretical buckling strength (the bifurcation point) of
an (geometrically) ideal and linear-elastic structure. This approach corresponds to the textbook
approach for elastic buckling analysis. In the case under consideration the reason for performing the
eigenvalue buckling analysis is two-fold: to have a comparison value for other approaches, and to
obtain a relevant buckling mode shape, to be used as an initial imperfection in the model of the fully
nonlinear analysis.
The first step of an eigenvalue buckling analysis is the calculation of the static solution for unit
load(s), i.e. small values of the loads for which the buckling mode shall be calculated. In the next
step the program calculates multiplication factors for these unit loads corresponding to the relevant
buckling modes and the buckling mode shapes.
Analysis Details
Example 6 / I-Check
Page
7.276 (A)
DBA
Design by Analysis
The relevant load case, for which the instability check has to be performed is LC 3:
Internal pressure in the inner chamber MPa PS
i
1 . 0 = ,
internal pressure in the outer chamber MPa PS
o
5 . 0 = ,
temperature in the inner chamber C TS
i
= 160 ,
temperature in the outer chamber C TS
o
=10 .
The modulus of elasticity used in the calculations was GPa E 191 = , a value which corresponds to
the mean temperature of the wall of the buckling-critical inner cylinder.
Since the program failed to calculate eigenvalue buckling modes for the case of internal pressure
applied to the outside wall of the jacket, and since the results were to be used as imperfection of the
geometry for the fully nonlinear approach only, the action used for the eigenvalue calculation was
external pressure acting on the inside wall of the jacket. The value of this unit external pressure
was MPa 7 . 0 , the calcualted multiplication factor was 7.18, and, therefore, the lowest eigenvalue
buckling pressure corresponding to the model and load used was MPa 02 . 5 . Figure 7.11.4-1 shows
the deformed shape of this first eigenvalue buckling mode, the deformation is scaled up in order to
show the (bifurcation) buckling mode shape clearly.
Figure 7.11.4-1
Analysis Details
Example 6 / I-Check
Page
7.277 (A)
DBA
Design by Analysis
4. Fully Nonlinear Analysis
A nonlinear buckling analysis is simply a nonlinear static analysis, which can be extended to a point
where the structure reaches its limit load or maximum load carrying capacity, or where simply the
admissibility of specified actions is shown. The latter simpler approach was followed here. In the
present analysis an initial imperfect geometry, geometric nonlinearities (2
nd
order theory) and a
linear elastic ideal plastic constitutive law was used.
With the exception of temperature, the partial safety factors for actions in the instability check are
equal to those of the GPD-check, i.e. the partial safety factor for internal pressure is 2 . 1 =
P
and
for external pressure (vacuum) the partial safety factor is 0 . 1 =
P
; for thermal stresses the partial
safety factor is 0 . 1 =
T
. Since the hydraulic pressure test can be performed as required, the partial
safety factor for the resistance of the structure against instability is 25 . 1 =
R
. Since only the
admissibility of the specified actions has to be shown , all characteristic values of the actions were
multiplied with the corresponding partial safety factors for the actions and the partial safety factor
for the resistance (see subsection 3.7.3 of section 3 - Procedures).
Thus, the following actions were applied:
internal pressure of MPa 75 . 0 25 . 1 2 . 1 5 . 0 = in the jacket,
external pressure of MPa 125 . 0 25 . 1 0 . 1 1 . 0 = on the inside of the main shell,
temperature distribution: as an approximation of the temperature distribution calculated in
the steady state thermal analysis, the following temperature distribution have been used for
the shell model: main shell outside of the jacket C T =160 , end rings of the jacket and
jackets outside wall C T =10 , main shell outside surface in jacket region C T = 20 , main
shell inside surface in jacket region C T = 6 . 102 . This approximate temperature
distribution, multiplied by the partial safety factor for the resistance 25 . 1 =
R
was applied
to FE-model.
To obtain the (initial) imperfect geometry of the structure, the results and the ANSYS 5.4
database of the eigenvalue buckling calculation were used. The macro IMPER generated an (initial)
imperfect geometry according to the first buckling mode shape with a maximum deviation of 20.15
mm from the perfect geometry; this value corresponds to the specified maximum out-of-roundness
of the main vessel cylindrical wall.
The fully nonlinear calculation was performed with temperature dependent elastic modulus and
thermal expansion coefficients, where the following interpolation knots were used in the input:
Elastic modulus: 0C: 198 GPa; 20C: 196 GPa; 100C: 190 GPa; 200C: 182 GPa.
Thermal expansion coefficient: 0C: 16.0e-6 1/K; 20C: 16.1e-6 1/K; 100C: 16.7e-6 1/K; 200C:
17.2e-6 1/K.
For the linear elastic ideal plastic constitutive law the following design material strength
parameter values, corresponding to the 1%-yield strength of X6CrNiTi18-10 according to prEN
10028-7, were used as interpolation knots: 0C: 240 MPa; 20C: 240 MPa; 100C: 208 MPa;
150C: 196 MPa; 200C: 186 MPa.
The analysis was performed using the Newton-Raphson method. The computation time until the
final load level was reached was 6 hours and 31 minutes on the Compaq Professional Workstation
5000 with two Pentium Pro processors and 256 MB RAM.
Analysis Details
Example 6 / I-Check
Page
7.278 (A)
DBA
Design by Analysis
Figure 7.11.4-2 shows
the distribution of the
Mises equivalent stress
of the whole structure,
and figure 7.11.4-3 of
the jacketed part of the
main shell, according
to the final load level.
Figure 7.11.4-4 shows
the corresponding
maxi-mum principal
plastic strain
distribution of the
whole structure.
Since the final load
level was reached via a
convergent solution of
the fully nonlinear FE
analysis, admissibility
of the actions under
consideration against
instability is shown.
F
igure 7.11.4-2
Figure 7.11.4-3
Analysis Details
Example 6 / I-Check
Page
7.279 (A)
DBA
Design by Analysis
Figure 7.11.4-4
Analysis Details
Example 6 / I-Check
Page
7.280 (A)
DBA
Design by Analysis
5. Handbook formulae approach
Load Case 3, Inner shell:
Axial stress resultant:
FE-calculation:
Axial stress resultant (per unit length)
mm N n
a
/ 66 . 22 =
This value results from the actions for load case 3, i. e. for characteristic values.
The partial safety factors for the internal pressure is 1.0, for the external 1.2. The partial
safety factor for thermal stresses is not specified, the value 1.0 is considered appropriate.
To avoid new calculations a conservative design axial stress of
mm N n
d a
/ 2 . 27 ) 66 . 22 ( ) 2 . 1 (
,
= =
is used.
Critical value:
Roarks Formulas for Stress and Strain, 6
th
edition, Table 35, Case 15:

5 . 0 2 2
)) 1 ( 3 /( ) ( = r e E n
crth
mm r mm e GPa E 1390 , 20 , 191 = = = , 3 . 0 =
According to the civil engineering code for steel structures DIN 18800 a reduction factor
of 1.0 can be used in this case of a small relative slenderness ratio:
A practical value of
mm N n n
th a pr a
/ 33266
, ,
= =
results; there follows:
Stability check for axial stress:
0008 . 0 /
, ,
=
pr a d a
n n
External pressure (difference):
Characteristic values of pressures:
internal pressure: MPa p
c i
1 . 0
,
=
external pressure: MPa p
c e
5 . 0
,
=
Design value of pressure difference:
MPa p
d d
7 . 0 ) 5 . 0 ( ) 2 . 1 ( ) 1 . 0 ( ) 0 . 1 (
,
= =
Analysis Details
Example 6 / I-Check
Page
7.281 (A)
DBA
Design by Analysis
Critical value:
Roarks Formulas for Stress and Strain, 6
th
edition, Table 35, Case 19b:
75 . 0 2 5 . 0 2
) 1 ( ) / ( ) / ( 807 . 0

= v r e lr e E p
crth
1492 = l
MPa p
crth
83 . 3 =
MPa p
cr
06 . 3 = with a knock-down factor 0.8
According to the civil engineering code for steel structures DIN 18800 the theoretical
value is given by 4.7 MPa, the reduction factor by 0.745. Thus
MPa p
pr d
5 . 3
,
= ,
and
2 . 0 /
, ,
=
pr d d d
p p
Stability check:
8 . 0 / 1 2 . 0 / /
, , , ,
= < = +
I pr d d d pr a d a
p p n n ,
with 25 . 1 =
I
, since the required pressure test can be performed for the relevant action,
the pressure difference.
Note: The influence of the temperature moment on the critical inner shell due to the
temperature gradient over the wall is not included in this result, but the safety
margin seems to be sufficient.
Note: The boundary conditions used in both buckling cases are:
Radial displacement at both ends of (inner) shell zero
No other constraints end sections may warp.
In the actual shell the radial displacement of the end section is restrained by the
annular end rings of the jacket (and the outer shell), but not totally. But, nevertheless,
the first boundary condition is considered to be a good one.
The axial displacement is restrained by the ends of the vessel and, to a minor extent,
by the annular end rings of the jacket and the outer shell. The second boundary
condition is considered to be a (poor) conservative one.
Analysis Details
Example 6 / I-Check
Page
7.282 (A)
DBA
Design by Analysis
1
Recommendations
Page
8.1
DBA
Design by Analysis
8 Recommendations
8.1 Overview
In this final Section, the experience the writers have gained from this study, and general and
specific recommendations for the aspiring analyst/designer or engineering manager are summarised.
To begin with, various recommendations concerning the methodology of design by analysis
specifically elastic-plastic finite element analysis are proposed. This is followed by a brief
discussion of software requirements (although it is noted that most commercial finite element
systems in common use in the industry have sufficient capability). Then a discussion on the
requirements expertise and engineering or technical knowledge is provided. It is argued that the
analyst plays a critical role in the design by analysis process, and that careful attention should be
paid to their appropriate training, ideally through Continuing Professional Development. Finally, a
few concluding remarks are made.
8.2 Methodology
Commercial finite element analysis systems which are capable of carrying out the analyses required
for design by analysis using the Direct Route are common (this is discussed further in sub-section
8.3). Current desktop PC-based computing hardware, running Windows 98 or NT, with twin
Pentium III processors and analysis software optimised to use them, and large fast multi-gigabyte
hard discs, are more than adequate for the majority of vessels and actions which need to undergo
design by analysis. The weak areas of the design by analysis procedure relates more to the expertise
of the analyst and the suitability (or correctness) of the analysis. The former will be discussed in
more detail in sub-section 8.4, while the latter is examined here.
To begin with, one crucial but misunderstood aspect of modern finite element analysis needs to be
emphasised:
Our practical knowledge of the behaviour of modern finite element formulations, and what
constitutes a suitable finite element model, is almost exclusively concerned with linear elastic
behaviour with small deformations. This (limited) core expertise is also the case with the majority
of practising finite element analysts. Put simply, a finite element model which may be suitable for
linear elastic, small deformation behaviour under simple actions may be wholly unsuitable for
elastic-plastic analysis, as well as large deformations and multiple actions. Even an adapted finite
element mesh (constructed during the kind of adaptive finite element procedures available in many
commercial systems) is only really valid for linear elastic, small deformation behaviour. For
example, in thin-walled vessels it is common for relatively small weak areas (related to the
formation of plastic hinges) to develop. When this occurs, finite element theory would recommend
re-meshing these areas if detailed plastic strain levels are required. Such a re-mesh is not common.
Similarly, experience with the effect of poorly shaped elements is almost exclusively based on
linear elastic analysis with limited actions. Most commercial systems provide various checks and
corrections for element shape, according to common industry guidelines. The validity of these
checks for elastic-plastic behaviour is largely unknown. It remains good advice to try to make the
element shape in regions of extensive plasticity close to its natural square shape, but this requires
analysts skill, and arguably could suggest a re-mesh once the plastic region is identified. Also,
2
Recommendations
Page
8.2
DBA
Design by Analysis
unless the analyst is familiar with the approximations and assumptions of the Classical Theory of
Plasticity, which is contained in the majority of commercial systems, he may be led to believe that
the strain levels being predicted for multiple actions are realistic. In regions which are essentially
compressive or highly constrained, this can be far from the truth! As a final example, it is often not
fully appreciated, or conveniently forgotten, that the majority of common commercial finite element
analysis systems are based on the displacement formulation, where displacement is interpolated
within an element in terms of the nodal values. As a consequence stresses are discontinuous from
element to element and in general equilibrium is only satisfied at nodes. When this is placed in the
context of a highly nonlinear material, this implies that the stresses are even less reliable in
nonlinear analysis than in linear, and that it is very easy for errors to propagate as load increases.
Very few commercial systems include mixed formulations, where both stress and displacement are
interpolated and equilibrium is satisfied, although these are common in the theoretical literature. It
is fairly easy to construct problems where the displacement formulation in commercial software
fails to converge to the correct solution for nonlinear analysis
[1]
.
The ten benchmark examples studied in Section 7, can for the most part, be modelled with well
structured finite element meshes. However, in the following some basic recommendations, or
warnings, are summarised. After some general comments, the discussion is split into two parts: the
first looks at more general modelling issues for nonlinear finite element analysis, while the second
deals with issues specific to this study:
8.2.1 General comments
The novice may only have a general idea of the role of advanced analysis and simulation in
engineering design. A common view is that the aim is to simulate, and thereby quantify, the actual
behaviour of the component or structure. This is far from the truth. In the design of pressure vessels
it has to be remembered at all times that the aim is to check the code requirements. As a
consequence the code rules tell us that small details, such as welds, may be omitted. Further, as
discussed in detail in Section 2, the results of stress categorisation have to be post-processed in a
particular manner the code stress limits apply to membrane and bending stress, and except for
fatigue ignore the peak. If the stress categorisation route is used, then the allowable loads are based
on shakedown and limit analysis, which assume perfect plasticity to introduce a failure mechanism
in the component. Perfect plasticity is a hypothetical material behaviour. Indeed the Classical
Theory of Plasticity is an approximation. In general any finite element model is an approximation,
and this is even more true for nonlinear behaviour. Since any model is an approximation,
engineering judgement, both in the modelling and assessment phase, is essential.
One approach, or modelling and assessment framework, for pressure vessel design by analysis
which the analyst could refer to is the following
[2]
:
Ensure there is a good understanding of how the finite element model behaves
Interpret the model behaviour into real structural behaviour
Ensure there is a good understanding of how the real structure behaves
Judge the fitness of purpose of the real structure
3
Recommendations
Page
8.3
DBA
Design by Analysis
This needs some more explanation:
To begin with, a clear distinction should be made throughout between the real structure on the one
hand and two different models of it on the other:
The first model is the physical model, quite often also called analysis model, and is deduced from
the real structure by an abstraction, or idealisation, process with regard to geometry, boundaries and
boundary conditions, constitutive laws, etc. This idealisation quite often also requires assumptions
on material properties or even constitutive laws, which are unknown and even not determinable for
the real structure - the real constitutive law of base metal or zones of weldments, the real
deviations from the ideal geometry and so on.
The second model is the mathematical model - a mathematical description of the physical model
using the basic principles of mechanics. In case of finite element analysis this mathematical model
is obtained by means of the FEA software.
Also one of two analysis situations can arise:
Firstly, if the starting point is a real existing structure, the actual geometry is known, especially the
actual thicknesses are known or can be determined. Nevertheless, it may be necessary to use a
different geometry in the physical model - not only because of the necessary idealisation, but also
because of the requirement in some checks to deal with safety/behaviour of the real structure for the
whole design life. Therefore, it may be necessary to use in some checks actual thicknesses minus
allowances for corrosion, erosion, i. e. for allowances specified in the design specification. In the
context of the European Standard obvious checks where the full specified thickness allowances
shall be taken into account are the checks against Global Plastic Deformation (GPD), Instability (I);
for reasons of simplicity the same physical model may be used in the checks against Fatigue (F),
and Progressive Plastic Deformation (PD) or in the Shakedown check (SD).
Secondly in a design situation the real structure is not known: the starting point is a virtual
structure, specified by the design drawings, material lists, directly or indirectly referred to
standards (for materials, tolerances of pre-products, shape deviations, allowed manufacturing
deviations or defects, etc.). If the starting point is the virtual model not only the allowances for
corrosion, erosion shall be deducted, but also those for material tolerances, manufacturing
tolerances, e. g. for dished ends, cones with knuckles, where the minimum specified thickness for
the knuckle should be used, a value which should have been specified in the drawing anyway.
In both cases it may also be necessary to take into account allowed shape deviations. In the
Standard, checks where this may be necessary are the Instability Check, and possibly the Fatigue
Check. Further, additional actions, not accounted for in the physical model, especially non-physical
ones, may play an important role in the behaviour of the real structure, for example some types of
local corrosion, including the various cracking types, embrittlement, etc.
Thus, while it should always be kept in mind that the behaviour being investigated is the behaviour
of the physical model of the real structure, the analysis or simulation tells us something about the
behaviour of the real structure, but should not be confused with that. If the analyst always keeps in
mind the distinction between the real structure, or the virtual one, and the models, an understanding
4
Recommendations
Page
8.4
DBA
Design by Analysis
of the behaviour of these models will help to better understand the behaviour of the real structure,
even in cases not taken into account in the model.
If we look at something, our brain uses our experience to transform the sense-impression into a
picture of something meaningful - at least it does that in normal situations. Similarly the analyst
does need (a lot of) experience to transform finite element results that is, results of the
mathematical model, into something meaningful with regard to the actual response of the real
structure under imposed actions and constraints. Transition from the physical model to the
mathematical one is made easy by software and requires just some vocational training in using the
software. The creation and assessment of the physical model requires much expertise and much
theoretical knowledge; it also requires knowledge of the software's capabilities, assumptions and
requirements. This can be clearly seen in the models used in the Instability Checks in Section 7.
Once the analysis has been carried out, the results from the physical model must be assessed in the
context of the actual vessel and the rules given in the standard. However it is at this stage that the
analyst very often makes a major blunder a situation which is in fact made worse by the fact that a
specific design code is being used! at first sight it might appear that the ideal result is a
single answer: Yes, the maximum stress is below the allowable the post-processors of most FE
programs are capable of sorting results so that the maximum and minimum of any calculated values
can be extracted easily. However by looking at these values alone the analyst would learn little
about how the structure works. Even worse, he would be unable to form an opinion about how well
his idealisation has simulated the real structure it is not possible to determine if an analysis
is right or wrong, only that the results are fit for their purpose
[2]
. Thus, the code checks are
done and, hopefully, satisfied very often that is the main test of the model!
It is often not easy to have a good understanding of how the real structure behaves and to relate this
to the models. This is especially true in the case of nonlinearity, and thermal actions in particular. In
any case, the first task facing the analyst is always to justify the assumptions made during
modelling, particularly those assumptions regarding constraints. These assumptions may also have a
physical basis: has an initial strain, or residual stress from welding, been justifiably ignored? Is the
sequence of loading important: the nonlinear behaviour of a pressure vessel which is heated then
pressurised, is much different if the load sequence is reversed. Does a change to the modelling
assumption have no effect, or only a trivial one which does not effect code compliance? If there is
an effect then further investigation, re-modelling and justification must be made. Then the influence
of the chosen mathematical choice of element type, mesh design, load modelling and so on -
should be examined. While a professional and competent analyst can build up a personal body of
understanding, this is most often for linear elastic behaviour, and this can only provide an indication
that the finite element models are suitable to begin with. In the following some general modelling
issues for nonlinear analysis are discussed:
8.2.2 General modelling issues
There are surprisingly few sources to which the designer/analyst can refer to give general help with
nonlinear finite element analysis, and elastic-plastic behaviour in particular. Probably the best
discussions have been given by Adams & Askenazi
[3]
, Cook
[4]
and NAFEMS
[5]
these are
summarised briefly here, but the reader is strongly advised to consult these sources directly. Written
5
Recommendations
Page
8.5
DBA
Design by Analysis
by experts, the advice given in theses sources are in basic agreement on the modelling issues which
arise for nonlinear finite element analysis.
A good summary of good modelling practice has been given by Adams & Askenazi
[3]
. Part of
Chapter 15: Nonlinear Analysis is reproduced here:
A few guidelines for meshing and model building are appropriate or more appropriate for nonlinear
problems. Building the model correctly in the first place can greatly enhance the speed of the solution and
the likelihood that convergence will be obtained. Most of these guidelines centre around the concept of
keeping the nonlinear model as simple as possible. Due to the length of a nonlinear solution compared to
a linear one, every attempt should be made to reduce the model size without compromising solution
accuracy. This is where your linear preliminary studies may be useful. Expect that you will run a
nonlinear model many times, both to achieve convergence and to adjust the final behaviour. Smart
modelling can save hours of run time. Keep the following points in mind while building the model for a
nonlinear solution.
Use symmetry wherever possible. This is a good idea in a linear model and a great idea in a nonlinear
model. The guidelines for using symmetry as described earlier in this book should be used, such as
avoiding symmetry in a nonlinear buckling problem or a when a dynamic solution is required.
Use beam, shell, or planar idealisations whenever possible. If a problem is marginal from the standpoint
of using an idealisation, it may still be worth-while to consider the simplification as a test model and learn
as much as you can from it. Region your model to use the nonlinear material model only where required.
A perfectly valid technique is to restrict the use of nonlinear elements to regions where plasticity is
expected. This may require you to mesh the model with multiple properties so that, in reality, two
separate materials are used. Chapters 5 and 7 discuss this technique in more detail. By limiting the use of
nonlinear elements to the areas that require it, you can speed up the model by forcing the solver to iterate
on only a subset of the mesh.
Refine and smooth the mesh in areas of high strain. Nonlinear solutions are sensitive to element
distortion, and the discontinuity these elements cause can force the solver to unnecessarily iterate a higher
strain, plastic solution when none should have been required. If an explicit nonlinear transient solution is
required, distorted elements may skew the time stepping algorithm unnecessarily as well. The mesh at any
contact region should be refined to capture the contact stresses that will be developed. As the contact area
gets smaller, the need for more refinement increases. You may be able to "un-refine" the mesh where
stress is low to reduce the model size as long as the overall model stiffness is not affected.
If large deformations are expected to distort elements such that their accuracy maybe called into
question, it may be worthwhile to manually distort the elements in the opposite direction somewhat before
starting the solution. In this way, the final shape will be closer to the ideal shape. Use your judgement to
determine if this is necessary for accuracy or even to prevent the solution from failing due to the presence
of highly distorted elements.
Always check your software documentation for element types allowed in a nonlinear solution. Many
codes restrict the use of elements in a nonlinear solution to a subset of those available in a linear solution.
The use of higher order elements is discussed below. Certain line, rigid, and speciality elements may be
restricted for use in linear or dynamic solutions only. Take the time to read your documentation to
understand which elements can and cannot be used in a nonlinear problem. Convergence problems can
sometimes be resolved simply by changing element types to those better suited for a nonlinear problem.
If you must use a solid model for a nonlinear solution, keep the following additional points in mind while
constructing the mesh.
Take another look at small or insignificant features in the model. Due to the speed of most linear
solutions, you may have developed the habit of leaving in some fillets or features simply because it took
more time to remove them than they added to the run time. While this is acceptable in a linear model, it
6
Recommendations
Page
8.6
DBA
Design by Analysis
can cause trouble in a nonlinear run. Consider that in addition to the increased length of a nonlinear
solution, many iterations may be required to achieve convergence. While this is not an issue in linear
analysis, any run time penalty due to an excessive mesh from unnecessary features will be paid for in
each iteration.
Even though current analysis software makes nonlinear analysis easy, much more attention must be
given to suitability of the model, its representation of the real component and interpretation of this
behaviour. It is harder to predict the structural response and a good understanding of the response
may develop only after carrying out several trial analyses. much more than in linear analysis,
the nature of a problem may become clear only after solving it. At the outset the types and extent of
the nonlinearity may not be apparent. Even if they are, the appropriate elements, mesh layout,
solution algorithms, and load steps may not be. Accordingly, an attempt to solve a nonlinear finite
element problem in one go is likely to fail
[4]
, or at worst be very misleading. A converged
solution, with appealing stress and strain contours may not be physically, or even mathematically,
correct! It is always advisable to make generous use of test cases and pilot studies before the main
analysis, then examine the effect of adjustments to all modelling assumptions. And above all, to ask
for help! All of this to get a feel for the behaviour of the model, and to test if it represents the real
component. The selection of suitable test cases and pilot studies is not easy for nonlinear analysis,
so justification of the analysis is even more difficult.
8.3 Software requirements
From the preceding discussion and Sections 3 & 7, it is clear that, as a basic requirement the
analysts should have access to finite element analysis software which includes:
1. Three dimensional and axisymmetric solid and shell elements
2. Thermal stress analysis
3. Small and large deflection elastic and elastic-plastic analysis
4. Instability and buckling capability
5. The ability to handle combined actions and complex load histories
6. A variety of conventional models of classical plasticity, including Tresca and von Mises initial
yield and some subsequent yield models (for example isotropic and kinematic hardening)
For some of the specialist procedures described in Section 3, the software should preferably also
include:
7. A macro language for enhanced post-processing, or to control procedures such as elastic
compensation
8. Built-in procedures for stress linearisation (which could of course be developed using a macro
language if necessary)
In general most modern commercial finite element software should be capable of the basic
requirements. A brief summary is given below:
ABAQUS (elastic, plastic and creep calculations with beam, shell or solid elements),
7
Recommendations
Page
8.7
DBA
Design by Analysis
ALGOR (elastic and plastic calculations with beam, shell or solid elements, it can compute limit
loads using inelastic calculations),
CASTEM 2000 (elastic, plastic and creep calculations with beam, shell or solid elements, it can
compute limit loads using elastic calculations),
SYSTUS (elastic, plastic and creep calculations with beam, shell or solid elements),
ANSYS (elastic, plastic and creep calculations with beam, shell or solid elements),
NASTRAN (elastic, plastic and creep calculations with beam, shell or solid elements),
CASTOR (elastic calculations with beam, shell or solid elements),
COSMOS (elastic, plastic and creep calculations with beam, shell or solid elements),
MAGICS (elastic calculations with beam, shell or solid elements),
CADSAP (elastic calculations with beam, shell or solid elements),
SAMCEF (elastic, plastic and creep calculations with beam, shell or solid elements),
FE-PIPE (elastic calculations with beam, shell or solid elements, especially for pressure vessel
components),
BOSOR4 (program for stress, buckling and vibration of complex shells of revolution).
A fuller comparison can be found in the text by Adams & Askenazi
[3]
.
8.4 Analyst requirements
As mentioned in the Introduction, Section 1, the expertise of the analyst/designer plays a critical
role in the whole design by analysis process. While the Standard itself, and to some extent most
commercial finite element software, undergo rigorous quality assessment, this is not the case with
the analyst. Indeed it is wholly possible for a relatively new and inexperienced engineer or analyst
to be allocated the task of design by analysis. Further, it is equally typical to team an experienced
pressure vessel designer with an experienced finite element analyst who has no specific familiarity
with the requirements of pressure vessel design, with neither having any real experience of design
by analysis, or inelastic analysis. This is just as much a recipe for disaster as the novice.
Typical professional training of engineering graduates across Europe does not examine inelastic
material behaviour and constitutive modelling, or inelastic stress analysis in any great detail (if at
all), and only a few cover the detailed behaviour of elastic-plastic components and structures, let
alone pressure vessel design. Training in practical finite element analysis is common, but tends to
emphasise modelling and the interface to CAD. Training in finite element theory beyond linear
elasticity is rare. As a consequence, analysts must gain the necessary experience for pressure vessel
8
Recommendations
Page
8.8
DBA
Design by Analysis
design by analysis on the job, hopefully apprenticed to an expert, and through Continuing
Professional Development Courses. That this additional training is rarely professionally accredited
has long been a concern of finite element educators
[6]
.
The experienced pressure vessel analyst and designer is uncommon in the industry. One of the
failings of modern engineering in general is that the gap between technical knowledge of
engineering principles and analysis software capabilities is growing! This has been expressed well
by Adams & Askenazi
[3]
(italics added by the writers for emphasis):
this literal explosion of speed and capabilities has been a mixed blessing for design engineers.
From the near-seamless, near-painless integration with CAD, a growing number of design analyst
non-specialists with a part-time interest in FEA has emerged. The growth of this segment of the
industry is accelerating rapidly as technology providers scramble to fill the need for easy-to-use
FEA software tools. Some of these tools are so easy to use, little thought is required to develop
stress contours on parts with complex geometry. With little thought comes little chance of accuracy.
The early analysts sweated every node and element and paid dearly for lengthy run times. Good
practices were developed to ensure boundary conditions and properties were well thought out
before a model was submitted. Hand calculations and results correlation increased users' awareness
of the capabilities and limitations of their tools. Today, new users tend to believe that any result that
looks right probably is right. Material properties are assigned carelessly and boundary conditions
are applied more out of convenience than based on actual environmental interactions. Is the role of
the analysis specialist doomed? Hardly! Most design analysts would agree that their usage of the
technology is, by the nature of their responsibilities, only surface level. Many do not want nor do
they have time to pursue more complex analysis types or assembly simulation with ambiguous
interactions. Engineers have come to expect the instant gratification provided by photo-realistic
CAD models and rapid prototyping systems. Consequently, most design analysts who learn to
appreciate the complexity of FEA hesitate to jump head first into a problem that requires lengthy
setup, and run times which often yield less than spectacular data
an editorial in the July 9, 1998 issue of MACHINE DESIGN magazine addressed the subject of
designers and FEA, This editorial's premise was essentially, 'Why not let designers do FEA without
understanding engineering theory? The discussion was supported by an example of a failed
aerospace project conducted by expert analysts that relied a little too heavily on analytical results.
The second example described an aerospace project completed by engineers who were admittedly
weak on theory and chose to use "make-and-break" methods instead of analysis to great success.
Based on these two data points, the author concluded that because expert analysts cannot always get
it right, designers without grounding in fundamentals can do no worse. This somewhat irresponsible
position is contradictory, assuming that the reader is objective in his/her views on the subject.
Taken to its extreme, one might conclude that because successful products have been developed by
individuals with no engineering background whatsoever and products developed by highly trained
and competent engineers have experienced failures, a basic engineering education should not be a
requirement for aircraft development!
Good users will readily admit they need to know more. Those who feel satisfied are probably
in trouble. Surprisingly, few design analysts model, analyse, and interpret results efficiently or
even accurately. However, most do not realise it unless a major mistake costs their company money

9
Recommendations
Page
8.9
DBA
Design by Analysis
The engineering manager thinking of embarking on design by analysis should ponder the above
quotes very carefully indeed!
So what are the educational requirements of the analyst?
The following are considered to be essential elements of a competent analysts basic training for
pressure vessel design by analysis:
1. An accredited undergraduate degree in mechanical/civil/plant engineering (preferably to the
highest SARTOR professional level)
2. A knowledge of the classical theory of plasticity, including multiaxial behaviour, subsequent
yield and standard elastic-plastic constitutive models
3. An understanding of the behaviour of standard components in the plastic range including the
basic concepts of limit analysis, failure mechanisms, plastic hinges and shakedown
4. A basic knowledge of finite element theory and common element formulations
5. A few years experience of linear elastic finite element analysis of complex structures and
components, working with a knowledgeable analyst
6. An understanding of the procedures of elastic-plastic finite element analysis
7. A basic knowledge of the behaviour and stress systems in common thin-walled pressurised
components
8. A basic knowledge of buckling in simple plates and shells
9. An understanding of the aims and requirements of the various national codes and standards,
including the European Standard
The following would be considered desirable elements:
10. A basic understanding of the behaviour of thin curved shells and the behaviour of more
complex pressurised components under mechanical and thermal loading
11. A basic understanding of modern constitutive models for elastic-plastic material behaviour,
and the limitations of the classical theory of plasticity
12. A basic understanding of finite element theory for nonlinear analysis, in particular both
material and geometrical nonlinearities
13. An understanding of the procedures for, but also the limitations of, finite element instability
and buckling analysis
14. A basic understanding of plasticity in the presence of large deformations, and the limitations
of associated constitutive and numerical models
15. Experience with a range of analysis tools and commercial finite element software
16. Basic, but first hand, experience of pressure vessel manufacture, construction, operation and
maintenance
These recommendations may seem to be rather rigorous, but it must be remembered that many
pressure vessels which need to be considered for design by analysis through the Direct Route may
be high integrity and consequently a high level of technical competence may be required. Of course
it may be argued that a team of engineers with a pooled competence may be sufficient, but in some
of the writers experience this may not be satisfactory. It is well understood in engineering
education that a higher level of technical knowledge and competence will be required by
tomorrows analyst and designer. This requirement is prompted not only by the availability, and
relative ease of use, of modern analysis tools, but also by the growing need for industrial
10
Recommendations
Page
8.10
DBA
Design by Analysis
competitiveness at the same time as increased public demand for safety. It is equally well
understood that only a basic introduction to a range of technical skills can be given in accredited
degree level engineering education, for example as detailed by SARTOR. The only way tomorrows
engineer (or indeed anyone in the workplace) can acquire these skills would be through lifelong
learning and Continuing Professional Development, both of which are seen as crucial. The question
of accreditation of engineering competence in the use of national codes and standards must be faced
at some time, as the level required increases.
8.5 Concluding remarks
The new approach to DBA, as laid down in CEN's prEN 13445-3, seems to be a major step forward
in design by analysis. Problems still contained in the present proposal have been identified in this
European research project, initiated by EPERC. This project has shown that the approach is sound,
that the approach gives much insight into the behaviour of the vessel, into the safety margins
against failure modes, and, therefore, that this approach can lead to improved designs and improved
in-service inspection procedures. This project has also shown that the time effort required is, even
with presently available hard and software, affordable.
It is clear to the writers of this document, that if the new European Standard recommendations for
Design by Analysis are to succeed and become more widely used, with the accompanying increased
safety and reliability, there is a real need for Continuing Professional Development Courses.
8.6 Literature
[1] J.N.C. Guerreiro, A.F.D. Loula & J.T. Boyle: Finite element methods in stress analysis for
creep General Lecture, Creep in Structures IV, Proceedings of IUTAM Symposium, Cracow,
1990. Ed M Zyczkowski, Springer, 1991
[2] NAFEMS: How to Interpret Finite Element Results National Agency for Finite Element
Methods & Standards, 1990
[3] V. Adams & A. Askenazi: Building Better Products with Finite Element Analysis Onward
Press, 1999
[4] R.D. Cook, D.S. Malkus, M.E. Plesha: Concepts and Applications of Finite Element Analysis,
John Wiley, 3
rd
Ed, 1994
[5] E. Hinton et al, Eds: NAFEMS Introduction to Nonlinear Stress Analysis National Agency for
Finite Element Methods & Standards, 1992
[6] J.T. Boyle et al, Eds: Finite Element Analysis; Education & Training Elsevier Applied
Science, 1991
1
Annex 1: Bibliography
Page
A1.1
DBA
Design by Analysis





In this Annex, three tables are presented as a literature review on design by analysis:

Table A1.1: Books concerning DBA, limit and shakedown analysis.
Table A1.2: DBA using the finite element method (FEM), numerical and/or analytical
procedures.
Table A1.3: Theory and/or numerical methods for limit and shakedown analysis.


