You are on page 1of 37

PRESSURE VESSEL TECHNOLOGY:

PAST, PRESENT and FUTURE



John Spence and John Darlaston
George Stephenson Lecture
1998











































The copyright of this paper is the property of the Institution of Mechanical Engineers. Apart from
any fair dealing for the purpose of private study, research, criticism or review, as permitted under
the Copyright Designs and Patent Act 1988, no part may be reproduced in any form or by means
without permission. Enquiries should be addressed to:

Director of Publications and Information Services, the Institution of Mechanical Engineers,
1 Birdcage Walk, London SW1H 9JJ, Telephone: + 44 (0)20 7222 7899

The Institution is not responsible for any statement contained in these papers. Data, discussion
and conclusions developed by the authors are for information only and are not intended for use
without independent substantiating investigation on the part of potential users. Opinions are
those of the author and not necessarily those of the Institution of Mechanical Engineers.

CONTENTS
0 BIOGRAPHIES
...
111
0 SYNOPSIS
0 PREAMBLE
0 THE GEORGE STEPHENSON LECTURE
1. THEPAST
1.1
1.2 Basic Design Rules
1.3 Computer Aided Design
1.4 Development of Standards
1.5 Inspection and Quality Assurance
George Stephenson and His Work
2. THE PRESENT
2.1 Design-by-Rule
2.2 Design-by-Analysis
2.3 Fatigue Design
2.4 Fracture
2.5
2.6 Current Standards
Materials, Manufacture, Inspection and Testing
3. THEFUTURE
3.1 Risk-Based Assessment
3.2 Equipment Qualification
3.3
3.4 The European Standards
3.5
3.6 The International Situation
The European Pressure Equipment Directive
The New European Unfired Pressure Vessel Standard
2,
V
1
5
5
7
13
16
18
19
19
19
20
21
21
22
24
0 CONCLUDI NG COMMENTS
25
0 REFERENCES
26
BIOGRAPHIES
PROFESSOR JOHN SPENCE
BSc ARCST MEng PhD DSc FEng FIMechE FRSE
Served an engineering apprenticeship with Stewarts and Lloyds
and later spent seven years with the Research Division of
Babcock and Wilcox Limited. He is a graduate of the Royal
College of Science and Technology, the University of
Strathclyde, and the University of Sheffield. At the University of
Strathclyde he holds the Trades House of Glasgow chair of
Mechanics of Materials and was until 1994 Head of the
Department of Mechanical Engineering; he is now Vice-Principal
with University-wide responsibilities for Quality Assurance of
all Learning and Teaching provision. His technical background is
in Structural Mechanics and Materials Behaviour with a special
interest in Pressure Vessels and Piping. He has over 100 papers
published in the technical press. He has served on various
national committees including BSI, DTI, Research Councils,
RAE Panels; current committees include EPSRC TOP, Advisory
Council for the Development of Royal Navy Personnel,
Accreditation Board of the Hong Kong Institution of Engineers
(HKIE) and the Scottish Higher Education Funding Council (SHEFC); he is immediate past chairman
of the Engineering Professors Council. He is currently President of the Institution of Mechanical
Engineers and has served on many of its main boards and committees. He has acted as External
Examiner for various undergraduate and postgraduate courses in the UK and overseas.
JOHN DARLASTON MBE
BSc(Eng) FIMechE FWeldI CEng
John Darlaston is a Fellow of the Institution of Mechanical
Engineers, the Welding Institute, a Member of the Institution of
Nuclear Engineers, a Chartered Engineer and a Member of the
Federation of European Engineers (FEANI). He was employed
with Nuclear Electric (CEGB), Berkeley Technology Centre for
over 30 years in the areas of research and structural mechanics
and, has served on several technical and professional committees
and groups including those specifically dealing with European
and international standards. He is a visiting professor of the East
China University of Science & Technology and an industrial
Fellow of mechanical engineering of the University of Bristol.
He has over 50 publications in journals and books and has
presented numerous conference papers both nationally and
internationally. John is currently Principal of Strutech
Consultancy, which deals with national, European, and
international activities in the field of pressure equipment including integrity assessment, plant life,
equipment qualification, experimental validation and standards. A principal activity is the Secretariat
for the European Unfired Pressure Vessel Standard. He is Chairman of the Pressure Systems Group of
the IMechE, Chairman of the International Council on Pressure Vessel Technology (Europe and Africa
Region), and Chairman of the UK Pressure Equipment Co-ordination Forum and Vice Chairman of
the European Pressure Equipment Research Council.
...
Ill
SYNOPSIS
Pressure vessel technology plays an important role in many branches of engineering. Equally many
aspects of engineering and science play an integral part in the technological area. It is an appropriate
subject for the George Stephenson Lecture since pressure vessels contributed considerably to the
developments with which he was associated.
The simple pressure vessel is called upon to perform many functions. The developments in the
technology associated with the early pressure vessels were generally based on practical experience. As
the demands on systems grew so the technology of pressure vessels benefited from an increasing
understanding of the underlying engineering and scientific principles. Present day pressure vessel
technology brings together many disciplines and the greater the requirements on the vessel the more
this is so. This increase in understanding can easily lead to applying complex approaches and standards
even to simple pressure vessels. There must always be a balance between the use of the vessel, the
essential safety requirements and the economics. The prime consideration must be to provide a vessel at
an economic cost, which will meet the performance and safety requirements. The developments in
legislation and standards are regularising all aspects of pressure vessel technology, providing safer and
more economic equipment.
The Lecture considers the past, present and future of pressure vessel technology.
PREAMBLE
The Institution of Mechanical Engineers was founded in Birmingham on 27 January 1847 and George
Stephenson was its first President. The 150th Anniversary Celebrations included a Meeting of the
Institution in Birmingham on 27 January 1997. The Pressure Systems Group was announced as the
Group of the Year on that occasion. The George Stephenson Lecture for 1998 is very much a
presentation on behalf of the Pressure Systems Group of the Institution.
V
THE GEORGE STEPHENSON LECTURE
The George Stephenson Research Fund was established in 1923 to commemorate the First President of
the Institution of Mechanical Engineers. One of the purposes of the fund is to assist the delivery of the
George Stephenson Lecture to a society outside Great Britain on subjects prescribed by Council.
Although the fund was used for other purposes the first presentation of the Lecture was made in
1987, delivered to audiences in Hong Kong and Singapore. The lecture is one of prestige and given by
speakers of eminence with some nine presentations to date.
Traditionally, the Lecture Paper is published within the relevant part of the Institution Proceedings
and the lecturer(s) is presented with a certificate to mark the occasion.
1. THEPAST
1.1 George Stephenson and his
It is appropriate to recognise the work of George
Stephenson Fig. 1. George Stephenson is known
as the Father of the Railways. This is
commemorated on the reverse side of the Bank
of England 25 note, which recognises his
contribution to the development of the railway
as a means of conveying goods and people. The
George Stephenson room at the Institution of
Mechanical Engineers in London contains many
items commemorating his life and work.
George Stephenson was born at Wylan in the
Tyne Valley in 1781 into a mining family, a man of
humble origins, who followed his father into the pit.
From the beginning he showed that his true talent
lay with machines and machinery. It was not long
before he was put in charge of a pumping engine.
Soon he was to become a Bankman with the
responsibility of working a winding engine that lifted men and coal between the pit bottom and the
surface, or bank. His reputation as an engineer grew steadily. The engineering world of the early
nineteenth century was very different from today. There was virtually no theoretical work and the
majority of engineers relied on practical experience. An agile brain was required together with the ability
to transfer the idea in the head to the work of the hands. George Stephenson had these abilities in full.
George Stephenson did not invent the railway or the steam locomotive. He did however have
tremendous drive, self-confidence and vision, which provided the essential catalyst that took Britain into
the Railway Age. Before his time, the distance a horse could travel in a day, which was about 25 miles,
bound mans horizons. The early steam engines were stationary affairs; it was Richard Trevithick, and
later George and Robert Stephenson, who exploited the idea of utilising the steam engine for the purpose
of self propulsion or locomotion. The early Trevithick locomotive designs were unreliable and prone to
blow-ups! George Stephenson was commissioned to rebuild a failed pumping engine and then to build a
locomotive for the Killingworth mine in the north east of England. This latter proved to be a success and
was followed by others. In 1823 he set up a locomotive manufacturing company with his son Robert and
named the company Robert Stephenson and Company. However, his reputation was greatly enhanced at
national level when he became the engineer for the Liverpool and Manchester Railway. The successful
completion of this line was the beginning of far reaching social, economic and industrial changes in the
years immediately before the formation of the Institution of Mechanical Engineers in 1847. By the end of
Stephensons life all of Britain had been opened up, towns and cities were connected, manufacture and
commerce flourished. The famous Stephensons Rocket is shown in Fig. 2.
Fig. 1 George Stephenson, First President of IMechE
1
Fig. 2 Stephensoni Rocket
George Stephenson and his colleagues developed
pressure vessels and pressure systems technology from
practical experience. There were no safety inspectors in
their day, but as responsible engineers they attempted
to ensure safety and, hopefully, their reputation.
