EPD/GEN/REP/0088/96

SINTAP/NE/007
Internal Use Only

FAILURE OF CRACKED
COMPONENTS
SINTAP TASK 2 REVIEW

By:
R A Ainsworth and J K Sharples*
with an Appendix by B Brickstad
and contributions from British Steel (UK), IWM (Germany), TWI (UK)
Nuclear Electric Ltd, UK
* AEA Technology plc, UK
SAQ, Sweden

Confidentiality classification:
Internal Use Only

Nuclear Electric Ltd

Barnwood
Glos. GL4 3RS

Failure of Cracked Components SINTAP Task 2 Review

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page i of iii

By: R A Ainsworth, Structural Integrity Branch

Issue 1: December 1996
I confirm this document has been subjected to verification and validation by internal review within Nuclear
Electric Ltd.
Date:

Verifier: Dr P J Budden, Assessment Technology Group,

Structural Integrity Branch, Barnwood

Approved for Issue:

Date:

Dr P Neumann, Manager, Structural Integrity Branch, Barnwood

SUMMARY
SINTAP (Structural Integrity Assessment Procedures for European Industry) is a Brite-Euram Project coordinated by British Steel with the objective of providing a unified structural integrity evaluation method for
European industry. The project is divided into 5 tasks dealing with: weld metal strength mismatch; failure of
cracked components; optimised treatment of data; secondary stresses; and procedure development. The first
activity is a state-of-the-art review and this report contains that review for task 2 - failure of cracked
components. This task covers a number of technical issues and for each of these the available information is
cited along with an identification of remaining problems and the extent to which these are being addressed
within SINTAP.

Key words: SINTAP, Fracture, Review

CONTENTS

Page

1.

INTRODUCTION .............................................................................................................................. 1

2.

STRESS INTENSITY FACTOR AND LIMIT LOAD SOLUTIONS ............................................ 2

2.1
2.2
3.

YIELD/TENSILE STRENGTH RATIO EFFECTS ......................................................................... 7

3.1
3.2

4.

General Outline of Procedures ........................................................................................... 11

Review of Main Aspects .................................................................................................... 12
Unresolved Issues and Scope for Further Work ................................................................ 14

CONSTRAINT ................................................................................................................................. 15
5.1
5.2
5.3
5.4
5.5
5.6

6.

Summary of Effects .............................................................................................................. 7

Future Work .......................................................................................................................... 9

LEAK BEFORE BREAK AND CRACK SHAPE DEVELOPMENT ........................................ 10

4.1
4.2
4.3

5.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page ii of iii
Stress Intensity Factor Solutions .......................................................................................... 2
Limit Load Solutions ............................................................................................................ 5

Constraint Parameters and Their Availability ....................................................................

Influence of Constraint on Material Toughness ................................................................
Local Approach Methodologies .........................................................................................
Defect Assessment Methodologies ....................................................................................
Validation of Methodologies ..............................................................................................
Further Work ......................................................................................................................

15
17
19
21
22
23

PRIOR OVERLOAD ....................................................................................................................... 24

6.1
6.2

Warm Pre-stressing Methodology ...................................................................................... 24

Limitations of Methods and Further Work ........................................................................ 25

7.

CONCLUDING REMARKS ........................................................................................................... 26

8.

ACKNOWLEDGEMENTS ............................................................................................................. 26

9.

REFERENCES ................................................................................................................................. 27
TABLES
Table 1

Basic stress intensity factor solutions needed for

assessment procedure ........................................................................................... 34

Table 2

Basic limit load solutions needed for assessment procedure .............................. 35

Table 3

Crack opening area solutions recommended in new

Appendix 9 for R6 ................................................................................................. 36

CONTENTS

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page iii of iii
Page

FIGURES
Figure 1

The leak-before-break diagram ............................................................................ 37