2
Annex 1: Bibliography
Page
A1.2
DBA
Design by Analysis
Table A1.1: Books concerned with DBA, limit and shakedown analysis


AUTHOR TITLE SOURCE
Bolt, S.E.; Bryson, J.W.; Corum,
J.M.; Gwaltney, R.C.
THEORETICAL AND EXPERIMENTAL STRESS ANALYSIS OF ORNL THIN-SHELL
CYLINDER-TO-CYLINDER MODEL NO. 3
Oak Ridge National
Laboratory - 5020
Bolt, S.E.; Bryson, J.W.; Gwaltney,
R.C.
THEORETICAL AND EXPERIMENTAL STRESS ANALYSIS OF ORNL THIN-SHELL
CYLINDER-TO-CYLINDER MODEL NO. 4
Oak Ridge National
Laboratory - 5019
Bolt, S.E.; Bryson, J.W.; Gwaltney,
R.C.
THEORETICAL AND EXPERIMENTAL STRESS ANALYSIS OF ORNL THIN-SHELL
CYLINDER-TO-CYLINDER MODEL NO. 2
Oak Ridge National
Laboratory - 5021
Burgreen,D. DESIGN METHODS FOR POWER PLANT C.P. Press, 1975
Burgreen,D. PRESSURE VESSEL ANALYSIS C.P. Press, 1979
Chen,W.F.; Han,D.J. PLASTICITY FOR STRUCTURAL ENGINEERS Springer, 1988
Corum, M.J.; Bolt, S.E.;
Greenstreet, W.E.; Gwaltney, R.C.
THEORETICAL AND EXPERIMENTAL STRESS ANALYSIS OF ORNL THIN-SHELL
CYLINDER-TO-CYLINDER MODEL NO. 1
Oak Ridge National
Laboratory - 4553
Gokhfeld, D.A.; Cherniavsky, O.F. LIMIT ANALYSIS OF STRUCTURES AT THERMAL CYCLING Sijthoff & Noordhoff, 1980
Kamenjarzh, J. LIMIT ANALYSIS OF SOLIDS AND STRUCTURES CRC-Press, Florida, 1996
Knig, J.A. SHAKEDOWN OF ELASTIC PLASTIC STRUCTURES Elsevier, 1987
Save, M.A.; Massonnet, C.E.; PLASTIC ANALYSIS AND DESIGN OF PLATES, SHELLS AND DISKS North-Holland, 1972
Sawczuk, A.; Jaeger, T. GRENZTRAGFHIGKEITSTHEORIE DER PLATTEN Springer, 1963
Zeman, J.L;
REPETITORIUM APPARATEBAU/GRUNDLAGEN DER
FERSTIGKEITSBERECHNUNG
Oldenbourg, 1992
3
Annex 1: Bibliography
Page
A1.3
DBA
Design by Analysis
Table A1.2: DBA using FEM, analytical and/or numerical procedures


AUTHOR TITLE SOURCE
Hamilton, J.; Boyle, J.T.,
Shi, J., Mackenzie, D.
SHAKEDOWN LOAD BOUNDS BY ELASTIC FINITE ELEMENT
ANALYSIS
PVP - Vol. 343, p. 205 f.f.
Zeman, Josef L.
SOME ASPECTS OF THE WORK OF THE EUROPEAN WORKING
GROUPS RELATING TO BASIC PRESSURE VESSEL DESIGN
IJPV&P, 70, p.3-10
Mackenzie, D.; Boyle,
J.T.;
A SIMPLE METHOD OF ESTIMATING SHAKEDOWN LOADS FOR
COMPLEX STRUCTURES
PVP-Vol. 265, p. 89-94
Hollinger, G.L. SUMMARY OF THREE DIMENSIONAL STRESS CLASSIFICATION 7.th ICPVT, Vol. 1, p. 454 f.f.
Hamilton, R.; Mackenzie,
D.; Shi, J.; Boyle, J. T.;
SIMPLIFIED LOWER BOUND LIMIT ANALYSIS OF PRESSURISED
CYLINDER/CYLINDER INTERSECTIONS USING GENERALISED
YIELD CRITERIA
IJPV&P, 67, p. 219-226
Mackenzie, D.; Nadarajah,
C.; Shi, J.; Boyle, J. T.;
SIMPLE BOUNDS ON LIMIT LOADS BY ELASTIC FINITE
ELEMENT ANALYSIS
IJPVT, Vol. 115, p. 27-31
Shi, J.; Mackenzie, D.;
Boyle, J. T.;
A METHOD OF ESTIMATING LIMIT LOADS BY ITERATIVE
ELASTIC ANALYSIS. III-TORISPHERICAL HEADS UNDER
INTERNAL PRESSURE
IPV&P, 53, p. 121-142
Yamamoto, Y.; Asada, S.;
Okamoto, A.
ROUND ROBIN CALCULATIONS OF COLLAPSE LOADS - A
TORISPHERICAL PRESSURE VESSEL HEAD WITH A CONICAL
TRANSITION
IJPVT, Vol. 119, p. 503 f.f.
Hollinger, G.L..; Hechmer,
J.L.
CODE STRESS CLASSIFICATION FOR EVALUATION BY 3D
METHODS
2-Part Workshop; February
4,1998 PVRC Meeting; San
Diego, CA
Mackenzie, D.; Boyle, J.T.
ASSESSMENT OF CLASSIFICATION PROCEDURES FOR FINITE
ELEMENT STRESSES
7th ICPVT, Vol.1, p. 346-358
Seibert, T.; Zeman, J.L.
ANALYTISCHER ZUVERLSSIGKEITSNACHWEIS VON
DRUCKGERTEN
Technische berwachung, Bd.
35 (1994), Nr. 5, p. 222 f.f.
Yeom, D.J.; Robinson, M.;
NUMERICAL ANALYSIS OF THE ELASTIC-PLASTIC BEHAVIOUR
OF PRESSURE VESSELS WITH ELLIPSOIDAL AND
TORISPHERICAL HEADS
IJPV&P, 65, p. 147-156
Porowski, J.S.; Kasraie, B.;
Bielawski, G.; O'Donnell,
W.J.; Badlani, M.L.
PRIMARY STRESS EVALUATIONS FOR REDUNDANT
STRUCTURES
PVP-Vol. 265, p. 185 f.f.
Tanaka, M.; Yamamoto,
Y.
2D AND 3D COLLAPSE EVALUATION OF SMALL RADIAL AND
OBLIQUE NOZZLE TO SPHERICAL SHELL INTERSECTIONS
to be presented at ASME PVP
1998
Boyle, J.T.; Hamilton, R.;
Mackenzie, D.;
A SIMPLE METHOD OF CALCULATING LOWER-BOUND LIMIT
LOADS FOR AXISYMMETRIC THIN SHELLS
IJPVT, Vol. 119, p. 236-242
4
Annex 1: Bibliography
Page
A1.4
DBA
Design by Analysis
Table A1.2: DBA using FEM, analytical and/or numerical procedures (continued)


AUTHOR TITLE SOURCE
Mackenzie, D.; Boyle,
J.T.; Hamilton, R.;
APPLICATION OF INELASTIC FINITE ELEMENT ANALYSIS TO
PRESSURE VESSEL DESIGN
8th IC PVT, Vol. 2, p. 109-115
Preiss, R.; Rauscher, F.;
Vazda, D.; Zeman, J.L.
THE FLAT END TO CYLINDRICAL SHELL CONNECTION - LIMIT
LOAD AND CREEP DESIGN
IJPV&P 75 (1998), p 715-726
Dixon, R.D.; Perez, E. H.
COMPARISON BETWEEN LINEAR-ELASTIC AND LIMIT
ANALYSIS METHODS FOR THE DESIGN OF HIGH PRESSURE
VESSELS
PVP-Vol. 344, p.43 f.f.
Mackenzie, D.; Boyle,
J.T.
COMPUTIONAL PROCEDURES FOR CALCULATING PRIMARY
STRESS FOR THE ASME B&PV CODE
PVP-Vol. 265, p. 177 f.f.
Okamoto, A.
PRIMARY STRESS EVALUATION PROCEDURE FOR NOZZLE-
TOCYLINDER JUNCTURE UNDER PRESSURE LOADING
to be presented at ASME PVP
1998
Nadarajah, C.;
Mackenzie, D.; Boyle,
J.T.;
LIMIT AND SHAKEDOWN ANALYSIS OF NOZZLE/CYLINDER
INTERSECTIONS UNDER INTERNAL PRESSURE AND IN-PLANE
MOMENT LOADING
IJPV&P, 68, p. 261-272
Lapsley, C.; Mackenzie,
D.; Tooth, A. S.;
LOCAL LOADING OF CYLINDERS: LIMIT ANALYSIS 8th ICPVT, Vol. 2, p. 55-59
Kalnins, A.; Updike, P.D.;
Park, I.
PLASTIC STRAINING AT NOZZLES IN PRESSURE VESSELS PVP-Vol. 353, p. 143 f.f.
Kalnins, A.; Updike, P.D.
EFFECT OF HYDROSTATIC PROOF TEST ON SHAKEDOWN OF
TORISPHERICAL HEADS UNDER CYLIC PRESSURE
PVP-Vol.353, p.217 f.f.
Mackenzie, D.; Boyle, J.
T.;
A METHOD OF ESTIMATING LIMIT LOADS BY ITERATIVE
ELATSIC ANALYSIS: I-SIMPLE EXAMPLES
IJPV&P, 53, p. 77-95
Nadarajah, C.;
Mackenzie, D.; Boyle,
J.T.;
A METHOD OF ESTIMATING LIMIT LOADS BY ITERATIVE
ELASTIC ANALYSIS. II-NOZZLE SPHERE INTERSECTIONS WITH
INTERNAL PRESSURE AND RADIAL LOAD
IJPV&P, 53, p. 97-119
Scavuzzo, R. J.; Lam, P.
C.; Gau, J.S.;
RATCHETING OF PRESSURIZED PIPING SUBJECTED TO SEISMIC
LOADING
PVP-Vol. 197, p. 77-84
Lietzmann, A.; Rudolph,
J.; Wei, E.;
FAILURE MODES OF PRESSURE VESSEL COMPONENTS AND
THEIR CONSIDERATION IN ANALYSES
Chemical Engineering and
Processing, 35, p. 287-293
Jin-Guang-Teng
PLASTIC COLLAPSE AT LAP JOINTS IN PRESSURIZED
CYLINDERS UNDER AXIAL LOAD
Journal of Structural
Engineering, Vol. 120, No. 1, p.
23-44
Porowski, J.S.; O'Donnell,
W.J.; Reid, R.H.;
WEIGHT-SAVING PLASTIC DESIGN OF PRESSURE VESSELS IJPVT, Vol. 119, p. 161-166
5
Annex 1: Bibliography
Page
A1.5
DBA
Design by Analysis
Table A1.2: DBA using FEM, analytical and/or numerical procedures (continued)


AUTHOR TITLE SOURCE
Miklus S.; Kosel, F.; PLASTIC COLLAPSE OF PIPE BIFURCATION IJPV&P, 48, p. 79-92
Naderan-Tahan, K.;
Robinson, M.;
PLASTIC LIMIT PRESSURES FOR NEIGHBOURING RADIAL
NOZZLES IN A SPHERICAL PRESURE VESSEL
J. of Process Mechanical
Engineering, Vol. 210, p. 75-78
Porter, A.M.; Martens, D.
A COMPARISON OF THE STRESS RESULTS FROM SEVERAL
COMMERCIAL FINITE ELEMENT CODES WITH ASME SECTION
VIII, DIVISION 2 REQUIREMENTS
http://www.dynamicanalysis.co
m/pvp96a.htm
Updike, D.P.; Kalnins, A.
ULTIMATE LOAD ANALYSIS FOR DESIGN OF PRESSURE
VESSELS
PVP-Vol.353, p. 289 f.f.
Teng, J.-G.
CONE-CYLINDER INTERSECTION UNDER PRESSURE:
AXISYMMETRIC FAILURE
Journal of Engineering
Mechanics, Vol. 120, No. 9, p.
1896 f.f
Tabone, C. J.; Mallett, R.
H.;
PRESSURE PLUS MOMENT LIMIT LOAD ANALYSIS FOR A
CYLINDRICAL SHELL NOZZLE
PVP - Vol. 120, p. 131-136
Shalaby, M.A.; Younan,
M.Y. A.
LIMIT LOADS FOR PIPE ELBOWS WITH INTERNAL PRESSURE
UNDER IN-PLANE CLOSING BENDING MOMENT
IJPVT, Vol. 120 , p. 35-42
Broyles, R.K.
FINDING DESIGN ACCETABILITY USING FINITE ELEMENT
ANALYSIS
PVP-Vol. 353, p. 151 f.f.
Babu, S.; Iyer, B.K.
INELASTIC ANALYSIS OF COMPONENTS USING A MODULUS
ADJUSTMENT SCHEME
IJPVT, Vol 120, p. 1-5
Seshadri, R.
ROBUST STRESS-CLASSIFICATION OF PRESSURE COMPONENTS
USING THE GLOSS METHODS
PVP-Vol. 265, p. 155 f.f.
Teng, J.-G.
CONE-CYLINDER INTERSECTION UNDER INTERNAL PRESSURE:
NONSYMMETRIC BUCKLING
Journal of Engineering
Mechanics, Vol. 121, No. 12, p.
1298 f.f.
Seshadri, R.; Fernando, C.
P. D.;
LIMIT LOADS OF MECHANICAL COMPONENTS AND
STRUCTURES USING THE GLOSS R-NODE METHOD
IJPVT, Vol. 114, p. 201-208
Rogalska, E.; Kakol, W.;
Guerlement, G.;Lamblin,
D.;
LIMIT LOAD ANALYSIS OF PERFORATED DISKS WITH SQUARE
PENETRATION PATTERN
IJPVT, Vol. 119, p. 122-126
Tong, R.; Wang, X.;
SIMPLIFIED METHOD BASED ON THE DEFORMATION THEORY
FOR STRUCTURAL LIMIT ANALYSIS - I. THEORY AND
FORMULATION
IJPV&P, 70, p. 43-49
Tong, R.; Wang, X.;
SIMPLIFIED METHOD BASED ON THE DEFORMATION THEORY
FOR STRUCTURAL LIMIT ANALYSIS - II. NUMERICAL
APPLICATION AND INVESTIGATION ON MESH DENSITY
IJPV&P, 70, p. 51-58
6
Annex 1: Bibliography
Page
A1.6
DBA
Design by Analysis
Table A1.2: DBA using FEM, analytical and/or numerical procedures (continued)


AUTHOR TITLE SOURCE
Teng, J.G.; Rotter, J.M., PLASTIC COLLAPSE OF RESTRAINED STEEL SILO HOPPERS
J. Construct. Steel Research , 14,
p. 139-158
Teng, J.-G.; Rotter M.
COLLAPSE BEHAVIOR AND STRENGTH OF STEEL SILO
TRANSITION JUNCTIONS. PART 2: PARAMETRIC STUDY
Journal of Structural
Engineering, Vol. 117, No.12, p.
3605 f.f.
Mello, R.M.; Griffin, D.S.;
PLASTIC COLLAPSE LOADS FOR PIPE ELBOWS USING INELASTIC
ANALYSIS
IJPVT, Aug. 1974, p.177-183
Teng, J.-G.; Rotter, J. M.
COLLAPSE BEHAVIOR AND STRENGTH OF STEEL SILO
TRANSITION JUNCTIONS. PART 1: COLLAPSE MECHANICS
Journal of Structural
Engineering, Vol.117, No.12, p.
3587 f.f.
Adinarayana, N.; Alwar,
R.S.;
ELASTOPLATIC STRESS AND STRAIN PREDICTIONS FOR
AXISYMMETRIC PRESSURE VESSEL BODIES UNDER
TRANSIENT THERMAL LOADS
Journal of Strain Analysis, Vol.
31, No. 2, p. 81-89
Wei, E; Joost, H.
BESTIMMUNG DER TRAGLAST VON BEHLTERN UNTER
ZUSTZLICHEN LASTEN MITTELS FEM
14. CAD-FEM User's Meeting,
1996, Bad Aibling, Germany,
Art. 1-9
Wei, E; Postberg, B.
MODELLIERUNG VON RATCHETTING-EFFEKTEN BEZOGEN AUF
KOMPONENTEN DER CHEMIE- UND KRAFTWERKSTECHNIK
15. CAD-FEM User`s Meeting,
1997, Fulda, Germany, Art. 2-11
Galletly, G.D.; Aylward,
R.W.;
PLASTIC COLLAPSE AND THE CONTROLLING FAILURE
PRESSURES OF THIN 2:1 ELLIPSOIDAL SHELLS SUBJECTED TO
INTERNAL PRESSURE
IJPVT, Vol. 101, 64-72
Radhamohan, S.K.;
Galletly, G.D.;
PLASTIC COLLAPSE OF THIN INTERNALLY PRESSURIZED
TORISPHERICAL SHELLS
IJPVT, Vol. 101, p. 311-319
Seshadri, R.;
Mangalaramanan, S., P.;
LOWER BOUND LIMIT LOADS USING VERIATIONAL CONCEPTS:
THE M -METHOD
IJPV&P, 71, p. 93-106
Wei, E.; Lietzmann, A.;
Rudolph, J.
ELASTISCHE UND ELASTISCH PLASTISCHE
FESTIGKEITSANALYSEN GEWLBTER BDEN MIT UND OHNE
STUTZEN IM KREMOENBEREICH
VGB Kraftwerkstechnik 75
(1995), Heft 6, p.. 549 f.f.
Zarrabi. K.;
PLASTIC COLLAPSE PRESSURES FOR DEFECTED CYLINDRICAL
VESSELS
IJPV&P; 60, p. 65-69
Hayakawa, T.; Yoshida, T.;
Mii, T.
COLLAPSE PRESSURE FOR THE SMALL END OF A CONE-
CYLINDER JUNCTION BASED ON ELASTISC-PLASTIC ANALYSIS
3rd ICPVT, Vol. 1, p. 149 f.f.
Wei, E.; Lietzmann, A.;
Rudolph, J.;
FESTIGKEITSANALYSE UND BEANSPRUCHUNGSBEWERTUNG
FR KOMPONENTEN DES DRUCKBEHLTERBAUS
Technische berwachung, Bd.
36, Nr. 11/12, p. 424-430
Webster, J.J.; Sahari, B.B.;
Hyde, T.H.;
ELASTIC-PLASTIC BEHAVIOUR OF A SIMPLY SUPPORTED
CIRCULAR PLATE SUBJECTED TO STEADY TRANSVERSE
PRESSURE AND CYCLIC LINEAR RADIAL TEMPERATURE
VARIATION
I.J.Mech.Sci., Vol. 29, No. 8, p.
553-544
7
Annex 1: Bibliography
Page
A1.7
DBA
Design by Analysis
Table A1.2: DBA using FEM, analytical and/or numerical procedures (continued)


Gerlach, H.D.; Ludeke,
M.
COMPARISON OF CONCEPTS FOR STRESS ANALYSIS OF
COMPONENTS IN PLANT TECHNOLOGY
10th MPA seminar, 11/12
October 1984, Stuttgart,
Germany
Gerlach, H.D.; Ludeke,
M.
INTERNATIONAL COMPARISON OF STRESS ANALYSIS
CONCEPTS FOR CHEMICAL PLANT COMPONENTS
6th ICPVT, Vol. 1 (Design and
Analysis), Beijing, 1988, 407-
421
Ringelstein, K.H.; Baues
COMPARISON AND APPLICATION OF INTERNATIONAL
TECHNICAL CODES, SPECIFICALLY PRESENTED WITH A VIEW
TO ASME BOILER AND PRESSURE VESSEL CODE AND THE
APPLICABLE GERMAN RULES, PART 3: DESIGN MINISTERY OF
RESEARCH AND TECHNOLOGY
Project no. RS 150 345, June
1980
Roche, R.
FACTEURS MECANIQUES ET METALLURIQUES DE LA
RUPTURE, NOTAMMENT DANS L'INDUSRIE NUCLEAIRE
17me colloque de mtallurgie
(Saclay, juin 1974)
Roche; R.
CRITERES MODERNES DAPPRECIATION DA LA SECURITE DES
APPAREILS A PRESSION, NOTAMMENT DANS LE DOMAINE
NUCLEAIRE
Confrence du 7 fvrier 1974,
salle Chaleil, 11 avenue Hoche,
Paris 8
Cousseran, P.; Lebey, J.;
Moulin, D.;Roche,
R.;Clement, G.
NOUVELLES EXPERIENCES SUR lEFFET DE ROCHET - UNE
REGLE PRATIQUE DE PREVENTION
IIIme Congrs national sur la
technologie des appareils
pression, AFIAP 1-2-3 octobre
1980
Roche, R.
SPECIALISATION MECANIQUE DES STRUCTURES APPLIQUEES
AU GENIE NUCLEAIRE
Note technique
EMT/SMTS/79/08, 26 janvier
1979
document FRAMATOME
FONDEMENTS DES REGLES LES DE PREVENTION DE
LENDOMMAGEMENT DES MATERIELS MECANIQUES -
INTRODUCTION AUX REGLES DE LA CONCEPTION ET
DANALYSE DU RCC-M
octobre 1991
Debaene, J.B. METHODES DANALYSE DES APPAREILS A PRESSION
Techniques de l Ingnieur,
Chapitre A846
Kroenke, W.C.; Addicott,
G.W.; Hinton, B.M.
INTERPREATION OF FINITE ELEMENT STRESSES ACCORDING
TO ASME SECTION III
ASME paper 75-PVP-63
(1975)
Hollinger, G.L.; Hechmer,
J.M.
THREE DIMENSIONAL STRESS CRITERIA PVRC 1991 Phase 1 Report
Hollinger, G.L.; Hechmer,
J.M.
3-D STRESS CRITERIA: GUIDELINES FOR APPLICATION PVRC 1995 Phase 2 Report
Pastor, T.P.; Hechmer,
J.L.
ASME TASK GROUP REPORT ON PRIMARY STRESSES PVP-vol. 277, ASME (1994)
Porowski, J.S.; Kasraie,
B.; Bielawski, G.;
O'Donnell, W.J.; Badlani,
M.L.
PRIMARY STRESS EVALUATIONS FOR REDUNDANT
STRUCTURES
PVP-vol. 265, ASME (1993)




8
Annex 1: Bibliography
Page
A1.8
DBA
Design by Analysis
Table A1.2: DBA using FEM, analytical and/or numerical procedures (continued)


Hollinger, G.L.; Hechmer,
J.M.
THREE DIMENSIONAL STRESS CRITERIA PVRC 1991 Phase 1 Report
Hollinger, G.L.; Hechmer,
J.M.
3-D STRESS CRITERIA: GUIDELINES FOR APPLICATION PVRC 1995 Phase 2 Report
Pastor, T.P.; Hechmer,
J.L.
ASME TASK GROUP REPORT ON PRIMARY STRESSES PVP-vol. 277, ASME (1994)
Porowski, J.S.; Kasraie,
B.; Bielawski, G.;
O'Donnell, W.J.; Badlani,
M.L.
PRIMARY STRESS EVALUATIONS FOR REDUNDANT
STRUCTURES
PVP-vol. 265, ASME (1993)
Zeman, J.L., Preiss; R.
THE DEVIATORIC MAP - A SIMPLE TOOL IN DESIGN
BYANALYSIS
IJPV&P, 76, p. 339-344
Preiss, R.
ON THE SHAKEDOWN ANALYSIS OF NOZZLES USING ELASTO-
PLASTIC FEA
IJPV&P, 76, p. 421-434
Save, M.
LIMIT ANALYSIS OF PLATES AND SELLS: RESEARCH OVER TWO
DECADES
J. Struct. Mech., 13 (3&4), p.
343-370
Bree, J.
PLASTIC DEFORMATION OF A CLOSED TUBE DUE TO
INTERACTION OF PRESSURE STRESSES AND CYCLIC THERMAL
STRESSES
I.J.Mech.Sci., Vol. 31, No.
11/12, p. 865-892
Kalnins, A.; Updike, D.P.
SHAKEDOWN AND STRESS RANGE OF TORISPHERICAL HEADS
UNDER CYCLIC INTERNAL PRESSURE
PVP-Vol. 2, p. 25-31
Hollinger, G.L.;
Hechmer, J.M.
ASSESSMENT OF THE ASME CODE STRESS LIMITS FOR 3D,
SOLID ELEMENT, FINITE ELEMENT ANALYSIS: SUMMARY OF
THE PVRC PROJECT
PVP-Vol. 360, ASME (1998)
Kalnins, A.; Updike,
P.D.
PLASTICITY AND CHANGING GEOMETRY IN PRESSURE
VESSEL DESIGN
PVP-360, ASME (1998)
Blachut, J.;
Ramachandra, L. S.;
Krishnan, P. A.
EXPERIMENTAL AND NUMERICAL INVESTIGATION OF
PLASTIC LOADS FOR INTERNALLY PRESSUREISED VESSEL
HEADS
PVP-360, ASME (1998)
Teng, J.G.
PLASTIC COLLAPSE BEHAVIOUR AND STRENGTH OF CONE-
CYLINDER AND CONE-CONE INTERSECTIONS
PVP-360, ASME (1998)
Mohamed, A.I.;
Megahed, M.M.;
Bayoumi, L.S.; Younan,
M.Y.A..
APPLICATION OF ITERATIVE ELASTIC TECHNIQUES FOR
ELASTIC-PLASTIC ANALYSIS OF PRESSURE VESSELS
IJPVT, Vol 121 (1999), p. 24-
29
Kalnins, A.; Updike,
P.D.
SHAKEDOWN OF TORISPHERICAL HEADS USING PLASTIC
ANALYSIS
IJPVT, Vol 120 (1998), p 431-
437
Sanal, Z.
GEOMETRISCH UND PHYSIKALISCH NICHTLINEARE ANALYSE
VON DRUCKBEHLTERN
Stahlbau 67 (1998), Heft 6



9
Annex 1: Bibliography
Page
A1.9
DBA
Design by Analysis
Table A1.2: DBA using FEM, analytical and/or numerical procedures (continued)


AUTHOR TITLE SOURCE
Zeman, J.L.
DIE VERBINDUNG MANTEL-EBENER BODEN UNTER
DRUCKEINWIRKUNG
Technische berwachung, Bd. 35
(1994), Nr. 11/12, p. 450 f.f.
Zeman, J.L
DIE VERBINDUNG MANTEL-EBENER BODEN MIT
UNVERSTRKTEM KONZENTRISCHEN AUSSCHNITT UNTER
DRUCKEINWIRKUNG
Technische berwachung, Bd. 36
(1995), Nr. 4, p. 153 f.f.
Zeman, J.L.
RATCHETING LIMIT OF FLAT END TO CYLINDRICAL SHELL
CONNECTIONS UNDER INTERNAL PRESSURE
IJPV&P, 68 (1996), p. 293-298
Poth, W.; Zeman, J.L.
GRENZTRAGFHIGKEIT DER ZYLINDER-KEGEL-VERBINDUNG
UNTER INNENDRUCKEINWIRKUNG
Konstruktion, 48, p. 219-223
Myler, P.; Robinson, M.
LIMIT ANALYSIS OF INTERSECTING CONICAL PRESSURE
VESSELS
IJPV&P, 18, p. 209-240
Updike, D.P.; Kalnins,
A.; Hechmer, J.
ON PRIMARY STRESSES IN CONICAL REDUCERS PVP-Vol. 210-2, p. 117 f.f.
Kalnins, A.; Updike, P.D. PRIMARY STRESSES FROM SIMPLE LAWS OF EQUILIRIUM PVP-Vol. 353, p. 259 f.f.
Kalnins, A.; Updike, D.P. ON PRIMARY STRESS CALCULATIONS PVP-Vol. 265, p. 167-176
Kalnins, A.; Updike, D.P.
ROLE OF PLASTIC LIMIT AND ELASTIC-PLASTIC ANALYSIS IN
DESIGN
PVP-Vol.210-2, p. 135 f.f.
Moreton, D.N.
THE RATCHETTING OF A CYLINDER SUBJECTED TO INTERNAL
PRESSURE AND ALTERNATING AXIAL DEFORMATION
Journal of Strain Aanalysis, Vol. 28,
No. 4, p. 277-282
Cinquini, C.; Zanon, P. LIMIT ANALYSIS OF CIRCULAR AND ANNULAR PLATES Ingenieur-Archiv , 55, p. 157-175
Save, M. LIMIT ANALYSIS AND DESIGN OF CONTAINMENT VESSELS
Nuclear Engineering and Design, 79,
p. 343-361
Dinno, K. S.; Al-Zabin,
S. A.
LIMIT ANALYSIS OF CIRCULAR STEEL SILOS
Bull. Int. Ass. Shells&Spatial
Structures, V 38, 3. Nov. 87, p 37-45
Leckie, F. A.; Penny,
R.K.
SHAKEDOWN LOADS FOR RADIAL NOZZLES IN SPHERICAL
PRESSURE VESSELS
I. J.Solids & Structures, Vol.3, p.
743-755
Guerlement, G.; Lamblin,
D.O.; Save, M. A:;
LIMIT ANALYSIS OF CYLINDRICAL SHELL WITH REINFORCING
RINGS
Engineering. Stuct., Vol. 9, p. 146-
156
10
Annex 1: Bibliography
Page
A1.10
DBA
Design by Analysis
Table A1.2 DBA using FEM, analytical and/or numerical procedures (continued)


AUTHOR TITLE SOURCE
Ghorashi, M.
LIMIT ANALYSIS OF CIRCULAR PLATES SUBJECTED TO
ARBITRARY ROTATIONAL SYMMETRIC LOADINGS
I. J. Mech. Sci., Vol. 36, No. 2, p. 87-
94
Srinivasan, M. A.;
LIMIT STRENGTH OF LOCALLY LOADED SPHERICAL DOMES:
RADIAL LOAD AT A CIRCULAR CUT-OUT
I. J. Solids & Stuctures, Vol. 24, No.
7, p. 723-734
Updike, D.P.; Kalnins, A.
LOCAL PLASTIC COLLAPSE IN STAYED TUBESHEETS WITH
REGULAR SUPPORT ARRAYS
PVP-Vol. 235, p. 195-202
Biron, A.; Veillon, J.
INFLUENCE OF HEAD THICKNESS ON YIELD PRESSURE FOR
CYLINDRICAL PRESSURE VESSELS
IJPVT, May 1974, p. 113-120
Foo, S. S. B.
ON THE LIMIT ANALYSIS OF CYLINDRICAL SHELLS WITH A
SINGLE CUTOUT
IJPV&P, Vol.49, p. 1-16
Cocks, A.C.F.
LOWER-BOUND SHAKEDWON ANALYSIS OF A SIMPLY
SUPPORTED PLATE CARRYING A UNIFORMLY DISTRIBUTED
LOAD AND SUBJECTED TO CYCLIC THERMAL LOADING
I.J.Mech.Sci, Vol.26, No.9, p.471-
475
Robinson, M.
LOWER-BOUND LIMIT PRESSURES FOR THE CYLINDER-
CYLINDER INTERSECTION: A PARAMETRIC SURVEY
IJPVT, Vol. 100, p. 65-73
McDonald, C.K.;
McDonald, R.E.
NONLINEAR EFFECTS ON THE DISCONTINUITY STRESSES IN A
CYLINDRICAL SHELL WITH A FLAT HEAD CLOSURE
IJPVT, February 1974, p. 44-46
Save, M. A.
LIMIT ANALYSIS AND DESIGN: AN UP-TO-DATE SUBJECT OF
ENGINEERING PLASTICITY
Conf. Title: Plasticity Today, June
1983, Udine, Italy; publ. by
Elsevier, p. 767-785
Karadeniz, S.; Ponter,
A.R.S.; Carter, K.F.;
THE PLASTIC RATCHETING OF THIN CYLINDRICAL SHELLS
SUBJECTED TO AXISYMMETRIC THERMAL AND MECHANICAL
LOADING
IJPVT, Vol. 109, p. 387-393
Jiang, W.
THE ELASTIC PLASTIC ANALYSIS OF TUBES-I: GENERAL
THEORY
IJPVT, Vol. 114, p. 213-221
Jiang, W.
THE ELASTIC-PLASTIC ANALYSIS OF TUBES-II: VARIABLE
LOADING
IJPVT, Vol. 114, p. 222-228
Jiang, W.
THE ELASTIC-PLASTIC ANALYSIS OF TUBES-III: SHAKEDOWN
ANALYSIS
IJPVT, Vol. 114, p. 229-235
Jiang, W.
THE ELASTIC PLASTIC ANALYSIS OF TUBES-IV: THERMAL
RATCHETTING
IJPVT, Vol. 114, p. 236-245
Cinquini, C.;
Guerlement, G.; Lamblin,
D.;
SHELL-STIFFENER INTERACTION. APPLICATION TO SIMPLY
SUPPORTED CYLINDRICAL SHELLS UNDER UNIFORM
PRESSURE
I. J. Solids & Structures, Vol. 21,
Nr. 5, p. 447-465
11
Annex 1: Bibliography
Page
A1.11
DBA
Design by Analysis
Table A1.3: Theory and/or numerical methods for limit and shakedown analysis


WRITER TITLE SOURCE
Rust, W.
MATERIALMODELLE FR ANSYS:
PULVERMETALLURGIE, ZYKLISCHE PLASTIZITT,
TRESCA
15.CAD-FEM Users' Meeting, Oktober
1997, Fulda, Germany, Art. 1-26
Franco, J.R.Q;
Ponter, A.R.S.
A GENERAL APPROXIMATE TECHNIQUE FOR THE FINITE
ELEMENT SHAKEDOWN AND LIMIT ANALYSIS OF
AXISYMMETRIC SHELLS. PART 2: NUMERICAL
APPLICATIONS
International Journal For Numerical Methods
in Engineering, Vol. 40, p. 3515-3536, 1997
Franco, J.R.Q.;
Ponter, A.R.S.
A GENERAL APPROXIMATE TECHNIQUE FOR THE FINITE
ELEMENT SHAKEDOWN AND LIMIT ANALYSIS OF
AXISYMMETRIC SHELLS: PART 1: THEORY AND
FUNDAMENTAL RELATIONS
International Journal for Numerical Methods
in Engineering, Vol. 40, p. 3495-3513, 1997
Morelle,P.; Fonder,
G.
SHAKEDOWN AND LIMIT ANALYSIS OF SHELLS - A
VARIATIONAL AND NUMERICAL APPROACH
Lecture Notes in Engineering 26. Publ. by
Springer-Verlag, p. 381-405, (Conf: Shell and
Spatial Structures, Louvain, Belg., 1986)
Stumpf, H.; Le
Khanh Chau
ON SHAKEDOWN OF ELASTOPLASTIC SHELLS
Quarterly of Applied Mathematics, Vol.
XLIX, No. 4, p. 781-793
Berak, E. G.;
Gerdeen, J. C.
A FINITE ELEMENT TECHNIQUE FOR LIMIT ANALYSIS OF
STRUCTURES
PVP - Vol. 124, p. 1-12
Borges, L. A.;
Zouain, N.; Huespe,
A. E.;
A NONLINEAR OPTIMIZATION PROCEDURE FOR LIMIT
ANALYSIS
Eur. J. Mech., A/Solids, 15, No. 3, p. 487-512
Esslinger, M.;
Geier, B.; Wendt,
U.
BERECHNUNG DER TRAGLAST VON
ROTATIONSSCHALEN IM ELASTOPLASTISCHEN BEREICH
Stahlbau , 3, p. 76-80
Tin-Loi, F.;
Pulmano, V. A.
LIMIT LOADS OF CYLINDRICAL SHELLS UNDER
HYDROSTATIC PRESSURE
J. of Structural Engineering, Vol. 117, No. 3,
p. 643-656
Biron, A.
REVIEW OF LOWER-BOUND LIMIT ANALYSIS FOR
PRESSURE VESSEL INTERSECTIONS
IJPVT, August 1977, p. 413-418
Staat, M.; Heitzer,
M.
LIMIT AND SHAKEDOWN ANALYSIS USING A GENERAL
PURPOSE FINITE ELEMENT CODE
Proc. of NAEFEMS Wordl Congress '97,
Vol.1, p. 522-533
Save, M.
ATLAS OF LIMIT LOADS OF METAL PLATES, SHELSS AND
DISKS
Elsevier