In passing, it is worth noting that the practical
experimental approach to engineering development
allowed British engineers to be extremely creative
during the eighteenth and nineteenth centuries.
Equally, the lack of a scientific foundation was possibly
the reason for the erosion of that pre-eminence in the
second half of the nineteenth century when other
countries had a more coherent and planned approach
to the application of new technology. Scottish
universities were teaching mechanical science in the
mid-eighteenth century and James Watt and his
contemporaries benefited. It was well into the
nineteenth century before engineering became
accepted as an academic subject in English universities;
the first professor of mechanical engineering, W Eaton
Hodgkinson, was appointed at University College in
1847 the same year as the Institution was formed.
1.2 Basic design rules
An example of one of the first welded pressure vessels stands in front of the Combustion Engineering
offices at Chattanooga in the USA. This vessel was grossly over designed, was cheap to manufacture,
and was only visually inspected, but represented a major step in the development of pressure vessel
technology. The design as such was based on the membrane stress not exceeding a factored value of the
tensile properties of the material. The procedure was based on years of experience. In the UK in 1854
the Manchester Steam Users Association was set up to . . . disseminate information about boiler
explosions. Over a 25-year period at the end of that century the UK Board of Trade investigated 1871
boiler explosions which had led to 732 deaths. Lloyds Register of Shipping issued the first design rules
Tentative requirements for fusion welded pressure vessels, in 1934. Experience was the major
contributor to the development of the early design rules. To some extent this is still the case although
many of the developments are based on analysis with validation testing.
Parallel developments took place in USA where the Polytechnic Club was formed in Hartford in 1854
with the aim to avoid boiler explosions through good materials, fine workmanship, careful operation and
periodic inspection. The first rules in USA were adopted in Massachusetts in 1907 but they were only
three pages long. Later, in 1915, the first ASME Boiler and Pressure Vessel Code was issued.
The next significant step in the development of design rules was the consideration of fatigue
resulting from the cyclic loading of the vessel. The maximum stress was limited to a proportion of the
fatigue limit of the material. Thus the elements of design began to develop.
(a) Failure Modes
As design rules were developed there was a conscious effort to avoid various modes of failure. These
can be stated typically as follows:
Bursting of the vessel wall
Tearing at a discontinuity
Excessive deformation
Buckling
Ratchetting
Brittle fracture at a defect
Fatigue and low cycle fatigue
Corrosion fatigue
Creep rupture at elevated temperature
Stress corrosion cracking and environmental effects
Vibration
2
The technical issues covered in the design rules for pressure vessels have been, and are still being,
developed to guard against anticipated failure modes. Fig. 3 shows a vessel tested at MPA Stuttgart
which is a good example of research work designed to examine failure modes. The technical issues
generally are related to the following:
Elastic stress analyses
Shakedown
Limit analysis
Thermal transients
Residual stress
Distortion
Dynamic and seismic responses
Manufacturing tolerances
Imperfections and defects
These issues are essentially the input to
the design philosophy, which guards against
the failure modes. The design codes, which
embody these design rules, usually include
consideration of design stresses, materials,
manufacture, inspection and testing. Fig. 3 The MPA (Stuttgart) Vessel
Elastic Stress Analysis
(b) Concepts of Design
Stress Concentration Factor Maximum Stress
Nominal or Membrane
Stress
SCF
For more than a decade, the Pressure Systems Group of the IMechE, in conjunction with the University of
Strathclyde, has run a series of courses on Pressure Vessel Design. The concepts of design are covered in
considerable detail in these courses. Several of the aspects brought out in these courses are used in this paper
Cracked Bodies
and duly acknowledged.3
Some topics, concepts
and characteristics which
are usually considered,
implicitly or explicitly, in
pressure vessel design, can
be identified as in Table 1. It
is helpful to think of each of
the topics as giving rise to
concepts, which in turn have
characterising parameters or
criteria.
In this context a concept
is a convenient way of
distilling a complex analysis
into a simple framework to
capture the essence of the
behavioural complexity in a
simple manner. This in turn
allows the situation to be
characterised by a parameter
(often a simple number)
and/or compared against
some design criterion. Note
that the complexity increases
as one goes down the list of
topics in Table 1.
LEFM
PYFM
GYFM
Table 1: Pressure Vessel Design Topics
Time Dependant
Topics I Concepts I Characterisations
I
Elastic Analogy
Reference Stress
Damage
Plasticity First Yield
Shakedown
Limit Load
Yield Pressure
Shakedown Pressure
Limit Pressure
Stress Intensity Factor
Toughness
Fatigue Crack Initiation
Crack Propagation
Cumulative Damage
Number of Cycles to Initiation
Number of Cycles for Propagation
Cumulative Damage
Steady State
Reference Stress/
Displacement/Time
Damage Parameters
Combinations Ratchetting
Creep/Fatigue
Fracture
Damage Summations
Bree Type Diagrams
The topics identified above have been developed and form the basis of the present design
approaches. Some of these are discussed in Section 2 of the paper.
3
1.3 Computer Aided Design
The development of the computer and the associated methods such as finite difference and finite
element have led to enormous steps forward in pressure vessel technology. Analysis has become the
dominant feature and is tending to replace experience and testing although the latter still have an
important part to play. One of the problems that existed in the early days of computer analysis was that
people with little understanding of practical engineering carried out the analysis. Although the
computer model would represent the loading, the geometry and simple material behaviour, there was
much more to the problem than the idealistic model. As the computing procedures have evolved so has
they way they have been used. The computer programmer now works with the engineer to develop a
computer method. Finite element methods are now tools to be used by engineers although
considerable experience is still required if one is to avoid misleading results. Computer methods allow
the engineer to investigate a range of variables within the design process. This contributes significantly
to the basic requirement of a safe design within an economic framework.
There is a second way in which the computer has transformed design. A number of companies have
computerised the design standards, i.e. the basic code calculations can be conducted via the computer
exactly as they would be by hand but more accurately, consistently and swiftly. This is especially useful
for iterative type calculation for routine work and for looking at different possibilities. Of course, it is
important to keep such programmes up-to-date as the standards change.
1.4 Development of Standards
A considerable amount of common material exists between the major pressure vessel standards. There
are also many parallels in their development and history. This is particularly so between BS 5500 4 and
ASME VIII Div. 1 5. It is sufficient here to focus on the British scene by way of example.
Tentative efforts to produce a British Standard specification for fusion welded vessels were made as
early as 1939. A serious start in the preparation of what became BS 1500 6 was made in 1940. This
resulted from approaches to BSI by the Institution of Chemical Engineers. It is interesting that
economics rather than safety drove this approach. The various BSI specifications of the day for welded
pressure vessels were rather more conservative than those issued jointly by the American Petroleum
Institute & the American Society of Mechanical Engineers. This was having an unfavourable effect on
the export trade.
The first draft British Standard ran to 94 pages and contained 56 graphs and sketches. Design stress
tables were prepared but were omitted from the first version as it was considered politic to concentrate
on the scope and format of the proposed code. It became apparent that a series of supporting
specifications for materials would be necessary. This was against BSI practice at the time. However, draft
documents were prepared for plate, forgings, tubes, etc.
On a historical point, in 1943, a current member of the IMechE Pressure Systems Group met in
London with other colleagues to discuss pressure vessel standardisation. The meeting was organised by
BSI. Over a period of a few meetings, the format and requirements of a master pressure vessel standard
were set down. The notes of these meetings show that BS 5500 and the European UFPV Standard follow
the lines proposed by this group for a master code.
In the 1950s the BS 1500 Standard 6 was published. Later on, in 1965, BS 1515 7 was published which
included a more detailed and advanced approach leading to higher allowable stresses. In the late 1960s
the Morrison Enquiry into the Pressure Vessel Industry proposed that all the pressure vessel interests of
BSI should be under one committee (IVE/-) to rationalise the activities. This led on to the formulation
of BS 5500, the master Pressure Vessel Standard, which was first issued in 1976. This standard is used
widely in many industries and by many countries and is updated annually and reissued every 3 years.
In BS 5500, as with the ASME Standard of that time, there were two approaches to design. Firstly
Design-by-Rule, which involves the use of formulae and rules to determine the vessel thickness and
keep the nominal stresses below the tabulated values. Secondly Design-by-Analysis, which involves
the use of stress analysis to keep certain calculated values below the allowable tabulated values. These
approaches are discussed in more detail later in the paper, together with some specific aspects of the
current design of pressure vessels.
4
1.5 Inspection and Quality Assurance
Inspection and quality assurance, in the broadest sense, have influenced the development of design
procedures and standards.
As design procedures developed it was recognised that there was a need to link the degree of inspection
to the duty and importance of the vessel. As with many issues in the field of pressure vessel technology this
was driven, to some extent, by the economics. For example, in BS 5500, as with other standards, there are
different categories of construction, namely 1, 2 and 3. Category 1 is the highest level and relates to the
general design levels in the Standard; it requires 100% NDT of all welded joints. By contrast, Category 3
requires visual inspection only but is restricted in material and has lower design stress levels.