Figure 2Recommended recharacterisation in R6 of defects at

breakthrough for predominantly tensile loading .................................................. 38
Figure 3Recommended recharacterisation in R6 of defects at
breakthrough for predominantly through-wall bending
loading .................................................................................................................. 39
Figure 4Normalised T-stress solutions of eqn (5) for a semi-elliptical
surface crack in a plate under uniform bending for various
ratios of crack depth, a, to plate thickness, t, and to crack
semi-length, c. Solutions are given as a function of
angle around the crack tip, where = 0 corresponds
to the deepest point and = /2 corresponds to the
plate surface .......................................................................................................... 40
Figure 5Normalised Q-stress solutions of eqn (9) for
three point bend specimens .................................................................................. 41
Figure 6Critical value of J as a function of T/y for 3PB and
CCT specimens for a mild steel at -50C (from Ref 52) ..................................... 42
Figure 7Critical value of J as a function of Q for 3PB and CCT
specimens for a mild steel at -50C (from Ref 53) .............................................. 42
Figure 8Ductile toughness of an A710 steel at various crack
extensions as a function of normalised T-stress (Ref 58) .................................... 43
Figure 9Crack tip opening displacement of an A710 steel
at various crack extensions as a function of
normalised T-stress (Ref 58) ................................................................................ 43
Figure 10

Ductile toughness of a A533B-1 steel plate at 20C

for a variety of specimens, all 20% side-grooved (Ref 59) ................................. 44

Figure 11

Data of Figure 10 plotted in terms of the constraint parameter

T Lr (= T/y) of eqn (5) with curve fits to the data ............................................ 44

Figure 12

The R6 failure assessment diagram ..................................................................... 45

Figure 13

Schematic of Jssy Approach .................................................................................. 45

Figure 14

Typical laboratory warm pre-stress cycles ........................................................... 46

Appendix 1

Review of Crack Shape Development ................................................................ A1

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 1 of 46

1.

INTRODUCTION
European industry makes increasing use of so-called "fitness for purpose" or "engineering critical
assessment" methods in order to assess the likelihood of failure of structures, or to specify
materials requirements and safe operating conditions. These methods rely on a detailed
knowledge of the relationships between the toughness of the materials of construction, the
presence of defects, and the stresses applied to a structure.
In this report, some aspects of these "fitness for purpose" methods are reviewed. For each aspect
considered, the review identifies what is well accepted, what is state of the art, and what needs to
be done both in verifying the state of the art and in further development. The review is not
exhaustive but addresses only those aspects within task 2 of the SINTAP (Structural Integrity
Assessment Procedures for European Industry) project. Other aspects are being addressed within
the other SINTAP tasks.
Currently, there are two self-contained defect assessment procedures: the British Standards
document PD6493 (Ref 1) and the R6 approach (Ref 2). Both procedures have been extensively
validated and shown to be safe, i.e. to err on the side of conservatism. However, to apply these
procedures to practical structures it is necessary to have certain basic information. This
information may be conveniently discussed by considering the two parameters Kr and Lr used in
R6. These are defined by

K r = K I (P, a) / K mat

(1)

Lr = P / PL (a, y )

(2)

Kr measures the proximity to linear elastic fracture and depends on the material fracture
toughness, Kmat, and on the linear elastic stress intensity factor KI, which in turn depends on the
magnitude of the applied loading, P, and the defect size, a. Lr measures the proximity to plastic
collapse with PL being the value of the plastic collapse load for a perfectly plastic material with
yield stress, y.
Clearly, solutions for KI and PL are required for a range of geometries, defect sizes and loading
conditions in order to calculate Kr and Lr and, therefore, to apply defect assessment procedures. In
view of the fundamental importance of these solutions, these are discussed first in Section 2 of
this review.
Although the limit load is defined in equation (2) in terms of a yield stress, materials used in
industry exhibit a wide range of hardening behaviour beyond initial yield. Such hardening can
lead to an increase in load bearing capacity beyond that for an elastic perfectly plastic material.
This is reviewed in Section 3 in terms of the effect of the yield to tensile strength ratio.