12
Annex 1: Bibliography
Page
A1.12
DBA
Design by Analysis
Table A1.3: Theory and/or numerical methods for limit and shakedown analysis (continued)


WRITER TITLE SOURCE
O'Donnell, W.J.;
Porowski, J.
YIELD SURFACES FOR PERFORATED MATERIALS
J. of Applied Mechanics, March 1973, p. 263
f.f.
Stein, E.; Zhang,
G.; Mahnken, R.
SHAKEDOWN ANALYSIS FOR PERFECTLY PLASTIC
AND KINEMATIC HARDENING MATERIALS
E. Stein: "Progress in Computational Analysis of
Inelastic Structures", CISM 1993, p.175-244
Oschatz, A.
MISES-FLIEBEDINGUNGEN FR
ROTATIONSSCHALEN
Technische Mechanik 5 (1984), Heft 1, p. 56 f.f.
Knig, J. A.
ON EXACTNESS OF THE KINEMATICAL APPROACH IN
THE STRUCTURAL SHAKEDOWN AND LIMIT
ANALYSIS
Ingenieur Archiv 52 (1982), p. 421-428
Koiter, W. T. GENERAL THEOREMS FOR ELASTIC-PLASTIC SOLIDS
Progress in Solid Mechanics, Vol. 1, p. 165-
221; Edited by I.N. Sneddon & R. Hill, North
Holland, 1960
Knig; J.A. SHAKEDOWN THEORY OF PLATES
Archiwum Mechaniki Stossowanej 5, 21 (1969),
p. 623-637
Hbel, H.
ELASTIC-PLASTIC CYLINDRICAL SHELL UNDER
AXISYMMETRIC LOADING-ANALYTICAL SOLUTION
IJPV&P, 29, p. 67-81
Polizzotto, C.
A STUDY ON PLASTIC SHAKEDOWN OF
STRUCTURES: PART II - THEOREMS
J. of Applied Mechanics, Vol. 60, p. 324-330
Polizzotto, C.
A STUDY ON PLASTIC SHAKEDOWN OF
STRUCTURES: PART I - BASIC PROPERTIES
J. of Applied Mechanics, Vol. 60, p. 318-323
Ploizzotto C.
ON THE CONDITIONS TO PREVENT PLASTIC
SHAKEDOWN OF STRUCTURES: PART II - THE
PLASTIC SHAKEDOWN LIMIT LOAD
J. of Applied Mechanics, Vol. 60, p. 20-25
Polizzotto C.
ON THE CONDITIONS TO PREVENT PLASTIC
SHAKEDOWN OF STRUCTURES: PART I - THEORY
J. of Applied Mechanics, Vol. 60, p. 15-19
Seshadri M.
LOWER BOUND LIMIT LOADS USING VARIATIONAL
CONCEPTS: THE M-alfa METHOD
Int. J. Ves. & Piping 71, 1997
Ponter, A.R.S.;
Carter, K.F.
LIMIT STATE SOLUTIONS, BASED UPON LINEAR
ELASTIC SOLUTIONS WITH A SPATIALLY VARYING
ELASTIC MODULUS
Computer methods in applied mechanics and
engineering
Ponter, A.R.S.;
Carter, K.F.
SHAKEDOWN STATE SIMULATION TECHNIQUES
BASED ON LINEAR ELASTIC SOLUTIONS
Computer methods in applied mechanics and
engineering
Plancq, D.
ETUDE ELASTIQUE ET ANALYSE LIMITE DES
PIQUAGES ET DES TES
Thesis (1997), Cetim
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.1
DBA
Design by Analysis
We are grateful to CEN for their permission to include the Clause 18, Annexes B and C of the draft
unfired pressure vessel standard prEN 13445-3, which constitute an essential reference for the
anlayses within this manual.
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.2
18 Detailed assessment of fatigue life
18.1 Purpose
18.1.1 This clause presents rules for the detailed fatigue assessment of pressure vessels and their
components which are subjected to repeated fluctuations of stress.
18.1.2 The assessment procedure assumes that the vessel has been designed in accordance with all other
requirements of this standard.
18.1.3 These rules are only applicable to the ferritic and austenitic steels specified in Part 2 of EN 13445.
18.1.4 These rules are not applicable to testing group 4 pressure vessels. For testing group 3 welded joints,
see the special provisions in 18.10.2.1.
18.1.5 This method is not intended for design involving elastic follow-up (See reference [1]).
18.2 Specific definitions
The following definitions are in addition to those in clause 3:
18.2.1 fatigue design curves: Curves given in this clause of
R
against N for welded and unwelded
material, and of
R
/R
m
against N for bolts.
18.2.2 discontinuity: A shape or material change which affects the stress distribution.
18.2.3 gross structural discontinuity: A structural discontinuity which affects the stress or strain
distribution across the entire wall thickness.
18.2.4 local structural discontinuity: A discontinuity which affects the stress or strain distribution locally,
across a fraction of the wall thickness.
18.2.5 nominal stress: The stress which would exist in the absence of a discontinuity.
NOTE 1: Nominal stress is a reference stress which is calculated using elementary theory of
structures. It excludes the effect of structural discontinuities (e.g. welds, openings and thickness
changes). See figure 18-1.
NOTE 2: The use of nominal stress is permitted for some specific weld details for which determination
of the structural stress would be unnecessarily complex. It is also applied to bolts.
NOTE 3: The nominal stress is the stress commonly used to express the results of fatigue tests
performed on laboratory specimens under simple unidirectional axial or bending loading. Hence,
fatigue curves derived from such data include the effect of any notches or other structural
discontinuities (e.g. welds) in the test specimen.
18.2.6 notch stress: The total stress located at the root of a notch, including the non-linear part of the
stress distribution.
NOTE 1: See figure 18-1 for the case where the component is welded, but notch stresses may
similarly be found at local discontinuities in unwelded components.
NOTE 2: Notch stresses are usually calculated using numerical analysis. Alternatively, the nominal or
structural stress is used in conjunction with the effective stress concentration factor, K
eff
.
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.3
18.2.7 equivalent stress: The uniaxial stress which produces the same fatigue damage as the applied
multi-axial stresses.
NOTE: It is calculated using a failure criterion; the Tresca criterion is applied in this clause.
18.2.8 stress on the weld throat: The average stress on the throat thickness in a fillet or partial penetration
weld.
NOTE 1: In the general case of a non-uniformly loaded weld, it is calculated as the maximum load per
unit length of weld divided by the weld throat thickness and it is assumed that none of the load is
carried by bearing between the components joined.
NOTE 2: The stress on the weld throat is used exclusively for assessment of fatigue failure by
cracking through weld metal in fillet or partial penetration welds.
18.2.9 stress range (
R
): The value from maximum to minimum in the cycle (see figure 18-2[withdrawn])
of nominal, principal or equivalent stress, depending on the component and as defined in this clause.
1 Nominal stress; 2 Structural stress; 3 Notch stress;
4 Extrapolation to give structural stress at potential crack initiation site.
Figure 18-1: Distribution of nominal, structural and notch stress at a structural discontinuity
1 One cycle; Stress range
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.4
18.2.10 structural stress: The linearly distributed stress across the section thickness which arises from
applied loads (forces, moments, pressure, etc.) and the corresponding reaction of the particular structural
part.
NOTE 1: Structural stress includes the effects of gross structural discontinuities (e.g. branch
connections, cone/cylinder intersections, vessel/end junctions, thickness change, deviations from
design shape, presence of an attachment). However, it excludes the notch effects of local structural
discontinuities (e.g. weld toe) which give rise to non-linear stress distributions across the section
thickness. See figure 18-1.
NOTE 2: For the purpose of a fatigue assessment, the structural stress shall be evaluated at the
potential crack initiation site.
NOTE 3: Structural stresses may be determined by one of the following methods: numerical analysis
(e.g. finite element analysis (FEA)), strain measurement or the application of stress concentration
factors to nominal stresses obtained analytically. Guidance on the use of numerical analysis is given
in reference [2].
NOTE 4: Under high thermal stresses, the peak stress rather than the linearly distributed stress should
be considered.
18.2.11 weld throat thickness: The minimum thickness in the weld cross-section.
18.3 Specific symbols and abbreviations
The following are in addition to those in clause 4.
C is the constant in equation of fatigue design curves for welded components;
D is the cumulative fatigue damage index;
E is the modulus of elasticity at maximum operating temperature;
F
e
, F
s
are coefficients;
f
b
is the overall correction factor applied to bolts;
f
c
is the compressive stress correction factor;
f
e
is the thickness correction factor in unwelded components;
f
ew
is the thickness correction factor in welded components and bolts;
f
m
is the mean stress correction factor;
f
s
is the surface finish correction factor;
f
t*
is the temperature correction factor;
f
u
is the overall correction factor applied to unwelded components;
f
w
is the overall correction factor applied to welded components;
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.5
g is the depth of groove produced by weld toe grinding;
K
eff
is the effective stress concentration factor given by equation 18.7-1
K
m
is the stress magnification factor due to deviations from design shape;
K
t
is the theoretical elastic stress concentration factor;
k
e
is the plasticity correction factor for stress due to mechanical loading;
k

is the plasticity correction factor for stress due to thermal loading;


M is the mean stress sensitivity factor;
m is the exponent in equations of fatigue design curves for welded components;
N is the allowable number of cycles obtained from the fatigue design curves (suffix i
refers to life under ith stress range);
n is the number of applied stress cycles (suffix i refers to number due to ith
stress range);
R is the mean radius of vessel at point considered;
R
min
is the minimum inside radius of cylindrical vessel, including corrosion allowance;
R
max
is the maximum inside radius of cylindrical vessel, including corrosion allowance;
R
z
is the peak to valley height;
r is the radius of groove produced by weld toe grinding;
S
ij
is the difference between either principal stresses (
i
and
j
) or structural principal stresses
(
struc,i
and
struc,j
) as appropriate;
t
max
is the maximum operating temperature;
t
min
is the minimum operating temperature;
t* is the assumed mean cycle temperature;

T
is the total strain range;
is the maximum principal stress range (suffix i refers to ith stress range);

eq
is the equivalent stress range (suffix i refers to ith stress range);

R
is the stress range obtained from fatigue design curve;

struc
is the structural stress range;

eq,l
is the equivalent stress range corresponding to variation of equivalent linear distribution;

eq,t
is the notch (or total) equivalent stress range;
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.6

eq,nl
is the stress range corresponding to variation of non-linear part of the stress distribution;
is the total deviation from mean circle of shell at seam weld;

1
is the offset of centre-lines of abutting plates;
is the angle between tangents to abutting plates at a seam;
is the direct stress or stress range as indicated (suffix w applies to weld);
( )

eq
op
is the equivalent stress due to operating pressure (for specific use in 18.4.6)

eqmax
is the maximum equivalent stress;

eqmin
is the minimum equivalent stress;

eq
is the mean equivalent stress;

eq, r
is the reduced mean equivalent stress for elastic-plastic conditions;

struc1
is a structural principal stress (1, 2 , 3 apply to the axes) at a given instant;

1
is a principal stress (suffices 1, 2 , 3 apply to the axes) at a given instant;

V1
,
V2
are stress ranges obtained in the example of reservoir cycle counting in 18.9.3;
is the shear stress or stress range as indicated (suffix w applies to weld);
18.4 Limitations
18.4.1 Where a vessel is designed for fatigue, the method of manufacture of all components,
including temporary fixtures and repairs, shall be specified by the manufacturer
18.4.2 There are no restrictions on the use of the fatigue design curves for vessels which operate at sub-
zero temperatures, provided that the material through which a fatigue crack might propagate is shown to be
sufficiently tough to ensure that fracture will not initiate from a fatigue crack.
18.4.3 These rules are only applicable to vessels which operate at temperatures below the creep range of
the material. Thus, the fatigue design curves are applicable up to 380 C for ferritic steels and 500 C for
austenitic stainless steels.
18.4.4 It is a condition of the use of these rules that all regions which are fatigue-critical are accessible for
inspection and non-destructive testing, and that in-service inspection shall be performed at not later than 20
% of the allowable fatigue life.
18.4.5 Corrosive conditions are detrimental to the fatigue lives of steels. Environmentally-assisted fatigue
cracks can occur at lower levels of fluctuating stress than in air and the rate at which they propagate can be
higher. The fatigue strengths specified do not include any allowances for corrosive conditions. Therefore,
where corrosion fatigue is anticipated and effective protection from the corrosive medium cannot be
guaranteed, a factor should be chosen, on the basis of experience or testing, by which the fatigue strengths
given in these rules should be reduced to compensate for the corrosion. If, because of lack of experience, it
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.7
is not certain that the chosen fatigue strengths are low enough, the frequency of inspection should be
increased until there is sufficient experience to justify the factor used.
18.4.6 For water conducting parts made from non-austenitic steels, operating at temperatures exceeding
200 C, conservation of the magnetite protective layer shall be ensured. This will be obtained if the stress
range on the surface in contact with water always fulfils the following conditions:
( ) ( )
eqmax eq
op
2
N/ mm + 200 (18.4-1)
( ) ( )
eqmin eq
op
2
N/ mm 600 (18.4-2)
NOTE: It is assumed that under the operating conditions at which the magnetite layer forms, there is
no stress in that layer.
18.4.7 Where vibration (e.g. due to machinery, pressure pulsing or wind) cannot be removed by
suitable strengthening, support or dampening, it shall be assessed using the method in this clause.
18.5 General
18.5.1 A fatigue assessment shall be made at all locations where there is a risk of fatigue crack
initiation.
NOTE: It is recommended that the fatigue assessment is performed using operating rather than
design loads.
18.5.2 In fatigue, welds behave differently from plain (unwelded) material. Therefore the
assessment procedures for welded and unwelded material are different.
18.5.3 Plain material might contain flush ground weld repairs. The presence of such repairs can
lead to a reduction in the fatigue life of the material. Hence, only material which is certain to be
free from welding shall be assessed as unwelded.
18.5.4 A typical sequence in the design of a vessel for fatigue is shown in table 18-1.
18.5.5 The fatigue life obtained from the appropriate fatigue design curves (for welded
components, unwelded components and bolts) for constant amplitude loading is the allowable
number of cycles.
18.5.6 For calculation of cumulative damage under variable amplitude loading, D is given by:
D
n
N
n
N
n
N
i
i
+ +

1
1
2
1
...... ...(18.5-1)
The following condition shall be met:
D 1 ...(18.5-2)
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.8
Table 18-1: Summary of fatigue assessment process
Task Comment Relevant clause(s)
1. Design vessel for static
loads
Gives layout, details, sizes Main Standard
2. Define fatigue loading Based on operating specification,
secondary effects identified by
manufacturer, etc.
18.5
3. Identify locations of vessel
to be assessed
Structural discontinuities, openings,
joints (welded, bolted), corners,
repairs, etc.
18.5
4. At each location, establish
stress range
a) Calculate stress ranges
b) Apply plasticity correction factors
where relevant
c) Apply overall correction factor
Welded: 18.6, 18.8 and
18.10.6;
Unwelded: 18.7, 18.8 and
18.11.2;
Bolts: 18.7.2,18.12.2.2.
5. At each location, establish
design stress range
spectrum
Perform cycle counting operation 18.9
6. Identify fatigue strength
data
a) Welded material
b) Unwelded material
c) Bolted material
18.10, tables 18-4 & 18-5
18.11
18.12
7. Note relevant implications
and inform rele-
vant manufacturing and
inspection personnel
a) Inspection requirements for
welds
b) Control of or assumptions
about misalignment
c) Acceptance levels for weld
flaws
Tables 18-4 or 18-5
18.10.4
18.10.5
8. Extract allowable fatigue
lives from fatigue design
and perform assessment
a) Welded material
b) Unwelded material
c) Bolts
d) Assessment method
18.10, table 18-7
18.11, table 18-10
18.12
18.5.5, 18.5.6
9. Further action if location
fails assessment
a) Re-assess using more refined
stress analysis
b) Reduce stresses by
increasing thickness
c) Change detail
d) Apply weld toe dressing
(if appropriate)
18.6 (welded), 18.7 (unwelded)
Table 18-4 or 18-5
18.10.2.2
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.9
18.6 Welded material
18.6.1 Stresses
For the assessment of simple attachments and aligned seam welds, principal stresses may be calculated
elastically from the nominal stresses.
In the case of weld metal in directly loaded fillet or partial penetration joints, but not butt joints (e.g. seams),
use is made of the nominal stress range on the throat as shown in 18.6.3.
For all other welded components, structural principal stresses shall be determined. They shall be:
- either calculated using elastic theory from the structural stresses at the potential crack initiation site,
taking account of all membrane, bending and shearing stresses;
- or deduced from strains measured on the vessel and converted to linear-elastic conditions.
Where the structural principal stress is obtained by detailed stress analysis (e.g. FEA) or by measurement, it
is to be determined from the principal stress which acts closest to the normal to the weld by extrapolation
using the procedures detailed in figure 18-3.
NOTE1: In arriving at the structural principal stress, it is necessary to take full account of the structural
discontinuities (e.g. nozzles) and all sources of stress. The latter may result from: global shape
discontinuities such as cylinder to end junctions, changes in thickness and welded-on rings;
deviations from intended shape such as ovality, temperature gradients, peaking and misaligned welds
(note some misalignment is already included in some of the fatigue design curves). Methods in this
clause and in the published literature (see references [3] - [7]) provide estimates of such stresses for
many geometries, or at least enable a conservative assessment to be made.
NOTE 2: For nozzles being assessed using principal stresses directly, three possible stress
concentrations due to structural discontinuities should be considered and estimates of the stresses
made as follows:
a) At the crotch corner, a stress concentration factor may be applied to the nominal hoop stress in
the shell to determine the maximum structural stress which is circumferential with respect to the
nozzle.
b) At the weld toe in the shell, the stresses in the shell acting in all radial directions with respect to
the nozzle should be considered in order to determine the maximum structural stress in the shell.
Stresses in the shell as a result of mechanical loading on the nozzle as well as pressure should be
considered.
c) At the weld toe in the branch, the maximum structural stress range in the branch should be
calculated. Again, the possibility of mechanical as well as pressure loading should be considered.
NOTE 3: Since the maximum range of stress on the weld throat can be expressed as a vector sum,
is the scalar value of the greatest vector difference between different stress conditions during the
cycle.
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.10
Locations of stresses for determination of structural stress by extrapolation to point of stress concentration
(weld toe in this case):
a) low bending stress component, gauge length 0,2e, linear extrapolation;
b) high bending stress component, stiff elastic foundation, gauge length 0,2e, quadratic
extrapolation;
c) gauge length > 0,2e, linear extrapolation
where "gauge length" refers to size of strain gauge or FE mesh.
Figure 18-3: Extrapolation to obtain structural stress from FEA or strain gauge results
18.6.2 Stress range in parent material and butt welds
18.6.2.1 Options
For the assessment of simple attachments and aligned seam welds, the nominal equivalent stress range
(see table 18-4a) and 18-4e)) or the nominal principal stress range (see table 18-5a) and 18-5e)) can be
used. This shall be calculated in the same way as structural stress ranges (see equations 18.6-4, 18.6-5 and
18.6-6) using nominal principal stresses instead of structural principal stresses.
For all other welded components, depending on the calculation method:
- either the principal stress range shall be determined from the range of the structural principal stresses
and used with table 18-5;
- or the equivalent stress range shall be calculated from the range of the equivalent stresses determined
from the structural principal stresses and used with table 18-4.
Tension stresses are considered positive and compression stresses negative. In both cases, an important
aspect is whether, under multiple load actions, the directions of the structural principal stresses remain
constant or not.
Where applicable, the elastically calculated principal or equivalent stress range shall be modified by the
plasticity correction factors given in 18.8.
NOTE: For welded components, the full stress range is used regardless of applied or effective mean
stress. The fatigue design curves incorporate the effect of tensile residual stresses; post-weld heat
treatment is ignored in the fatigue analysis.
18.6.2.2 Equivalent stress range
eq
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.11
18.6.2.2.1 Structural principal stress directions constant
When the structural principal stress directions are constant,
eq
shall be calculated as follows.
The variation with time of the three structural principal stresses shall be established. The variation with time
of the three principal stress differences shall be calculated as follows:
S
12 2

struc1 struc
...(18.6-1)
S
23 2 3

struc struc
...(18.6-2)
S
31

struc3 struc1
...(18.6-3)
Applying Tresca's criterion,
eq
is:
( )

eq
max ; ;
max min max min
S S S S S S
12 12 23 23 31max 31min
...(18.6-4)
NOTE: A typical example is shown in figures 18-4(a) and (b).
eq
is twice the greatest shear stress
range and occurs on one of the three planes of maximum shear.
(a): Typical variation with time of the structural principal stresses
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.12
(b): Variation with time of the principal stress differences and the resulting
eq
(c): For unwelded components, the variation with time of the difference between the structural
principal stresses which determine
eq
(i.e.
struc1
and
struc3
in this case) and
the resulting mean
eq
Figure 18-4: Typical example of stress variation when the principal stress directions remain constant
18.6.2.2.2 Structural principal stress directions change
When the structural principal stress directions change,
eq
shall be calculated as follows.
Determine the variation with time of the six stress components (three direct and three shear) with
reference to some convenient fixed axes. For each stress component, calculate the maximum
variation. The structural principal stress ranges, S
12
, S
23
and S
31
, are then calculated from the
resulting stress variations.
eq
is determined as before from equation 18.6-4.
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.13
18.6.2.3 Principal stress range
18.6.2.3.1 Application
If the potential fatigue crack initiation site is at the weld toe or on the surface of the weld, the structural stress
range in the material adjacent to the weld is required for the fatigue assessment. Since
struc3
= 0, use is
made of the two structural principal stresses
struc1
and
struc2
acting essentially (i.e. within 45) parallel and
normal to the direction of the weld respectively, on each material surface.
18.6.2.3.2 Structural principal stress directions constant
Where the directions of the structural principal stresses remain fixed, is determined as follows.


struc1 struc1max struc1min
= - ...(18.6-5)


struc2 struc2ma struc2min
= -
x
...(18.6-6)
NOTE: Both principal stress ranges may need to be considered, depending on their directions.
18.6.2.3.3 Structural principal stress directions change
When the structural principal stress directions change during cycling between two load conditions, shall
be calculated as follows.
Determine the three stress components (two direct and one shear) at each load condition with reference to
some convenient fixed axes. For each stress component, calculate the maximum difference between the
stresses. Calculate the principal stresses from the resulting stress differences.
NOTE: Both principal stress ranges may need to be considered, depending on their directions.
Where cycling is of such a complex nature that it is not clear which two load conditions will result in the
greatest value of , they shall be established by carrying out the above procedure for all pairs of load
conditions. Alternatively, it is conservative to assume that is the difference between the algebraically
greatest and smallest principal stresses occurring during the whole loading cycle regardless of their
directions, and to assume the lowest classification for the detail (see Table 18.5).
18.6.3 Stress range in directly loaded fillet or partial penetration welds
is the maximum range of stress on the weld throat.
Where stress cycling is due to the application and removal of a single load,
( )
=
w w
2 2
1 2
+
/
...(18.6-7)
where
w
is the normal stress range on the weld throat and
w
is the shear stress range on the weld throat.
Where stress cycling is due to more than one load source, but the directions of the stresses remain fixed,
is determined from the maximum range of the load per unit length of the weld.
Where the direction of the stress vector on the weld throat changes during the cycle between two extreme
load conditions, is the magnitude of the vector difference between the two stress vectors.
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.14
Where cycling is of such a complex nature that it is not clear which two load conditions will result in the
greatest value of , then the vector difference should be found for all pairs of extreme load conditions.
Alternatively, it is conservative to assume:


= ( - ) + ( - ) + ( - )
/
[ ]
max min max min
2 2 2 1 2
1max 1min 2 2
...(18.6-8)
where
1
and
2
are the two components of shear stress on the weld throat.
Directly loaded fillet and partial penetration welds do not need to be assessed if the effective weld throat
thickness is such that the stress range in the weld does not exceed 0,8 times the stress range in the plate.
Welds not subject to direct loading (e.g. attachments) do not need to be assessed if the effective weld throat
is at least 0,7 times the thickness of the thinner part joined by the weld.
18.7 Unwelded components and bolts
18.7.1 Unwelded components
18.7.1.1 Stresses
For unwelded components, equivalent effective notch stresses only shall be determined. They shall be
calculated using structural principal stresses which incorporate the full effect of gross and local structural
discontinuities. The equivalent effective notch stress can be obtained by the use of K
eff
given by:
( )
K
K
K
eff
t
t
struc
D
+

+
1
15 1
1 0 5
,
,

(18-7.1)
where
D
=
R
for N 2.10
6
cycles for unwelded material.
The effective stress concentration factor K
eff
is used in subclause 18.11.2 for the calculation of the overall
correction factor.
If the notch stresses are calculated directly by analysis (e.g. FEA) or determined experimentally (e.g strain
gauges), the structural and peak stresses should be separated (as described in Annex C) to give the total
stress as follows:

total struc peak
+ (18.7-2)
The theoretical stress concentration factor shall be calculated as follows:
struc
peak
struc
total
t
1

+ K (18.7-3)
If the total stress is calculated directly by analysis (e.g. FEA) the model shall include the notch in sufficiently
fine details.
If the equivalent notch stresses are determined directly by analysis (e.g. FEA) the model must include the
notch in sufficiently fine detail. If they are determined experimentally (eg. strain gauges), measurements
must be made within the notch, or sufficiently close to enable the notch stress to be established by
extrapolation (see reference [2]). Strains shall be converted to stresses assuming linear elastic conditions
and, for this case, no plasticity correction is required.
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.15
The equivalent stress range
eq
and equivalent mean stress
eq
shall be determined. Two methods are
given for this depending on whether, under multiple load actions, the directions of the structural principal
stresses remain constant or not. Tension stresses are considered positive and compression stresses
negative.A
18.7.1.2 Principal stress directions constant
When the principal stress directions remain constant,
eq
shall be determined per 18.6.2.2.1 and equation
(18.6-4).
The corresponding mean equivalent stress
eq
Bis the maximum value of the average of the two structural
principal stresses,
struc,i
and
struc,j
, which produced
eq
. Thus:
( ) ( )
[ ] eq
struc,i struc, j struc,i struc, j

=
1
2
+ + +
max min
...(18.7-4)
NOTE: A typical example is shown in figure18-4(c).
eq
is twice the mean value of the direct stress,
averaged over time, normal to the plane of maximum shear stress range.
18.7.1.3 Principal stress directions change
When the principal stress directions change,
eq
shall be calculated as detailed in 18.6.2.2.2.
eq
and

eq
Care determined from equations (18.6-4) and (18.7-4) respectively.
18.7.2 Bolts
For bolts, is the maximum nominal stress range arising from direct tensile and bending loads on the core
cross-sectional area, determined on the basis of the minor diameter. For pre-loaded bolts, account may be
taken of the level of pre-load, with based on the full applied load rather than the fluctuating portion of that
load. For a bolt pre-tensioned to its minimum proof load, may be based on 20 % of the maximum applied
load.
NOTE: The fatigue design curve for bolts takes account, for any form of thread, of the stress
concentrations at the thread root.
18.8 Plasticity correction factors
18.8.1 Elastic-plastic conditions
For any component, if the calculated pseudo-elastic structural stress range for both welded joints and
unwelded parts exceeds twice the yield strength of the material under consideration, i.e. if
eq,l
p0,2/ t*
> 2R ,
see note, it shall be multiplied by a plasticity correction factor. The correction factor for mechanical loading is
k
e
and for thermal loading it is k

.
NOTE: This applies to ferritic steels; for austenitic steels, use R
p1,0/t*
.
18.8.1.1 Mechanical loading
For mechanical loading, the corrected stress range is k
e
, where:
DBA
Design by Analysis
Annex 2: Clause 18
Draft CEN prEN 13445-3
Page
A2.16
k A
R
e
eq,l
p0,2/ t
+

_
,

1
2
1
0

*
...(18.8-1)
where: A
0
= 0,5 for ferritic steels with 800 1000 R
m
2
N/ mm ( ) ;
= 0,5 for ferritic steels with R
m
2
N/ mm 500( ) and for all austenitic steels (see .........note in
18.8.1);
=
( )
0 4
500
3000
, +
R
m
for ferritic steels with 500 800 R
m
2
N/ mm ( ) .
The procedure for determining the mean equivalent stress to allow for elastic-plastic conditions is
shown in figure 18-5 and applied in 18.11.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.17
DBA
Design by Analysis
(*) For unwelded parts, or values are notch stresses or stress ranges
(**)This applies to ferritic steels; for austenitic steels, use R
p1,0/t*
.
Figure 18-5: Modifications to mean equivalent stress to allow for elastic-plastic conditions due to
mechanical loadings
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.18
DBA
Design by Analysis
18.8.1.2 Thermal loading
In the case of a thermal stress distribution which is non-linear through the material thickness, both the non-
linear and the equivalent linear stress distributions shall be determined for each stress component. Using the
linearised stress range
eq,l
, k

shall be calculated by:

k
=
,
, +
,
R
0 7
0 5
0 4

eq,l
p0,2/ t*

_
,

...(18.8-3)
The corrected stress range shall be either k

.

eq,l
for welded joints or k

.

eq,t
for unwelded zones.
18.8.1.3 Elastic-plastic analysis
If the total strain range
T
(elastic plus plastic) due to any source of loading is known from theoretical or
experimental stress analysis, correction for plasticity is not required and
= E .
T
...(18.8-4)
18.9 Fatigue action
18.9.1 Loading
18.9.1.1 All sources of fluctuating load acting on the vessel or part shall be identified.
NOTE: Such loads are: fluctuations of pressure; variations in contents; temperature transients;
restrictions of expansion or contraction during temperature variations; forced vibrations; and variations
in external loads. Account shall be taken of all operational and environmental effects defined in the
purchase specification.
18.9.2 Simplified cycle counting method
18.9.2.1 Loads shall be grouped into specific loading events. Loading events shall be independent of each
other and shall be considered separately.
18.9.2.2 A loading specification shall be prepared stating for each loading event the stress range (calculated
from 18.5, 18.6, 18.7 and 18.8 as appropriate for the component and load) and number of cycles for each
load.
As shown in figure 18-6 and table 18-3, the stress ranges shall be plotted or tabulated against number of
cycles. The loading with the lowest number of cycles shall be plotted or tabulated at the top and the cycles
summed as shown.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.19
DBA
Design by Analysis

n
0
n
4
n
3
n
2
n
1
c
4
c
3
c
2
c
1
combined stress range
n number of applied cycles
c
4
cycles of
4
+
3
+
2
+
1
c
3
cycles of
3
+
2
+
1
c
2
cycles of
2
+
1
c
1
cycles of
1
Figure 18--6 Simplified counting method
NOTE: An example is shown in table 18-3
Table 18-3: Example of determination of stress cycles using simplified cycle counting method
Individual loadings Loading events
Loading Stress range No of cycles Example Number Stress range No of cycles
4
4
n
4
Full pressure
range
A
4
+
3
+

2
+
1
c
4
= n
4
3
3
n
3
Temperature
difference
B
3
+
2
+

1
c
3
= n
3
- n
4
2
2
n
2
Pressure
fluctuation
C
2
+
1
c
2
= n
2
-n
3
-
n
4
1
1
n
1
Mechanical
loading
D
1
c
1
= n
1
-n
2
-n
3
-n
4
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.20
DBA
Design by Analysis
18.9.3 Reservoir cycle counting method
18.9.3.1 As an alternative to the simplified counting method given in 18.9.2, the more accurate reservoir
cycle counting procedure may used.
18.9.3.2 For each loading event, derive the variation with time of either the structural principal stress or the
equivalent stress as specified in 18.6 or 18.7 for the component.
18.9.3.3 Plot the peak and trough values for two occurrences of the event as shown in figure 18-7.
18.9.3.4 Mark the highest peak stress in each cycle and join the two peaks together with a straight line. If
there are two or more equal highest peaks in a cycle, mark only the first such peak in the occurrence.
18.9.3.5 Join the two marked points and consider only that part of the plot which falls below this line, like the
section of a full reservoir.
18.9.3.6 Drain the reservoir from the lowest point leaving the water that cannot escape. If there are two or
more equal lowest points, drainage may be from any one of them.
18.9.3.7 List one cycle having a stress range,
V1
, equal to the vertical height of water drained.
18.9.3.8 Repeat the step in 18.9.3.7 successively with each remaining body of water until the reservoir is
emptied, listing one cycle at each draining operation.
18.9.3.9 List all the individual stress ranges in descending order of magnitude,
V1
,
V2
,
V3
,
V4
etc. Where
two or more cycles of equal stress range occur, record them separately. This provides the design stress
range spectrum.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.21
DBA
Design by Analysis
Figure18-7: Reservoir cycle counting method
18.10 Fatigue strength of welded components
18.10.1 Classification of weld details
18.10.1.1 Use of the tables
Welds shall be classified to tables 18-4 and 18-5 according to whether the stress range is calculated from
equivalent or principal stresses. The sketches in tables 18-4 and 18-5 indicate the potential mode of
cracking corresponding to the position and direction of the fluctuating stress shown.
All deviations from the ideal shape (misalignment, peaking, ovality etc.) shall be included in the
determination of the stresses.
NOTE1: A detail may appear several times in the tables because of the different modes in which it
might fail.
In general, fatigue strength depends on: the direction of the fluctuating stress relative to the weld
detail; the locations of possible fatigue crack initiation at the detail; the geometrical arrangement and
proportions of the detail; and the methods of manufacture and inspection.
NOTE2: The fatigue life of a vessel or part of a vessel may be governed by one particular detail.
Therefore, the classes of other details which experience the same fatigue loading need be no higher.
For example, the potentially high class attainable from perfectly-aligned seams may not be required if
overall fatigue life is governed by fillet welds.
18.10.1.2 Classification of weld details to be assessed using equivalent stress range
Weld details and their corresponding classes for use in assessments based on equivalent stress range are
given in table 18-4. The classification refers either to fatigue cracking in the parent metal from the weld toe
or end, which shall be assessed using
eq
in the parent metal adjacent to the potential crack initiation site,
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.22
DBA
Design by Analysis
or to fatigue cracking in the weld itself from the root or surface, which shall be assessed using
eq
in the
weld.
Since
eq
has no direction, the class indicated in table 18-4 refers to the least favourable stressing direction
for the particular weld detail and mode of fatigue cracking shown.
Table 18.4. Class of weld details for use with structural equivalent stress range
Table 18.4(a) Seam welds
Detail
No.
Joint type Sketch of detail Class Comments
Testing
group 1
or 2
Testing
group 3
1.1 Full penetration butt weld
flush ground, including
weld repairs
Fatigue cracks usually initiate at
weld flaws
90 71 Weld proved free from surface-
breaking flaws and significant
sub-surface flaws (see annex
18xx) by non-destructive testing.
f
ew
=1
1.2 Full penetration butt weld
made from both sides or
from one side on to
consumable insert or
temporary non-fusible
backing
1:3
80 63 Weld proved free from
significant flaws (see annex
18xx) by non-destructive testing
and, for welds made from one
side, full penetration.
In case of misalignment, see
clause 18.10.4.
1.3
1:3
e
e
80
63
63
40
Weld proved free from
significant flaws by non-
destructive testing (see annex
18xx).
Effect of centre-line offset
included in calculated stress
Class includes effect of centre-
line offset of e/10 and therefore
its effect is neglected when
calculating the structural stress.
For other cases of misalignment,
see detail 1.2.
1.4