The process was defined in a more formal manner in the mid-1970s when the experiences from the
space industry were taken up by other industries. At this time the broader subject of quality assurance
started to make its mark with pressure equipment. Quality assurance can be defined as 'all those
planned and systematic actions necessary to provide adequate confidence that a structure, system or
component will perform satisfactorily in service'. Inspection during manufacture and in service is an
important issue. Equally, the quality assurance approach throughout design, manufacture, inspection
and operation of a vessel provided ensures that the safety requirements are met.
2. THE PRESENT
The situation today is that there is a good provision of national standards relating to basic pressure
vessel design. Many of these have developed independently but all owe a great deal to the ASME codes.
Pressure vessel technology is crucially dependent on design; this section covers the two main design
approaches, namely Design-by-Rule and Design-by-Analysis and touches on more specialised areas
such as fatigue and fracture which are generally covered less well in the basic design standards.
2.1 Design-by-Rule
The main philosophy of virtually all National Design Standards is based on the concept of Design-by-
Rule. Indeed, the approach has been enshrined in Pressure Systems Design Standards for so long that it
is part of the culture of the industry. Essentially it involves relatively simple calculations to arrive at
basic scantlings via an allowable standardised design stress. Thereafter component details are dealt with
by strict adherence to specific rules delineated in the Standard. The origin of the rules is not always
obvious but they usually incorporate considerable experience.
(a)
Tabulated values of allowable design stresses are given for various materials and temperatures. For
example, in BS 5500 the basic yield and ultimate stress is given for each material, and where relevant,
design values for a range of temperatures up to 480" C. Materials, other than those specifically listed in
the design tables, are allowable provided the design values are derived in a manner consistent with those
in the Standard.
The basis of the design allowables is given in BS 5500, Annex K, and are to be derived from actual
material data from tensile tests. Stated simply, the design stress (f) is typically the lower of
Material Properties and Allowable Design Stresses
- Re Rm srt
or - or -
1.5 2.35 1.3
where
Re is the minimum value of the specified yield strength for the grade of steel concerned at room
temperature or at temperature, Re (T)
R, is the minimum tensile strength (i.e. ultimate stress) at room temperature
S,, is the mean value of the stress required to produce rupture in time t at temperature T
5
Slightly different values may apply particularly for austenitic steels. The actual allowable values are
tabulated in BS 5500.
In passing, it should be noted that, although basic material properties and the allowable design
stress tend to be taken for granted, their importance should not be forgotten. They dictate the basic
scantlings and consequently have a strong influence on the total costs.
(b)
As for other national standards simple formulae are given in BS 5500, section 3, for standard geometries
which when used with the design stress f leads to the minimum thickness of the vessel or component.
The basic idea of Design-by-Rule is that once the leading scantlings are fixed in this way the designer
simply obeys the rules laid down in the procedures for specified components such as nozzles, heads,
flanges, etc. The methodology or reasoning behind the rules will not always be apparent. However, this
is the most common approach used in all national design codes.
One of the simplest examples of Design-by-Rule is the idea of area replacement for nozzles. The
origin of the idea is obscure. Simply expressed one replaces the area cut away by the cross-section of the
hole in the shell and relocates it around the hole close to the cut out. The area replaced is in addition to
any thickness required to meet the basic pressure strength of either shell or nozzle. Notice it is an area
replacement rather than a volume replacement. The idea is clearly seen by reference to FiE. 4.
The basic concept and examples of Design-by-Rule
I
ial Set-through branch
,r- - i- - -
- 7--
I
( b ) Set-an branch
l
lc) Opening
The area Y should not be less than the area X
Tand t are the thicknesses calculated by the equations for
pressure loading only in 3.5 1 to 3.5.3 for the appropriate
shape of sections, N is the smallerof the t wo values 2 5r,
and 2 5T,. and S IS the greater of the t wo values
IT,+ 75) mm and d, / 2
Fig. 4 Area Replacement
-
The disposition of the replaced area is important.
Although the stress field is increased local to the opening it
dies out in a relatively short distance so that to be effective
the replaced material needs to be close to the edge of the
opening. This distance is usually a function of @? where R
and T are the radius and thickness of the particular shell
respectively. The multiplier used with d a is not exact and
different national standards use different factors. However,
the point is that, whatever the origin of the method, the
designer simply obeys the rules. The area replacement method
has largely fallen into abeyance since it has been superseded by
more elegant approaches (it is still used in ASME Div. 1). The
current pressure area method in Annex F of BS 5500 has
similarities to the area replacement method which is has
displaced. It is also a good Design-by-Rule example.
In the design section 3 in BS 5500 the openings and
nozzle designs are based upon either a constant stress
concentration factor of 2.25 or a shakedown type criteria,
each of which apply in different geometric regimes. These
criteria have allowed the construction of simple rules based
on factors provided in graphs or tables. Some of the work
(nozzles in spheres) is based on thin shell theory and some
(large nozzles in cylinders) is empirically based coupled with
simple shakedown calculations. However, the designer is
simply required to satisfy the rules in the appropriate section
without necessarily knowing the background.
In the main design section of a typical design standard it
will not be obvious what the underlying philosophy or
criteria are (maximum stress limitation, shakedown, limit
load, etc) nor what evidence has been used to construct the
rules or the graphdtables (theoretical analysis, experimental
data or a mixture of both). There may also be factors of
ignorance built in where the data is uncertain.
The main strength of the Design-by-Rule approach is
that it has the great benefit of simplicity and consistency and
is backed up by experience but it has obvious limitations.
6
The approach cannot be easily extended either to different geometries outwith those covered by the
Standard or additional loadings beyond the normal pressure case.
Unless such additional loads are rather small it is unclear how the situation can be tackled logically
since the criteria used in Design-by-Rule are in general not well specified. Arguably also, it leaves the
designer in the dark regarding the limitations of the method, which does not foster a good approach to
design.
2.2 Design-by-Analysis
Essentially Design-by-Analysis is based on the idea that if a proper stress analysis can be conducted
then a better, less conservative, assessment of the design can be made than would otherwise be the case
by the usual approach of Design by Rule. The philosophy originated in the 1960s in America although
there was encouragement and interest from the UK from the beginning. The motivation was driven by
the sophisticated design work being conducted in the nuclear industry during that period. There were
many pressure vessel features which were not covered directly by the existing methods in the available
standards and consequently there was a large investment in detailed stress analysis not only in the USA
but in the UK and elsewhere. These analyses were primarily thin shell type analyses; it is therefore
important to note that the Design-by-Analysis ideas developed against this background of thin shell
analysis and in particular the analysis of discontinuity effects including thermal discontinuities.
It was suggested that different types of stress had different degrees of importance and this led to the
idea of categorisation of stress. Once allocated to categories the stresses are cast in the form of stress
intensities to reflect the Tresca yield criteria and then compared with specified stress limits which are
set at different levels for the different stress categories. The methodology was first incorporated into the
ASME Boiler and Pressure Vessel Code Section I11 and Section VIII Division 2 in 1968 and then a little
later into BS 5500 as Appendix A (now Annex A). Many other countries have now adopted the same
basic approach.
Annex A, frequently referred to as Design-by-Analysis, is intended to cover design where the
loadings and components are not covered by the rules of the design section 3. The various aspects of the
Design-by-Analysis approach are outlined in what follows.
(a) Multiaxial Stress States
In real vessels the stress system will usually be two or three dimensional, i.e. there will be stresses in
different directions. It is well known that in the presence of multiaxial stress states, yielding is not
governed by the individual components of stress but by some combination of all the stress components.
The most common theories of yielding for metallic type materials are the von Mises criterion and Tresca
criterion. Most Design-by-Rule approaches simply use the individual maximum stress values (although
increasingly there are exceptions to this) but the Design-by-Analysis approach uses a more accurate
representation of the multiaxial state in order to gauge its proximity to yield. To date, pressure vessel
standards use the Tresca criterion. If ol, o2 and o3 are the principal stresses at a point in a component,
the Tresca criterion states that yield will take place when the maximum shear stress reaches the shear
stress in a tensile specimen at yield, i.e.
(o1-o2)/2 or (o,-o,)/~ or (o,-o,)/~ = +0,/2
The stress differences denoted by S12,S23,S31 are simply twice the shear stresses; then
A new term called equivalent intensity of combined stress or the Stress Intensity, S is defined as
i.e. S = max ( I S,, I, I S23 I ,I S311) so that the Tresca criterion reduces to
the maximum absolute value of the stress difference,
s = Oy
7
(b) Stress Categories
Certain types of stresses may be more significant than others and these are assigned to different
categories with different levels of importance having different stress limits.
The ASME chose the following categories:
A. Primary Stress
(i) General Primary Membrane Stress, fm (P,)
(ii) Local Primary Membrane Stress, f, (P,)
(iii) Primary Bending Stress, f b (Pb)
B. Secondary Stress, f, (Q)
C. Peak Stress, f p (F)
The symbols shown are those for the BS with ASME symbols in brackets. It seems rather
The chief characteristics of each category are described in.BS 5500, Annex A and summarised below:
(i) Primary stress is a stress developed by the imposed loading, which is necessary to satisfy the
laws of equilibrium between external and internal effects. The basic characteristic of a primary
stress is that it is not self-limiting. If a primary stress exceeds the yield strength of the material
through the entire thickness, the prevention of failure is entirely dependent on the strain-
hardening properties of the material.