For pressurised components, a defect may grow in such a way as to cause, in the first instance, a
stable detectable leak of the pressure boundary rather than a sudden, disruptive break. A leakbefore-break argument is aimed at demonstrating that leakage of fluid through a crack in the wall
of a pipe or vessel can be detected prior to the crack attaining conditions of instability at which
rapid crack extension occurs. Such arguments are a useful complement to arguments that a crack
will not penetrate a pressure boundary and are reviewed in Section 4.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 2 of 46
Although the PD6493 and R6 procedures have been shown to be conservative, there is a growing
need to quantify the safety factors inherent in the use of engineering critical assessment
procedures. One area where such a safety factor is known to exist is in the measurement of the
fracture toughness, Kmat, in equation (1). This is typically carried out on small-scale specimens
containing a deep crack and tested under predominantly bending loads. In contrast, structures
such as pipelines, tanks and pressure vessels are often under predominantly tensile loading and
contain only shallow defects as might arise from fatigue cracking or fabrication problems.
Transfer of results from standard small-scale tests to structures therefore incorporates a variable
safety factor due to constraint, which should ideally be quantified. Developments in this area
are reviewed in Section 5.
As written, equations (1) and (2) refer to single application of a load P. However, components are
often subjected to a variable load history and, in particular, to a prior overload or proof test. This
prior overload can affect the material properties to be used in a subsequent defect assessment and
this is discussed in Section 6.
Following the reviews of the various inputs to a defect assessment in Sections 2-6, which include
the future work required to develop defect assessment methods in these areas, some concluding
remarks are contained in Section 7.
2.

STRESS INTENSITY FACTOR AND LIMIT LOAD SOLUTIONS

2.1

Stress Intensity Factor Solutions

For mode I loading, the stresses ij close to a crack tip, calculated elastically, may be written as

ij =

KI
gij ( ) + T 1i 1 j + 0( r )
(2 r )

(3)

as r 0, for polar co-ordinates (r,) centred at the crack tip. Here, gij are angular functions of
, ij is Kronecker's delta and the second-order T stress term can be regarded as the stress parallel
to the crack flanks. This term is considered as a measure of constraint in Section 5. In this
section, attention is focused on the stress intensity factor, KI.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 3 of 46

2.1.1

Calculation of Stress Intensity Factors

Solutions for stress intensity factors (SIFs) can be obtained using analytical or numerical methods.
Closed form solutions obtained by analytical methods are available for some cracked bodies,
especially those where the geometry is simple or the body is infinite. However, most solutions for
SIFs have been obtained by numerical methods with the finite element method mainly used for
this purpose.
Values of SIF are usually published in terms of non-dimensional geometry factors, Y,
Y = K I / ( a )

(4)

where is some convenient applied stress. In some cases interpolation formulae are provided for
Y as a function of parameters such as crack aspect ratio, normalised crack depth or
radius/thickness ratio of a cylinder. Such formulae are convenient where calculations are required
for a range of crack sizes; for example, when addressing fatigue or stable crack growth or
calculating margins on crack size.
In practice, components are subjected to a range of loading conditions including thermal and
residual stresses. Therefore, SIF solutions are often required as a function of uncracked body
elastic stresses normal to the prospective crack plane rather than simply as a function of a nominal
applied stress as in equation (4). For non-uniform stresses the influence and weight function
methods can be used. These functions are available for some cracked bodies. In applying
influence functions the uncracked body stresses have to be approximated by polynomials. Such
an approximation is not necessary if weight functions are used.
2.1.2

Handbook Solutions for Stress Intensity Factors

Many SIF solutions can be found in handbooks (Refs 3-7), some of which (Refs 3,5) describe in
more detail the calculation methods discussed above. For frequently used geometry and load
conditions, documents on defect assessment methods (Refs 1,2,8 & 9) can also be used and these
contain references to other solutions. There are also simplified SIF formulae for very general use
in codes (Refs 1,10,11) which, however, are necessarily rather conservative.
Most solutions can be found for mode I loading, while for modes II and III solutions are available
for simplified geometries only. For numerical solutions estimates of their accuracy are usually
given. Despite the advantage of newer editions, for some cases more information can be found in
older handbooks, for example cracks in sheets with stiffeners in (Ref 4).
Many exact solutions and solutions with high accuracy can be found in Tada's handbook (Ref 5).
Most are for two-dimensional cracks in semi-infinite solids or in plates. The handbook includes
solutions for multiple cracks, cracks under point loads and modes II and III loading, cracks at
holes and penny-shaped cracks.