80
71
63
56
Weld proved free from
significant flaws (see annex
18xx) by non-destructive testing
30
> 30
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.23
DBA
Design by Analysis
Table 18.4(a) Seam welds cont'd...
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
or 2
Testing
1.5 Full penetration butt
side without backing
63
40
40
If full penetration can be
assured.
If inside cannot be visually
inspected and full penetration
cannot be assured.
In all cases
In case of misalignment, see
clause 18.10.4.
1.6 Full penetration butt
welds made from one
side onto permanent
backing.
56
40
40
Circumferential seams only (see
5.7) Minimum throat = shell
thickness.
Weld root pass inspected to
ensure full fusion to backing.
Single pass weld
In all cases
1.7 Joggle joint
56
40
40
Circumferential seams only (see
5.7) Minimum throat = shell
thickness.
Weld root pass inspected to
ensure full fusion to backing.
Single pass weld
In all cases
Cont'd...
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.24
DBA
Design by Analysis
Table 18.4(b) Shell to head or tubesheet
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
group 1
or 2
Testing
group 3
2.1 Welded-on head
71
80
32
63
63
40
63
63
32
63
63
40
Head plate must have
adequate through-thickness
properties to resist lamellar
tearing.
Full penetration welds made
from both sides:
- as-welded
- weld toes dressed (see
18.10.2.2)
Partial penetration welds
made from both sides:
- based on stress range on
weld throat
- weld throat 0,8 x shell
thickness
Full penetration welds made
from one side without back-up
weld:
- if the inside weld can be
visually inspected and is
proved to be free from overlap
or root concavity.
- if the inside cannot be
visually inspected and full
penetration cannot be
assured.
- in all cases
2.2 Welded-on head with
relief groove
80
63
40
63
40
Weld proved free from
significant flaws (see annex
18XX) by NDT.
Full penetration welds: made
from both sides, or from one
side with the root pass ground
flush.
Made from one side:
- if the inside weld can be
visually inspected and is
proved to be free from weld
overlap and root concavity.
- if the inside cannot be
visually inspected.
- in all cases.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.25
DBA
Design by Analysis
18.4(b) Shell to head or tubesheet cont'd ...
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
group 1
or 2
Testing
group 3
2.3 Set-in head (a)
(b)
(c)
71
80
32
63
63
40
63
63
32
63
40
Full or partial penetration welds
made from both sides.
Refers to fatigue cracking from
weld toe in shell:
- as-welded;
- weld toes dressed (see
18.10.2.2).
Partial penetration welds made
from both sides:
- refers to fatigue cracking in
weld, based on weld throat
stress range.
- weld throat 0,8 x head
thickness.
Full penetration weld made
from one side without back-up
weld:
- if the inside weld can be
visually inspected and is
proved to be free from overlap
or root concavity.
- if the inside cannot be visually
inspected.
- in all cases.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.26
DBA
Design by Analysis
Table 18.4(c) Branch connections
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
group 1
or 2
Testing
group 3
3.1 Crotch corner
Crack radiates from corner
into piece, sketches show
plane of crack
100 100 Assessment by the method
for unwelded parts is the
normal approach. However,
simplified assessment using
class 100 according to
clause 18.11.2.2 is allowed.
few = 1.
3.2 Weld toe in shell
71
80
63
71
63
63
63
Full penetration welds:
- as welded
- weld toes dressed (see
18.10.2.2)
- in all cases
Partial penetration welds:
- weld throat 0,8 x thinner
thickness of connecting
walls, as welded
- weld toes dressed (see
18.10.2.2)
3.3 Stressed weld metal 32 32 Fillet and partial penetration
welds
3.4 Weld toe in branch 71
80
63
As-welded
Weld toes dressed (see
18.10.2.2).
In all cases
en = branch thickness in
equation 18.10-6
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.27
DBA
Design by Analysis
Table 18.4(d) Jackets
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
group 1
or 2
Testing
group 3
4.1 Jacket connection
weld with shaped
sealer ring
63
40
71
40
56
Full penetration weld to be
proved free from significant
flaws (see annex 18xx) by
non-destructive testing
Welded from one side:
- multi-pass weld with root
pass inspected to ensure full
fusion;
- single pass weld.
- in all cases.
Welded from both sides or
from one side with back-up
weld.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.28
DBA
Design by Analysis
Table 18.4(e) Attachments
Detail
No.
Joint type Sketch of detail Class for use with: Comments
Structural
equivalent
stress
Nominal
equivalent
stress
Testing
group 1,
2, 3
Testing
group 1,
2, 3
5.1 Attachment of any
shape with an edge
fillet or bevel butt-
welded to the surface
of a stressed member,
with welds continuous
around the ends or not
71
71
56
50
For details with welds
continuous around
ends, one class
increase if weld toes
dressed (see
18.10.2.2)
L 160mm, t 55mm
L > 160mm
5.2 Attachments of any
shape with surface in
contact with stressed
member, with welds
continuous around
ends or not
71
71
71
56
50
45
For details with welds
continuous around
ends, one class
increase if weld toes
dressed (see
18.10.2.2)
L 160mm, W
55mm
L > 160mm, W
55mm
L > 160mm, W
55mm
5.3 Continuous stiffener
71
71
56
50
For full penetration
welds, one class
increase if weld toes
dressed (see
18.10.2.2).
t 55mm
t > 55mm
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.29
DBA
Design by Analysis
Table 18.4(f) Supports
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
group 1
or 2
Testing
group 3
6.1 Support on either
horizontal or vertical
vessel
71
80
71
80
As-welded.
Weld toe in shell dressed (see
18.10.2.2)
6.2 Trunnion support 71
80
71
80
As-welded.
Weld toe in shell dressed (see
18.10.2.2)
6.3 Saddle support 71
80
71
80
As-welded.
Weld toe in shell dressed (see
18.10.2.2)
6.4 Skirt support
71
80
56
71
80
56
Welded from both sides:
as-welded;
weld toe in shell dressed
(see 18.10.2.2).
Welded from one side
6.5 Leg support (with or
without reinforcing
pad) with fillet weld
to vessel continuous
all around.
71 71
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.30
DBA
Design by Analysis
Table 18.4(g) Flanges and pads
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
group 1
or 2
Testing
group 3
7.1 Full penetration butt
welded neck flange or
compensation flange
with welding lug.
80
63
40
63
40
Weld proved free from
surface-breaking and
significant sub-surface flaws
(see annex 18xx) by non-
destructive testing.
Weld made from both sides
or from one side with back-up
weld or onto consumable
insert or temporary backing.
Weld made from one side:
- if full penetration can be
assured
- if the inside cannot be
visually inspected.
- in all cases.
7.2 Welded flange
71
80
63
32
63
63
63
32
Full penetration welds:
- as-welded;
- weld toe dressed (see
18.10.2.2)
Partial penetration welds:
- weld throat 0,8 x shell
thickness;
- weld throat < 0,8 x shell
thickness.
7.3 Set-in flange or pad
71
80
63
32
63
63
63
32
Full penetration weld:
- as-welded;
- weld toe dressed (see
18.10.2.2).
Fillet welded on both sides:
- weld throat 0,8 x shell
thickness.
- weld throat < 0,8 x shell
thickness.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.31
DBA
Design by Analysis
Table 18.4(g) Flanges and pads concluded
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
group 1
or 2
Testing
group 3
7.4 Set-in flange or pad,
welded from both sides
63
32
63
32
weld throat 0,8 x shell
thickness.
weld throat < 0,8 x shell
thickness.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.32
DBA
Design by Analysis
Table 18.5. Class of weld details for use with principal stress range
Table 18.5(a) Seam welds
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
group 1
or 2
Testing
group 3
1.1 Full penetration butt
weld flush ground,
including weld repairs
Fatigue cracks usually initiate
at weld flaws
90
90
71
71
Weld proved free from
surface-breaking flaws and
significant sub-surface flaws
(see annex 18xx) by non-
destructive testing.
few = 1.
1.2 Full penetration butt
weld made from both
sides or from one side
on to consumable
insert or temporary
non-fusible backing
80
80
80
63
63
71
Weld proved free from
significant flaws (see annex
18xx) by non-destructive
testing. In case of
misalignment, see clause
18.10.4.1.
1.3
e
63
80
80
40
63
63
Class includes effect of
centre-line offset of e/10,
due to thickness change.
Effect of misalignment to
be included in calculated
stress.
For other cases of
misalignment, see detail
1.2.
1.4 80
71
80
63
56
71
Weld proved free from
significant flaws (see annex
18xx) by non-destructive
testing.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.33
DBA
Design by Analysis
Table 18.5(a) Seam welds cont'd..
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
group 1
or 2
Testing
group 3
1.5 Full penetration butt
welds made from one
side without backing 80
63
40
71
40
Weld proved to be full
penetration and free from
significant flaws (see annex
18xx) by non-destructive
testing.
If full penetration can be
assured.
If inside cannot be visually
inspected.
In case of misalignment,
see clause 18.10.4.
1.6 Full penetration butt
welds made from one
side onto permanent
backing
(a)
63
56
40
63
40
40
Circumferential seams only
(see clause 5.7).
Backing strip to be
continuous and, if attached
by welding, tack welds to be
ground out or buried in main
butt weld, or continuous
fillet welds are permitted.
Minimum throat = shell
thickness. Weld root pass
shall be inspected to ensure
full fusion to backing.
Single pass weld.
(b)
63 63
Circumferential seams only
(see clause 5.7).
Backing strip attached with
discontinuous fillet weld.
1
ew
f
1.7 Joggle joint
63
56
40
63
40
40
Circumferential seams only
(see clause 5.7).
Minimum throat = shell
thickness.
1
ew
f
Weld root pass shall be
inspected to ensure full
fusion.
Single pass weld.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.34
DBA
Design by Analysis
Table 18.5(b) Shell to head or tubesheet
Detail
No
Joint type For principal stresses acting essentially normal to the weld
Sketch of detail Class Comments
Testing
group 1
or 2
Testing
group 3
2.1 Welded-on head
71
80
63
32
63
40
63
63
63
32
40
40
Head plate must have
adequate through-thickness
properties to resist lamellar
tearing.
Full penetration welds made
from both sides:
- as-welded;
- weld toes dressed (see
18.10.2.2).
Partial penetration welds
made from both sides:
- refers to fatigue cracking in
shell from weld toe
- refers to fatigue cracking in
weld, based on stress range
on weld throat
Full penetration welds made
from one side without back-
up weld:
- if the inside weld can be
visually inspected and is
proved free from weld
overlap and root concavity.
- if the inside cannot be
visually inspected.
2.2 Welded-on head with
relief groove
80
63
40
63
40
Full penetration welds
Made from one side with the
inside weld ground flush
Made from one side:
- if the inside weld can be
visually inspected and is
proved free from weld
overlap and root concavity.
- if the inside cannot be
visually inspected.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.35
DBA
Design by Analysis
Table 18.5(b) Shell to head or tubesheet cont'd..
Detail
No
Joint type For principal stresses acting essentially normal to the weld
Sketch of detail Class Comments
Testing
group 1
or 2
Testing
group 3
2.3 Set-in head (a)
(b)
(c)
71
80
32
71
63
56
63
63
32
71
63
40
Full penetration weld made
from both sides: refers to
fatigue cracking from weld toe
in shell:
- as-welded;
- weld toes dressed (see
18.10.2.2).
Partial penetration welds
made from both sides:
- refers to fatigue cracking in
weld, based on weld throat
stress range
- refers to fatigue failure in
shell.
- refers to fatigue failure in
head.
Full penetration weld made
from one side
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.36
DBA
Design by Analysis
Table 18.5(c) Branch connections
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
group 1
or 2
Testing
group 3
3.1 Crotch corner
Crack radiates from corner.
Sketches show plane of crack.
100 100 Assessment by the
method for unwelded
parts based on
equivalent stress is the
normal approach.
However, simplified
assessment, using class
100, according to clause
18.11.2.2, still based on
equivalent stress, is
allowed
few = 1.
3.2 Weld toe in shell
71
80
63
63
71
63
Full penetration welds:
- as-welded;
- weld toes dressed (see
18.10.2.2)
Partial penetration welds
3.3 Stressed weld metal Continuous weld stressed
along its length
Weld metal stressed normal to
its length
71
71
32
71
71
32
Based on stress range
parallel to weld on weld
cross-section
few = 1.
Full penetration weld
Partial penetration weld
Based on stress range
on weld throat. few = 1.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.37
DBA
Design by Analysis
Table 18.5(c) Branch connections cont'd..
Detail
No.
Joint type Sketch of detail Class
Comments
Testing
group 1
or 2
Testing
group 3
3.4 Weld toe in branch 71
80
63
71
As-welded;
Weld toes dressed (see
18.11.2.2)
e
n
= branch thickness in
Eq. 18.10-6
Table 18.5(d) Jackets
Detail
No
Joint type For principal stresses acting essentially normal to the weld
Sketch of detail Class Comments
Testing
group 1
or 2
Testing
group 3
4.1 Jacket connection
weld with shaped
sealer ring
63
40
71
40
56
Full penetration weld to be
proved free from
significant flaws (see
annex 18xx) by non-
destructive testing
Welded from one side:
- multi-pass weld with root
pass inspected to ensure
full fusion;
- single pass weld.
- in all cases
Welded from both sides or
from one side with back-up
weld.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.38
DBA
Design by Analysis
Table 18.5(e) Attachments
Detail
No.
Joint type Sketch of detail Class for use with: Comments
Structural
equivalent
stress
range
Nominal
equivalen
t stress
range
Testing
group 1, 2,
3
Testing
group 1, 2,
3
5.1 Attachment of any shape
with an edge fillet or
bevel - butt welded to the
surface of a stressed
member, with welds
continuous around the
ends or not
Stresses acting essentially
parallel to weld:
71
71
56
50
For details with welds
continuous around ends,
one class increase if
weld toes dressed (see
18.10.2.2)
L 160mm
L > 160mm
f
ew
= 1.
Stresses acting essentially normal
to weld:
71
71
56
50
One class increase if
weld toes dressed (see
18.10.2.2)
t 55mm
t > 55mm
f
ew
= 1.
5.2 Attachments of any shape
with surface in contact
with stressed member,
with welds continuous
around ends or not
71
71
71
56
50
45
For details with welds
continuous around ends,
one class increase if
weld toes dressed (see
18.10.2.2)
L 160mm, W 55mm
L > 160mm, W 55mm
L > 160mm, W > 55mm
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.39
DBA
Design by Analysis
Table 18.5(e) Attachments cont'd..
Detail
No.
Joint type Sketch of detail Class for use with: Comments
Structural
equivalent
stress
range
Nominal
equivalent
stress
range
Testing
group 1
or 2
Testing
group 3
Testing
group 1
or 2
Testing
group 3
5.3 Continuous
stiffener
Stresses acting essentially
parallel to weld:
Stresses acting essentially
normal to weld
80
71
71
71
71
71
71
71
80
71
56
50
71
71
56
50
Based on stress
range parallel to
weld in stiffener.
f
ew
= 1.
Full penetration
weld.
Partial penetration
weld.
For full
penetration welds,
one class increase
if weld toes
dressed (see
18.10.2.2).
t 55mm
t> 55mm
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.40
DBA
Design by Analysis
Table 18.5(f) Supports
Detail
No
Joint type For principal stresses acting essentially normal to the weld
Sketch of detail Class Comments
Testing
group 1
or 2
Testing
group 3
6.1 Support on either
horizontal or vertical
vessel
71
80
71
80
As-welded;
Weld toe in shell dressed (see
18.10.2.2)
6.2 Trunnion support 71
80
71
80
As-welded;
Weld toe in shell dressed (see
18.10.2.2)
6.3 Saddle support 71
80
71
80
As-welded;
Weld toe in shell dressed (see
18.10.2.2)
6.4 Skirt support
71
80
56
71
80
56
Welded from both sides:
As-welded;
Weld toe in shell dressed (see
18.10.2.2).
Welded from one side.
6.5 Leg support (with or
without reinforcing pad)
with fillet weld to vessel
continuous all around.
a
b
a) 80
b) 71
a) 80
b) 71
Refers to fatigue cracking in the
shell.
Refers to fatigue cracking in the
leg.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.41
DBA
Design by Analysis
Table 18.5(g) Flanges and pads
Detail
No
Joint type For principal stresses acting essentially normal to the weld
Sketch of detail Class Comments
Testing
group 1
or 2
Testing
group 3
7.1 Full penetration butt
welded neck flange or
compensation flange
with welding lug.
80
63
40
63
40
Weld proved free from surface-
breaking and significant sub-
surface flaws (see annex 18xx)
by non-destructive testing.
Weld made from both sides or
from one side with back-up weld
or onto consumable insert or
temporary backing.
Weld made from one side:
- if full penetration can be
assured;
- if the inside cannot be visually
inspected.
7.2 Welded flange
a
a
b
a
a) 71
a) 80
a) 63
b) 32
a) 63
a) 63
a) 63
b) 32
Full penetration welds:
as-welded;
weld toes dressed (see
18.10.2.2.
Partial penetration welds:
- refers to fatigue cracking from
weld toe;
- refers to fatigue cracking in
weld, based on stress range on
weld throat.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.42
DBA
Design by Analysis
Table 18.5(g) Flanges and pads contd
Detail
No
Joint type For principal stresses
acting essentially
normal to the weld
Sketch of detail Class Comments
Testing
group 1
or 2
Testing
group 3
7.3 Set in flange or pad
71
80
63
32
63
63
63
32
Full penetration weld:
- as-welded
- weld toes dressed (see
18.10.2.2).
Fillet welded from both sides:
- refers to fatigue cracking
from weld toe
- refers to fatigue cracking in
weld, based on stress range
on weld throat .
7.4 Set-in flange or pad,
welded from both
sides a
b
a) 63
a) 32
b) 71
a) 63
a) 32
b) 71
Refers to fatigue cracking
from weld toe
Refers to fatigue cracking in
weld, based on stress range
on weld throat .
Based on hoop stress in shell
at weld root. few = 1.
18.10.1.3 Classification of weld details to be assessed using principal stress range
Weld details and their corresponding classes for use in assessments based on principal stress range are
given in table 18-5.
The fatigue strengths of weld details for which the relevant potential failure mode is by fatigue cracking from
the weld toe or weld surface are expressed in terms of the principal stress range on the parent metal surface
adjacent to the crack initiation site (see 18.6.2.3.1).
Short or discontinuous welds, where the relevant potential failure is by fatigue cracking from the weld end or
weld toe into the parent metal, are assessed on the basis of the maximum principal stress range, , and
classified on the basis that the weld is orientated in the least favourable direction with respect to .
Continuous welds (e.g. seams, ring stiffener welds) may be treated differently if the maximum principal stress
range acts in the direction which is within 45 of the direction of the weld. Then, the weld can be classified as
being parallel to the direction of loading with respect to the maximum principal stress range and normal to
the direction of loading with respect to the minimum principal stress range.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.43
DBA
Design by Analysis
18.10.1.4 Exclusions
The classification tables do not include any bolts which are welded. The assessment method in this clause is
not applicable to such bolts.
18.10.2 Change of classification
18.10.2.1 Welds in testing group 3
Welds in testing group 3 shall be assessed according to tables 18-4 and 18-5.
18.10.2.2 Weld toe dressing
Fatigue cracks readily initiate at weld toes on stressed members partly because of the stress concentration
resulting from the weld shape but chiefly because of the presence of inherent flaws. The fatigue lives of
welds which might fail from the toe can be increased by locally machining and/or grinding the toe to reduce
the stress concentration and remove the inherent flaws.
The classification of fillet welds (including full penetration welds with reinforcing fillets) may, where indicated
in tables 18-4 and 18-5, be raised when dressing of the toe is carried out according to the following
procedure. Tables 18-4 and 18-5 include the revised class.
Figure 18-7: Weld toe dressing
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.44
DBA
Design by Analysis
The weld toe is machined using a rotating conical tungsten-carbide machining burr. In order to ensure that
weld toe flaws are removed, the required depth of machining is 0,5 mm below any undercut (see figure 18-
7). The area should be inspected using dye penetrant or magnetic particle. Such inspection is facilitated if
the machined toe is ground using emery bands, a measure which also improves fatigue life. The resulting
profile should produce a smooth transition from the plate surface to the weld, as shown in figure 18-8, with all
machining marks lying transverse to the weld toe.
Toe dressing only affects the fatigue strength of a welded joint as regards failure from the weld toe. The
possibility of fatigue crack initiation from other features of the weld (e.g. weld root in fillet welds) should not
be overlooked.
Weld toe dressing cannot be assumed to be effective in the presence of any corrosive environment which
can cause pitting in the dressed region.
18.10.2.3 Dressing of seam welds
Dressing or flush grinding of the seam welds justifies an upgrade from Class 80 to Class 90. A fatigue
strength higher than Class 90 cannot be justified because of the possible presence of weld flaws which are
too small for reliable detection by non-destructive inspection methods but are of sufficient size to reduce the
fatigue strength of the joint.
The detrimental effect of misalignment can, to some extent, be alleviated by weld toe dressing (see
18.10.2.2).
Previously buried flaws revealed by dressing, which could reduce the fatigue strength of the joint, should be
assessed (see 18.10.5).
18.10.3 Unclassified Details
Details not fully covered in tables 18-4 and 18-5 shall be treated as Class 32 unless superior resistance to
fatigue is proved by special tests or reference to relevant fatigue test results. To justify a particular design

R
-N curve, tests must be performed on specimens which are representative of the design, manufacture
and quality of the relevant detail in the actual vessel. Test stress levels must be chosen to result in lives no
more than 2x10
6
cycles and the geometric mean fatigue life obtained from tests performed at a particular
stress range must not be less than the life from the
R
-N curve at that stress multiplied by the factor F from
table 18-6.
Table 18-6: Fatigue test factor F related to 97,7% probability of survival
Number of test results F
1
2
3
4
10
12,5
10,5
9,8
9,4
8,8
18.10.4 Deviations from design shape
Discontinuities and departures from the intended shape of a vessel (i.e. "misalignments") will cause local
increases in pressure-induced stresses in shells, as a result of secondary bending, and hence reduce fatigue
life. This is true even if the allowable assembly tolerances given in Part 4 of this standard are met.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.45
DBA
Design by Analysis
Departures from intended shape include misalignment of abutting plates, an angle between abutting plates,
roof-topping where there is a flat at the end of each plate, weld peaking and ovality (see figure 18-9). In
most cases these features cause local increases in the hoop stress in the shell but deviations from design
shape associated with circumferential seams cause increases in the longitudinal stress.
Figure 18-9: Deviations from design shape at seam welds
NOTE : When stresses greater than yield arise as a result of deviation from design shape, the
pressure test will lead to an improvement in the shape of the vessel due to plastic deformation.
However, vessels made from materials with yield strengths considerably higher than the specified
minimum are less likely to benefit in this way. The beneficial effect of the pressure test on the shape of
the vessel cannot be predicted and therefore if some benefit is required in order to satisfy the fatigue
analysis, it is necessary to measure the actual shape after pressure test. Similarly, strain
measurements to determine the actual stress concentration factor should be made after pressure test.
The influence of misalignment must be considered at the design stage using one of the following
approaches. In each case, the aim is to deduce assembly tolerances which are consistent with the required
fatigue life.
a) Assume values for misalignment, calculate the resulting secondary bending stresses, and include them
in the calculation of structural stress for the detail under consideration. Adopt the class from table 18-4 or
18-5 and check the fatigue life. If unacceptable, tighten some or all of the tolerances to meet the required
life.
b) For a detail of nominal class C
1
, determine the class actually needed to meet the required fatigue life,
C
2
. Then, the allowable increase in stress due to misalignments is K
m
= C
1
/C
2
. Assembly tolerances
which result in K
m
C
1
/C
2
can then be deduced.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.46
DBA
Design by Analysis
A conservative estimate of K
m
is:
K A A A A
m
+ + + + 1
1 2 3 4
...(18.10-1)
where
A
1
caters for misalignment and is given by:
A
e
e
e e
x
x x
1
1 1
1 2
6

_
,

_
,

n1
n
n n
...(18.10-2)
in which

1
is the offset of the centre lines of abutting plates;
e
n1
e
n2
where e
n1
and e
n2
are the nominal thicknesses of the two abutting plates;
x is 1,5 for a sphere or circumferential seam in a cylinder and 0,6 for a longitudinal seam in a cylinder.
A
2
caters for ovality in cylinders and is given by:
( )
( )
A
R R
e
P
E
R
e
2
2
3
3
1
1
2
2

_
,

1
]
1
1
1
1
1
1
max min

n
...(18.10-3)
where R is the mean radius
A
3
caters for poor angular alignment of plates in spheres and is given by:
A
R
e
3
0 5
49

_
,

n
,
...(18.10-4)
where is the angle between tangents to the plates, at the seam (in degrees);
A
4
caters for local peaking and is given by:
A
e
4
6


n
...(18.10-5)
where is the deviation from true form, other than above, and other terms are defined in figure 18-8.
NOTE: This estimate of A
4
ignores the beneficial reduction of the peaking due to pressure and is
therefore conservative. Corrections due to non-linear effects, which reduce A
4
, are permissible [11].
In the case of seam welds, the incorporation of a transition taper at a thickness change does not affect the
value of A
1
.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.47
DBA
Design by Analysis
Equation 18.10-1 will overestimate K
m
if local bending is restricted, for example: at short shape
imperfections, when there will be a stress redistribution around the imperfection; at imperfections in short
cylindrical vessels, which can get support from the ends; adjacent to attachments which stiffen the shell.
However, special analysis must be performed to justify lower K
m
values.
The effect of departures from design shape for which K
m
2 can be ignored if the weld toe is burr machined
using the procedure given in 18.10.2.1.
18.10.5 Welding flaws
Fatigue cracks can propagate from welding flaws and, therefore, depending on the required fatigue life, the
flaws tolerated in Parts 4 and 5 of this standard may or may not be acceptable. Thus, in fatigue-loaded
vessels the following apply:
a) Planar flaws are unacceptable.
b) Acceptance levels for embedded non-planar flaws and geometric imperfections are given in annex in
preparation.
NOTE: All other flaws can be assessed using an established fitness-for-purpose flaw assessment
method, such as that in reference [8]. The fatigue strengths of welds containing flaws can be
expressed in terms of the classification system in 18.10.1. Thus, they can be readily compared with
those of other weld details.
18.10.6 Correction factors
18.10.6.1 To take account of material thickness e
n
> 25 mm, f
ew
shall be calculated as follows:
ew
n
f
=
25
e
0 25 ,

_
,
...(18.10-6)
For e
n
> 150 mm, the value of f
ew
for e
n
= 150 mm applies.
NOTE: In all cases, fatigue cracking from the toe of the weld in the stressed member is being
considered. Thus, the correction is not required (i.e. f
ew
= 1) for some details, see tables 18-4 and 18-
5.
18.10.6.2 For operating temperatures above 100 C, f
t*
is given by:
- for ferritic materials:
f t t
t *


103 15 10 15 10
4 6 2
, , * , * ...(18.10-7)
- and for austenitic materials:
f t
t *


1043 4 3 10
4
, , * ...(18.10-8)
where
t t t * , ,
max min
+ 075 0 25 ...(18.10-9)
NOTE: temperatures in equation (18.10-9) are all in degree Celsius.
f
t*
is illustrated in figure 18-10.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.48
DBA
Design by Analysis
18.10.6.3 The overall correction factor for welded components, f
w
, shall be calculated as follows:
f f f
w ew t*
...(18.10-10)
f
t*
Mean cycle temperature t*, C
Figure 18-10: Correction factor f
t*
18.10.7 Fatigue design curves
Fatigue strength is expressed in terms of a series of
R
-N curves in figure 18-11, each applying to
particular construction details. The curves are identified by the fatigue strength value
R
(N/mm
2
) at fatigue
life N = 2x10
6
cycles.
The design curves have the form as shown in figure 18-12 and conform to the equation:
N =
C
m
R

...(18.10-11)
where m and C are constants whose values are given in table 18-7.
Different values apply for fatigue lives up to 5x10
6
cycles and for lives above 5x10
6
cycles. For constant
amplitude loading, the endurance limit (i.e. stress range below which the fatigue life can be assumed to be
infinite) corresponds to the stress range at 5x10
6
cycles. The corresponding stress range for variable
amplitude loading is that at 10
8
cycles.
NOTE: Alternative curves and constant amplitude endurance limits are permissible if they can be
justified. For lives above 2x10
6
cycles the curves, which are consistent with reference [9], are
conservative.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.49
DBA
Design by Analysis
To obtain the permissible number of load cycles, N, at a specified stress range,
eq
or , the following
shall be calculated.
N
f

_
,

C
eq
w
m

...(18.10-12)
or
N
f

_
,

C
w
m

...(18.10-13)
Alternatively, for use as a design curve to obtain the allowable stress range for a specified number of applied
load cycles, n,

eq R w
m
w
C



or
f
=
n

f
1/

_
,

...(18.10-14)
NOTE1: The curves have been derived from fatigue test data obtained from appropriate laboratory specimens, tested
under load control or, for applied strains exceeding yield (low cycle fatigue), under strain control. Continuity from the low
to high cycle regime is achieved by expressing the low cycle fatigue data in terms of the pseudo-elastic stress range (i.e.
strain range multiplied by elastic modulus, if necessary corrected for plasticity (see 18.8)). Such data are compatible with
results obtained from pressure cycling tests on actual vessels.
NOTE2: The fatigue strength design curves are approximately three standard deviations of log N below the mean curve,
fitted to the original test data by regression analysis. Thus, they represent probability of failure of approximately 0,5 %
with 99 % confidence.
10
100
1000
10000
1,0E+02 1,0E+03 1,0E+04 1,0E+05 1,0E+06 1,0E+07 1,0E+08
N

(
N
/
m
m

)
32 40 45 50 56 63 71 8 0 9 0 10 0
E=2,09x10
5
N/mm
2
2
1
Figure 18-11: Fatigue design curves for welded components: (1) curves for assessing variable
amplitude loading; (2) For constant amplitude loading, endurance limit =
R
at 5x10
6
cycles.
N.B.: For N>2x10
6
cylces, alternative curves and
R
values are permissible, see clause 18.10-6.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.50
DBA
Design by Analysis
Figure 18-12: The form of the fatigue design curves for welded components
Table 18-7: Coefficients of the fatigue design curves for welded components
Class Constants of
R
- N curve* Stress range at N cycles,
N/mm
2
For 10
2
< N < 5x10
6
For 5x10
6
< N < 10
8
m C m C 5 x 10
6
10
8
100
90
80
71
63
56
50
45
40
32
3,0
3,0
3,0
3,0
3,0
3,0
3,0
3,0
3,0
3,0
2,00 x 10
12
1,46 x 10
12
1,02 x 10
12
7,16 x 10
11
5,00 x 10
11
3,51 x 10
11
2,50 x 10
11
1,82 x 10
11
1,28 x 10
11
6,55 x 10
10
5,0
5,0
5,0
5,0
5,0
5,0
5,0
5,0
5,0
5,0
1,09 x 10
16
6,41 x 10
15
3,56 x 10
15
1,96 x 10
15
1,08 x 10
15
5,98 x 10
14
3,39 x 10
14
2,00 x 10
14
1,11 x 10
14
3,64 x 10
13
74
66
59
52
46
41
37
33
29,5
24
40
36
32
29
26
23
20
18
16
13
* For E = 2,09x10
5
N/mm
2
18.11 Fatigue strength of unwelded components
18.11.1 Correction factors
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.51
DBA
Design by Analysis
18.11.1.1 Surface finish correction factor
To take account of surface finish, f
s
shall be calculated as follows:
f F
N
s s

( , ln , ) 0 1 0 465
...(18.11-1)
where
( ) ( ) F R R R
s z m z
+ 1 0 056 0 289
0 64 0 53
, ln ln , ln
, ,
...(18.11-2)
and R
z
is the peak-to-valley height (m).
If not specified, the manufacturing-related peak-to-valley heights in table 18-8 shall be used in equation
18.11-2.
For polished surfaces with a peak-to-valley height R
z
< 6 m, assume f
s
= 1. Values of f
s
for as-rolled plate
are given in figure 18-13.
Table 18-8 Base values for peak-to-valley heights
Surface condition R
Z
, m
Rolled or extruded
Machined
Ground, free of notches
200
50
10
Figure 18-13: Correction factor f
s
for as-rolled plates

18.11.1.2 Thickness correction factor
For wall thicknesses 25 mm < e
n
150 mm, f
e
is:
f F
N
e e

( , ln , ) 01 0 465
...(18.11-3)
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.52
DBA
Design by Analysis
where

,
_

n
e
e
25
=
F
0,182
...(18.11-4)
For e
n
> 150 mm, the value of f
e
for e = 150 mm applies.
18.11.1.3 Correction factor to take account of the influence of mean stress
18.11.1.3.1 Elastic range
For
* p0,2/t
l eq,
2R < and
eq max
< R
p0,2/t*
, the mean stress correction factor f
m
for N 2x10
6
cycles is to
be determined for rolled and forged steel as a function of the mean stress sensitivity M from:
( )
5 , 0
R
R
2
1
2
1
1
1
]
1

,
_

+
+

M
M M
f
m
...(18.11-5)
when
M) + (
R
R
R
1 2
* 0,2/t p



or

,
_

+
+

R
R
m
M
M
M
f

2
3 1
3 / 1
...(18.11-6)
when
R
M) + (
* 0,2/t p
R
R
1 2

where for rolled and forged steel:


M =
R
- 0 00035 01 , ,
m
...(18.11-7)
For N 2x10
6
cycles, f
m
shall be taken from figure 18-14.
NOTE: f
m
is independent of stress range.
18.11.1.3.2 Partly plastic range
For
eq,l
p0,2/ t *
< 2R and
eq max
> R
p 0,2/t*
, equation (18.11-5) or (18.11-6) shall also be used to
determine f
m
, although the reduced mean equivalent stress, as calculated from equations (18.11-8) or
(18.11-9) shall be used instead of
eq
. See figure 18-5 .
If
eq
> 0 ,
2
R
=

eq
* 0,2/t p
r eq,

...(18.11-8)
If
eq
< 0 ,

R
2
=
* 0,2/t p
eq
r eq,

...(18.11-9)
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.53
DBA
Design by Analysis
Figure 18-14: Correction factor f
m
to take account of mean stress in unwelded material for N > 2x10
6
cycles
18.11.2 Overall correction factor for unwelded components
18.11.2.1 Main procedure
The overall correction factor for unwelded components, f
u
, shall be calculated as follows:
f
f f f f
K
u
s e m t*
eff


...(18.11-9)
in which f
s
, f
e
, and f
m
are given in 18.11.1.1 - 3 respectively; and f
t*
is given in 18.10.5.2.
18.11.2.2 Simplified Procedure
A simplified procedure for the fatigue assessment of unwelded steel is permissible using the class 100
design data for welded components, independently of material static strength or surface finish. The data are
used in conjunction with equation 18.10-12, with f
w
replaced by f
u
.
If the applied stress is partly compressive, it is permissible to assume that the relevant value of
eq
is the
sum of the tensile component and 60 % of the compressive component. Thus, for mean stress
eq
the
correction factor f
u
becomes f f f
e t c

*
/ K
eff
in which:
c
f
= -
2
125 ,

eq
R

_
,

...(18.11-10)
f
e
is given in 18.11.1.2 and f
t*
in 18.10.5.2.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.54
DBA
Design by Analysis
18.11.3 Design data
The fatigue strengths of unwelded components are expressed in terms of a series of
R
-N curves, each
applying to a particular tensile strength of steel, as given in figure 18-15.
The fatigue design curves in figure 18-15 have the form:
N =
-
R
+
2
4 6
10
0 63 115
4
, .
, ,
R m

1
]
1
1
...(18.11-11)
for lives up to 2x10
6
cycles. For N 2x10
6
cycles, values of
R
are given in table 18-10. For cumulative
damage calculations using equation 18.5-1, the curves are linear for N = 2x10
6
to 10
8
cycles, and have the
form:

1
1
]
1

eq
m
10
92 . . 7 , 2
R
= N ...(18.11-12)
Values of
R
at (and beyond) 10
8
cycles for selected tensile strengths are included in table 18-10.
To obtain the allowable number of load cycles, N, at a specified stress range
eq
,
N
f
-
R
+
2

_
,

4 6
10
0 63 115
4
,
, ,

eq
u
m

...(18.11-13)
Alternatively, for use as a design curve to obtain the allowable stress range for a specified number of load
cycles, n,

eq R u m u

f =

n
+
R
- f

_
,

4 6
10
0 63 115
4
,
, , ...(18.11-14)
NOTE1: The curves have been derived from fatigue test data obtained from unnotched polished
ferritic and austenitic rolled and forged steel specimens at room temperature, under alternating (mean
load = 0) load control or, for applied strains exceeding yield (low-cycle fatigue), strain control.
NOTE2: Compared with the mean curve fitted to the original data, the curves incorporate safety
factors of 10 on fatigue life and 1.5 on stress range.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.55
DBA
Design by Analysis
Table 18-10: Stress range
R
for N 2x10
6
cycles for unnotched test bars of ferritic and austenitic
rolled and forged steels at room temperature and zero mean stress
Tensile strength
R
m
,
N/mm
2

R
= constant, N/mm
2
N 2x10
6
N 10
8
(for cumulative damage calculations)
400 275 162
600 400 236
800 525 310
1000 650 385
100
1 000
10 000
1,0E+02 1,0E+03 1,0E+04 1,0E+05 1,0E+06 1,0E+07 1,0E+08
Fatigue life, N, cycles
S
t
r
e
s
s

r
a
n
g
e
,

R

,
N
/
m
m
2
R
m
(N/mm
2
)
600
400
800
1000
Figure 18-15: Fatigue design curves for unwelded ferritic and austenitic forged and rolled steels
(mean stress = 0)
18.12 Fatigue strength of steel bolts
18.12.1 Scope
These rules apply only to axially-loaded steel bolts. They do not apply to other threaded components such as
flanges, ends or valves.
18.12.2 Correction factors
18.12.2.1 For bolt diameters > 25 mm, the correction factor f
e
shall be calculated using equation 18.11-3,
with e
n
put equal to the bolt diameter. For bolt diameters 25 mm, f
e
= 1.
18.12.2.2 Overall correction factor for bolts
f
b
shall be calculated as follows:
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.56
DBA
Design by Analysis
f f f
b e t*
...(18.12-1)
in which f
e
is given in 18.12.2.1 and f
t*
is given in 18.10.5.2.
18.12.3 Design data
The fatigue strength of axially loaded bolts is expressed in terms of the ratio:
maximum nominal stress range
nominal ultimate tensile strength of bolt material
m
=
R

The single design curve



R
N
3
285
R
m

_
,
...(18.12-2)
with an endurance limit at 2x10
6
cycles, shown in figure 18-15, is used for any thread form (machined,
ground or rolled) and core diameters up to 25 mm. However, regardless of the actual tensile strength of the
bolt material, R
m
should never be assumed to exceed 785 N/mm
2
.
Figure 18-16: Fatigue design curve for bolts
NOTE: The design curve has been derived from fatigue test data obtained from axially-loaded
threaded connections. The design curve is three standard deviations of log N below the mean curve,
fitted to the original test data by regression analysis. Thus, the curve represents a failure probability of
approximately 0,5 % with 99 % confidence.
To obtain the allowable number of load cycles, N, at a specified stress range, :
N 285
R
.
f
3

_
,

m b

...(18.12-3)
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.57
DBA
Design by Analysis
Alternatively, for use of the design curve to obtain the allowable stress range, , for a specified number of
load cycles, n,
f R
285
n
f
R
1/ 3

_
,
.
b m b
...(18.12-4)
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.58
DBA
Design by Analysis
Annex B (informative)
Direct route for design by analysis
(This annex refers to clause 5).
B.1 Purpose
Design-by-analysis (DBA) provides rules for the design of any component under any action. It may be used:
- as an alternative to design-by-formula (see 5.4.1),
- as a complement to design-by-formula for:
- cases not covered by that route;
- cases involving superposition of environmental actions;
- where DBA is required, e.g. by local authorities where a potential major hazard is
involved or for environmental reasons;
- exceptional cases where the manufacturing tolerances given in clause 5 of Part 4
are exceeded.
.
NOTE1: In the last item, any deviations beyond tolerance limits shall be clearly documented.
NOTE2: The method given in this clause is used in structural design, e.g. Eurocode No.3 Design of
Steel Structures. Some aspects are applied in the Danish code for pressure vessels, DS 458.
It is pre-supposed that this annex will be used with conformity assessment modules G, B1+D, or
B1+F.
B.2 Specific definitions
The following definitions are in addition to those in clause 3.
B.2.1 action: Physical influence which causes stress and/or strain in a structure.
B.2.2 application rule: Procedure to determine whether a principle is satisfied.
B.2.3 characteristic value: Representative value which takes account of the variation of an action.
B.2.4 coefficient of variation: Measure of statistical dispersion (standard deviation divided by mean);
B.2.5 design check: Assessment of a component for a load case by means of an application rule.
B.2.6 effect: Response (e.g. stress, strain, displacement, resultant force or moment, equivalent stress
resultant) of a component to a specific action.
B.2.7 load case: A combination of coincident actions.
B.2.8 loading type: Classification of loading based on frequency and duration.
B.2.9 limit state: Structural condition beyond which the design performance requirements of a component
are not satisfied.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.59
DBA
Design by Analysis
B.2.10 partial safety factor: Factor which depends on the limit state and is applied to either an effect or a
resistance to obtain a design value.
B.2.11 principle: General statement, definition or requirement for a given failure mode for which there is no
alternative, unless specifically stated otherwise.
B.2.12 resistance: the limiting value for a given limit state of an action or effect.
B.2.13 structure: All load carrying parts relevant to the component, e.g. the whole vessel, its load carrying
attachments, supports and foundations.
B.3 Specific symbols and abbreviations
The following symbols and abbreviations are in addition to those in clause 4.
B.3.1 Subscripts
d is design
I is i
th
value
inf is lowest (infimum)
j is j
th
value
k is k
th
value
E is effect
G is permanent action
P is pressure action
Q is variable action
R is resistance
sup is highest (supremum)
X is exceptional action
B.3.2 Symbols
A is action (general term)
A
5
is minimum rupture elongation
a is any structural dimension
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.60
DBA
Design by Analysis
D is fatigue damage
E is effect (general term)
Ep is plastic modulus
R is resistance (general term)
RM is design strength parameter
is stress
is strain
is partial safety factor
is a combination factor for pressure and variable actions.
B.4 Method
The method comprises the following stages:
a) specify the relevant failure mode and limit state, taking account of the loading type, see
B.5;
NOTE: There may be more than one failure mode.
b) specify the principle, see B.6;
c) select an appropriate application rule, see B.6;
d) using the detailed information in B.9, carry out the design check as follows:
- define the load case and specify the actions, see B.7.1;
- determine the characteristic value and calculate the design value of each action, see
B.7.2 and B.7.3;
- calculate the effect of the actions, see B.7.4;
- calculate the resistance of the component, see B.8;
- indicate whether or not the principle is satisfied.
e) if the principle is not satisfied, repeat the design check using amended loading, geometry
or
material.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.61
DBA
Design by Analysis
B.5 Failure modes and limit states
The main failure modes are listed in table B.5-1 with the relevant limit states. The latter are
classified according to whether the loading is short term, long term or cyclic.
Individual failure modes only are given in table B.5-1. Combinations of failure modes, e.g.
e.g. fatigue - plastic rupture, creep - plastic rupture, creep - fatigue, shall be considered separately.
A limit state is classified as either an ultimate or a serviceability limit state.
An ultimate limit state is a structural condition (of the component or vessel) beyond which the safety of
personnel could be endangered.
NOTE1: Ultimate limit states include: failure by gross plastic deformation; rupture caused by
fatigue; collapse caused by instability of the vessel or part of it; loss of equilibrium of the vessel or any
part of it, considered as a rigid body, by overturning or displacement; and leakage which affects
safety.
NOTE2: In the case of collapse, some states prior to collapse are considered as collapse and also
classified as ultimate limit states.
A serviceability limit state is a structural condition (of the component or vessel) beyond which the service
criteria specified for the component are no longer met.
NOTE1: Serviceability limit states include:
- deformation or deflection which adversely affects the use of the vessel (including the proper
functioning of machines or services), or causes damage to structural or non-structural elements;
- leakage which affects efficient use of the vessel but does not compromise safety or cause an
unacceptable environmental hazard.
NOTE2: Depending upon the hazard, leakage may create either an ultimate or a serviceability limit
state.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.62
DBA
Design by Analysis
Table B.5-1: Classification of failure modes and limit states
Loading type
Short term Long term Cyclic
Failure mode Single
application
Multiple
application
Single
application
Multiple
application
Brittle fracture
Ductile rupture
3)
Excessive deformation 1
4)
Excessive deformation 2
5)
Excessive deformation 3
6)
Excessive local strains
7)
Instability
8)
U
U
S, U
1)
U
S
U
U, S
2)
Progressive plastic def.
9)
Alternating plasticity
10)
U
U
Creep rupture
Creep-Excessive def. 1
11)
Creep-Excessive def. 2
12)
Creep-Excessive def. 3
13)
Creep instability
Erosion, corrosion
Environmentally assisted
cracking
14)
U
S, U
1)
U
S
U, S
2)
S
U
Creep
Creep-Excessive def. 1
11)
Creep-Excessive def. 2
12)
Creep-Excessive def. 3
13)
Creep instability
Erosion, corrosion
Environmentally assisted
cracking
14)
U
S, U
1)
U
S
U, S
2)
S
U
Fatigue
Environmentally assisted
fatigue
U
U
U indicates ultimate limit state. S indicates service limit state.
1) In case of risk due to leakage of content (toxic, inflammable, steam etc.)
2) In case of sufficient post-instability load carrying capacity
3) Unstable gross plastic yielding or unstable crack growth
4) Excessive deformations at mechanical joints.
5) Excessive deformations resulting in unacceptable transfer of load.
6) Excessive deformations related to service restraints.
7) Resulting in crack formation or ductile tearing, by exhaustion of material ductility
8) Elastic, plastic, or elastic-plastic
9) Progressive plastic deformations (or ratcheting)
10) Alternative plasticity (see also clause 6)
11) Creep-Excessive deformation at mechanical joints
12) Creep-Excessive deformation resulting in unacceptable transfer of load
13) Creep-Excessive deformation related to service restraints
14) Stress corrosion cracking (SCC), Hydrogen induced cracking (HIC), Stress orientated hydrogen induced cracking (SOHIC).
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.63
DBA
Design by Analysis
B.6 Principles, application rules and design checks
For each failure mode, a single principle is stated to ensure that the limit state is not exceeded. The
principle requires that for any load case either the:
- combination of the design actions does not exceed the design resistance
or
- the design effect of the design actions does not exceed the design resistance.
Coincident actions are combined in a load case. All relevant load cases shall be considered.
For each principle, one or more application rules are given to indicate different means by which an
assessment can be made. The most relevant application rule or rules shall be selected. It is
permissible to use other application rules provided they accord with the relevant principle and are at
least equivalent with regard to the resistance, serviceability and durability of the vessel.
The assessment of a component against a load case by means of an application rule comprises a
design check. The principles, application rules and design checks are specified in detail in B.9.
NOTE: Proof testing or non-destructive testing , additional to that specified in Part 5, should be
specified which is appropriate to the stress level in the component and the failure mode.
B.7 Actions
B.7.1 Classification
Actions are classified into the following four types:
- permanent;
- variable (other than temperature, pressure and actions related to them deterministically, i.e. not
involving probability);
- exceptional (see 5.3.5 and 6.1.1);
- temperature, pressure and actions related to them deterministically.
NOTE1: Mechanical, physical, chemical or biological actions may have an influence on the safety of a
vessel. However, in DBA, only those which cause stress or strain are considered. Examples are:
volume forces (e.g. self-weight), surface forces (pressures, surface loadings, etc.), singular forces
(resultants representing e.g. imposed surface forces), line forces, point forces, temperature changes,
displacements imposed on the vessel at connections, foundations, due to e.g. temperature changes,
settlement.
NOTE2: Examples of permanent actions are: self-weight of a structure and associated fittings,
ancillaries and fixed equipment.
NOTE3: Examples of variable actions are: imposed loads, wind or snow loads
NOTE4: Examples of exceptional actions are: actions on secondary containment due to failure of
primary containment or exceptional earthquake actions.
NOTE5: Actions which may be either permanent or variable are: temperature changes, imposed loads
or displacements. Temperature changes have a dual role in that they may cause stress in the
structure and also change its resistance.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.64
DBA
Design by Analysis
NOTE6: Environmental attack (whether internal or external) may reduce the safety or serviceability of
a vessel. This should be taken into account in the selection of materials, provision of additional wall
thickness (see 5.2.2), or specification of appropriate material parameters in the determination of
resistance (see B.8.2) .
Although operating pressures and temperatures are variable actions, they have special characteristics with
regard to their variation in time, random properties, etc. Because there is usually a strong correlation
between operating pressure and temperature, they shall be considered to act simultaneously, and the
pressure - temperature dependence shall be defined appropriately.
NOTE: Pressure-temperature dependence may be stated either in the form of coincident pairs or in
the form of a functional relationship between fluid pressure and temperature.
With actions which consist of permanent and variable parts, the parts shall be considered
individually.
Variable actions may include actions of quite different characteristics, e.g.:
- actions which are related to pressure and/or temperature in a deterministic way. These shall be
combined in the pressure/temperature action and the relationship, exact or approximate, shall be
used.
-actions which are not correlated with pressure or temperature but have well defined (bounded)
extreme values;
- actions, like wind loads, which can be described only as stochastic (i.e. random) processes and
are not correlated with pressure or temperature.
B.7.2 Characteristic values
The requirements for determining the characteristic values of different types of action are given in
table B.7-1 and in the following.
The characteristic values of pressure and temperature describe the pressure-temperature regime that
envelops those pressures and temperatures which can occur under reasonably forseeable conditions,
see figure B.7-1.
The following characteristic values shall always be specified:
- the upper characteristic value of the pressure (P
sup
)
- the lower characteristic value of the pressure (P
inf
)
- the upper characteristic value of the temperature (T
sup
)
- the lower characteristic value of the temperature (T
inf
).
The self-weight of the structure and of non-structural parts may be calculated on the basis of
nominal dimensions and mean unit masses.
For wind and snow loadings, the values specified in relevant codes may be used.
In load cases where thermal stresses (constant or transient) have an influence on the safety of the structure,
the characteristic values of coincident pressure / temperature shall be the extreme values of operating
pressure and temperature that can reasonably be expected to occur under normal operating conditions over
the life of the vessel.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.65
DBA
Design by Analysis
Table B.7-1: Characteristic values for different types of action
Action Coefficient
of variation
Symbol Characteristic value
Permanent 0,1
(note 1)
G
k
(note 2)
Mean of extreme values
Permanent
>0,1
(note 3)
G
k, sup
G
k, inf
(note 2)
Upper limit with 95% probability
of not being exceeded (see note
4);
Lower limit with 95%
probability of being exceeded
(see note 4).
Variable 0,1
(note1)
Q
k
(note 2)
Mean of extreme values
Variable >0,1 Q
k
(note 2)
97% percentile of extreme value
in given period (see note 5)
Exceptional - - Shall be individually specified
Pressures and
temperatures -
P
sup
T
sup
P
inf
T
inf
Reasonably forseeable highest
pressure (see note 7)
Reasonably forseeable highest
temperature
Reasonably forseeable lowest
pressure (see note 6)
Reasonably forseeable lowest
temperature
NOTE1: The mean of the extreme values may also be used when the difference between the
reasonably foreseeable highest value and the lowest one is not greater than 20% of their arithmetic
mean value.
NOTE2: The k subscript in table B.7-1 indicates that there are usually several actions in a load case
and they are individually numbered.
NOTE3: Also applies where the actions are likely to vary during the life of the vessel (e.g. some
superimposed permanent loads)
NOTE4: If a statistical approach is not possible, the highest and lowest credible values may be used.
NOTE5: For variable actions which are bounded, the limit values may be used as characteristic
values.
NOTE6: This value is usually either zero or -1,0 (for vacuum conditions).
NOTE7: This may be, for example, the set pressure of the relief valve.
For temperature values which are not environmentally imposed and in cases where a combination of
P
sup
and T
sup
is uneconomic, it may be necessary to specify characteristic pressure - temperature
pairs, e.g. (P
sup,i
, T
sup,i
), (P
inf,i
, T
inf,i
), which determine an envelope of the (P, T) - regime of the
reasonably foreseeable extreme values, see figure B.7-1.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.66
DBA
Design by Analysis
P
T
1
2
3
4
5
Psup1
Psup 2
Pinf 3
Pinf 4
Psup 5
Tinf 4
Tsup 1
Tsup 2
Tsup 3
Tinf 5
Tinf 1
Figure B.7-1: Typical plot of coincident temperatures and pressures
NOTE: Characteristic values should be stated clearly in the technical documentation.
B.7.3 Design value
The design value A
d
of an action is expressed in general terms as:
A A
d A
... (B.7-1)
where A is the characteristic value of the action (permanent, variable, exceptional or pressure) and
A
is the
relevant partial safety factor as given in B.9 for the considered limit state.
NOTE:
A
takes account of the following:
- the possibility of non-conservative deviation of the actions from their characteristic values;
- the uncertainty of the models which describe the physical phenomena for the action and effect;
- uncertainty in any stochastic models of the action;
- whether the action has a favourable or an unfavourable effect. (For example, in one load case the
action due to the weight of a component might be opposing the pressure force and therefore has a
favourable effect on reducing stress. In another, the weight might be acting with the pressure and so
has an unfavourable effect. In the two load cases,
A
would have a different value).
B.7.4 Design effect
For each load case, the effect of all the design actions is combined to give the design effect. This is a
function of the design actions ( A
d
) and the dimensions ( a
d
). It is expressed in general terms as:
E E A a E P G Q X a
d d d P G Q X d
( , ,....) ( , , , , ......) ... (B.7-2)
The calculation requires both geometric data and material properties. For geometric data, nominal values for
individual dimensions and properties, rather than minimum values, may be used. For strength related data,
R
eH
, R
p
, R
m
etc., the minimum guaranteed values specified in the material codes or material data sheets
shall be used in the calculations. For the other properties, e.g. modulus of elasticity or coefficient of linear
thermal expansion, nominal or mean values may be used.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.67
DBA
Design by Analysis
For components clad on the inside and subjected to internal pressure, the inside surface for pressure action
shall be taken as the nominal inner face of the cladding. For components clad on the outside and subjected
to external pressure, the outside surface for pressure action shall be taken as the outer face of the base
metal.
B.8 Design resistance
B.8.1 Resistance
For the calculation of resistance:
- the nominal values of the geometric data may be used with the exception of thicknesses for which
the nominal values minus the allowances shall be used;
- the minimum guaranteed values shall be used for strength data, i.e.: R R R R
eH p0,2/ t p1,0/ t m/ t
, , , ;
- for other properties, e.g. modulus of elasticity, coefficient of linear thermal expansion, nominal or
mean values may be used.
The resistance shall be determined by increasing in the same proportion all the design actions in the
load case.
NOTE: When all the design actions in the load case are increased in the same proportion, the path in
the action space is a straight line passing through the origin and through the point the co-ordinates of
which represent the design actions of the load case.
B.8.2 Design resistance
The design resistance of a component is expressed in general terms as
R
R
d
R

...(B.8-1)
where
R
is the partial safety factor for the resistance.
NOTE:
R
takes account of the following:
- the possibility of a non-conservative deviation in the material properties or geometrical data;
- the possibility of inaccuracies in the model for the calculation of resistance;
- the possibility of deterioration not explicitly accounted for;
- the failure mode, or failure modes under consideration, and the use in some cases of a strength
parameter which is only approximate for the failure mode;
- the level of hazard which is considered acceptable.
In some design checks, the design resistance is obtained directly as a function of the values of material
strength parameters divided by
R
, e.g.
R R
R
a
eH
d d
R

_
,

, ,.... ...(B.8-2)
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.68
DBA
Design by Analysis
B.9 Design checks
B.9.1 General
A design check shall be carried out against a limit state for one of the following circumstances:
- normal situations where normal conditions apply;
- special situations where conditions for construction, erection, repair or testing apply;
- exceptional situations, see 5.3.5.
All relevant requirements of all the following design checks shall be fulfilled.
Subclause B.9.2 applies to failure by gross plastic deformation (GPD) in either operation or test. Design
details which would cause severe strain concentration or elastic follow-up shall not be present.
NOTE: Avoidance of severe stress concentrations is specified because only the effects of excessive
local yielding are included in subclause B.9.2.
The other subclauses apply as follows. For failure by progressive plastic deformation (PD), see B.9.3; by
instability, see B.9.4; by fatigue, see B.9.5, and by overturning and global displacement (rigid body motions),
see B.9.6.
NOTE1: The design checks in the following are not exhaustive.
NOTE2: In some cases, it may be necessary to investigate additional limit states. For example, with
austenitic stainless steel, failure by GPD should be checked (as an ultimate limit state) but leakage
may also need to be checked (as either an ultimate or a serviceability limit state), see table 5.B.9-3.
B.9.2 Gross Plastic Deformation (GPD)
B.9.2.1 Principle
The principle is as follows:
For any load case, either the combination of the design actions shall not exceed the design
resistance:
A R
d d
...(B.9-1)
or the design effect of the actions shall not exceed the design resistance:
E R
d d
...(B.9-2)
In either case, the design resistance shall be obtained from calculations assuming:
- proportional increase of all design actions, see note in B.8.1;
- first order theory;
- a linear-elastic ideal-plastic material or a rigid ideal-plastic material;
- Tresca's yield criterion (maximum shear stress criterion) and associated flow rule;
- design strength parameter RM as specified in B.9.2.5 or B.9.2.6;
- partial safety factor
R
as specified in B.9.2.5 or B.9.2.6.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.69
DBA
Design by Analysis
B.9.2.2 Application rule 1 - lower-bound limit load
NOTE: Application rule 1 comprises the whole of this subclause.
The design action shall be less than the lower-bound limit load divided by
R
.
To avoid possible computational difficulties when using a computer, a linear-elastic linear-hardening material
with a plastic modulus Ep equal to E/10 000 (or a similar small value) may be used, instead of the linear
ideal plastic one, see figure B.9-1.
The lower bound limit is given by the tangent-intersection, see figures B.9-2 and B.9-3.
If there is no maximum in the region of principal strains less than + 5 %, the greatest tangent intersection
value shall be used with one tangent through the origin, the other through a point where the maximum
principal strain does not exceed + 5 %.
Figure B.9-1 Simplified stress-strain model for computation of the lower bound limit load
deflexion deflexion
action action
Figure B.9-2: Limit load determination: plot Figure B.9-3: Limit load determination: plot

Annex 2:
Draft CEN prEN 13445-3
Page
A 2.70
DBA
Design by Analysis
of action versus deflection initially linear of action versus deflection initially non-linear
B.9.2.3 Application rule 2 - maximum primary stress intensity
NOTE: Application rule 2 comprises the whole of this subclause.
The maximum primary stress intensity in the structure shall not be greater than
RM

R
.
NOTE: Stress intensity is the uniaxial stress equivalent to the multi-axial stress state and is defined as
twice the maximum shear stress. It is therefore is the difference between the algebraically largest
principal stress and the algebraically smallest at a point.
A primary stress field is any stress field which just satisfies the equilibrium equations (at any point
throughout the structure). The main characteristic of a primary stress field is that it is not self-limiting.
The main problem of determining a primary stress field corresponding to a given imposed load is that
it has no unique solution.
In cases of structures, where the concept of stress resultants is applicable, e.g. beams, plates, shells, this
requirement may be verified in terms of stress resultants (generalised stresses) and local (technical) limit
loads. Examples of local (technical) limit load sets and the equations for the allowable resultants are given in
table B.9-1.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.71
DBA
Design by Analysis
Table B.9-1. Equations for allowable stress resultants (Tresca yield criterion)
1. Beam, rectangular cross-section, bending moment M
B2
(direction of axis of symmetry), normal
force F
N1
(direction of beam axis)
Limit bending moment M
B2pl
= RM a
2
a
1
2
/ (4
R
) ... (B.9-3)
Limit normal force F
N1pl
= RM a
2
a
1

/
R
... (B.9-4)
Allowable stress resultants |M
B2
|/ M
B2pl
+ (F
nl
/ F
N1pl
) 1 ... (B.9-5)
(See figure B.9-4 for a plot of this equation).
2. Plate, thickness e, bending moments (per unit length) m1, m2, m12
Limit bending moment m
pl
= RM e
2
/ (4R ) ... (B.9-6)
Allowable stress resultants m
pl
( m
1
+ m
2
) - m
1
m
2
+ m
12
m
pl
... (B.9-7)
- m
pl
( m
1
+ m
2
) - m
1
m
2
+ m
12
m
pl
... (B.9-8)
( m
1
- m
2
) + 4 m
12
m
pl
... (B.9-9)
(See figure B.9-5 for a plot of the surface formed by these equations).
3. Plate, thickness e, rotational symmetry, bending moments (per unit length) m
r
, m

Limit bending moment m
pl
= RM e
2
/ (4
R
) ... (B.9-10)
Allowable stress resultants |m
r
| m
pl
... (B.9-11)
|m

| m
pl
... (B.9-12)
|m
r
- m

| m
pl
... (B.9-13)
(See figure B.9-6 for a plot of these equations).
4. Shell, thickness e, rotational symmetry, bending moments (per unit length) m
S
, m

(=
m
S
), membrane forces (per unit length) n
S
, n

Limit bending moment m
pl
= RM e
2
/ (4
R
) ... (B.9-14)
Limit normal force n
pl
= RM e


/
R
... (B.9-15)
Allowable stress resultants |n
s
|/n
pl
1 ... (B.9-16)
|n

|/n
pl
1 ... (B.9-17)
|n
s
- n

| /n
pl
1 ... (B.9-18)
|m
s
|/ m
pl
+(n
s
/n
pl
) 1 ... (B.9-19)
2|m
s
|/ m
pl
+ 2(n

/n
pl
-1) + 2(n

/n
pl
- 2n
s
/n
pl
-1) 2 ... (B.9-20)
(See figure B.9-7 for a plot of the surface formed by these
equations).
5. Shell, special case of 4, without moment
Limit normal force n
pl
= RM e

/
R ... (5.B.9-21)
Allowable stress resultants
|n
s
| n
pl
... (B.5-22)
|n

| n
pl
... (B.5-23)
|n
s
- n

| n
pl
... (B.5-24)
(See figure B.9-8 for a plot of these equations).
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.72
DBA
Design by Analysis
Figure B.9-4
Figure B.9-5
Figure B.9-6
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.73
DBA
Design by Analysis
Figure B.9-7
Figure B.9-8
B.9.2.4 Particular requirements
In design checks against GPD, structural strength may only be attributed to cladding in the case of
integrally-bonded cladding, see 5.5.2.
B.9.2.5 Design check against failure by GPD in operation
a) Partial safety factors against actions shall be as given in table B.9-2.
Table B.9-2: Partial safety factors against actions for GPD for load cases in operation
Action Condition Partial safety factor
Permanent For actions with an unfavourable effect
G
135 ,
Permanent For actions with an favourable effect
G
10 ,
Variable For unbounded variables
Q
15 ,
Variable For bounded variable actions and limit values
Q
10 ,
Pressure For actions without a natural limit
P
12 ,
Pressure For actions with a natural limit, e.g. vacuum
P
10 ,
If only part of the pressure is subject to a natural limit, e.g. static head, this part may be multiplied by
p
= 1,0
and the remainder by
p
= 1,2.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.74
DBA
Design by Analysis
b) Combination rules shall be as follows.
All permanent design actions shall be included in each load case.
Each pressure design action shall be combined with the most unfavourable variable design action.
Each pressure design action shall be combined with the corresponding sum of the variable design actions;
the design values of stochastic actions, see B.7-1 and table B.7-1, may be multiplied by the combination
factor = 0,9.
NOTE: Since it is most unlikely that all the variable actions would be at their maximum together, they
may each be multiplied by = 0,9 when their total is combined with pressure.
Favourable variable actions shall not be considered.
c) Design material strength parameter (RM) and partial safety factor (
R
) shall be as given in table
B.9-3.
Table B.9-3: RM and
R
for GPD for load cases in operation
Material RM
R
Ferritic
1
steel R R
eH p0,2/ t
or
1,25 for
R
R
p0,2/ t
m/ 20
0 8 ,
1,5625
R
R
p0,2/ t
m/ 20

_
,

otherwise
Austenitic steel
(30%<A
5
<35%)
R
p1,0/ t
1,25
Austenitic steel
( A
5
35%)
R
p1,0/ t
(see note)
1,25 for
R
R
m/ t
p1,0/ t
20 ,
225
0 5
,
,

R
R
m/ t
p1,0/ t
for 20 25 , ,
R
R
m/ t
p1,0/ t
1,0 for
R
R
m/ t
p1,0/ t
25 ,
Steel castings R
p0,2/ t
1,58 for
R
R
m/ 20
p0,2/ t
158 ,
25 , R
R
p0,2/ t
m/ 20
otherwise
1
Steel other than austenitic steel as per 6.3 and 6.4 for structural parts other than bolts
NOTE: The strain at this material strength is large and it is advisable to check against leakage.
B.9.2.6 Design check against failure by GPD in a hydraulic test
a) Partial safety factors against actions shall be as given in table B.9-4.
Table B.9-4: Partial safety factors for GPD load cases in hydraulic test
Action Condition Partial safety factor
Permanent For actions with an unfavourable effect
G
135 ,
Permanent For actions with an favourable effect
G
10 ,
Pressure -
P
10 ,
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.75
DBA
Design by Analysis
Variable actions need not be considered
b) Combination rules shall be as follows.
All permanent design actions shall be included in each load case.
In cases where more than one test is applied, e.g. multi-chamber vessels, each pressure case shall be
included.
c) RM and
R
shall be as given in table B.9-5.
Table B.9-5: RM and
R
for GPD load cases in hydraulic test
Material RM
R
Ferritic
1
steel R R
eH p0,2
or 1,05
Austenitic steel
(30%< A
5
<35%)
R
p1,0
1,05
Austenitic steel
( A
5
35%)
R
p1,0
1,05 for
R
R
m
p1,0
1905 ,
20 , R
R
p1,0
m
otherwise
Steel castings R
p0,2
1,33 for
R
R
m/ 20
p0,2/ t
158 ,