Secondary stress is a stress developed by the self-constraint of a structure. It must satisfy an
imposed strain pattern rather than being in equilibrium with an external load. The basic
characteristic of a secondary stress is that it is self-limiting. Local yielding and minor distortions
can satisfy the discontinuity conditions or thermal expansion, which cause the stress to occur.
(iii) Peak stress is the highest stress in the region under consideration. The basic characteristic of a
peak stress is that it causes no significant distortion and is objectionable mostly as a possible
source of fatigue failure.
The sub-division of the primary category into membrane and bending is justified on the basis that
the level of the allowable stress can be higher for the bending case since gross distortion is not imminent
when the surface stress levels reach yield in the bending case. Primary bending arises for example in flat
plate type structures, which are subjected to pressure such as in a flat end in a cylindrical pressure
vessel. Local primary membrane stress is more difficult to envisage, but may arise from a local loading
although it may also arise due to pressure loading at discontinuities.
unfortunate that different symbols were used.
(ii)
(c) Allowable Stress Limits
In setting the stress limits attention is concentrated on three modes of failure. These are paraphrased as
follows.
0 Avoidance of gross distortion or bursting
0 Avoidance of ratchetting
0 Avoidance of fatigue
and they will be considered in turn.
(i)
Primary Stress Limits: Avoidance of Gross Distortion or Bursting
For an elastic perfectly plastic stress strain law the pressure vessel wall would be fully plastic when
the membrane stress reached the yield stress level. Understandably most standards require a
reasonable margin to avoid this situation and the maximum membrane stress at design conditions is
governed by the allowable design stress f which in general is restricted to 0,/1.5. This is t hej r s t of
t he stress limits.
8
When both direct and bending stresses are
N
present the avoidance of gross distortion or
bursting is slightly more complex. The objective
is to have a reasonable margin on the situation
where locally the full wall section of the vessel
becomes plastic. The ASME approach was to
a direct end force and a bending moment and
relate this to what happens in the wall of a
pressure vessel. If one considers a cylindrical
vessel then the appropriate direct stress is in the
circumferential direction, the thickness is the full
thickness of the vessel and the beam width is any
convenient short axial length along the vessel.
2h, subjected to a direct end load N and a
bending moment M. If the material is elastic, perfectly plastic with yield stress oy, then with N
tensile, yield first occurs in tension in the outer fibre when
2h
consider a simple rectangular beam element with
N
N
Fig.
shows such a beam, width b, depth Fig. 5 Beam Element with Direct and Bending Actions
If the loads are increased beyond this combination there is redistribution of stress since the
stress cannot increase anywhere above o Gradually the portion which reaches oy will increase
from the outer surface(s) inward until a situation pertains where the stress is everywhere equal to
+oy or -oy through the thickness as shown in Fig. 6. Any combination of loads which produces
such a condition of full plasticity is known as a limit state. The conditions for the limit state are
easily derived from equilibrium since the stress distribution must balance the applied loading. Thus
Y
We may now draw the initial yield condition and the limit condition in Fig. 7 in non-
The limit for a beam in pure bending
(N = 0) is given by ML = oybh2. Notice that
the first yield moment acting alone is My = 5
o,bh2 so that ML/MY = 1.5 as is well known
for the beam.
Fig. 7 can be redrawn in an alternative
form where the quantity N/2bh may be
interpreted as the elastic membrane stress,
om and the quantity 3M/(2bh2) may be
interpreted as the elastic bending stress, (sb, at
the outer fibre. The maximum stress is then
dimensional load space.
t uy
DIRECT + BENDING
1
-- - - f--..# =$z
1
ELASTIC UP TO FIRST YIELD
o( max) = om + o b
-. .p ~ I the initial yield is given by
om + (sb = oy
and the limit condition can be written as
CfY
BEYOND FIRST YIELD FULLY PLASTIC
Fig. 6 Stress Distributions under Elastic and Plastic Conditions
2
9
I .O
2
3
-
0
//// //////
REGION DESIGN
/
/
/
/
M
t uybh2
I + / YIELD
urn
-
N
1 .o 2 6 h ~ y
Fig. 7 Load Combinations for Yield and Full Plasticity
um+b
--
--
I (JY QY
1 .o
0, + 0, so,
0
1 .o
2
-
2
J 2
0, 5-uy
Avoid Gross Distortion 3
Fig. 8 Design Limits to
-
QY
N
2bh
3 M
2bh2
-
--
These equations are plotted as o(max)
and 0, in Fig. 8, which alters the shape of
the diagram. It is important to realise that it
is essentially the same diagram. It was in this
form that ASME presented the diagram.
The design limit on the membrane
stress has already been mentioned. It was
decided to limit the allowable value of
o(max) to the yield stress 0,. The design
limits therefore are
2
3
om 5 - 0,
and are shown in Fig. 8.
In addition, the membrane stress locally is
allowed to increase to yield under certain
restricted conditions. Although contributions
to this stress may have the characteristics of a
secondary stress for convenience it is classed
as primary (fL).
In stress category parlance these primary
stress limits then become
2
3 y
f , I - 0
f, soy
f , +f b soy
f, + f b 5 (Sy
(ii)
10
Secondary Stress Limits: Avoidance of Ratchetting OY Repeated Plastic Straining
The allowable levels of secondary stress (in addition to primary stress) are related to the avoidance
of repeated plastic straining or ratchetting, which is sometimes termed incremental collapse. If a
structure is repeatedly loaded to progressively higher levels one can imagine that at some highly
stressed region a stage will be reached when plastic strain will accumulate during each cycle of load.
Except for very special cases this situation should be avoided. However, some initial plastic
deformation is judged permissible during the first (few) cycle(s) of load provided the situation
settles down to elastic behaviour for subsequent loading cycles. This is termed shakedown
behaviour and the highest loading or stressing where shakedown can be achieved is used as the
limit for secondary stresses.
Consider Fig 9, which is intended to represent a possible loading on the outer fibre of a
beam, which in turn may represent an element from a pressure vessel. It may be helpful to think
of a local thermal loading on a shell; for example, if a cool spot is induced on a warm shell the spot
will be strained in tension. Consider then that the surface fibre is strained in tension to a value E~
(the strain range) assumed somewhat greater than the yield strain cy so that the stress strain curve
follows the path OAB. Again an elastic perfectly plastic material is assumed.
A
SiRAlN HISTORY BEYOND YIELD
Fig. 9 Design Li mi t s t o Avoid Ratcheting
If we now assume that the nature of the loading
is such as to return through the same strain range to
zero strain, then the fibre will have a residual
compressive stress (0, -EER) as shown at point C.
On any subsequent loading this residual
compression must first be removed before the
material goes into tension, i.e. the path CB is
followed and the elastic range has been increased in
size from OA to CB. Provided this stress at C in the
outer fibre is less than the yield stress the behaviour
is said to have exhibited shakedown. When the strain
range is increased to D so that on unloading the
residual stress coincides with the compressive yield
stress (E), a total elastic stress range (ED) of 20, is
attained. This is the maximum possible elastic range.
In a design context it was considered that the
limit of 20,, could be regarded as a threshold beyond
which some plasticity action would progress.
Accordingly it was decided that E E ~ , interpreted as
an elastically calculated stress range oR, could be
allowed to be equal to the shakedown limit, i.e.
This allowable stress limit is used f or all secondary type stresses regardless of origin. Thus
f , +f b+f g 5 20, f, + f, + fg I 20,
(iii) Avoidance of Fatigue Failure
As already mentioned the peak stress is the highest stress in the region under consideration. That
part of the peak stress, which is additional to the general stress levels evaluated by shell theory for
example, may arise from local effects such as the detailed weld profile or a small hole or filet.
Whereas these effects may be often ignored in general design they need to be considered in fatigue.
It was considered in ASME and until recently in BS 5500, that the peak stress intensity should be
limited by the value obtained from the fatigue S-N curves provided. The general approach in BS
5500 has been modified and this will be discussed later.
( d) Relationship of the Stress Limits t o the Stress Categories
The allowable stresses in the Standard are not expressed in terms of the yield strength as indicated above
but rather as multiples of the tabulated design value f. In BS 5500 the design is governed by both yield
and ultimate stress but the limits can be transferred to multiples of f by assuming that f = oy/1.5. The
design allowables are summarised in the following table.
Stress Intensity Allowable Equivalent
Stress 1 Yield
General primary membrane, f,
Local primary membrane, f,
Primary membrane plus bending
Primary plus secondary
(f, + fb) or (fL + fb)
(f, + f b + f,) or (fL + f b + f,)
f
1.5f
1.5f
3f
11
The Standard also summarises the stress limits and the various categories in a so-called hopper
There are a number of special cases for limited application, for example.
For attachments and supports the limits are
diagram in Annex A.