Zahoor's handbook in 3 volumes covers cases of cracked pipes including cracks in piping tees and
elbows. Solutions are provided for a wide range of geometry parameters but the accuracy of some
solutions is not very high.
Murakami's handbook, also in 3 volumes, is the largest collection of SIF solutions. Volumes 1-2,
published in 1987, are divided into 18 chapters. Several chapter titles give some idea of the
structure of the handbook: Fracture Mechanics Test Specimens; Finite Width Plate Containing

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 4 of 46
Two-Dimensional cracks; Cracks in a Circular Plate or Cylinder; Cracks at Stress Concentration;
Three-Dimensional Surface and Interior Cracks; Cracks in Welded Joints; Cracks in Residual
Stress Field. Volume 3, published in 1991, has the same structure and can be regarded as a
supplement containing newer results.
2.1.3

Solutions for Complex Geometries and Flaws

In practice, complex geometries are often handled by using solutions for simple geometries to
which the actual one can be approximated. However, the corresponding stress fields, as
determined by finite-element analysis for example, are generally non-linear and hence KI needs to
be determined by influence or weight function techniques. Such functions are, therefore, needed
for surface, embedded and through-thickness cracks and basic requirements are listed in Table 1.
For many of the cases listed, solutions are available in the literature but for others further work is
needed as discussed in Section 2.1.4.
For offshore structures, stress intensity factor solutions are required for tubular joints. A limited
number of solutions are available, particularly from finite-element analyses of Y joints with semielliptical surface cracks (See Appendix L of the draft revision to Ref 1). More generally, estimates
can be obtained using plate solutions.
For through-wall axial and circumferential cracks in cylinders, comprehensive stress intensity
factor solutions have recently been reported (Ref 12). These solutions were obtained from threedimensional finite-element calculations for cylinders with mean radius to wall thickness ratios
ranging from 3 to 100. The solutions cover membrane (or internal pressure), global bending and
through-wall bending loads.

2.1.4

Further Work Needed

Although the stress intensity factor handbooks are valuable sources of information, a collation of
SIF solutions is needed for efficient use of defect assessment methods. Such a collation should
take account of additional solutions not included in handbooks and address a limited number of
solutions most suitable for practical use regarding modelling of geometry and loading of real
components, accuracy, and validity for a wide range of geometry parameters. This limited list is
indicated in Table 1 which also includes areas where further work is needed beyond extraction of
results from the literature. It may be noted that work is needed, in particular, to increase the order
of available influence functions to address the highly non-linear stresses which occur in locations
with stress concentrations and in the residual stress fields associated with welded joints. For this
last application, it may be noted that stress intensity factor solutions for self-balancing through
wall
residual
stress

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 5 of 46

distributions representative of
those which might arise from some welding processes are
given in (Ref 13) for through-wall defects. These stress intensity factors are dependent on the
wall thickness rather than the length of the through-wall crack.
2.2

Limit Load Solutions

In this section, attention is focused on the limit load PL. Methods for generating limit loads are
briefly discussed in Section 2.2.1. Available solutions for homogeneous components are
reviewed in Section 2.2.2 and for components containing mismatched welds in Section 2.2.3. For
through-wall cracks, the limit load is often the so-called global collapse load, ie. the rigid plastic
limit load of the structure, calculated for a rigid-plastic material with a yield stress equal to y.
For surface or embedded flaws it is also possible to define a local collapse load which is the load
needed to cause plasticity to spread across the remaining ligament, calculated for an elasticperfectly plastic material with a yield stress y. This distinction between local and global collapse
loads is discussed in Section 2.2.4. Finally areas where further work is needed are reviewed in
Section 2.2.5.
In this section, limit loads in the region of a flaw are reviewed but it should be recognised that
when a component is being assessed, the possibility of collapse elsewhere in the structure should
also be investigated. For such investigations, plastic collapse loads for undefective components
are needed and these are contained in a number of textbooks and have, for example, been
reviewed by Save (Ref 14).