1
Steel other than austenitic steel as per 6.3 and 6.4 for structural parts other than bolts
B.9.3 Progressive Plastic Deformation (PD)
B.9.3.1 Principle
On repeated application of the action cycle described below progressive plastic deformation shall not occur
for:
- first order theory
- a linear-elastic ideal-plastic material
- Mises' yield criterion (maximum distortion energy criterion) and associated flow rule, and
NOTE: The Tresca yield criterion is also permitted as it is more conservative.
- design strength parameters RM are as specified in subclause B.9.3.4
- partial safety factors on actions are as table B.9-2.
B.9.3.2 Application rule
NOTE: The application rule comprises the whole of this subclause.
At a local structural discontinuity (see note2), where the stress state at each point of the construction is given
in isometric coordinates (see note3) by the principal stresses
1
,
2
and
3
, see figure B.9-9, the following
applies:
a) Constant principal stress direction
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.76
DBA
Design by Analysis
The diameter of the deviatoric mapping of each path of the stress states, determined for each point of the
stress-concentration-free construction (see note4) using linear elasticity theory for the action cycle (see
B.3.4a), shall be not greater than 2RM (see B.9.4b and note 5).
The diameter of a curve is given by the largest distance between any two points of the curve.
b) Varying principal stress directions (see note6)
For each point P on the stress-concentration-free construction, a stress state
i,j
(P, t
o
) corresponding to
a "point in time" of the action cycle (when the conditions are one of the extremes for the cycle) shall be
selected.
For each (arbitrary) choice of t
o
and corresponding
i,j
(P, t
o
) the diameter of the deviatoric mapping of
the path of the stress differences
i,j
(P, t) -
i,j
(P, t
o
), determined for each point of the stress-
concentration-free construction by means of linear elasticity theory for the action cycle, shall not be
greater than 2RM.
In the deviatoric mapping of the path of the stress states, the identity of each principal stress axis shall be
maintained throughout the whole action cycle.
NOTE1: The requirement given in this application rule is only a necessary condition for the fulfilment
of the principle, but, together with the requirements in B.9.2 against GPD, it is considered also to be
sufficient to achieve the principle's goal.
NOTE2: A local structural discontinuity is a geometric or material discontinuity which affects the stress
or strain distribution through a fractional part of the wall thickness. The stress distribution associated
with a local discontinuity causes only very localised types of deformation or strain and has no
significant effect on the gross structural deformations. Examples are small fillet radii, small
attachments and partial penetration welds.
NOTE3: This plot is known as a deviatoric map. The path obtained marks the variation of the effect
(with time) or another quantity on the deviatoric map.
NOTE4: The equivalent stress-concentration-free construction is an idealised construction without
local structural discontinuities.
NOTE5:In cases where the stress path under investigation corresponds to a cyclic action superposed
on a non-negligible permanent (external) action, the assessment shall also be made using the zero
stress point (0,0,0) on the deviatoric map.
NOTE6: For a case in which the directions of the principal stresses at the point being considered
change during the action cycle, a more general procedure is required.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.77
DBA
Design by Analysis
1
2 3
1
2
3
120
120
Fig. B.9-9: Deviatoric mapping of a stress state
B.9.3.3 Particular requirements
In a design check against PD, cladding shall be considered in both thermal analysis and stress analysis.
However, when the cladding is of the integrally-bonded type and its nominal thickness is not more than 10%
of the total thickness of the component, the cladding may be neglected.
B.9.3.4 Design checks
a) Action cycle
Characteristic values of permanent actions, and characteristic values or functions of pressure-temperatures,
shall be combined with the most unfavourable variable action in an action cycle, which shall encompass all
reasonably foreseeable combinations.
b) Design material strength parameters
1) Steel other than austenitic steel as per 6.3 and 6.4 for structural parts other than bolts
RM is given by the arithmetic mean of R R
eH p0,2/ t
or , for the highest and lowest calculation temperatures
at the position under consideration during the whole action cycle.
2) Austenitic steels as per 6.3 and 6.4
RM is given by the arithmetic mean of R
p1,0/ t
for the highest and the lowest calculation temperatures at the
position under consideration during the whole action cycle.
3) Steel castings
RM is given by the arithmetic mean of R R
eH p0,2/ t
or , for the highest and lowest calculation temperatures
at the position under consideration during the whole action cycle.
B.9.4 Instability (I)
B.9.4.1 Principle
The principle comprises the following.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.78
DBA
Design by Analysis
The design resistance to external pressure is the lower bound of the expected range of failure pressures
based on experimental observation. The experiments shall include the effect of shape deviations. The results
will normally be correlated by a theoretical model with an experimentally determined reduction factor. Such a
theoretical model will cover buckling failure in the elastic range and comparison of a calculated stress with
yield stress, and may include the effect of shape imperfections. Tolerances on the design shape shall ensure
that imperfections are kept within the range covered by the experimental data.
The partial safety factor for the resistance is 1,25 provided that the pressure test (external) as called for in
Part 5 is to be carried out. In the absence of such a pressure test, the partial safety factor shall be 1,5.
B.9.4.2 Application rule
The application rule comprises the following.
Fulfilment of the requirements given in clause 8 suffices as a check against stability for pressure action.
B.9.4.3 Particular requirements
In design checks against instability, no structural strength shall be attributed to the cladding.
B.9.5 Fatigue failure (F)
B.9.5.1 Principle
The principle is stated as follows.
The design value of the damage indicator Dd, obtained for the design functions of pressure / temperature
and variable actions shall not exceed unity.
B.9.5.2 Application rule
Fulfilment of the requirements given in clause 18 suffices as a check against fatigue loading.
B.9.5.3 Particular requirements
In a design check against fatigue, cladding shall be considered with respect to both thermal analysis and
stress analysis. However, when the cladding is of the integrally-bonded type and the nominal thickness of
the cladding is not more than 10% of the total thickness of the component, the presence of the cladding may
be neglected.
B.9.6 Static equilibrium (SE)
B.9.6.1 Principle
The principle is stated as follows.
The design effect of the destabilising actions shall be smaller than the design effect of the stabilising actions.
B.9.6.2 Application rule
The application rule is the same as the principle.
B.9.6.3 Particular requirements
None
B.9.6.4 Design checks
a) Partial safety factors against actions shall be as given in table B.9-2.
Annex 2:
Draft CEN prEN 13445-3
Page
A 2.79
DBA
Design by Analysis
For the verification of static equilibrium, stabilising (favourable) actions shall be represented by lower design
values and destabilising (unfavourable) actions by upper design values.
Permanent actions shall be represented by appropriate design values, depending on whether the stabilising
and destabilising effects result from:
- the favourable and unfavourable part of a single permanent action and/or
- different permanent actions.
The self-weight of any unrelated structural or non-structural elements made of different construction
materials should be treated as a separate permanent action.
The self-weight of a homogeneous structure should be treated as a single permanent action consisting of
separate favourable and unfavourable parts.
The self-weight of essentially similar parts of a structure (or of essentially uniform non-structural elements)
may also be treated as separate favourable and unfavourable parts of a single permanent action.
b) Combination rules shall be as follows.
For stabilising effects, only those actions, which can reliably be assumed to be present in the situation
considered, shall be included in the relevant combination.
Variable actions shall be applied where they increase the destabilising effects but omitted where they would
increase the stabilising effects.
Account shall be taken of the possibility that non-structural elements might be omitted or removed.
The favourable effect of variable action shall not be taken into account.
Where uncertainty of a value of a geometrical dimension significantly affects the verification of static
equilibrium, this dimension shall be represented in this verification by the most unfavourable value that it is
reasonably possible for it to reach.
c) RM and
R
shall be as given in table B.9-3.
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.80
Annex C (informative)
STRESS CATEGORISATION ROUTE FOR DESIGN BY ANALYSIS
(This annex refers to clause 5)
C.1 Purpose
This annex gives rules concerning design by analysis using stress classification. It applies to pressure vessels in all
testing groups.
The method described, known as stress analysis , involves the interpretation of stresses calculated on an elastic
basis at any point in a part of a vessel, and then verification of their admissibility by means of appropriate assessment
criteria.
It applies to pressure vessels in all Testing Groups.
It may be used:
- as an alternative to design-by-formula(see 5.4.1),
- as a complement to design-by-formula for:
- cases not covered by that route;
- cases involving superposition of environmental actions;
- where DBA is required, e.g. by local authorities where a potential major hazard is involved or for
environmental reasons;
- exceptional cases where the manufacturing tolerances given in clause 5 of Part 4 are exceeded.
NOTE 1: In the last item, any deviation beyond tolerance limits shall be clearly documented.
- as an alternative to the design-by-analysis direct route, according to annex B.
It may be used for a component or even a part of a component.
In all cases, all relevant requirements of this annex shall be fulfilled for that component or part.
The minimum thickness for pressure loading only, shall not be less than required by 7.4.2 for cylindrical shells, 7.4.3
for spherical shells, 7.5 for dished ends, and 7.6.4 for conical shells.
Fatigue failure is not covered by this annex. When required, fatigue assessment shall be performed according to
clause 18.
Failure by elastic or elastic-plastic instability (buckling) is not covered by this annex. When the analysis reveals
significant compression stresses, the risk for buckling must be assessed separately.
These rules do not apply in the range of temperatures where there is a risk of failure by creep-rupture, i.e. when the
value of the nominal design stress is governed by the creep characteristics of the material.
It is presupposed that this annex will be used with conformity assessment modules G, B1+D, or B1+F.
C.2 Specific definitions
The following definitions are in addition to those in clause 3 :
C.2.1 gross structural discontinuity: a structural or material discontinuity which affects the stress or strain
distribution across the entire wall thickness over a region of significant area.
NOTE: Examples of gross structural discontinuities are end-to-cylindrical shell or conical shell-to-
cylindrical shell junction, flange-to-cylindrical shell junction, an opening in a shell, the junction of two
cylindrical shells of different diameter, thickness or material, or a stiffener-to-shell junction.
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.81
C.2.2 local structural discontinuity: a discontinuity which only very locally affects the stress or strain
distribution, across a fraction of the thickness of the wall.
NOTE 1: Stresses resulting from such a discontinuity can only cause highly localized strains and
consequently have no significant influence on the global behaviour of the wall.
NOTE 2: Examples of local structural discontinuities are: small radius fillets, weld toes, non penetrated
zones in partial penetration welds.
C.2.3 primary stresses: stresses which satisfy the laws of equilibrium of external forces and moments
(loads).
NOTE 1: Regarding the mechanical behaviour of a structure, the basic characteristic of this type of
stress is, that in case of high (non admissible) increment of external loads, the deformations upon full
plastification of the section considerably increase without being self-limiting.
NOTE 2: Regarding primary stresses distinction shall be made between membrane stresses (P
m
, P
L
)
and bending stresses (P
b
) with respect to their distribution across the cross-section governing the load-
bearing behaviour. Primary membrane stresses (P
m
) are defined as the average value of the
respective stress components distributed over the section governing the load-bearing behaviour
defined by the supporting line segment (see C.4.4). Primary bending stresses (P
b
) are defined as
primary stresses distributed linearly across the considered section and proportionally to the distance
from the neutral axis.
NOTE 3: Regarding the distribution of membrane stresses along the wall, distinction shall be made
between general primary membrane stresses (P
m
) and local primary membrane stresses (P
L
). At
discontinuities, primary membrane stresses in shells are classified as local if the equivalent membrane
stress exceeds 1,1 times the nominal design stress f and if the region in which this value is exceeded
remains within the length of 10 , R e
a
in the meridional direction. Two adjacent regions of local
primary membrane stresses which exceed 1,1 times the nominal design stress f must be at a distance
of at least 25 , R e
a
in meridional direction. Here, R is the mid-surface radius of curvature and e
a
the
wall analysis thickness.
Discrete regions of local primary membrane stresses, (e.g. those resulting from concentrated loads
acting on brackets), where the equivalent membrane stress exceeds 1,1 times the nominal design
stress f, shall be spaced so that there is no overlapping of these regions.
NOTE 4: General primary membrane stresses are distributed in the structure such that no essential
redistribution of load occurs as a result of yielding. In the case of local primary membrane stresses,
yielding will cause such redistribution of loads.
C.2.4 secondary stresses : stresses developed by constraints due to geometric discontinuities, by the
use of materials of different elastic moduli under external loads, or by constraints due to differential thermal
expansions.
NOTE 1: Only stresses that are distributed linearly across the cross-section are considered to be
secondary stresses. For non linearly distributed stresses, the secondary stresses are those of the
equivalent linear distribution.
NOTE 2: With respect to the mechanical behaviour of the structure, the basic characteristic of
secondary stresses is that they lead to plastic deformation when equalizing different local distortions in
the case of excess of the yield strength. Characteristic for a secondary stress is that it is self-limiting,
i.e. local flow deformation leads to a limitation of the stress.
NOTE 3: Secondary stresses may be of membrane type (Q
m
) or bending type (Q
b
). Yet, in most
cases, distinction between both is not necessary, because criterion 5C.7.3 requires only consideration
of their sum (Q
m
+ Q
b
). Satisfaction of another criterion which needs separate consideration of the
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.82
secondary membrane stress (Q
m
) is only necessary when instability phenomena are likely to occur
(see note 2 to Table C-2).
C.2.5 peak stress : that part of stress which is additive to the respective primary and secondary
stresses, to form the total stress.
NOTE 1: Peak stresses do not cause any noticeable distortion and are only important to fatigue and
brittle fracture in conjunction with primary and secondary stresses.
NOTE 2: Peak stresses also comprise deviations from nominal stresses at hole edges within tube-hole
fields due to pressure and temperature, in which case the nominal stresses shall be derived from
equilibrium of forces considerations.
C.3 Specific symbols and units
Table C-1 : Symbols, descriptions and units
Symbol Description Unit

ij
stress tensor components of a given stress category, due to an
individual load.
N/mm

ij
stress tensor components of a given stress category, due to
superposition of all loads acting simultaneously (at a given instant)
N/mm

1
,
2
,
3
principal stresses of the tensor of components
ij
N/mm

eq
equivalent stress according to the shear stress theory (Tresca) N/mm

ij
tensor of stress differences between two loading conditions N/mm
()
1
,()
2
,()
3
principal stresses of the tensor of components
ij
N/mm

eq
equivalent stress range according to the shear stress theory (Tresca) N/mm
h length of the supporting segment mm
P
m
general primary membrane stress N/mm
P
L
local primary membrane stress N/mm
P
b
primary bending stress N/mm
Q secondary membrane + bending stress. N/mm
Q
m
secondary membrane stress N/mm
Q
b
secondary bending stress N/mm
F peak stress N/mm
R mean radius for the region, measured perpendicular to the shell wall mm
C.4 Representative stresses
C.4.1 Equivalent stress
The equivalent stress
eq
is a scalar quantity defined in accordance with the maximum shear stress theory
from the stress tensor of components
ij
, obtained by summation of all stresses
ij
of same category
generated by the various loads to be considered simultaneously.
The determination of the equivalent stress shall be performed as follows:
a) Calculate the principal stresses
1
,
2
,
3
of the tensor of components
ij
.
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.83
b) The equivalent stress is given by the relation:

eq
= max {
1

2
,
2

3
,
3

1
} ...(C.4.1-1)
C.4.2 Equivalent stress range
The equivalent stress range
eq
is a scalar quantity defined in accordance with the maximum shear stress
theory, from the variation of the stress tensor of components
ij
between two normal operating conditions.
The determination of the equivalent stress range shall be performed as follows:
a) Calculate the values (
ij
)
a
and (
ij
)
b
of the components of the stress tensor
ij
for loading conditions a
and b respectively.
b) Calculate the components
ij
of the tensor which represents the stress variation between loading
conditions a and b:

ij
= (
ij
)
a
(
ij
)
b
...(C.4.2-1)
c) Calculate the principal stresses ()
1
, ()
2
, ()
3
of the tensor of components
ij
.
In the case where the principal directions of the tensor
ij
are the same in both conditions a and b, these
principal stresses may be directly calculated from the difference between the principal stresses of the
tensors of components (
ij
)
a
and (
ij
)
b
:
()
1
= (
1
)
a
(
1
)
b
()
2
= (
2
)
a
(
2
)
b
...(C.4.2-2)
()
3
= (
3
)
a
(
3
)
b
NOTE : principal stress ranges may be used directly for fatigue assessment (see clause 18)
d) The equivalent stress range between loading conditions a and b is given by the relation:

eq
= max {()
1
()
2
,()
2
()
3
,()
3
()
1
} ...(C.4.2-3)
NOTE: Criterion C.7.3-1 requires that the maximum value of
eq
be found. When more than one load
are applied and vary independently, and/or when principal directions change, identification of the two
load conditions a and b that maximize
eq
may be difficult ; a trial and error calculation process may
there be necessary.
C.4.3 Total stress - elementary stresses
The total stress tensor is the symmetrical tensor whose components are the six elementary stresses
ij
determined on an elastic basis by mean of a calculation or experimental method in accordance with the
requirements of C.4.5.
These stresses shall be expressed in a set of local coordinates designated O, X
1
, X
2
, X
3
and referenced to
the supporting line segment defined in C.4.4.
Axis X
3
is that containing the supporting line segment; the origin O is located at the mid-point of the
supporting line segment; the abscissa of a point of this segment is designated x
3
(see figure C-1).
The total stress tensor shall be determined, at a given point, for each load which has to be taken into
account.
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.84
1 : supporting line segment
2 : gross structural discontinuity
Figure C-1 : Supporting line segment and local axes
in which elementary stresses are expressed.
C.4.4 Decomposition of stresses
C.4.4.1 Supporting line segment
The decomposition of the elementary stresses, outlined hereafter, shall be carried out across the wall
thickness along a segment which is referred to as the supporting line segment.
The supporting line segment, of length h, is the smallest segment joining the two sides of the wall (see
figure C-1). Outside of gross structural discontinuity regions, the supporting line segment is normal to the
wall mean surface; its length h, is then equal to the analysis thickness of the wall.
C.4.4.2 Membrane stress
The membrane stress tensor is the tensor whose components
ij
,
m
, constant along the supporting line
segment, are equal to the average value of the elementary stresses
ij
along this supporting line segment:
( )

ij ij h
h
h
dx
m
=

1
3
2
2
...(C.4.4-1)
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.85
C.4.4.3 Bending stress
The bending stress tensor is the tensor whose components
ij
,
b
, varying linearly across the thickness
of the wall, are given by the formula:
( )

ij ij h
h
x
h
x dx
b
=

12
3
3
3 3
2
2
...(C.4.4-2)
For a stress analysis in accordance with this annex, only maximal values of
ij
,
b
equal and of opposite sign
on each side of the wall, i.e. at both ends of the supporting line segment, shall be considered. For this case:
( )

ij ij h
h
h
x dx
b
=

6
2
3 3
2
2
...(C.4.4-3)
C.4.4.4 Linearized stress
The linearized stress tensor is the tensor whose components
ij
,
l
are equal to:

ij
,
l
=
ij
,
m
+
ij
,
b
...(C.4.4-4)
C.4.4.5 Nonlinearity stress
The nonlinearity stress tensor is the tensor whose components
ij
,
nl
are equal to:

ij
,
nl
=
ij

ij
,
l
=
ij
[
ij
,
m
+
ij
,
b
] ...(C.4.4-5)
Figure C-2 shows the decomposition of the elementary stresses outlined above. In order to avoid possible
confusion between global and local bending stresses, an example of application of the stress decomposition
to the particular case of longitudinal stresses in a cylindrical shell subjected to an external bending moment
is illustrated in figure C-3.
1 : supporting line segment
2 : membrane stress
ij
,
m
3 : bending stress
ij
,
b
4 : nonlinearity stress
ij
,
nl
Figure C-2 : Decomposition of elementary stresses
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.86
1 : longitudinal stress distribution along the shell cross section
2 : longitudinal stress distribution along the thickness of the wall
3 : membrane stress:

22
16
,
( )
( )
m
e i
e
4
i
4


=
+

M D D
D D
4 : bending stress:

22
16
,
( )
( )
b
e i
e
4
i
4


=

M D D
D D
(on each side of the wall)
Figure C-3 : Decomposition of the longitudinal stress
for the particular case of a cylindrical shell subject to an external bending moment M
(for this particular case, the longitudinal stress
22
is a principal stress)
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.87
C.4.5 Requirements relating to the methods for determining stresses
C.4.5.1 Assumption of linear elasticity
Elementary stresses shall be determined in accordance with the assumptions of linear elasticity:
- material behaviour is linear elastic in accordance with Hookes law
- material is isotropic
- displacements and strains are small (first order theory).
C.4.5.2 Selection of methods for determining stresses
The choice of the method used for determining stresses is under the responsibility of the manufacturer.
This method may be numerical, analytical or possibly experimental.
The following requirements relate only to methods for determining stresses by calculation.
When the vessel studied is built of components which can be classified as shells and plates, calculation
methods that describe the state of these components using global mechanical parameters (i.e. generalised
deformations and stress resultants in a section, corresponding to linear strain and stress distribution across
the thickness of the wall) are generally acceptable.
This is certainly so for :
- vessels for which a fatigue analysis in accordance with clause 18 is not required,
- vessels or vessel parts for which such an analysis is required but does not necessitate evaluation of
peak stresses (e.g. all cases where the critical fatigue zones are located in welded joints),
- vessels or vessel parts for which evaluation of peak stresses for use in clause 18 can be carried out
using suitable stress concentration factors, applied to the linearized stresses derived from these
methods.
The analysis of thick wall vessels or of thick parts of vessels, particularly under thermal loads, may require
the use of refined models (two or three dimensional continuous medium permitting analysis of actual non-
linear stress or strain distributions across the thickness of the wall).
In all cases, accuracy or conservatism of the methods used shall be adequate to ensure good
representativeness of the calculated stresses with regard to those required for the analysis. In this respect,
the use of tested and recognized practices is recommended.
C.5 Classification of stresses
Stresses determined by analysis shall be classified in accordance with the different categories whose
definitions are given in C.2. In some cases, interpretation of these definitions may be problematical and, to
a large extend, depends on the analysts judgement.
In order to limit this difficulty, Table C-2 prescribes the classification to be used for a certain number of
configurations covering most of the common cases.
Information given in this table refers to stresses calculated in accordance with the requirements of C.4.5.
For the analysis of particular geometrical arrangements or loadings, for which the classifications proposed in
these tables would not be suitable, departure from them is permissible, so long as the alternative
classifications are justified by means of direct reference to the definitions given in C.2.
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.88
Table C-2 - Classification of stresses in some typical cases
ORIGIN OF STRESS
REGION TYPE
Mechanical loads
Thermal
VESSEL
COMPONENT
UNDER
CONSIDERATION
OF
STRESS
Pressure
Local load
acting in the
vicinity of the
point under
consideration
Other
mechanical
loads
1)
loads;
restrained or
imposed
displacement
s
Region far from

ij
,
m P
m
P
L
P
m
Q
m
2)
Cylindrical,
any gross structural
discontinuity

ij
,
b Q
b
3)
P
b
P
b
Q
b
spherical,
conical,
or toroidal
Vicinity of a junction
with another shell,
an end, a flange;

ij
,
m
P
L
P
L
P
L
Qm
shell vicinity of an opening
(with or without nozzle)
4) 5)

ij
,
b
Q
b
P
b
4)
Q
b
Q
b
Central region
outside

ij
,
m P
m
P
L
P
m
Q
m
2)
Torispherical
the vicinity
of an opening

ij
,
b P
b
P
b
P
b
Q
b
or ellipsoidal
dished end
Central region
in the vicinity of
an opening (with or

ij
,
m
P
L
P
L
P
L
Q
m
without nozzle)
4) 5)
;
peripherical region
6)

ij
,
b Q
b
P
b
4)
Q
b
Q
b
Region far from any
gross structural
discontinuity;

ij
,
m
P
m
P
m
P
m
Q
m
2)
Flat end,
plane wall
vicinity of an
opening (with or
without nozzle)
4)

ij
,
b
P
b
P
b
4)
P
b
Q
b
Vicinity of edges

ij
,
m P
m
P
m
P
m
Q
m
or of a stiffener
ij
,
b Q
b
P
b
Q
b
Q
b
Isolated

ij
,
m
8)
P
L
or P
m
7)
P
L
or P
m
7)
P
L
or P
m
7)
Q
m
Perforated
ligament
ij
,
b
8)
Q
b
P
b
Q
b
Q
b
wall
(shell or plate)
Ligament
in a multiple

ij
,
m
8)
P
m
P
m
P
m
Q
m
9)
and close
perforation region

ij
,
b
8)
P
b
P
b
P
b
Q
b
Region far
from junction

ij
,
m P
m
P
L
P
m
Q
m
2)
to vessel wall
ij
,
b Q
b
3)
P
b
P
b
Q
b
Nozzle
Vicinity of the
junction to a shell

ij
,
m P
L
P
L
P
L
Q
m
or a dished end
5)

ij
,
b Q
b
P
b
Q
b
Q
b
Vicinity of the
junction to a flat end

ij
,
m P
L
P
L
P
L
Q
m
or plane wall
10)

ij
,
b P
b
or Q
b
11)
P
b
Q
b
Q
b
Notes
1)
to
11)
: see next page
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.89
Notes to Table C-2 :
1) The other mechanical loads are the overall mechanical loads acting on whole vessel, and the local loads
acting far from the point under consideration (a local load acts far from a point when this point is located
outside the region of local primary membrane stress possibly generated by this load).
2) For regions far from gross structural discontinuities, the classification of membrane stresses due to
thermal loads or to restrained or imposed displacements in category Q
m leads to plastic deformations
occurring in these regions during the early loading cycles, at any point where the equivalent primary +
secondary membrane stress is greater than the yield stress of the material.
With regard to the failure modes covered by the rules of this annex, the strength of the vessel is not affected
by these plastic deformations; however, due to these deformations, the use of stresses calculated on an
elastic basis is not correct in assessing the risk of elastic or elastic-plastic instability (buckling).
Consequently, if there are regions of the vessel where this risk of instability shall be considered and if this
risk may be increased by the redistribution of stresses associated with the plastic deformations mentioned
above, such plastic deformations shall not be permitted.
This requirement is met by ensuring that, in the regions far from any gross structural or loading
discontinuity, the equivalent primary + secondary membrane stress (
eq
)
(P+Q)m (equivalent stress
corresponding to [(
ij
)
P
m
or (
ij
)
P
L
] + (
ij
)
Q
m
) satisfies the relationship:
(
eq
)
( P+Q)
m
1,5 f ...(C.6-1)
3) P
b
when the shell is not axisymmetric (example: oblique conical shell, cylinder of elliptic cross section).
4) For an opening with nozzle, the nozzle loads acting on the shell come under the heading other
mechanical loads, whether they are due to purely mechanical loadings (self-weight) or thermal loadings
(thermal expansion).
5) See figure C-4.
6) For a torispherical end, although there are two different peripherical discontinuities (spherical shell-
toroidal shell and toroidal shell-cylindrical shell junctions), the stress distribution is generally such that only
one single local primary membrane stress region occurs in the knuckle.
Where relative dimensions and thicknesses of spherical, toroidal, and cylindrical components are such that
two such regions occur, the classification given here for the peripherical region applies in the vicinity of
each discontinuity; the intermediate region is to be classified as region far from any gross structural
discontinuity and the rules relating to the spacing of local primary membrane stress regions shall be
satisfied.
7) P
m
for a flat wall. P
L
for a wall which is not flat.
8) For this particular case, the stress value to be retained is the average value across the ligament width.
9) The effect of the perforations shall be taken into account in stress calculation.
10) For an opening with nozzle in a flat end or a flat wall, for which the concept of local primary membrane
stress region has no meaning, the meridional extent of the local primary membrane stress region which
may occur at the nozzle base shall, for the nozzle, be measured from the outside surface of the end or of the
wall.
11) P
b
when the strength of the nozzle is taken into account for the calculation of stresses acting in the flat
end or the flat wall; if not, Q
b
. The first solution, conservative for the nozzle, is only interesting in practice if
taking into account the strength of the nozzle leads to a significant decrease of the flat end or flat wall
thickness.
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.90
1 : shell
2 : nozzle
3 : nozzle base region
4 : thickness transition regions
5 : limits of local primary membrane stress region
This figure shows the case of an opening with nozzle in a cylindrical shell. It also applies to the case of an
opening with nozzle in a spherical, conical or toroidal shell or in the central region of a dished end (R
m
is
the circumferential mean curvature radius) It applies as well to the case of an opening without nozzle (for
this case e
a,n
= 0).
It does not apply to an opening in a flat end or a flat wall; in such cases, see note 10 of Table C-2.
When the level of stress acting in the vicinity of the opening is such that a local primary membrane stress
region occurs, the extent of this region, measured taking account of both sides of the nozzle-shell
discontinuity, shall satisfy the condition:
l l
R e r e
s n
m a,s m a,n
+
+ . .
2
...(C.6-2)
Possible thickness transitions which may occur between a reinforced part and an unreinforced part of the
nozzle and/or of the shell do not usually involve local primary membrane stress regions.
Where, for particular geometrical or loading arrangements, such regions occur in the vicinity of these
transitions, the conditions in C2.3 (local primary membrane stress) relating to local regions shall be met,
particularly the condition relating to the spacing of these regions in relation to the adjacent local primary
membrane region at nozzle base.
Figure C-4 : Opening in a shell
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.91
C.6 Stress analysis procedure
The procedure to be followed for a stress analysis is the following:
- Step 1: For each point of the region under study, calculate the elementary stresses resulting from each
load acting on the vessel wall for each loading condition to be considered.
These calculations shall be carried out in accordance with the requirements C.4.5.
The loading conditions to be considered are:
- the loading conditions of all types (normal operation, exceptional operation, proof test) for which the
stress level may be determinant through assessment criteria C.7.2 (step 7).
- the normal operating conditions between which the stress variation may be determinant through the
assessment criteria C.7.3-1 (step 9).
- Step 2: Decompose the stresses
ij
calculated above, in accordance with the requirements of C.4.4, into:
- membrane stress:
ij
,
m
,
- bending stress:
ij
,
b
.
The bending stress to be taken into account for the analysis is the stress on both sides of the wall i.e., at
the two ends of the supporting line segment (two equal values with opposite signs).
- Step 3: In accordance with the directives of C.5, classify these stresses into the different categories
defined in C.2:
- general primary membrane stress (P
m
),
- local primary membrane stress (P
L
),
- primary bending stress (P
b
),
- secondary membrane stress (Q
m
),
- secondary bending stress (Q
b
).
Following this classification the stress
ij
,
m
is designated (
ij
)
P
m
, (
ij
)
P
L
, or (
ij
)
Q
m
, and the stress
ij
,
b
is
designated (
ij
)
P
b
or (
ij
)
Q
b
.
- Step 4: Calculate the sum of the stresses classified in this way for the set of loads acting simultaneously in
the loading condition under consideration.
Stresses resulting from this summation are designated: (
ij
)
P
m
, (
ij
)
P
L
, (
ij
)
P
b
,
(
ij
)
Q
m
, (
ij
)
Q
b
- Step 5: From this, deduce:
a) the primary membrane stress, general or local (depending on the point under consideration):
(
ij
)
P
m

or (
ij
)
P
L
.
b) the total primary stress (
ij
)
P
:
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.92
(
ij
)
P
= [(
ij
)
P
m
, or (
ij
)
P
L
] +

(
ij
)
P
b
...(C.6-1)
c) the primary + secondary stress (
ij
)
P+Q
:
(
ij
)
P+Q
= [(
ij
)
P
m
, or (
ij
)
P
L
] +

(
ij
)
P
b

+ (
ij
)
Q
m
+

(
ij
)
Q
b
...(C.6-2)
- Step 6: According to C.4.1 calculate the following equivalent stresses:
- (
eq
)
P
m
, equivalent to stresses (
ij
)
P
m
, or, depending on point under consideration, (
eq
)
P
L
,
equivalent to stresses (
ij
)
P
L
,
- (
eq
)
P
, equivalent to stresses (
ij
)
P
- Step 7: Verify the admissibility of these equivalent stresses with respect to criteria in C.4.2.
- Step 8: For each set of two normal operating loading conditions which may be determinant, calculate the
range of the primary + secondary stress (
ij
)
P+Q
and then, as indicated in C.4.2, calculate the
corresponding equivalent stress range (
eq
)
P+Q
.
The set of loading conditions to be retained is that which results in the greatest value of (
eq
)
P+Q
.
- Step 9: Verify the admissibility of the equivalent resulting stress range (
eq
)
P+Q
with respect to criteria
C.5.3.
C.7 Assessment criteria
C.7.1 General
The whole design shall basically meet the stress criteria given in C.7.2 and C.7.3. These criteria are
illustrated diagrammatically in Table C-2.
Relaxation of criterion C.7.3-1 is possible in some cases, under the conditions given in C.7.4.
When compressive stresses occur, buckling shall be assessed. For external pressure, see applicable rules
in clause 8.
NOTE: Functional requirements may set limitations on the allowable deformations.
C.7.2 Limitation of equivalent primary stresses
The equivalent primary membrane stresses shall for all loading conditions satisfy the relationships:
(
eq
)
Pm
f ...(C.7.2-1)
(
eq
)
P
L