The membrane stress intensity
Membrane + bending stress intensity
5 1.2f (0.80,)
I 2f (1.330,)
For nozzles and openings
Membrane + bending stress intensity 5 2.25f (1 SO,)
The latter limit is obviously intended to be consistent with the earlier idea of a maximum SCF of
2.25 (albeit now for multiaxial stress). Notice that in the case of these special limits there is no need to
classify the behaviour into primary and secondary, etc.
(e)
Comments on the Design-by-Analysis Approach
Some cautionary words are necessary for the unwary. Confusion arises because of the tendency to
denote the stress intensity in a particular category by the symbol for that category, e.g. f b is the stress
intensity for the primary bending stress category. However, (fL + f, + fs) is not the sum of the local
primary stress intensity, the primary bending stress intensity and the secondary stress intensity. It is the
stress intensity evaluated from the principal stresses after the stresses for each category have been added
together in the appropriate way. This can be summarised by simply stating
Only add stresses
Do not add stress intensities
The Design-by-Analysis approach is closely related to its origins, which are rooted in thin shell
discontinuity analyses. Even where shell theory is used there are difficulties of classification in many
cases. With the use of modern analysis methods, such as the finite element method, difficulties increase.
With solid continuum elements an analyst can produce reasonably accurate stress information for
complex geometries. These stresses may vary non-linearly through the thickness. For assessment
purposes it is usually necessary to linearise the stress distribution and separate membrane and bending
effects. One is tempted to say that the membrane stress (intensity) so derived is equivalent to f, (or f,)
and that the bending stress (intensity) corresponds to the secondary category f Unfortunately things
are not always so simple. Firstly, the linearisation procedure is itself sugject to a number of
uncertainties. Secondly, the bending component in general may include primary bending as well as
secondary bending.
In practice, the approach used tends to assume that the membrane stress intensity is primary (which
is acceptable) and to assume that the bending stress intensity is secondary (which may not be
conservative). In critical situations the intelligent designer may wish to impose hidher own
conservatism at this point.
However, it is worth noting that alternative methods may be forthcoming, which would by-pass
the categorisation problem or at least simplify its interpretation. The Standard allows the design to be
based on limit load analysis with a suitable factor where the factor has to be the same as for the main
shell, i.e. 1.5. Some recent work, 8 proposes an elastic compensation method which uses successive
elastic finite element analysis, in which the elastic modulii of elements are modified, to obtain estimates
of the limit load. Design may proceed directly with a factor on load without detailed consideration of
the stresses. The approach seems promising. If it can be successfully extended to complex loading
situation it could provide a relatively simple alternative to the current classification route. If it could be
further extended to cover structures with cracks it would also simplify structural integrity assessments.
It will be clear from the above that the shakedown concept has had a great influence on the design
codes not only on the Design-by-Rule sections but also in the Design-by-Analysis sections. Yet this
key concept is dealt with in a superficial way. It is important to note that the stress range for shakedown
is taken as 20, although in general this is not necessarily the case except for thermal loading. In
12
arriving at the shakedown condition a certain amount of permanent plastic strain is experienced.
Unfortunately the use of code rules or stress limits do not give any indication of the strain levels
involved. Care is required where higher alloy material, especially with welds, are present in highly
constrained regions.
The Design-by-Analysis approach is an attempt to provide a systematic method for general design
to cover any load and geometry combination. It rests on the assumption that a proper elastic stress
analysis of sufficient accuracy can be conducted. (This could be a big assumption.) Thereafter it
categorises the stresses and casts them into a Tresca-type framework of stress intensities for comparison
with certain allowables. The greatest problem lies in the classification of the stresses and this has not yet
been satisfactorily resolved particularly where modern methods of analysis are employed. Although
alternative approaches are being investigated they are not yet far enough advanced to be used routinely.
The categorisation problems are increasingly being highlighted. If adequate guidance is not
forthcoming soon the whole method will need a radical overhaul.
2.3 Fatigue Design
The possibility of failure by fatigue cracking needs to be considered in the design of any vessel, or part
of a vessel, which experiences fluctuating loading in service. BS 5500 provides fatigue design rules in
Annex C. These first appeared in the 1996 issue of the Standard, having previously been included as an
Enquiry Case. It is radically different from the version in the previous Appendix C, which was based on
rules developed in the early 1960s in the USA and incorporated in the ASME codes. The change came
about because of the realisation that the old rules were not suitable for designing welded vessels,
together with the fact that quite different fatigue rules were being used to design other types of welded
structure.
Usually a fatigue assessment is performed when material thicknesses and design details have already
been selected. In BS 5500, fatigue design data are provided for different weld details in the form of S-N
curves, relating fluctuating stress range and fatigue life. The steps in the fatigue assessment are as
follows:
(9
(ii)
(iii)
For
Identify design details which will introduce stress concentrations and are therefore potential
sites for fatigue cracking;
identify all repeated or cyclic loads and hence stresses to be experienced by the details during
service operation of the vessel; and
using the appropriate design S-N curves in the Standard, deduce the fatigue lives of the details
and compare them with the required design life.
the realistic situation in which it is necessary to consider several sources of fatigue loading,
which give rise to variable amplitude stress cycling, the fatigue damage induced is determined using the
linear cumulative damage rule (Miner's rule):
where ni is the number of stress cycles arising from the various load sources and corresponding to stress
ranges Si, Ni is the fatigue life obtained from the appropriate design curve at Si and i = 1,2,3, etc.
Guidance is given in the Standard regarding how to evaluate the different cycle stress ranges.
(a) Justification of the Method
The basis of the method used in the previous version (Appendix C), and still used in ASME, was that
the fatigue performance of a structure can be represented by an S-N curve obtained from fatigue tests
on smooth, polished specimens machined from the material of interest. Such 'material' S-N curves are
then used in conjunction with stress concentration or fatigue strength reduction factors, which
represent the stress concentration effect of geometric discontinuities in the structure.
13
The fundamental weakness of this approach is the assumption that the 'material' S-N curve is
relevant to all cases of potential fatigue failure in components. This is not the case, chiefly because of the
need to distinguish between the fatigue crack initiation and fatigue crack propagation processes. The
behaviour of the smooth specimen is dominated by fatigue crack initiation, whereas the initiation
process may be relatively insignificant in the life of a real structural detail, depending on the severity of
the stress concentration it introduces.
Another misconception concerns the influence on fatigue life of the static strength of the material.
Fatigue data obtained from smooth polished specimens clearly show an increase in fatigue strength with
increase in material tensile strength. As a result, some codes (e.g. ASME) provide higher design S-N
curves for high strength steels. However, the benefit of material strength relates only to crack initiation
and the benefit of material strength seen in smooth specimens is not reflected in the fatigue performance
of notched or welded material. Consequently, the fatigue rules in the ASME codes are unconservative
for high strength steels. In ASME provision is made for using smooth specimen test data but one should
be careful with this approach.
Another significant difference between the fatigue performance of smooth and notched
components relates to the fatigue limit. The fatigue limit for smooth polished specimens bears little
relation to the behaviour of notched or welded components.
(b) Basis of the Approach
The 'new' rules in BS 5500 adopt the approach used in many all other design codes for structures, where
weld details are separated into different classes, the weld details in a particular class having similar fatigue
strengths, each with its own design S-N curve, Fig. 10. The S-N curves are derived from data obtained
from fatigue tests on specimens incorporating actual design details, such as welded joints. The important
difference is that the S-N curve already allows for the stress concentration effect of the detail itself.
Consequently, 'Peak' Stress considerations are no longer required except perhaps in non-welded areas.
The design curves were derived from
t
E 100
> 300
a statistical analysis of a large body of
published data obtained from simple
welded specimens in steel. The mean
minus two standard deviations S-N
curves, corresponding to approximately
97.7% probability of survival, are given
as the design curves. All of the curves can
be expressed in the form:
S, mN = A
where Sr is the stress range, N is the
number of cycles and A is a constant. For
10/02. i ' i ii~~c,io,~ i
i ' j i3'.cjos i t i l ? Y O 8 fatigue endurances UP to 10'cycles all six
Number of cycl es, N
design curves (D to W) for weld details
are parallel, with m = 3. A seventh curve,
class C, is given for unwelded material
and this has a slope of m = 3.5.
Limitations on the design curves arise for elevated temperature (greater than 375" C for ferritic
steels, 430" C for austentic steel, and 100C for aluminium alloys) or corrosive conditions. Account
may also need to be taken of the thickness of the part considered, by multiplying the fatigue life
obtained from a curve by
Fig. 10 Fatique Desi gn Cur ves, Stress versus Nu mb e r of Cycles
(22) 0.75
where e is the relevant thickness. This allows for the fact that the fatigue strength of some weld details
decreases with increase in plate thickness.
14
(c) Stresses
Whenever fatigue design of a structure is based on S-N data, it is important to ensure that the stress,
which is calculated, is the same as the stress used to express the fatigue data. Two problems commonly
arise. Firstly, the detail under consideration is situated in a region of stress gradient. The stress
calculated will vary depending on the precise position chosen. Secondly, stress fields in real structures
are rarely unidirectional. Thus, details on the shell of a vessel will experience biaxial stress and the stress
direction may change, even during a single cycle. By contrast, most fatigue data have been obtained
under relatively simple unidirectional loading conditions.