2.2.1

Methods for Obtaining Limit Load Solutions

There are a number of methods for obtaining limit load solutions including the following:The forces and moments, or their equivalent elastically calculated stresses acting over the
gross section containing the flaw may be treated using a generalised plate model. Such
an approach is common in Codes but care must be taken when the stress normal to the
crack plane is not the dominant stress component; as occurs for circumferential cracks in
pressurised cylinders, for example.
Established plastic limit load analysis or lower bound limit analysis may be used.
Non-linear finite-element analysis may be used.
Elastic finite element analysis may be used with a lower bound limit load deduced by
invoking the lower bound limit load theorem. The accuracy of this approach may be
improved by iteratively modifying the elastic stiffness of individual elements and
convergence of the method to the true limit load may be examined by involving the upper
bound theorem of limit analysis (Ref 15).

A scale model of the structure may be tested, taking care not only that the flawed
structure and loading are modelled correctly, but also that the model fails by plastic
mechanisms. Care needs to be taken in interpreting limit loads in tests as a result of
material work hardening beyond yield or ductile crack growth prior to collapse.
More information on some of these methods is given in Appendix 2 of R6
and in (Ref 16).

(Ref

2)

2.2.2

Available Limit Load Solutions for Homogeneous Components

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 6 of 46

A review by Miller (Ref 16) contains solutions for flaws in plates, bars, cylinders, spheres, pipe
bends and some shell intersection geometries. Further solutions are contained in (Ref 9).
However, it should be recognised that the accuracy of many solutions is not known and (Ref 16)
lists a number of alternative solutions for some cases.
In a similar manner to Table 1, a list of basic limit load solutions required for a practical
assessment procedure is given in Table 2. These solutions are required as functions of external
loads rather than local stresses as local thermal and residual stresses do not influence plastic
collapse. The basic solutions may then be supplemented by specific solutions for elbows and
tubular joints, although such solutions are often limited to specific component and flaw
dimensions. For example, global collapse loads for ship structural details are given in (Ref 17)
including advice for treating stiffened members.
2.2.3

Available Limit Load Solutions for Mismatch Welds

For defects in welds with a mismatch in tensile properties from the surrounding base material, the
mismatch limit load, which takes account of this strength difference, is an important input to
defect assessment procedures (Ref 18). Such limit loads may be calculated by the established
methods described in Section 2.2.1. While solutions for defective mismatched welds are not
widely available, results have been obtained for centre cracked plates, three point bend specimens,
and for fully circumferentially cracked cylinders under axial load and tension. These solutions are
summarised in (Ref 18) and more recently in (Ref 19) and cover a range of mismatch ratios.
When the size of an overmatched weld is large compared to the remaining ligament ahead of a
defect, the limit load may approach that of the defective geometry made totally of weld metal.

2.2.4

Global and Local Solutions

The distinction between so-called global and local collapse loads for part-penetrating defects
has been described above. As the ligament thickness ahead of a part-penetrating defect tends to
zero, the local limit load tends to zero. However, failure of the ligament need not correspond to
overall yielding as the component may be able to sustain a fully penetrating defect.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 7 of 46

The local limit load is less than or equal to the global limit load. Therefore, in assessments its
use generally leads to conservative results. However, in leak-before-break cases, for example, a
more realistic assessment is required and, therefore, the global limit load may be preferred.
It is important to recognise that the limit load is often used explicitly or implicitly to estimate
crack tip parameters such as J or COD. Therefore, the choice of limit load solution can be based
on whether the local or global solution provides the more accurate estimate. Some available
solutions for J have been listed by Chell et al (Ref 20) and include semi-elliptical surface flaws in
plates subjected to tension and bending and in cylinders subjected to internal pressure. As
demonstrated in (Ref 20), an appropriate limit load can be chosen to estimate J at both the surface
and deepest points of such defects. The global limit load often appears to provide the better
estimate of J but can lead to non-conservative assessments.
The use of a reference load to estimate J has also been developed by Gilles et al (Refs 21-25).
These workers have examined surface defects in pipes and elbows under pressure, tension,
bending and combined loadings. The results lead to limit loads which can be used to define Lr
by eqn (2) within the R6 method (Ref 2) or to define J-estimation schemes. The results of (Refs
21-25) are of importance in view of the practical geometries considered.
2.2.5