1,5 f ...(C.7.2-2)
(
eq
)
P
1,5 f ...(C.7.2-3)
The value of f to be retained shall be that consistent with the type of loading condition considered (normal
operation, exceptional operation, proof test), and shall be taken at the calculation temperature of that
condition.
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.93
Table C-3 : Illustration of assessment criteria
Stress Categories
Primary stress Secondary
General
membrane
stress
Local
membrane
stress
Bending
stress
membrane + + bending
stress
Peak stress
Description
(For practical
examples, see
Table C-2)
Primary mean
stress calculated
across the wall
thickness without
taking into
account
discontinuities
and stress
concentrations.
Caused only by
mechanical loads.
Primary mean
stress calculated
across the wall
thickness taking
into account large
discontinuities, but
not stress
concentrations.
Caused only by
mechanical loads.
Primary stress
component
proportional to the
distance from the
controid of the solid
wall section. Does
not include
discontinuities
and stress
concentrations.
Caused only by
mechanical loads
Self-equilibrating stress
necessary to satisfy the
continuity of the structure.
Occurs at large
discontinuities, but does
not include stress
concentrations.
Can be caused by both
mechanical loads and
thermal effects.
a) Addition to
primary or
secondary
stress because
of stress
concentration.
b) Certain thermal
stresses which
may cause
fatigue, but not
distortion.
Symbol P
m
P
L
1) P
b
Q
(Q
m
+ Q
b
)
F
(eq)
Pm
f
(eq. C.7.2-1)
2)
assessment
againts
static loading
(eq)
PL
1,5f
(eq. C.7.2-2)
(eq)
P+Q
3 f
(eq. C.7.3-1)
3)
7)
_______
= design loads
= operating
loads
(eq)
P
1,5 f
(eq. C.7.2-3)
2)
fatigue
assessment
(only if
required)
Assessment
4)
based on :
5)
7)
(eq)
P+Q
or
max (
i
)
or (eq)
P+Q+F
6)
7)
1) P
L
= P
m
does not occur at the point in question.
2) In assessment criteria (C.7.2-1) to (C.7.2-3), the value of the nominal design stress f shall be that relevant for the loading
condition under consideration (normal operation, exceptional operation, proof test), as defined in clause 6.
3) If (
eq
)
P+Q
is greater than 3f, see C.7.6
4) Fatigue assessment shall consider all the applied cycles of various ranges, each of them being characterized by its own
relevant stress range (see notes 5 and 6). Clause 18 (detailed fatigue assessment) shall normally be used.
5) The primary + secondary stress range (named structural stress range in clause 18 on detailed fatigue assessment)
applies to assessment of welded joints. In that case, either the equivalent stress range (
eq
)
P+Q
or the maximum principal
stress range max(
i
) may be used.
6) The primary + secondary + peak stress range (named notch stress range in clause 18 on detailed fatigue assessment)
applies to assessment of unwelded parts.
7) It should be observed that, depending on the model used, the computer programs usually give directly the primary + secondary
stresses (P + Q) or the primary + secondary + peak stresses (P + Q + F).
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.94
C.7.3 Limitation of equivalent stress ranges resulting from primary + secondary stresses
The equivalent stress range resulting from variation of primary + secondary stresses between two any
normal operating conditions shall at all points satisfy the relationship:
(
eq
)
P+Q
3 f ...(C.7.3-1)
The value of f to be retained shall be that corresponding to loading conditions of normal operating type,
and shall be taken at the following temperature:
t* = 0,75t
max
+ 0,25t
min
...(C.7.3-2)
where t
max
and t
min
are respectively the higher and the lower of the calculation temperatures of the two
normal operating conditions considered.
C.7.4 Alternative to limitation of equivalent stresses and equivalent stress ranges
Deviations from the preceding limitations of equivalent stresses and equivalent stress ranges are possible if
it is proved by other means that the component meets the required safety margin against gross plastic
deformation and progressive plastic deformation stated in annex B (e.g. by tests on the component, plastic
analysis, or the like).
C.7.5 Limitation of primary stresses in case of tri-axial state of stress
Where the stress analysis leads to a tri-axial state of stress, the following condition shall be satisfied
additionally whenever the smallest tensile principal stress exceeds half the highest tensile principal stress,
to avoid brittle failure caused by the limited ductility in such stress states:
max (
1
;
2
;
3
) R
p/t
...(C.7.5-1)
where R
p/t
is the value of the yield stress relevant for determination of f (either R
p0,2/t
or R
p1,0/t
) at
calculation temperature.
This value can be exceeded, if it is shown by a fracture mechanics analysis that higher values can be
accepted.
C.7.6 Simplified elastic-plastic analysis
The equivalent stress range resulting from variation of primary + secondary stresses between two normal
operating conditions is allowed to exceed 3f on condition that
a) (
eq
)
P+Q
3 f ...(C.7.6-1)
where (
eq
)
P+Q
is the equivalent same stress range, calculated without taking into account bending
stresses of thermal origin.
b) a detailed fatigue analysis according to clause 18 is performed. In this analysis, (
eq
) shall be
multiplied by the appropriate plasticity correction factor, as determined from that annex (Detailed
assessment of fatigue life).
c) the material is such that R
p
< 0,8 R
m
, R
p
being here the value of the yield stress relevant for
determination of f (either R
p0,2
or R
p0,1
) at room temperature.
d) the absence of risk of incremental collapse by thermal stress ratchet in regions of general primary
membrane stress is established according to C.7.7.
DBA
Design by Analysis
Annex 2: Annex C
Draft CEN prEN 13445-3
Page
A2.95
C.7.7 Prevention of incremental collapse resulting from thermal ratcheting
C.7.7.1 General
The thermal ratcheting phenomena is the mechanism of incremental collapse which may occur in certain
conditions under the effect of cyclic thermal loads associated with a permanent pressure action.
It results in a plastic deformation which increases by about the same amount at each cycle and quickly
leads to an unacceptable value.
Meeting the criterion C.7.3-1 guarantees the absence of thermal ratcheting.
Paragraph C.7.7.3 provides a rule which, for the particular case of a linear thermal gradient, enables the
absence of thermal ratcheting to be guaranteed when the condition C.7.3 is not met.
This rule applies to the regions of general primary membrane stress. The absence of thermal ratcheting in
these regions ensures the absence of thermal ratcheting in discontinuity regions.
C.7.7.2 Specific parameters
(
eq
)
P
m
,P
is the equivalent general primary membrane stress due to pressure alone.
(
eq
)
(P+Q),T
is the equivalent primary + secondary stress range of the stress due to thermal load
C.7.7.3 Assessment criterion
For an axisymmetric shell under constant pressure and subject to a linear thermal gradient across the
thickness of the wall, there is no risk of failure by incremental collapse due to thermal ratcheting if, in
regions of general primary membrane stress, the following relationships are satisfied:
- for 0 x 0,5 :
y 1/x ...(C.7.7-1)
- for 0,5 x 1 :
y 4(1 x) ...(C.7.7-2)
where x and y are defined respectively as:
x
f
=
( )
,