In general, a fatigue assessment is performed on the basis of the structural stress, calculated using
elastic theory taking account of all effects but excluding the stress concentration effect of the detail
itself. Alternatively, it may be derived from measurements on the actual vessel.
Static design to BS 5500 involves the use of the stress intensity. However, since it is necessary to
take account of both the magnitude and direction of the fluctuating stress, fatigue design is based on the
relevant maximum principal direct stress range and stress intensity if not used at all.
( d)
Fatigue Analysis of Weld Details
The appropriate design S-N curve for a particular weld is chosen using a classification system. Each
constructional detail is placed into one of six classes, designated D to W as in other Standards (e.g. BS
5400 and BS 7608), corresponding to six fatigue design curves. The classification depends upon the
following:
(1) The direction of the fluctuating stress relative to the detail;
(2) the location of possible crack initiation at the detail;
( 3 ) the geometric arrangement and proportions of the detail; and
(4) the methods of manufacture and inspection.
Thus, more than one class may apply for a given weld detail, e.g. Fig. 11. Sketches in Annex C
indicate the potential mode of fatigue cracking considered and the position and direction of relevant
fluctuating stress. Clearly, it is important for the designer to consider all possible sites for fatigue
cracking. Class D is used to cover parts of the vessel which are nominally unwelded on the basis that
repair welds may be required anywhere. A seventh design curve, class C, is provided for unwelded
material.
Fig. 11 Possible Detail Classifications i n
t he same Joint depending on t he
Mode of Failure Considered
(e) Fatigue Life Improvement
An innovation in Annex C is the introduction of recommendations about the use of weld toe machining
or grinding to improve fatigue strength. Fatigue cracks readily initiate at weld toes chiefly because of
the presence of inherent crack like flaws. Removal of these flaws and blending of the weld toe may
allow the use of a higher class in the analysis. Grinding below the plate surface is crucial to ensure
complete removal of the crack like flaws, which provide sites for fatigue crack initiation at the toe.
( ' Simplified Fatigue Analysis
As has already been mentioned, the fatigue loading on many pressure vessels is not severe.
Consequently, simplified approaches are provided.
15
Annex C gives a two-stage simplified fatigue analysis method. If either stage is satisfied, the need
for a more detailed fatigue analysis is avoided. The first stage is a highly conservative requirement based
on limitation of the total number of cycles of fatigue loading. The criterion is intended to act as a 'sieve'
through which effectively statically-loaded pressure vessels should pass. The second method is less
conservative but requires more input and involves the application of a cycle counting method.
The first simplified fatigue analyses is based on the assumption that every load cycle produces a
stress range 3f and that this occurs at a low fatigue strength class G weld detail (at minus three standard
deviations). A modified design stress ff may be chosen usually less than f.
Allowing for E, the elastic modulus for the material at the maximum operating temperature in
N/mm2, and applying the thickness correction, the total allowable number of cycles N becomes
6x109 ( . E 5) (T) 0'75
2 09x10
N 2 7
ff
The second simplified method is similar to the first but requires the designer to establish separately
the numbers of stress cycles (ni) arising from the various load sources (pressure, temperature and
mechanical). These are then used with endurances Ni in cumulative damage calculations (Miner's rule).
The thickness correction is again introduced. The following equation must be satisfied:
where i = 1, 2, 3, etc, Ni values are obtained from the appropriate fatigue design curve, adjusted where
necessary for elastic modulus. The design curves are used directly, with allowance for the difference
between them and mean minus three standard deviation curves being incorporated in the constant (0.6).
A Simplified and conservative cycle counting procedure is given, based on the rainflow method. If the
simplified approaches cannot be satisfied a full analysis is required.
(g) Assessment of Shape Imperfections
In earlier versions of BS 5000 if the vessel complied with the manufacturing tolerances no account
needed to be taken of the stresses arising due t o the tolerances. However, the current fatigue method
requires the stresses due to departures from the design shape to be included in the fatigue analysis even
where the departures are within those allowed in the Standard. Appropriate stress concentration factors
are given for ovality, misalignment, weld peaking, etc.
(h) Future Needs
Even though the new fatigue rules in BS 5500 contain quite extensive changes to the previous rules,
there are still a number of areas which will need to be reviewed and ultimately revised. The main ones
are high temperature effects, bolting, environmental effects and further information on the detail
required in stress analysis.
2.4 Fracture
Over several decades the subject of fracture mechanics has received considerable attention. Associated
with this has been the realisation that it is possible to live with defects. The links between fracture
assessment and inspection now provide a significant input to the consolidation of the subject of
pressure vessel technology as discussed later in the paper.
The recognition of fracture analysis was based on linear elastic fracture mechanics (LEFM)
procedures. In the field of structural analysis attention was mainly focussed on the LEFM
considerations because of the difficulty in applying the more sophisticated post yield (PYFM) methods.
With the wider use of fracture mechanics and the need to apply the methods to tougher materials
attention turned to PYFM assessment methods. The whole subject was opened up leading to the
development of more user-friendly approaches.
16
The so-called two criteria approach 9 received considerable attention around the world. In the
original form this stated that failure occurs when the applied load reaches the lower of either the load to
cause brittle fracture based on LEFM or a collapse load dependant on the flow stress and the structural
geometry: A sizeable transition exist between the two criteria and it was shown that this could be
described by an equation developed from the Bilby, Cottrell, Swindon 10 strip yield model. This was all
in the late 1970s.
Since then the methodology has been developed beyond recognition taking into account such
aspects as ductile tearing, as well as thermal and residual stress. In order to make full use of the methods
considerable effort has gone in to the development of procedures for the determination of material
properties and the evaluation of fracture parameters defining the behaviour of cracks. Procedures based
on the two criteria approach have been adopted in the British Standard BS7910 11 the second draft of
which has recently completed the stage of public comment.
BS 7910 provides the choice of three levels of assessment:
Level 1 is a simplified method applicable when the information on materials and properties is
limited
Level 2 is the normal assessment route
Level 3 is for ductile materials and enables a tearing resistance analysis to be performed.
Assessment is generally carried out using a failure assessment diagram, Fig 12. The vertical axis of
the diagram is the ratio of the applied conditions to the conditions required to cause failure in the same
fracture terms. The horizontal axis is the ratio of the applied load to that required to cause plastic
collapse. The calculation for the flaw provides the co-ordinates of the assessment point. This is plotted
and compared with the assessment line to determine the acceptability of the flaw. If the assessment
shows a flaw to be unacceptable, it is possible that acceptability can still be demonstrated by improving
the quality of the input data and applying a higher assessment level.
0
Fig. 12 The Failure Assessment Di agram
The approaches discussed are concerned with structures operating below the creep range. The UK
power industry in particular, has brought together the work on high temperature crack growth and
formulated procedures which are user friendly. 12
17
2.5 Materials, Manufacture, Inspection and Testing
The safety of a p-ressure vessel depends on good design, selection of materials, controlled fabrication,
inspection and testing followed by operation within the specified design conditions. Much attention is
given to the development of design codes and standards and often insufficient attention is paid to the
other important elements.
Among the important issues is that of the over pressure test. With many vessels the test provides the
principal demonstration that the vessel will meet the design requirements. In other cases care has to be
exercised regarding the meaning of the over pressure test. A good example of this is the nuclear reactor
pressure vessel. During operation and in emergency shutdown conditions, the vessel sees more than just
pressure loading. In operation the combination of the pipe system stresses plus the pressure loading can
produce stresses higher than those of induced in the over pressure test. Under emergency shutdown
conditions, the thermal stresses can be severe and depending on the state of the system there may or
may not be additional pressure loading.
Other aspects of the over pressure test need to be considered. If a defect is present it is possible for
it to be extended as a result of the over pressure. With high technology vessels, acoustic emission
techniques are sometimes used to indicate such growth. Those opposed to pressure testing often state
that the pressure test only shows that the vessel is not leaking. The conditions for the over pressure test
must be considered and defined at the design stage, otherwise difficulties may occur. The Sizewell 'A'
nuclear power station heat exchanger provides an example of this. The consequent failure is shown in
Fig. 13. The hydraulic over pressure test was carried out under conditions where the vessel was
Fig. 13
18
Sizewell Vessel Failure at Proof Test
supported on trestles. The additional local loading induced by the supports or the weight of the water
was not accounted for and may have contributed to the failure of the vessel during the test. A general
design requirement is that all anticipated loading conditions must be considered.
2.6 Current Standards
Much of what has been said so far in the section has been based on BS 5500. As mentioned, the ASME
Boiler and Pressure Vessel code, and those of other countries, have not only influenced the development
of BS 5000 but have made a very significant contribution to the pressure vessel technology scene. In the
not too distant future, BS 5500 together with the pressure vessels of other EU National Standards
Bodies will have to be withdrawn. This is because of the Pressure Vessel Equipment Directive and the
legislation associated with the new European Standards (see section 3.3). It is likely that BS 5500 will be
maintained by BSI, not as a standard but as a pressure design publication thus circumnavigating the EU
regulations. Whether the document will be maintained technically remains to be seen as the group of
people likely to be involved in such a task are those involved in the Standard.