eq Pm,P
15
...(C.7.7-3)
y
f
=
+
( )
,

eq (P Q),T
15
...(C.7.7-4)
The value of f to be retained shall be that applying to normal operating conditions, at the maximum
calculation temperature reached during the cycle.
Annex 3: Input Listings
Page
A 3.1
DBA
Design by Analysis
Group member A&AB, FE software ANSYS 5.4
A 3.1 Example 1.1
Model input file check against PD
/prep7
*set,ri,250.4 !inner radius shell!
*set,tz,101.6 !wall thickness shell!
*set,ra,(ri+tz) !outer radius shell!
*set,tp,101.6 !thickness plate!
*set,r,101.6 !fillet!
*set,lg,751.2 !model height!
*set,bf,20 !parameters for finer mesh at high stressed region!
*set,lz,(2*bf)
*set,lp,bf
*set,tz1,(3*bf)
*set,t1,8 !meshing parameters!
*set,t2,8
*set,t3,8
*set,t4,6
*set,t5,3
*set,t6,25
*set,e,((lg-tp-r-2*lz)/t6)
*set,t7,8
*set,t8,10
*set,t9,6
local,11,1,(ri-r),(tp+r)
csys,0
*set,hil,(ri-r)
k,1,hil,(tp-tz1) !modelling!
k,2,(ri+tz1),(tp-tz1)
k,8,(ri+tz1),(tp+r)
csys,11
k,3,(r+bf),-45
k,4,(r+bf),-90
k,5,r,-45
k,6,r,-90
k,7,(r+bf),0
k,9,r,0
l,4,3,t3
l,3,7,t3
l,6,5,t3
l,5,9,t3
csys,0
l,4,6,t1
l,3,5,t1
l,7,9,t1
l,1,2,t3
l,2,8,t3
l,2,3,t2
l,1,4,t2
l,8,7,t2
a,1,2,3,4
a,4,3,5,6
a,2,8,7,3
a,3,7,9,5
*set,hil2,(tz1-bf)
k,10,(ri+tz1),(tp+r+lz+hil2)
k,11,ri,(tp+r+lz+hil2)
k,12,ri,(tp+r+lz)
k,13,(ri+bf),(tp+r+lz)
k,14,(ri+tz1),lg
k,15,ri,lg
l,13,12,t1
l,10,11,t1
l,14,15,t1
l,9,12,t4
l,7,13,t4
l,8,10,t4
Annex 3: Input Listings
Page
A 3.2
DBA
Design by Analysis
l,12,11,t2
l,13,10,t2
l,11,15,t6
l,10,14,t6
a,7,13,12,9
a,7,8,10,13
a,10,11,12,13
a,10,14,15,11
k,16,(ri+tz1),0
k,17,ra,0
k,18,ra,(tp-tz1)
k,19,ra,(tp+r)
k,20,ra,(tp+r+lz+hil2)
k,21,ra,lg
l,2,16,t9
l,17,18,t9
l,16,17,t7
l,2,18,t7
l,8,19,t7
l,10,20,t7
l,14,21,t7
l,18,19,t3
l,19,20,t4
l,20,21,t6
a,2,16,17,18
a,2,18,19,8
a,8,19,20,10
a,10,20,21,14
k,22,(hil-lp),tp
k,23,(hil-lp),(tp-bf)
k,24,(hil-lp-hil2),(tp-tz1)
k,25,(hil-lp-hil2),tp
k,26,0,tp
k,27,0,(tp-tz1)
l,6,22,t5
l,4,23,t5
l,1,24,t5
l,23,24,t2
l,22,25,t2
l,22,23,t1
l,24,25,t1
l,26,27,t1
l,25,26,t8
l,27,24,t8
a,4,6,22,23
a,1,4,23,24
a,24,23,22,25
a,24,25,26,27
k,28,0,0
k,29,(hil-lp-hil2),0
k,30,hil,0
l,27,28,t9
l,24,29,t9
l,1,30,t9
l,28,29,t8
l,29,30,t5
l,30,16,t3
a,24,27,28,29
a,1,24,29,30
a,1,30,16,2
mp,ex,1,2.12e5 !material properties!
tb,bkin,1
tbdat,1,255,0
et,1,42,,,1 !axisymmetric 4-node element PLANE42!
asel,all
amesh,all
csys,0
nsel,s,loc,x,0 !boundary conditions!
dsym,symm,x
csys,0
nsel,s,loc,y,lg
d,all,uy
Annex 3: Input Listings
Page
A 3.3
DBA
Design by Analysis
allsel
save
Solution input file - check against PD
/solu
*set,pi,150 !maximum (theoretical) internal pressure!
lsel,s,,,21
lsel,a,,,19
lsel,a,,,16
lsel,a,,,4
lsel,a,,,3
lsel,a,,,33
lsel,a,,,37
lsel,a,,,41
sfl,all,pres,pi !apply internal pressure on selected lines!
allsel
arclen,on,1,0.001 !use the arc-length method!
nsubst,30 !first substep with 5 MPa!
arctrm,u,10,481,ux !termination criterion!
neqit,30
/output,out,txt !output file for debug purposes!
outres,all,all
solv
A 3.2 Example 1.2
Model Input file check against PD
/prep,7
*set,t1,8 !defining mesh parameters!
*set,t2,16
*set,t3,8
*set,t4,6
*set,t5,65
*set,t6,35
*set,t7,8
k,1,0,25.4 !modeling the geometry!
k,2,0,0
k,3,199.2,0
k,4,199.2,25.4
k,5,217,25.4
k,6,217,17.4
k,7,225,25.4
k,8,225,17.4
k,9,225,0
local,11,1,225,50.8
k,10,25.4,-45
k,11,33.4,-45
k,13,25.4,0
k,14,33.4,0
k,15,50.8,0
k,17,127,0
csys,0
k,12,275.8,0
k,16,352,0
k,18,250.4,58.8
k,19,258.4,58.8
k,20,250.4,76.2
k,21,275.8,76.2
k,22,352,76.2
k,23,250.4,476.2
k,24,275.8,476.2
k,25,352,476.2
l,23,24,t1
l,24,25,t2
l,23,20,t5
l,24,21,t5
l,25,22,t5
l,20,21,t1
l,21,22,t2
Annex 3: Input Listings
Page
A 3.4
DBA
Design by Analysis
l,15,17,t2
l,12,16,t2
l,16,17,t3
l,17,22,t4
l,15,21,t4
l,12,15,t3
l,19,21,t7
l,18,20,t7
l,14,19,t4
l,13,14,t1
l,13,18,t4
l,14,15,t7
csys,11
l,13,10,t3
l,10,7,t3
l,14,11,t3
l,11,8,t3
csys,0
l,10,11,t1
l,7,8,t1
l,8,9,t7
l,11,12,t7
l,9,12,t7
l,6,8,t4
l,5,7,t4
l,6,5,t1
l,3,4,t1
l,3,6,t7
l,3,9,t4
l,2,3,t6
l,1,4,t6
l,1,2,t1
l,4,5,t7
l,18,19,t1
al,1,3,6,4
al,4,7,5,2
al,6,15,39,14
al,16,19,12,14
al,12,8,11,7
al,8,13,9,10
al,27,13,19,22
al,28,27,23,26
al,34,26,29,33
al,33,31,38,32
al,29,25,30,31
al,17,16,39,18
al,17,20,24,22
al,23,24,21,25
al,32,36,37,35
et,1,42,,,1,,2 !axisymmetric 4-node element PLANE42!
mp,ex,1,212000 !material properties!
tb,bkin,1
tbdat,1,255,0
asel,all
amesh,all !meshing!
nsel,s,loc,x,0 !define boundary conditions!
dsym,symm,x
nsel,s,loc,y,476.2
d,all,uy,0
allsel
Solution Input file check against PD
/solu
*set,pi,100 !maximum (theoretical) internal pressure!
lsel,s,,,3
lsel,a,,,15
lsel,a,,,18
lsel,a,,,20
lsel,a,,,21
lsel,a,,,30
Annex 3: Input Listings
Page
A 3.5
DBA
Design by Analysis
lsel,a,,,38
lsel,a,,,36
sfl,all,pres,pi !apply internal pressure on selected lines!
allsel
arclen,on,1,0.001 !use the arc-length method!
nsubst,50 !first substep with 2 MPa!
arctrm,u,10,2487,uy
neqit,30
/output,out,txt !output file for debug purposes!
outres,all,all
solv
A 3.3 Example 1.3
Model input file check against PD
/prep7
*set,t1,38 !!defining mesh fineness parameters
*set,t2,100
*set,t3,10
*set,t4,8
*set,t5,12
*set,t6,12
k,1,0,0 !!modelling geometry
k,2,221.5,0
k,3,221.5,43
k,4,0,43
k,5,141,0
k,6,141,43
k,7,156,43
k,8,184,15
k,9,212,43
k,10,184,0
k,11,166,43
k,12,184,25
k,13,202,43
k,14,184,43
local,11,1,184,43 !!local CS
local,21,0,200,43 !!local CS
k,15,2,4.6575
k,16,0,16
k,17,0,20
k,18,0,30
k,19,0,400
k,20,21.5,400
k,21,21.5,30
k,22,12,20
csys,0
l,4,6,t1
l,1,5,t1
l,18,19,t2
l,20,21,t2
l,13,15
l,15,16
l,16,17
l,13,9,t3
l,9,3,t4
l,9,22,t5
l,3,21,t5
l,17,22,t3
l,22,21,t4
l,17,18,t4
l,19,20,t3
l,1,4,t6
csys,11
k,23,18,-45
k,24,18,-135
k,25,28,-135
k,26,28,-45
l,11,24,t6
Annex 3: Input Listings
Page
A 3.6
DBA
Design by Analysis
l,24,12,t6
l,12,23,t6
l,23,13,t6
l,7,25,t6
l,25,8,t6
l,8,26,t6
l,26,9,t6
csys,0
l,3,2,t6
l,23,26,t3
l,26,2,t4
l,2,10,t6
l,10,8,t4
l,8,12,t3
l,5,10,t6
l,24,25,t3
l,25,5,t4
l,6,7,t4
l,7,11,t3
l,5,6,t6
lfillt,5,6,10
al,1,16,2,36 !!areas plate (No 1 9)
al,36,33,21,34
al,21,32,17,35
al,31,29,22,33
al,22,30,18,32
al,28,27,23,29
al,23,26,19,30
al,27,25,9,24
al,8,20,26,24
l,18,21,10
al,8,10,12,7,6,37,5 !!areas shell (No 10 13)
al,9,11,13,10
al,13,38,14,12
al,38,4,15,3
et,1,42,,,1 !!4-node PLANE42 elements, keyoption axisymm.
tb,bkin,1 !!material properties plate
mp,ex,1,212000 !!Modulus of elasticity
tbdat,1,245,0 !!yield strength 245MPa, lin.-elastic id.-plastic material
tb,bkin,2 !!material properties shell
mp,ex,2,212000 !!Modulus of elasticity
tbdat,1,255,0 !!yield strength 255MPa, lin.-elastic id.-plastic material
asel,s,,,1,9,1 !!meshing plate
mat,1
amesh,all
allsel
asel,s,,,10,13,1 !!meshing shell
mat,2
amesh,all
allsel
nsel,s,loc,y,443
d,all,uy,0 !!boundary condition at the undisturbed end of the shell no vertical displacement
allsel
nsel,s,loc,x,0
dsym,symm,x !!symmetry boundary condition in the centre of the plate
allsel
save
Solution input file - check against PD
/solu
solcon,on
*set,pi,30 !!max. (theoretical) value of internal pressure
lsel,s,,,3
lsel,a,,,14
lsel,a,,,7
lsel,a,,,6
lsel,a,,,37
lsel,a,,,5
lsel,a,,,20
lsel,a,,,19
Annex 3: Input Listings
Page
A 3.7
DBA
Design by Analysis
lsel,a,,,18
lsel,a,,,17
lsel,a,,,35
lsel,a,,,34
lsel,a,,,1
sfl,all,pres,pi !!pressure load on selected lines
allsel
arclen,on,1,0.001 !!use the arclength method
nsubst,30
arctrm,u,10,52,uy !!termination criterion
neqit,30
/output,out,txt !!output to file for debug purposes
outres,all,all
solv
save
A 3.4 Example 1.4
Model Input file check against PD
/prep7
*set,t1,12 !defining mesh parameters!
*set,t2,100
*set,t3,10
*set,t4,8
*set,t5,12
*set,t6,12
*set,t7,12
*set,t8,6
*set,t9,15
*set,t10,5
*set,t11,50
*set,t12,18
k,1,100,0 !!modelling the geometry!
k,2,221.5,0
k,3,221.5,43
k,4,100,43
k,5,141,0
k,6,141,43
k,7,156,43
k,8,184,15
k,9,212,43
k,10,184,0
k,11,166,43
k,12,184,25
k,13,202,43
k,14,184,43
local,11,1,184,43
local,21,0,200,43
k,15,2,4.6575
k,16,0,16
k,17,0,20
k,18,0,30
k,19,0,400
k,20,21.5,400
k,21,21.5,30
k,22,12,20
local,31,0,100,0
k,101,0,25
k,102,25,0
k,105,25,25
k,106,18,0
k,107,8.85,-15.67
k,108,0,-3.5
k,109,-1.4,-13.46
k,110,-1.4,-35
k,111,3.725,-35
k,112,8.85,-35
k,113,-1.4,-40.125
k,114,3.725,-40.125
k,115,-1.4,-45.25
Annex 3: Input Listings
Page
A 3.8
DBA
Design by Analysis
k,116,8.85,-45.25
k,117,-1.4,-150
k,118,8.85,-150
csys,0
l,18,19,t2
l,20,21,t2
l,13,15
l,15,16
l,16,17
l,13,9,t3
l,9,3,t4
l,9,22,t5
l,3,21,t5
l,17,22,t3
l,22,21,t4
l,17,18,t4
l,19,20,t3
csys,11
k,23,18,-45
k,24,18,-135
k,25,28,-135
k,26,28,-45
l,11,24,t6
l,24,12,t6
l,12,23,t6
l,23,13,t6
l,7,25,t6
l,25,8,t6
l,8,26,t6
l,26,9,t6
csys,0
l,3,2,t6
l,23,26,t3
l,26,2,t4
l,2,10,t6
l,10,8,t4
l,8,12,t3
l,5,10,t6
l,24,25,t3
l,25,5,t4
l,6,7,t4
l,7,11,t3
l,5,6,t6
l,18,21,10
l,6,4,t12
l,4,101,t6
l,101,105,t12
l,105,6,t6
l,102,105,t6
l,102,5,t6
l,1,101,t6
l,102,106
l,106,107,t7
l,107,112,t9
l,1,108
l,108,109
l,109,110,t9
l,110,111,t8
l,111,112,t8
l,111,114,t10
l,113,114,t8
l,114,116,t8
l,112,116,t10
l,115,116,t8
l,115,117,t11
l,116,118,t11
l,117,118,t8
l,110,113,t10
l,113,115,t8
lfillt,35,36,3
lfillt,3,4,10
lfillt,42,43,7
Annex 3: Input Listings
Page
A 3.9
DBA
Design by Analysis
lfillt,43,44,10.5
l,1,31
l,109,34,t7
lesize,44,,,t9
al,37,38,35,60,36 !areas plate A1 A11!
al,64,42,39,37,41
al,40,33,38,39
al,30,18,31,33
al,29,14,32,18
al,28,26,19,30
al,27,15,29,19
al,25,24,20,26
al,20,23,16,27
al,24,22,7,21
al,23,21,6,17
al,6,8,10,5,4,61,3 !areas shell A12-A15!
al,7,9,11,8
al,11,34,12,10
al,34,2,13,1
al,64,45,46,65,63,43,62 !areas nozzle A12-A21!
al,47,48,49,44,65
al,49,50,52,53
al,48,58,51,50
al,51,59,54,52
al,54,55,57,56
et,1,42,,,1 !axisymmetric 4-node element PLANE42!
tb,bkin,1 !material properties plate!
mp,ex,1,212000
tbdat,1,245,0
tb,bkin,2 !material properties shell!
mp,ex,2,212000
tbdat,1,255,0
tb,bkin,3 !material properties nozzle!
mp,ex,3,212000
tbdat,1,265,0
asel,s,,,1,11,1 !meshing plate!
mat,1
amesh,all
allsel
asel,s,,,12,15,1 !meshing shell!
mat,2
amesh,all
allsel
asel,s,,,16,21,1 !meshing nozzle!
mat,3
amesh,all
allsel
esel,s,elem,,2823 !refine the mesh at high stressed regions!
esel,a,elem,,2822
esel,a,elem,,2819
esel,a,elem,,2825
esel,a,elem,,2826
esel,a,elem,,2848
esel,a,elem,,2850
esel,a,elem,,2851
esel,a,elem,,2863
esel,a,elem,,2870
esel,a,elem,,2876
esel,a,elem,,2878
esel,a,elem,,2885
esel,a,elem,,2886
esel,a,elem,,2890
esel,a,elem,,2891
esel,a,elem,,2895
esel,a,elem,,2898
esel,a,elem,,2906
esel,a,elem,,2909
esel,a,elem,,2915
esel,a,elem,,2916
esel,a,elem,,2917
esel,a,elem,,2920
esel,a,elem,,2942
Annex 3: Input Listings
Page
A 3.10
DBA
Design by Analysis
esel,a,elem,,2987
esel,a,elem,,3004
esel,a,elem,,3052
esel,a,elem,,3053
eref,all,,,3
allsel
nsel,s,loc,y,443 !defrine boundary conditions!
d,all,uy,0
allsel
save
Solution Input file check against PD
/solu
*set,pi,30 !maximum (theoretical) internal pressure!
lsel,s,,,1
lsel,a,,,12
lsel,a,,,5
lsel,a,,,4
lsel,a,,,61
lsel,a,,,3
lsel,a,,,17
lsel,a,,,16
lsel,a,,,15
lsel,a,,,14
lsel,a,,,32
lsel,a,,,31
lsel,a,,,35
lsel,a,,,60
lsel,a,,,36
lsel,a,,,41
lsel,a,,,45
lsel,a,,,46
lsel,a,,,47
lsel,a,,,58
lsel,a,,,59
lsel,a,,,55
sfl,all,pres,pi !internal pressure on selected lines!
allsel
lsel,s,,,57
sfl,all,pres,-(4.572*pi) !longitudinal stress of nozzle!
allsel
arclen,on,1,0.001 !use the arc-lenght method!
nsubst,30 !first substep with 1 MPa!
arctrm,u,5,1,uy
neqit,15
/output,out,txt !output file for debug purposes!
outres,all,all
solv
A 3.5 Example 2
Model Input file check against PD
/prep,7
*set,t1,8 !defining mesh parameters!
*set,t2,16
*set,t3,8
*set,t4,6
*set,t5,65
*set,t6,35
*set,t7,8
k,1,0,25.4 !modeling the geometry!
k,2,0,0
k,3,199.2,0
k,4,199.2,25.4
k,5,217,25.4
k,6,217,17.4
k,7,225,25.4
Annex 3: Input Listings
Page
A 3.11
DBA
Design by Analysis
k,8,225,17.4
k,9,225,0
local,11,1,225,50.8
k,10,25.4,-45
k,11,33.4,-45
k,13,25.4,0
k,14,33.4,0
k,15,50.8,0
k,17,127,0
csys,0
k,12,275.8,0
k,16,352,0
k,18,250.4,58.8
k,19,258.4,58.8
k,20,250.4,76.2
k,21,275.8,76.2
k,22,352,76.2
k,23,250.4,476.2
k,24,275.8,476.2
k,25,352,476.2
l,23,24,t1
l,24,25,t2
l,23,20,t5
l,24,21,t5
l,25,22,t5
l,20,21,t1
l,21,22,t2
l,15,17,t2
l,12,16,t2
l,16,17,t3
l,17,22,t4
l,15,21,t4
l,12,15,t3
l,19,21,t7
l,18,20,t7
l,14,19,t4
l,13,14,t1
l,13,18,t4
l,14,15,t7
csys,11
l,13,10,t3
l,10,7,t3
l,14,11,t3
l,11,8,t3
csys,0
l,10,11,t1
l,7,8,t1
l,8,9,t7
l,11,12,t7
l,9,12,t7
l,6,8,t4
l,5,7,t4
l,6,5,t1
l,3,4,t1
l,3,6,t7
l,3,9,t4
l,2,3,t6
l,1,4,t6
l,1,2,t1
l,4,5,t7
l,18,19,t1
al,1,3,6,4
al,4,7,5,2
al,6,15,39,14
al,16,19,12,14
al,12,8,11,7
al,8,13,9,10
al,27,13,19,22
al,28,27,23,26
al,34,26,29,33
al,33,31,38,32
al,29,25,30,31
Annex 3: Input Listings
Page
A 3.12
DBA
Design by Analysis
al,17,16,39,18
al,17,20,24,22
al,23,24,21,25
al,32,36,37,35
et,1,42,,,1,,2 !axisymmetric 4-node element PLANE42!
mp,ex,1,212000 !material properties!
tb,bkin,1
tbdat,1,255,0
asel,all
amesh,all !meshing!
nsel,s,loc,x,0 !define boundary conditions!
dsym,symm,x
nsel,s,loc,y,476.2
d,all,uy,0
allsel
Solution Input file check against PD
/solu
*set,pi,100 !maximum (theoretical) internal pressure!
lsel,s,,,3
lsel,a,,,15
lsel,a,,,18
lsel,a,,,20
lsel,a,,,21
lsel,a,,,30
lsel,a,,,38
lsel,a,,,36
sfl,all,pres,pi !apply internal pressure on selected lines!
allsel
arclen,on,1,0.001 !use the arc-length method!
nsubst,50 !first substep with 2 MPa!
arctrm,u,10,2487,uy
neqit,30
/output,out,txt !output file for debug purposes!
outres,all,all
solv
A 3.6 Example 3.1
Coarse Model Input file check against PD
/prep7
!geometry parameters!
a=982.5 !half shell length
h=537.5 !shell radius + nozzle length
dr=500 !mean diameter shell
ds=371 !mean diameter nozzle
b=675 !thick part of shell
t1=4 !thickness intersection
t2=10 !thickness outer shell
t3=25 !thickness flat end nozzle
t4=35 !thickness flat ends shell
!cylinder and nozzle!
k,1,0,0,dr/2
k,2,dr/2,0,0
k,3,0,0,-dr/2
k,4
k,5,0,2*a
larc,1,2,4,dr/2
larc,2,3,4,dr/2
l,4,5
k,6,0,a,h
k,7,0,a-ds/2,h
k,8,ds/2,a,h
k,9,0,a+ds/2,h
k,10,0,a,0
larc,7,8,6,ds/2
larc,8,9,6,ds/2
Annex 3: Input Listings
Page
A 3.13
DBA
Design by Analysis
l,6,10
adrag,1,2,,,,,3
adrag,4,5,,,,,6
!vert. cyl. for evaluation cross-section!
k,17,0,a-ds/2-12,h
k,18,ds/2+12,a,h
k,19,0,a+ds/2+12,h
larc,17,18,6,ds/2+12
larc,18,19,6,ds/2+12
adrag,17,18,,,,,6
boptn,keep,yes
ainp,1,5
adel,5,,,1
ainp,1,6
adel,6,,,1
!horiz. cyl. for evaluation cross-section!
k,25,0,0,dr/2+12
k,26,dr/2+12,0,0
larc,25,26,4,dr/2+12
adrag,18,,,,,,3
ainp,3,5
ainp,4,5
adel,5,,,1
!create areas for different mesh finess!
k,31,0,b,dr/2
k,32,dr/2,b,0
k,33,0,b,-dr/2
k,34,0,b,0
k,35,0,2*a-b,dr/2
k,36,dr/2,2*a-b,0
k,37,0,2*a-b,-dr/2
k,38,0,2*a-b,0
larc,31,32,34,dr/2
larc,32,33,34,dr/2
larc,35,36,38,dr/2
larc,36,37,38,dr/2
boptn,keep,no
asbl,1,18
asbl,6,20
asbl,2,19
asbl,8,21
numm,kp
k,41,0,a,dr/2
k,42,dr/2,a,0
larc,41,42,10,dr/2
asbl,7,28
numm,kp
!overlap nozzle and shell!
aovlap,3,4,10,8
adel,14,15,1
adel,7,11,4
!element area for evaluation cross-section
asbl,17,17
asbl,16,24
asbl,13,23
asbl,12,22
!flat ends!
k,51,,1965
k,52
k,53,,a,h
csys,1
a,13,12,51
a,11,12,51
a,1,2,52
a,2,3,52
a,7,8,53
a,8,9,53
csys,0
!Element type SHELL93!
et,1,93
mp,ex,1,210125 !Elastic modulus!
r,1,t1
Annex 3: Input Listings
Page
A 3.14
DBA
Design by Analysis
r,2,t2
r,3,t3
r,4,t4
!Meshing!
!evaluation cross-section!
real,1
lesize,24,,,12
lesize,36,,,12
lesize,22,,,12
lesize,23,,,12
lesize,37,,,12
lesize,17,,,12
lesize,13,,,4
lesize,48,,,4
lesize,47,,,4
lesize,14,,,4
lesize,41,,,4
lesize,32,,,4
amesh,11
amesh,14
amesh,3
amesh,7
!nozzle left part!
lesize,34,,,10,3
lesize,4,,,12
lesize,40,,,12,0.33
mshkey,1
amesh,13
!shell-upper-centre-left part!
lccat,18,33
amesh,8
!shell-upper-outer-left part!
real,2
lesize,25,,,10
amesh,5
!shell-lower-outer-left part!
lesize,2,,,7
amesh,6
!nozzle right part!
real,1
lesize,5,,,12
lesize,16,,,10,3
amesh,10
!shell-upper-centre-right part!
lccat,20,35
amesh,4
!shell-lower-centre part!
lccat,33,35
amesh,9
!shell-upper&lower-outer-right part!
real,2
lesize,8,,,10
amesh,1
amesh,2
!end plates shell!
real,4
mshkey,2
lesize,39,,,8
lesize,49,,,8
lesize,50,,,8
lesize,51,,,8
lesize,52,,,8
lesize,53,,,8
amesh,12
amesh,15
amesh,16
amesh,17
!end plate nozzle!
real,3
lesize,54,,,8
lesize,55,,,8
lesize,56,,,8
Annex 3: Input Listings
Page
A 3.15
DBA
Design by Analysis
amesh,18
amesh,19
numm,all
!Symmetry b.c.!
nsel,s,loc,x,0
dsym,symm,x
allsel
!support nodes b.c.!
d,3620,uy,0
d,3620,uz,0
d,3374,uz,0
!coupling for moment load!
nsel,s,loc,z,h
cp,1,rotx,all
allsel
!material strength parameter!
tb,bkin,1
tbdat,1,272,0
save
Coarse Model Solution file limit load check against PD
/solu
solcon,on
f,3824,mx,7822200
pr=1.2
asel,s,,,10,12,1
asel,a,,,13,14,1
asel,a,,,16,17,1
sfa,all,,pres,-pr !orientation of shell elements!
allsel
asel,u,,,10,12,1
asel,u,,,13,14,1
asel,u,,,16,17,1
sfa,all,,pres,pr
allsel
nropt,auto
autots,on
nsubst,5,100,5
neqit,20
time,1
/output,out1,txt !output file for debug purposes!
outres,all,all
solv
save
Submodel Input file check against PD
/filnam,fein
/prep7
!elements and material properties!
et,1,solid45
mp,ex,1,210125
tb,bkin,1
tbdat,1,272,0
!creating the half model!
wpoffs,,982.5 !Submodel C.S. identical with Coarse Model C.S.!
cylin,183.5,187.5,150,280,-90,0
wprot,,-90
cylin,248,252,-220,0,-30,-90
wpst
vdel,all
asel,u,,,7,12,5
asel,u,,,3,4
asel,u,,,9,10
adel,all
alls
aovl,3,9
aovl,4,10
adel,1,3,2,1
adel,2,8,6,1
afill,9,11,2 !inner fillet!
Annex 3: Input Listings
Page
A 3.16
DBA
Design by Analysis
afill,6,5,4 weld fillet!
v,32,35,36,29,8,5,6,7
v,32,35,36,29,34,31,27,33
a,34,10,9,31
a,14,15,33,27
va,17,12,7,16,15,3,4
!meshing the half model!
lesi,8,,,4
lesi,11,,,4
lesi,44,,,50
esha,2
vmesh,1
lesi,10,,,4
vmesh,2
acca,7,12
vmesh,3
!generate the other half of the model by refelection!
local,31,0,0,982.5
vsym,y,all
csys,0
numm,node
numm,kp
nsel,s,loc,x,0
dsym,symm,x
allsel
!write nodes on cut boundaries to a file!
asel,s,,,11,12
asel,a,,,31,32
asel,a,,,7,24,17
nsla,s,1
nwrite
allsel
save
Submodell solution file limit load - check againstPD
*use,feinm.txt !creating the supmodel with the model input file!
fini
save,fein,db
/post1
resu,grob,db !resume database of corresponding coarse model!
cbdof,,,,,,,,,1 !acivates cut boundary interpolation for submodelling shell to solid!
fini
/prep7
resu,fein,db !resume database of submodell!
/inp,,cbdo !interpolate the boundary conditions to the cut boundaries!
/inp,,cbdo,,:cb33 !interpolate the boundary conditions to the cut boundaries!
fini
/solu
pi=1.2 !pressure load!
asel,s,,,1,3,1
asel,a,,,23,28,5
asel,a,,,34
sfa,all,,pres,pi
allsel
nropt,auto
autots,on
nsubst,5,200,5
neqit,50
outres,all,all
time,1
/output,outfein,txt !output file for debug purposes!
solv
save
Annex 3: Input Listings
Page
A 3.17
DBA
Design by Analysis
A 3.7 Example 3.2
Model input file check against PD
/prep7
!properties beam element!
et,1,4
r,1,100,10000/12,10000/12,10,10
mp,ex,1,210125
mp,nuxy,1,0.3
!element SOLID45!
et,2,45
!mat. properties shell!
tb,bkin,2
tbdat,1,234,0
mp,ex,2,210125
!mat. properties nozzle!
tb,bkin,3
tbdat,1,343,0
mp,ex,3,210125
!creating the half model!
local,11,1,,,,,-90
ri1=142.5 !inner radius shell!
t1=15 !thickness shell!
ro1=ri1+t1
z1=100 !0.5 * length of shell!
ri2=10 !inner radius nozzle!
t2=7.5 !thickness nozzle!
str=1 !inside fillet!
ro2=ri2+t2
z2=ri1+t1+100
r=6.5 !weld fillet!
csys,1
k,1,ro1,65,z1
k,2,ro1,65
kgen,2,1,2,1,,25
a,1,2,4,3
csys,11
k,5,ro2,-90
k,6,ro2
kgen,2,5,6,,,,z2
a,5,6,8,7
/view,,1,1,1
/pnum,area,1
/pnum,kpoi,1
/pnum,line,1
aplot
aovlap,1,2
aplot
adele,3,4,1,1
afillt,5,6,r
csys,1
k,22,ri1,65,z1
k,23,ri1,65
kgen,2,22,23,1,,25
a,22,23,5,4
csys,11
k,24,ri2,-90
k,25,ri2
kgen,2,24,25,,,,z2
a,24,25,9,6
aovlap,4,5
adele,6,7,1,1
afillt,8,9,str
csys,0
lgen,2,25,,,-ri2,-z2/2
a,20,21,10,5
asba,6,7
k,100,,ro1+r
a,16,19,100
asba,4,6
Annex 3: Input Listings
Page
A 3.18
DBA
Design by Analysis
a,5,16,21,12,30,25
a,20,11,29,28,10,19
a,19,10,5,16
a,21,12,11,20
va,4,8,5,10,11,2,12,6
a,5,16,7,6
a,6,7,8,9
a,8,9,10,19
va,11,13,7,15,1,14
a,1,3,4,22
a,1,2,23,22
a,11,20,2,23
a,12,21,3,4
va,9,12,18,17,16,19,3
/pnum,volu,1
vplot
esize,,4 !number of element divisions along lines, if not specified otherwise!
flst,2,3,5,orde,3
fitem,2,5
fitem,2,8
fitem,2,10
accat,p51x
flst,2,3,4,orde,3
fitem,2,8
fitem,2,16
fitem,2,28
lccat,p51x
flst,2,3,4,orde,3
fitem,2,6
fitem,2,15
fitem,2,29
lccat,p51x
accat,16,17
lccat,5,9
lccat,1,4
csys,1
lgen,2,1,5,4,,-155
l,1,13
l,22,15
l,2,14
l,23,17
v,1,22,15,13,2,23,17,14
vplot
!element division on specified lines!
a=10
lesi,24,,,a
lesi,20,,,a
lesi,31,,,a
lesi,36,,,a
lesi,25,,,a
lesi,17,,,a
lesi,15,,,5
lesi,16,,,5
lesi,28,,,5
lesi,29,,,5
lesi,6,,,1
lesi,8,,,1
lesi,13,,,12,.33
lesi,14,,,12,3
lesi,19,,,12,.33
lesi,23,,,12,.33
lesi,1,,,5
lesi,5,,,5
lesi,4,,,5
lesi,27,,,10,.33
lesi,44,,,16
lesi,45,,,16
lesi,46,,,16
lesi,47,,,16
!use mapped meshing!
mshkey,1
Annex 3: Input Listings
Page
A 3.19
DBA
Design by Analysis
!meshing!
vsel,s,,,2
type,2
mat,3
vmesh,all
eplot
allsel
vsel,u,,,2
vplot
type,2
mat,2
vmesh,all
allsel
csys,0
!generate the other half of the model by reflection!
vsymm,z,all
nummrg,all
eplot
!create beam element!
k,100,0,z2,0
k,101,0,z2+10,0
nkpt,,100 !node No. 4701!
nkpt,,101 !node No. 4702!
type,1
real,1
mat,1
e,4701,4702
eplot
!symmetry b.c.!
nsel,s,loc,x,0
dsym,symm,x
allsel
!b.c. for dummy end of beam!
d,4702,uz,0
!b.c. at the ends of the shell!
csys,1
nsel,s,loc,z,100
nsel,u,loc,y,90
nsel,u,loc,y,-90
nrotat,all !rotate the nodal CS!
d,all,uy,0 !hoop displacements constraint!
nsel,r,loc,y,16.563
nsel,r,loc,x,ro1
d,all,uz,0 !longitudinal displacement constraint!
allsel
nsel,s,loc,z,-100
nsel,u,loc,y,90
nsel,u,loc,y,-90
nrotat,all !rotate the nodal CS!
d,all,uy,0 !hoop displacement constraint!
allsel
!create rigid region!
csys,0
nsel,s,loc,y,z2
cerig,4701,all,uy
allsel
save
Solution input file - check against PD states 1 and 2
/solu
!moment load!
f,4701,mx,35555
solcon,on
nropt,auto
autots,on
nsubst,2,10,2
neqit,50
outres,all,all
time,1
/output,out1,txt
Annex 3: Input Listings
Page
A 3.20
DBA
Design by Analysis
solv
save
!moment load!
f,4701,mx,35555
!internal pressure!
pi=2.6
asel,s,,,7
asel,a,,,9
asel,a,,,10
asel,a,,,23
asel,a,,,30
asel,a,,,36
asel,a,,,40
asel,a,,,47
asel,a,,,28
asel,a,,,29
asel,a,,,5
asel,a,,,8
sfa,all,,pres,pi
allsel
!longitudinal stress in nozzle!
asel,s,,,14
asel,a,,,39
sfa,all,,pres,-0.4848*pi
allsel
!longitudinal stress shell!
asel,s,,,16
asel,a,,,22
asel,a,,,42
asel,a,,,46
sfa,all,,pres,-4.5125*pi
allsel
solcon,on
nropt,auto
autots,on
nsubst,5,25,5
neqit,50
outres,all,all
time,2
/output,out2,txt
solv
save
A 3.8 Example 4
Model and solution input file GPD check
/prep7
sm=138.5 ! define sm as yield stress
nu=0.3 ! define poissons ratio
ex=183.6e3 ! define elastic modulus
!geometry
k,,
k,,,427.5
k,,995
k,,,-1567.5
k,,800
k,,800,400
k,,995,200
k,,1225,-550
circle,4,1995,,8,90,2
circle,4,1800,,8,90,2
l,5,6
l,3,7
linter,5,3
circle,15,195,,3,180,1
linter,6,3
linter,14,1
k,,,600
Annex 3: Input Listings
Page
A 3.21
DBA
Design by Analysis
l,1,21
linter,1,2
lsel,s,,,6,17,11
lsel,a,,,11,16,5
lsel,invert
ldel,all,,,1
allsel
ksll,s,1
ksel,invert
kdel,all
allsel
k,,995,-1000
l,3,1
k,,
arotat,17,,,,,,22,2,90,1
arotat,1,11,16,6,,,22,2,35,1
arotat,4,5,7,8,,,22,2,55,1
nummrg,all
arotat,2,20,19,18,15,,22,2,90,1
local,11,1,838.05,,,,90,180
csys,11
k,,156.95
l,3,23
k,,156.95,,432.5 !length of nozzle
l,23,24
adrag,36,,,,,,37
asel,s,,,4,5
asel,a,,,15
asba,15,all,,,keep
asel,s,,,4,5
asel,a,,,16
asba,all,16
allsel
adel,15,,,1
adel,18,,,1
linter,3,8
linter,42,48
nummrg,all
ldiv,38,0.43
csys,0
l,24,25
al,16,38,14
!meshing
mp,ex,1,ex
mp,nuxy,1,nu
r,1,10
r,2,6
r,3,8.5
r,4,15
et,1,93
eshape,2
real,1
lesize,49,,,6,0.25
lesize,47,,,6
lesize,48,,,7,0.25
lesize,8,,,12,0.33
lesize,13,,,7
lccat,47,49
amesh,20
lesize,6,,,5
lesize,31,,,10
lesize,20,,,12,0.33
lesize,27,,,12,0.33
amesh,9,11,2
lesize,28,,,6
lesize,29,,,2
lesize,30,,,12,9
lesize,12,,,7,4
amesh,8,12,4
amesh,19
lesize,17,,,6
lesize,2,,,6
Annex 3: Input Listings
Page
A 3.22
DBA
Design by Analysis
lccat,13,6
eshape,1
amesh,1
lesize,26,,,6
amesh,10
eshape,2
lesize,10,,,7,4
amesh,13
amesh,7
amesh,3
real,2
lesize,15,,,12,9
lesize,4,,,12,9
lesize,1,,,12,9
lesize,9,,,7,4
amesh,6,14,8
amesh,2
real,3
lesize,38,,,6
lesize,14,,,6
lccat,47,3
lccat,14,38
lesize,44,,,12,3
lesize,45,,,12,0.25
amesh,17
real,4
ldel,40
eshape,1
lesize,16,,,6
amesh,4
!normalise element normals
enorm,683
!apply boundary conditions
csys,0
nsel,s,loc,z,0
dsymm,symm,z,0
local,13,1,,,,,90,180
nsel,s,loc,z,-1000
dsymm,symm,y,0
allsel
d,1840,ux,0
!non linear solution
/solu
pred,on
tb,bkin,1,1
tbdata,1,sm,0
p1=0.10
time,p1
sfe,all,1,press,,p1
lswrit
autots,on
outres,all,all
/com start=0.01,min=0.005,max=0.01
deltim,0.003,0.001,0.005
p2=1.0
time,p2
sfe,all,1,press,,p2
lswrit
lssolve,1,2
save
A 3.9 Example 5
Model input file- transient thermal analysis
/prep7
k,1 !modelling the structure!
k,2,65.55
k,3,109.55
k,4,600,1039.23
Annex 3: Input Listings
Page
A 3.23
DBA
Design by Analysis
k,5,109.33,1300
k,6,65.55,1300
k,7,0,1007
k,8,0,1060
k,9,1060
k,10,1107
l,2,6
l,3,5
l,1,4
larc,8,10,1,1107,
larc,9,7,1,1060
lptn,all
lsel,s,,,12,16
lsel,a,,,18,21
lsel,a,,,6,9
lsel,a,,,24
ldel,all
allsel
l,14,18
local,11,0,65.55,1005.053
k,20,0,166.57
k,21,4.29,174
k,22,6.29,174
k,23,22.29,163.61
k,24,44,129.471
k,25,18.6,183.11
k,26,4.29,224
k,27,18.6,224
k,28,46,24.5
k,29,96,111.1
l,28,29
lptn,2,23
ldel,5
ldel,6
k,30,29.55,99.97
l,19,30
lptn,2,10
ldel,8
l,26,27
l,26,21
l,27,25
l,21,20
l,25,23
l,23,24
lptn,5,13
ldel,14
lptn,10,11
ldel,5
l,21,22
l,22,23
lfillt,15,12,23.5
k,37,28,163.61
l,23,37
lptn,14,19
ldel,24
lfillt,18,7,6
lfillt,3,7,6
ldel,4
lfillt,9,12,24.67
l,31,36
l,18,39
k,43,4.29,324
k,44,18.6,324
l,43,44
l,26,43
l,27,44
ldel,16
al,28,29,2,30
al,2,8,5,11,20,23,12,4,9
l,20,32
al,10,16,15,21,20,11,5
al,13,17,26,1,6,24,18,16
Annex 3: Input Listings
Page
A 3.24
DBA
Design by Analysis
al,1,6,24,14,7,19,27
al,27,25,22,3
!element type!
et,1,55,,,1 !axisymmetric elements!
!material properties!
!P265GH!
mptemp,,20,100,200,300,400 !temperatures!
mp,dens,1,7.85e-9 !density!
mpdata,c,1,,461e6,479e6,499e6,517e6,536e6 !specific heat!
mpdata,kxx,1,,51,50.8,48.7,45.8,42.5 !thermal conductivity!
!11CrMo9-10!
mp,dens,2,7.85e-9
mpdata,c,2,,461e6,479e6,499e6,517e6,536e6
mpdata,kxx,2,,34.9,37.3,38.2,37.8,36.6
!meshing!
lesi,22,,,12
lesi,27,,,12
lesi,3,,,60,.4
lesi,25,,,60,.4
mat,2
amesh,6
lcomb,17,13
lcomb,1,6
lcomb,1,24
lesize,26,,,15
lesize,16,,,20
lesize,13,,,40
lesize,1,,,20
lesize,18,,,20
mat,2
amesh,4
esize,3.5
mshkey,2
mat,2
amesh,5
lesi,11,,,8
lesi,21,,,3
lesi,15,,,10
lesi,10,,,5
mat,2
amesh,3
lesi,2,,,6
lesi,8,,,20
lesi,4,,,3
lesi,9,,,15
lesi,23,,,3
lesi,12,,,8
mat,1
amesh,2
lesi,28,,,6
lesi,29,,,24,3
lesi,30,,,24,3
mat,1
amesh,1
!refining the mesh!
esel,s,,,601,602
esel,a,,,661,662
esel,a,,,745
esel,a,,,747
esel,a,,,1201,1203
esel,a,,,1295,1299
esel,a,,,1330,1332
esel,a,,,1335
esel,a,,,1338
esel,a,,,1344,1346
esel,a,,,1360
esel,a,,,1364,1365
esel,a,,,1375,1376
esel,a,,,1381,1382
esel,a,,,1390,1392
esel,a,,,1409,1410
esel,a,,,1414
Annex 3: Input Listings
Page
A 3.25
DBA
Design by Analysis
esel,a,,,1436,1437
esel,a,,,1441
esel,a,,,1449,1451
esel,a,,,1481,1482
esel,a,,,1499
esel,a,,,1560
esel,a,,,1562,1563
esel,a,,,1580,1583
esel,a,,,1589,1591
erefine,all,2
allsel
eplot
allsel
eplot
save
Solution input file transient thermal analysis
/solu
antype,trans !transient analysis!
outres,all,all
autots,on
nsubst,10,30,10
bfunif,temp,325 !uniform temperature at the beginning of the cold medium injection!
time,0.1
kbc,1 !stepped loading!
lsel,s,,,13
lsel,a,,,10
lsel,a,,,8
lsel,a,,,29
nsll,s,1
sf,all,conv,10.8,80 !cold media injection, heat transfer coefficient, bulk temperature!
allsel
lsel,s,,,25,26
nsll,s,1
sf,all,conv,1.16,325 !hemispherical part, heat transfer coefficient, bulk temperature!
allsel
solve
autots,on
nsubst,10,30,10
time,1
solve
autots,on
nsubst,10,30,10
time,10
solve
autots,on
nsubst,10,30,10
time,100
solve
autots,on
nsubst,10,30,10
time,600
solve
autots,on !end of cold medium injection!
nsubst,10,30,10
time,600.1
kbc,1
lsel,s,,,13
lsel,a,,,10
lsel,a,,,8
lsel,a,,,29
nsll,s,1
sf,all,conv,1.16,325
allsel
lsel,s,,,25,26
nsll,s,1
sf,all,conv,1.16,325
allsel
solve
autots,on
nsubst,10,30,10
Annex 3: Input Listings
Page
A 3.26
DBA
Design by Analysis
time,601
solve
autots,on
nsubst,10,30,10
time,610
solve
autots,on
nsubst,10,30,10
time,700
solve
autots,on
nsubst,10,30,10
time,1200
solve
Solution input file linear-elastic cyclic analysis
/solu
/output,out,txt
time,1e-6 !pressure load only!
tunif,325
pi=11.71
lsel,s,,,25,26
lsel,a,,,13
lsel,a,,,10
lsel,a,,,8
lsel,a,,,29
sfl,all,pres,pi
allsel
lsel,s,,,28
sfl,all,pres,-2.213*pi
allsel
solv
save
*do,i,1,10 !calculate stress states at time interval 0.01 s 0.1 s!
time,(0.01*i)
ldread,temp,,,(0.01*i),,,rth
solv
save
*enddo
*do,j,1,10 !calculate stress states at time interval 0.1 s 1 s!
time,(0.1+0.09*j)
ldread,temp,,,(0.1+0.09*j),,,rth
solv
save
*enddo
*do,k,1,10 !calculate stress states at time interval 1 s 10 s!
time,(1+0.9*k)
ldread,temp,,,(1+0.9*k),,,rth
solv
save
*enddo
*do,l,1,10 !calculate stress states at time interval 10 s 100 s!
time,(10+9*l)
ldread,temp,,,(10+9*l),,,rth
solv
save
*enddo
*do,m,1,10 !calculate stress states at time interval 100 s 600 s!
time,(100+50*m)
ldread,temp,,,(100+50*m),,,rth
solv
save
*enddo
*do,n,1,10 !calculate stress states at time interval 600 s 600.1 s!
time,(600+0.01*n)
ldread,temp,,,(600+0.01*n),,,rth
solv
save
*enddo
*do,a,1,10 !calculate stress states at time interval 600.1 s 601 s!
Annex 3: Input Listings
Page
A 3.27
DBA
Design by Analysis
time,(600.1+0.09*a)
ldread,temp,,,(600.1+0.09*a),,,rth
solv
save
*enddo
*do,b,1,10 !calculate stress states at time interval 601 s 610 s!
time,(601+0.9*b)
ldread,temp,,,(601+0.9*b),,,rth
solv
save
*enddo
*do,c,1,10 !calculate stress states at time interval 610 s 700 s!
time,(610+9*c)
ldread,temp,,,(610+9*c),,,rth
solv
save
*enddo
*do,d,1,10 !calculate stress states at time interval 700 s 1200 s!
time,(700+50*d)
ldread,temp,,,(700+50*d),,,rth
solv
save
*enddo
Cyclic elastic plastic calculation Solutions macros
Note: the sequence of the states is given by the macro named control, the macro first concerns the startup, the macro cyc 1 the
thermal cycle, and the macro cyc 0 the shutdown.
Macro control
!start-up
/output,first.txt
first
!
!thermal cycle
/output,cyc111.txt
cyc1
!
!shutdown cycle
/output,cyc10.txt
cyc0
!
!three thermal cycles
/output,cyc211.txt
cyc1
/output,cyc212.txt
cyc1
/output,cyc213.txt
cyc1
!
!shutdown cycle
/output,cyc20.txt
cyc0
!
!three thermal cycles
/output,cyc311.txt
cyc1
/output,cyc312.txt
cyc1
/output,cyc313.txt
cyc1
!
!shutdown cycle
/output,cyc30.txt
cyc0
/output
Makro first
solcon,on
!reference step, T = 20C, p = 0 bar !
Annex 3: Input Listings
Page
A 3.28
DBA
Design by Analysis
Zeit=1e-6
time,Zeit
tunif,20
BF,all,temp,20
outres,all,last
nsubst,1
!
!pressure load and
!temperature in phase
Zeit=1000
BF,all,temp,325
pi=11.71
lsel,s,,,25,26
lsel,a,,,13
lsel,a,,,10
lsel,a,,,8
lsel,a,,,29
sfl,all,pres,pi
allsel
lsel,s,,,28
sfl,all,pres,-2.213*pi
allsel
nropt,auto
autots,on
nsubst,10,500,10
neqit,30
outres,all,last
time,Zeit
!/output,out1,txt
solv
save
Makro cyc0
BF,all,temp,20
pi=0
lsel,s,,,25,26
lsel,a,,,13
lsel,a,,,10
lsel,a,,,8
lsel,a,,,29
sfl,all,pres,pi
allsel
lsel,s,,,28
sfl,all,pres,-2.213*pi
allsel
nropt,auto
autots,on
nsubst,10,500,10
neqit,30
!!outres,all,last
Zeit=Zeit+1000
time,Zeit
!/output,out17,txt
solv
save
Makro cyc1
BF,all,temp,325
pi=11.71
lsel,s,,,25,26
lsel,a,,,13
lsel,a,,,10
lsel,a,,,8
lsel,a,,,29
sfl,all,pres,pi
allsel
lsel,s,,,28
sfl,all,pres,-2.213*pi
allsel
nropt,auto
autots,on
Annex 3: Input Listings
Page
A 3.29
DBA
Design by Analysis
nsubst,10,500,10
neqit,30
!outres,all,last
Zeit=Zeit+1000
time,Zeit
!/output,out21,txt
solv
save
ldread,temp,,,0.05,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!!outres,all,last
Zeit=Zeit+0.05
time,Zeit
!/output,out2,txt
solv
save
ldread,temp,,,0.1,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+0.05
time,Zeit
!/output,out3,txt
solv
save
ldread,temp,,,0.55,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+0.45
time,Zeit
!/output,out4,txt
solv
save
ldread,temp,,,1,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+0.45
time,Zeit
!/output,out5,txt
solv
save
ldread,temp,,,5.5,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+4.5
time,Zeit
!/output,out6,txt
solv
save
ldread,temp,,,10,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+4.5
time,Zeit
!/output,out7,txt
Annex 3: Input Listings
Page
A 3.30
DBA
Design by Analysis
solv
save
ldread,temp,,,55,,file,rthnropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+45
time,Zeit
!/output,out8,txt
solv
save
ldread,temp,,,200,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+145
time,Zeit
!/output,out9,txt
solv
save
ldread,temp,,,600,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+400
time,Zeit
!/output,out10,txt
solv
save
ldread,temp,,,601,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+1
time,Zeit
!/output,out11,txt
solv
save
ldread,temp,,,610,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+9
time,Zeit
!/output,out12,txt
solv
save
ldread,temp,,,655,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+45
time,Zeit
!/output,out13,txt
solv
save
ldread,temp,,,700,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
Annex 3: Input Listings
Page
A 3.31
DBA
Design by Analysis
!outres,all,last
Zeit=Zeit+45
time,Zeit
!/output,out14,txt
solv
save
ldread,temp,,,1000,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+300
time,Zeit
!/output,out15,txt
solv
save
ldread,temp,,,1200,,file,rth
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+200
time,Zeit
!/output,out16,txt
solv
save
!uniform temperature!
BF,all,temp,325
nropt,auto
autots,on
nsubst,3,50,3
neqit,30
!outres,all,last
Zeit=Zeit+800
time,Zeit
!/output,out16,txt
solv
save
A 3.10 Example 6
Model input file shakedown check linear-elastic cycle
/prep7
!geometry parameters!
s1=26 !thickness dished end!
r1=2800 !outer radius cylinder!
r2=280 !fillet dished end!
h1=605.58 !height dished end!
h=2250
x=r1-h-s1
!element type!
et,1,42,,,1 !axial-symmetric elements!
!material properties!
mptemp,,10,20,100,160 !temperatures!
mpdata,ex,1,,1.97e5,1.96e5,1.9e5,1.852e5 !elastic modulus!
mpdata,alpx,1,,16.05e-6,16.1e-6,16.7e-6,17e-6 !thermal expansion coefficient!
!reference temperature!
tref,20
!modelling the structure!
local,11,1,,-x
k,1,r1,30
k,2,r1+s1,30
k,3,r1,90
k,4,r1+s1,90
l,1,3
l,2,4
Annex 3: Input Listings
Page
A 3.32
DBA
Design by Analysis
local,12,0,,h-h1
k,5,(r1/2)-s1
k,6,r1/2
kgen,2,5,6,1,,h1
/pnum,kpoi,1
/pnum,line,1
l,5,7
l,6,8
lcsl,all
ldele,5,6,1,1
ldele,11,12,1,1
ldele,13,15,2,1
lfillt,7,9,r2
lcomb,8,14
lcomb,10,16
lfillt,8,10,r2
lplot
lcomb,7,1
lcomb,9,1
lcomb,8,2
lcomb,10,2
lplot
local,13,1,r1/2+125-5,2250-605.85-145-13.5
local,14,1,r1/2+5.75,2250-605.85-145-13.5-5.75
local,15,1,r1/2+5.75,2250-605.85-145+5.75
local,16,1,r1/2+125+8+8.48,500+7+8.48
local,17,1,r1/2+125+8+8.48,500-7-8.48
local,18,1,r1/2+16.97,10+16.97a
csys,13
k,20,5
k,21,5,90
k,22,5,180
l,20,21
l,21,22
csys,14
k,30,5.75,90
k,31,5.75,180
l,30,31
csys,15
k,40,5.75,-90
k,41,5.75,-180
l,40,41
csys,18
k,50,16.97,-90
k,51,16.97,-180
l,50,51
csys,16
k,60,8.48,-90
k,61,8.48,-180
l,60,61
csys,17
k,70,8.48,90
k,71,8.48,180
l,70,71
csys,0
kgen,2,31,41,10,-20
kgen,2,30,40,10,10
kgen,2,22,,,-(13.5-5)
kgen,2,21,,,,13.5-5
kgen,2,20,,,8
kgen,2,9,,,,13.5
kgen,2,13,,,,13.5
kgen,2,9,14,5,-10
kgen,2,13,20,7,,-10
kgen,2,61,71,10,-8
kgen,2,60,70,10,10
kgen,2,51,,,-20
kgen,2,50,,,,-10
kgen,2,28,,,-(16.97+20)
kgen,2,60,70,10,65-8.48
kgen,2,28,50,22,110-16.97
kgen,2,34,35,1,15
Annex 3: Input Listings
Page
A 3.33
DBA
Design by Analysis
kgen,2,36,37,1,8
kgen,2,5,,,6
kgen,2,42,,,-10,17.32
l,42,43
lcsl,all
kgen,2,44,,,30
l,44,45
lcsl,all
ldele,14,17,3,1
ldele,12,,,1
kplot
a,3,4,46,44
a,44,46,6,42
a,42,6,41,2
a,2,41,40,30,31,1
a,30,40,8,7
a,7,8,17,16
a,16,17,14,9
a,9,14,12,21,22
a,21,12,15,13,20
a,20,13,18,19
a,19,18,61,23
a,23,61,60,70,71,24
a,60,70,26,25
a,25,26,33,32
a,24,71,39,37
a,37,39,38,36
a,35,34,28,50
a,50,28,29,27,51
a,1,31,51,27
aplot
!meshing!
esize,,4
lesi,11,,,50,.33
lesi,10,,,50,.33
lesi,13,,,1
lesi,15,,,1
lesi,17,,,12
lesi,12,,,12
lesi,20,,,8
lesi,18,,,8
lesi,24,,,20
lesi,26,,,20
lesi,28,,,8
lccat,14,6
lccat,19,5
amesh,1,7,1
lsel,s,line,,30,35,1
lesi,all,,,8
lsel,all
lesi,4,,,16
lesi,3,,,16
lccat,28,30
lccat,33,34
lesi,38,,,6,.33
lesi,36,,,6,3
lesi,37,,,4
lesi,42,,,8
lesi,44,,,8
lccat,40,8
lccat,43,9
lesi,39,,,150
lesi,41,,,150
lesi,50,,,10,.33
lesi,48,,,10,3
lesi,65,,,200
lesi,64,,,200
lccat,61,62
lesi,7,,,10
lesi,60,,,15,3
lesi,58,,,15,.33
lesi,61,,,6
Annex 3: Input Listings
Page
A 3.34
DBA
Design by Analysis
lesi,53,,,100
lesi,51,,,100
amesh,all
eplot
allsel
!boundary conditions!
dl,55,,symm
dl,58,,symm
dl,61,,symm
dl,72,,symm
dl,1,,symm
save
Solution file shakedown check linear-elastic cycle
/solu
/output,out,txt !output file for debug purposes!
!thermal stress only!
ldread,temp,,,1,,,rth
time,1
solv
save
! state 1!
bfdele,all,temp
tunif,160
lsel,s,line,,11
lsel,a,line,,13
lsel,a,line,,17
lsel,a,line,,20
lsel,a,line,,65
lsel,a,line,,62
nsll,s,1
sf,all,pres,1.3
allsel
lsel,s,line,,57
lsel,a,line,,60
lsel,a,line,,7
lsel,a,line,,64
lsel,a,line,,5
lsel,a,line,,23
lsel,a,line,,26
lsel,a,line,,29
lsel,a,line,,32
lsel,a,line,,4
lsel,a,line,,3
lsel,a,line,,38
lsel,a,line,,41
lsel,a,line,,44
lsel,a,line,,53
lsel,a,line,,56
nsll,s,1
sf,all,pres,0.5
allsel
time,2
solv
save
!state 2!
bfdele,all,temp
tunif,160
sfdele,all,pres
lsel,s,line,,11
lsel,a,line,,13
lsel,a,line,,17
lsel,a,line,,20
lsel,a,line,,65
lsel,a,line,,62
nsll,s,1
sf,all,pres,1.3
allsel
time,3
solv
Annex 3: Input Listings
Page
A 3.35
DBA
Design by Analysis
save
!state 3!
sfdele,all,pres
lsel,s,line,,11
lsel,a,line,,13
lsel,a,line,,17
lsel,a,line,,20
lsel,a,line,,65
lsel,a,line,,62
nsll,s,1
sf,all,pres,-0.1
allsel
lsel,s,line,,57
lsel,a,line,,60
lsel,a,line,,7
lsel,a,line,,64
lsel,a,line,,5
lsel,a,line,,23
lsel,a,line,,26
lsel,a,line,,29
lsel,a,line,,32
lsel,a,line,,4
lsel,a,line,,3
lsel,a,line,,38
lsel,a,line,,41
lsel,a,line,,44
lsel,a,line,,53
lsel,a,line,,56
nsll,s,1
sf,all,pres,0.5
allsel
bfdele,all,temp
ldread,temp,,,1,,,rth
time,4
solv
save
!state4!
sfdele,all,pres
lsel,s,line,,11
lsel,a,line,,13
lsel,a,line,,17
lsel,a,line,,20
lsel,a,line,,65
lsel,a,line,,62
nsll,s,1
sf,all,pres,1.3
allsel
lsel,s,line,,57
lsel,a,line,,60
lsel,a,line,,7
lsel,a,line,,64
lsel,a,line,,5
lsel,a,line,,23
lsel,a,line,,26
lsel,a,line,,29
lsel,a,line,,32
lsel,a,line,,4
lsel,a,line,,3
lsel,a,line,,38
lsel,a,line,,41
lsel,a,line,,44
lsel,a,line,,53
lsel,a,line,,56
nsll,s,1
sf,all,pres,0.5
allsel
bfdele,all,temp
ldread,temp,,,1,,,rth
time,5
solv
save
Annex 3: Input Listings
Page
A 3.36
DBA
Design by Analysis
Solution input file thermal analysis
/solu
lsel,s,line,,11
lsel,a,line,,13
lsel,a,line,,17
lsel,a,line,,20
lsel,a,line,,65
lsel,a,line,,62
nsll,s,1
sf,all,conv,1.16,160 !heat transfer coefficient, bulk temperature inner vessel!
allsel
lsel,s,,,57
lsel,a,,,60
lsel,a,,,7
lsel,a,,,64
lsel,a,,,5
lsel,a,,,23
lsel,a,,,26
lsel,a,,,29
lsel,a,,,32
lsel,a,,,4
lsel,a,,,3
lsel,a,,,38
lsel,a,,,41
lsel,a,,,44
lsel,a,,,53
lsel,a,,,56
nsll,s,1
sf,all,conv,14.4,10 !heat transfer coefficient, bulk temperature - jacket!
allsel
/psf,conv,,2
/pbc,all,1
nplot
solve
save
fini
I_Check: Model and solution input files eigenvalue bucking and fully nonlinear analysis
!Eigenvalue buckling analysis!
/prep7
rn=293 !knuckle radius of dished end!
re=2813 !sphere radius of dished end!
rc=1390 !radius cylindrical main shell!
he=592.85 !height of dished end!
hra=510 !distance outer ring symmetry plane!
h1=145 !height of cylindrical part outside of jacket!
hj=1499.15 !height of jacket!
wj=139 !width of jacket!
wri=120 !width of inner ring!
wra=69 !width of outer ring!
te=26 !thickness dished end!
tc=20 !thickness cylinder!
tp=13.5 !thickness end plate of jacket!
tj=8 !thickness jacket!
tri=20 !thickness inner ring!
tra=14 !thickness outer ring!
local,11,1,,,he+h1+hj-re
k,1,re
k,2,,,re
k,3
larc,1,2,3,re
csys,0
k,4,rc
k,5,rc,,he+h1+hj
l,4,5
lptn,1,2
ldel,3
ldel,6
lfillt,4,5,rn
Annex 3: Input Listings
Page
A 3.37
DBA
Design by Analysis
k,9,rc,,hj
k,10,rc+wj,,hj
l,9,10
k,11,rc+wj,,hra
k,12,rc+wj+wra,,hra
k,13,rc+wj
k,14,rc+wri
l,10,11
l,11,12
l,11,13
l,4,14
k,15
arotat,4,1,5,8,,,2,15,180
arotat,2,3,7,,,,2,15,180
arotat,6,,,,,,2,15,180
asbl,3,31 !divide area by line!
asbl,7,35
wpoffs,,,hj+h1 !working plane offset!
asbw,3 !divide area by working plane!
asbw,17
et,1,93 !8-node shell element!
mp,ex,1,191100 !elastic modulus, mean value!
r,1,te !thickness dished end!
r,2,tc !thickness main shell outside of jacket!
r,3,tc !thickness jacketed part of main shell!
r,4,tp !thickness plate of jacket!
r,5,tj !thickness jacket!
r,6,tri !thickness inner ring!
r,7,tra !thickness outer ring!
!meshing!
esize,100 !global element size!
real,1
asel,s,,,1,7
asel,u,,,4
amesh,all
allsel
real,2
asel,s,,,20,21
amesh,all
allsel
real,3
asel,s,,,18,19
amesh,all
allsel
real,4
asel,s,,,9,12,3
amesh,all
allsel
real,5
asel,s,,,10,11
asel,a,,,13,14
amesh,all
allsel
real,6
asel,s,,,4,8,4
amesh,all
allsel
real,7
asel,s,,,15,16
amesh,all
allsel
asel,u,,,4,8,4
arsym,z,all !2nd part of the model by symmetry reflection!
allsel
nummrg,kp
nummrg,node
!boundary conditions!
nsel,s,loc,y,0
dsym,symm,y
allsel
nsel,s,,,3056
d,all,uz,0
Annex 3: Input Listings
Page
A 3.38
DBA
Design by Analysis
d,all,ux,0
allsel
nsel,s,,,2066
d,all,uz,0
allsel
eplot
!pressure load only jacketed part of main vessel under external pressure!
asel,s,,,18,19
asel,a,,,35,36 !inner shell!
sfa,all,1,pres,-0.7
allsel
/psf,pres,norm,2
aplot
/solu !calculate static solution!
pstres,on
solv
fini
/solu !calculate the eigenvalue buckling solutions!
antype,buck
bucopt,subs,2 !calculate the first 2 eigenforms!
mxpand,2 !expanding the solution!
solv
save
fini
!Fully nonlinear analysis!
imper,1,20.15 !use macro imper, mode 1, max. displ. 20.15 mm!
/prep7
sfdel,all,pres !delete pressure load!
sfadel,all,,pres
eplot
/psf,pres,norm,2
!New Pressure load!
esel,s,real,,2 !cylinder main vessel outside of jacket!
sfe,all,1,pres,,0.75
allsel
esel,s,real,,1 !dished end!
sfe,all,1,pres,,0.75
allsel
esel,s,real,,4 !plates of jacket!
sfe,all,1,pres,,0.75
allsel
esel,s,real,,5 !jacket!
sfe,all,1,pres,,0.75
allsel
esel,s,real,,3 !cylinder jacketed part of main vessel!
sfe,all,1,pres,,-0.875
allsel
!temperature load!
esel,s,real,,3 ! cylinder jacketed part of main vessel!
bfe,all,temp,1,128.25,25
allsel
esel,s,real,,6 !inner ring!
bfe,all,temp,1,25
allsel
esel,s,real,,4 !plates of jacket!
esel,a,real,,5 !jacket!
esel,a,real,,7 !outer rings!
bfe,all,temp,1,12.5
allsel
esel,s,real,,1 !dished ends!
esel,a,real,,2 ! cylinder main vessel outside of jacket!
bfe,all,temp,1,200
allsel
!temp. dependent material properties!
mptemp,,0,20,100,200
mpdata,ex,1,1,1.98e5,1.96e5,1.90e5,1.82e5
mpdata,alpx,1,1,16e-6,16.1e-6,16.7e-6,17.2e-6
tb,bkin,1,5
tbtemp,10
Annex 3: Input Listings
Page
A 3.39
DBA
Design by Analysis
tbdat,1,240,0
tbtemp,20
tbdat,1,240,0
tbtemp,100
tbdat,1,208,0
tbtemp,150
tbdat,1,196,0
tbtemp,200
tbdat,1,186,0
/solu
antype,static
solcon,on
nlgeo,on !use nonlinear geometry!
nropt,auto
autots,on
nsubst,10,100,10
neqit,30
outres,all,all
/output,out,txt
solv
save
Macro imper
/post1
set,1,arg1
/nopr
nsort,u,sum,,1
*get,umx,sort,,max
umul=arg2/umx
/com Anz. Knoten
*get,nnod,node,,count
*dim,ukn,array,nnod,3
ikn=0
ik=0
:a01
*get,ikn,node,ikn,nxth
*if,ikn,eq,0,:e01
ik=ik+1
*get,ukn(ik,1),node,ikn,u,x
*get,ukn(ik,2),node,ikn,u,y
*get,ukn(ik,3),node,ikn,u,z
*go,:a01
:e01
fini
/prep7
modmsh,deta
ikn=0
ik=0
:a02
*get,ikn,node,ikn,nxth
*if,ikn,eq,0,:e02
ik=ik+1
*get,kxkn,node,ikn,loc,x
*get,kykn,node,ikn,loc,y
*get,kzkn,node,ikn,loc,z
kx=(kxkn+(ukn(ik,1)*umul))
ky=(kykn+(ukn(ik,2)*umul))
kz=(kzkn+(ukn(ik,3)*umul))
n,ikn,kx,ky,kz
*go,:a02
:e02
/gopr
finish

You might also like