The ASME codes are a dominant force around the world. BS 5500 is a respected and well used
Standard in many countries. There is therefore considerable interest in the approaches in the new
European Standard and the extent to which the EU Directives and Standards will affect the current
situation. As discussed later, I S0 is re-activating its interest in the unfired pressure vessel scene.
3. THEFUTURE
The future of pressure vessel technology promises to be somewhat different from the past three
decades. Here we have seen major advances in analysis tools such as finite elements and fracture
mechanics, the understanding of fatigue and creep processes, non-destructive testing, fabrication
techniques, and the evaluation of material degradation. These advances in knowledge and
understanding have provided the basis for the development of approaches which provide for more
economic design and manufacture, longer life, more effective inspection, monitoring maintenance and
repair of pressure vessels.
In the short term it is likely that pressure vessel technology will be concerned with the
establishment and acceptance of the methodologies and approaches which regularise the developments
of the past three decades. In parallel with this the influence of new materials will introduce a new
dimension. The developments in codes and standard particularly the European Standards will introduce
new approaches that could have an influence on the market for pressure vessels.
In the medium term, attention may well be focussed on the integration of the various aspects of PV
technology under the headings of Risk Assessment, Equipment Qualification & Integrated Logistics.
Work in these areas is already underway. These headings take the technology in an integrated manner
from the design stage through manufacture, inspection, repair, and replacement to life extension.
Although the fundamentals will continue to be developed, the rate of development is likely to be rather
less than in the past 50 years.
The future appears to lie in the integrated technology approach leading to increased reliability and
safety with greater economy in cost and effort.
Some of the topics, which are likely to influence the future of pressure vessel technology, are
considered in this section
3.1 Risk-Based Assessment
The idea of Risk Based Assessment is to reduce the overall risk exposure on plant. This is achieved by
focussing inspection, assessment and maintenance on those locations where the risk is perceived to be
high. Risk can be defined as the product of likelihood of an unexpected failure and the consequence of
the event. The approach is sometimes termed Risk Based Inspection. Fig. 14 illustrates the concept.
19
Components for
Inspection
Failure Mode Analysis
&
Identify Integrity Parameters
+
No Threats to Integrity
(Condition Assessment *
Low Likelihood
Integrity Parameters)
Likelihood Analysis
f
Timing of Next Inspection
(Deterioration)
b
Condition Fit for Purpose
Fig. 14 Risk-based Inspection Process (Ref: 14)
\
r
iation /
\ WVufrk /
ASME has supported the development of more formalised approaches 13 to the subject whereas in the
United Kingdom custom and practice have developed the practice further. The methodology results in a
set of recommendations on how much preventative effort is required. The inspection and the subsequent
analysis of the data quantify the condition of the item by determining the likelihood of failure and calls for
corrective action if necessary. In this way it reduces the likelihood of an unexpected failure.
An important issue is that of failure consequence. Various attempts have been made to identify risk
categories. The system used by Brunei Shell Petroleum Company is a good example 14.
There are three categories:
Category A functional failure of this type represents a potential health, safety and/or health
risk as well as disrupting production in many cases.
Category B where functional failure is likely to impact availability and therefore production or
operating costs.
Category C where functional failure affects neither health & safety nor short-term operations.
Limited loss of functionality and the only direct cost is repair.
The interpretation and associated examples will vary according to the industry concerned. As the
plant will generally contain pressure vessels the implications of RBA will involve current pressure
vessel technology.
The Risk Based approach simply provides a regulated approach to ensuring plant is safe. The
elements of the methodology are technical approaches well established in the relevant fields, which
include inspection, analysis, monitoring, and materials behaviour.
3.2 Equipment qualification
Many topics have been developed from experiences in the nuclear industry. Equipment qualification is
one such topic. Equipment is classified as active or passive in terms of the function it performs.
Pressure equipment is usually passive. The pressure vessels in nuclear plant have received enormous
attention over the years and consequently much of the contemporary technology has resulted from this
activity.
The term Equipment Qualification is used in a number of ways. Encouragement has been given
recently to using the term to express the integrity and performance of systems over the entire lifetime.
20
At the design stage consideration is given to aspects such as inspectability in service, monitoring the
component state including material degradation, monitoring the operating conditions, maintenance,
repair and i-eplacement, and life extension. It can be seen that all these are interacting and, depending on
the component should receive varying levels of attention. The extent of equipment qualification
required will depend on the Risk as defined earlier.
As with Risk Assessment the methodologies used are generally well established. The EQ
approach simply sets out a procedure to provide for maximum availability whilst ensuring adequate
safety throughout the design lifetime. Given that the other steps are in place, the evaluation of life
extension is straightforward, because of the data available, and would not represent a major step. Risk
Assessment discussed in the previous section would be an integral part of Equipment Qualification.
EQ is not just applicable to plant presently at the design stage. The approach is being applied to
existing plant by back fitting and building up the relevant databases. This is limited under some of the
Regulations introduced as historical data is excluded. Operating experience can be used in design and
operating reviews. The responsibility is on the supplier to state the lifetime and the maintenance regime
for the component. When the component is in operation it is the plant responsibility to manage the
Equipment Qualification programme.
3.3 The European Pressure Equipment Directive
The European Parliament and the European Council formally approved the Pressure Equipment
Directive 15 in May 1997. This Directive harmonizes the laws of the fifteen member countries of the
European Union relating to pressure equipment. It is part of the programme for the elimination of
barriers to trade. As such the purpose of the PED is to ensure the free placing on the market and putting
into service of the equipment concerned. At the same time it permits a flexible regulatory environment
and allows European industry to develop new techniques thus increasing the international
competitiveness. The Directive covers a wide range of products.
The timetable for the implementation of the PED is as follows:
May 1997
November 1999
May 2002
Between November 1999 and May 2002, manufacturers have a choice of the PED or their national
The Directive defines the Essential Requirements regarding safety and other provisions without
reducing existing levels of protection within the community. Compliance with the Essential
Requirements is necessary under the Directive in order to ensure the safety of the equipment or
assemblies. A number of conformity assessment procedures are available to the manufacturer by which
conformity with the Directive can be declared and the CE marking can then be affixed to the product.
Approved by Council of Ministers
Introduced as Law in Member States
Obligatory in the European Union
legislation.
3.4 The European Standards
The European harmonized standards are one means of conforming to the essential requirements of a
directive. A harmonized standard is a technical specification adopted by the European Committee on
Standardization (CEN) and by CENELEC. The European Commission has given CEN a mandate to
undertake a programme of work in the field of pressure equipment. CEN is recognised formally by the
Commission as being competent to adopt harmonized standards at the request of the Commission.
Harmonized European Standards are not mandatory and other standards may co-exist alongside
then without the presumption of conformity. National standards will be withdrawn on publication of
the relevant CEN standard.
The programme of European Standards in the field of pressure equipment is enormous and covers
approximately 800 adopted and draft Standards. There are about 300 candidate harmonized product
21
standards, 250 candidate harmonized materials standards and 250 harmonized supporting standards
covering welding and non-destructive examination.
The principal committees and product standards are:
CEN / TC 54
CEN / TC69 Valves
CEN / TC74 Flanges
CEN / TC267 Industrial Piping
CEN / TC269
The Unfired Pressure Vessel Standard is currently at the stage of being submitted for public
comment within the EU. This is known as the CEN Inquiry and will extend over a period of six
months. It is expected that between seven and ten thousand comments will result from the inquiry. The
aim is to complete all the stages in the preparation of the Standard such that it will be formally adopted
at the time the PED becomes obligatory.
Unfired Pressure Vessels
Shell and Water Tube Boilers
3.5 The New European Unfired Pressure Vessel Standard
Many clauses in the Unfired Pressure Vessel Standard are based on the approaches in BS 5500 together
with significant contributions from the German and French Standards. There are some interesting
similarities and differences between the enUFPV and BS 5500.
The draft UFPV Standard is made up as follows:
Part 1 General
Part 2 Materials
Part 3 Design
Part 4 Manufacture
Part 5 Inspection and Testing
Part 6 Safety Systems
Part 7 Cast Iron Vessels
Additional parts are being developed on: Refrigeration Vessels, Hydropneumatic Vessels and
Expansion Vessels and there is also a series of horizontal supporting standards
EN764 Pressure Equipment
Part 1 Pressure and Volume
Part 2 Symbols
Part 3 Role of Parties
Part 4
Part 5
Part 6 Operating Conditions
Some of the features of the Unfired Pressure Vessel Standard are now considered.
Safety factors are often critical in defining the vessel thickness and therefore influence economic
considerations. Usually, the lower the safety factor the higher the complexity of the design approach
together with the levels of inspection and testing. In the alternative design approach the enUFPV takes
the UTY1.875. This is a considerably lower factor than that based on 2.35 in BS 5500 (or 4 in ASME
Div 1).
This difference in UTS requirements is shown by consideration of a vessel, with a common set of
pressure volume requirements. The thickness, when designed to the enUFPV alternative approach,
would be 18.75mm, whereas, to BS 5500 it would be 23.5mm. In both standards there is an over riding
Delivery Conditions for Materials
Compliance and Inspection Documentation for Materials
22
condition taking into account 2/3 Yield with the lower of the values controlling the thickness. There is
an incentive here to use high yield materials where the yieldhltimate ratio is in the range 0.64 to 0.80.
Great care-is required with this approach where fatigue is a relevant failure mode.
The main design part of the Standard uses the Design-by-Formula approach (i.e. same as Design-
by-rule). Some of the details here are significantly different from the British or American Standards. For
example, on dished ends the established British methods produce lower allowable pressures than are
currently used in some EU countries such as Germany. Consequently, a new set of curves has been
developed, based on the German experience, in order to be competitive.
For flanges the designer will have a choice of two methods. The Taylor Forge approach will be in
the main body of the Standard while the East German method, referred to as the CEN method, is being
included in the annex. The basic approach using the Taylor Forge method is very close to that in BS
5500 and ASME.
The section on openings in shells is based on the pressure area method. The method has been used
satisfactorily in European codes but has not been used in the UK or USA. However, the current BS
5500 has included this as an alternative in Annex F in anticipation of it being adopted as the basis of the
new European Standard. The principle, which is mentioned in Section 2.2, is illustrated in Fig 15, where
a certain area of the vessel in the region of the opening multiplied by the design stress is equated to the
cross-sectional area of the vessel in the same region multiplied by the pressure. This is a simplified limit
load type approach with the yield stress factored to the design level.
The fatigue methods are very close to those in BS 5500 Annex C as described here in section 2.3.
There are two clauses on fatigue in the draft Standard. The first is the Simplified Method for fatigue
assessment. This covers just pressure cycles. The other is the Detailed Method which takes into
account various types of loading with a comprehensive analysis. The same approach using a
classification of weld details is used as in Annex C but the classification is a little different (and different
between the simplified and the detailed method); a number based classification is used (31-100) but
otherwise the general approach is similar to BS 5500 except that more detail is provided.
The Design-by-Analysis section allows two separate methods to be used. There is a Direct
approach to DBA and an alternative approach to DBA. The latter is the familiar stress categories
based approach similar to that described in section 2.3 herein. The Direct approach is a little different. It
PA, = G y (Afm + Afp + Am)
Fig. 15 Pressure Area Design Met hod for Nozzl es
23
uses the idea of principles and application rules coupled with partial safety factors. Design checks have
to be made against the expected failure modes; the failure modes are similar to the ones listed in section
2 here, namely gross plastic deformation, progressive plastic deformation (ratchetting) plus instability
and fatigue. The first of these is satisfied by a limit load type calculation to ensure the design resistance
(the resistance or limit state divided by the partial safety factor) is greater than or equal to the design
actions (loads). The second design check on progressive plastic deformation is, again, the shakedown
condition (3f or 20,) but in this case the von Mises criterion is recommended although Tresca is
permitted.
The control of materials is an important issue. The Standard includes the general materials
approved for use. In addition, the European materials data sheet describes a material not included in the
material standard but approved for use under a European Approval for Materials (EAM). The EAM are
technical approvals for non-EN standardized materials which are divided into;
-
-
The requirements for prevention of Brittle Fracture are covered by one of the following three
Met hod 1 A code of practice (German) developed from operating experience but limited to
certain thicknesses (30mm and 60mm) for which experience exists.
Met hod 2 A code of practice developed from fracture mechanics principles and operating
experience. (Very similar to BS 5500 Ann.D.)
Met hod 3 The application of fracture Mechanics analysis.
In Manufacture, the tolerances are very similar to those in BS 5500 for out of roundness and for
offset of middle lines of plates for welding. CEN does not have any of the step changes in tolerances
that occur in BS 5500. Weld details are taken from prEN1708-1 and the welding procedure specification
should be in accordance with EN288 Part 2. Procedures are specified for: Production test plates,
forming, post weld heat treatment and repairs.
The inspection philosophy is based on the Conformity Assessment Modules of the PED. The menu
of modules are applicable to the risk categories. The manufacturer always has the choice between using
a module which involves quality systems and one which does not. For those modules involving quality
systems there is far less involvement of the notified body. This is different from BS 5500 where there is a
clear task for the inspection Authority and from the various ASME system where quality systems are
required.
restricted EAM covering new materials
unrestricted EAM based on existing specifications of a CEN or non-CEN country
methods:
3.6 The International Situation
In the mid-19707s, I S0 attempted to produce a pressure vessel standard. Although a draft (ISODIS
2694) was produced in 1973, it was essentially abandoned although the committee structure was kept in
place. In fact, the first issue of BS 5500 was delayed pending the possible adoption of the International
Standard. At the end of 1997 a meeting was held in Tokyo of ISO/TC11 to discuss a USA proposal to
produce a Pressure Vessel Standard. After some considerable debate based on a proposal from Japan, it
was agreed that what was required was a facilitating standard similar in form but not as comprehensive
as the EU Directive. The purpose of this facilitating standard would be to define the essential
requirements of a standard from any region used to design a pressure vessel. The content of such a
standard would be to define something similar to the essential requirements in the EU Directive. There
would be no internationally harmonized standards system but any standard meeting the requirements
could be used whether it is ASME, or CEN. An I S 0 Committee has been set up, under TC11, to
develop such a standard. The title of the standard will be Specification for International Pressure
Equipment Codes and Standards (SIPECS). Any country should be able to have its code or standard
harmonised under SIPECS but would be required to translate the document into English. The
CEN/TC269 Draft standard on shell boilers will be submitted after CEN publication for fast track
approval through the I S0 system. Other CEN standards including the unfired pressure vessels
standard will follow.
24
One misunderstanding regarding the European Pressure Equipment Directive is on the use of non
harmonized standards. Any recognised standard can be used providing it meets the essential
requirements of the PED.
CONCLUDING COMMENTS
The subject of pressure vessel technology is central to industry as a whole. The balance of safety with a
sound economic approach is a major consideration to the designer, manufacturer and operator. The
legislation and standards must be such that these requirements are met.
Starting with the industrial revolution, or maybe earlier, more than two centuries of progress has
brought us to the stage where the significant issues are recognised and understood and formal
procedures have been laid down to regulate pressure vessel technology. The future appears to lie in the
recognition and establishment of the integration of the total technology.
Pressure Vessel Technology is a good example of practical engineering built on secure design
foundations. George Stephenson would surely have approved of these developments.
25
REFERENCES
1. Denton A A. George Stephenson Lecture. IMechE, 1996.
2. Pullin J. Progress through mechanical engineering. Quiller Press, London, 1997.
3. Pressure Vessels Design. Concepts & Principles. J Spence & A S Tooth. Published Chapman & Hall London,
4. BS 5500: Specification for unfired fusion welded pressure vessels. BSI, UK, 1997.
5. ASME Boiler and Pressure Vessel Code, Section VIII, Div 1, Div 2. American Society of Mechanical
6. BS 1500: Specification for unfired fusion welded pressure vessels. BSI (Withdrawn).
7. BS 1515: Fusion welded pressure vessels for use in chemical petroleum and allied industries. BSI, (Withdrawn)
8. Mackenzie D & Boyle J T. A simple method of estimating limit loads by iterative elastic analysis, I - Simple
examples, I1 -Nozzle sphere intersections, 111- Torispherical leads. Int. Jn1 of Press. Vess. & Piping, 1992.
9. Townley C H A & Dowling A R, 1975 The effects of defects on structural failure: a two criteria approach. Int
Jnl of Press. Vess. & Piping 3, 77, 1975.
1994.
Engineers, New York, USA, 1995.
10. Bilby B A, Cottrell A H , & Swinden K H, 1963. Proc Roy Soc, A272306,1963.
11. BS 7910: 1997 Guide on methods for assessing the acceptability of flaws in fusion welded vessels.
12. Nuclear Electric. Assessment Procedure R5 An assessment procedure for the high temperature response of
13. ASME Research Task Force on Risk Based Inspection Guidelines. General Document CRTD 20-1, ASME,
14. Hagemeijer PM & Kerkveld G. Risk Based Assessment of Pressure Systems Proc Inst Mech Eng, Vol 212,
15. Directive 97/23/EC of the European Parliament and of the Council of 29 May 1997. On the approximation of
structures. Issue 2, Nuclear Electric plc, Barnwood, Glos, UK, 1994.
Washington DC, 1991.
Part E, 1998.
the laws of the Member States concerning pressure equipment.
Printed in Great Britain by Latimer Trend C Company Ltd, Plymouth
Institution of Mechanical Engineers
1 Birdcage Walk
London
SWl H 9JJ
United Kingdom
Phone: +441712227899
Fax: +441712224557
E-Mail: enquiries@imeche.0rg.uk
Internet: http://www.imeche.org.uk
Also at
Northgate Avenue
Bury St Edmunds
IP32 6BN
United Kingdom
Telephone: +44 1284 763277
Fax: +44 1284 704006
IMechE is a registered charity number 206882
promoting mechanical engineeringTM

You might also like