Further Work Needed

A collation of limit load solutions is needed for efficient use of defect assessment methods. In a
similar manner to Table 1 for SIF solutions, a limited list of solutions most suitable for practical
use is given in Table 2 for limit loads. This table indicates areas where further work is needed
beyond extraction of results from the literature. It can be seen that an important area for further
work is provision of advice on the use of the global or local limit loads. It is also important to
quantify the accuracy of the available solutions.

3.

YIELD/TENSILE STRENGTH RATIO EFFECTS

The trend towards the optimisation of the useful weight of structures has led to the use of
increased strength material. In this context high strength ferritic steels (y >450 MPa) have a
significant potential contribution which still remains largely unrealised. This is predominantly
due to design code limitations, the upper allowable limit of yield stress/ultimate stress ratio (Y/T)
being particularly severe. (Ref 26) presents a review of the current literature on the origins,
causes and structural significance of high Y/T ratios in steels. In this section, the broad
conclusions of this review are summarised and areas for further work are identified.

3.1

Summary of Effects
The significance of the yield/ultimate tensile strength ratio on the fracture behaviour of steels has
been investigated by means of a literature review (Ref 26).
The

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 8 of 46

principal areas assessed are the origins of the concern over Y/T ratio, the inter-relationship
between tensile parameters, the structural significance of the Y/T ratio and its treatment in
assessment codes. The main findings of the review are summarised in the paragraphs below.
Limits to the Y/T ratio were introduced into design codes based on the behaviour of firstgeneration high strength steels and the notion that a high Y/T value equates to poor fracture
performance. Modern steels give higher elongation values for a given strength level and Y/T ratio
and the initial concern is of lower relevance to modern steels.
Modern ferritic steels produced via controlled rolling or quenching and tempering generally have
Y/T in the range 0.8 - 0.95 compared to 0.65 - 0.75 for normalised steels. A high Y/T ratio is
generally associated with a low work hardening rate (high value of n in eqn (6) below); the
relationship is however neither linear nor consistent.
Current design code limits for Y/T vary between 0.67 and 0.90. There is general agreement that
values of up to 0.85 are satisfactory in conventional structural applications and values up to 0.95
in specific cases. However, these limits have not yet found their way into design codes.
Of the various parameters applicable to the post-yield regime, the yield tensile ratio, strain
hardening exponent, local elongation (Lders strain) and strain at UTS are the most relevant
parameters for structural integrity assessments.
Numerous estimates of n are available; most rely on the assumption that n can be correlated with
yield stress. Such correlations are promising, particularly when different forms are used for
different strengthening mechanisms. Furthermore, the strain at UTS appears to give a reasonable
estimate of n.
The presence of a crack modifies the shape of the load-deflection curve; crack depth and n dictate
the extent of this. The yield point may be suppressed and a Lders band not obtained. The latter
effect can however be observed in the CTOD strain response where a plateau of CTOD can be
achieved beyond yield in steels showing a Lders plateau in the conventional tensile test.
The significance of Y/T in buildings and bridges is only relevant for cases of earthquake
resistance in the former and plastic design in both structures. Design rotation capacities
(maximum/yield rotation) of connections are typically 3 for general plastic design and 7 for severe
earthquake design. The rotation capacity tends to decrease with increasing Y/T ratio although
geometry and thickness also have a major influence.
For tension members gross section yielding rather than net section fracture is the preferred failure
mode. Achievable elongation in the presence of holes such as bolt holes is very sensitive to Y/T.
For tapered members, strong sensitivity is only noted above Y/T of about 0.85. However,
industry experience suggests steels with values of Y/T up to 0.95 can be used without problems.

In the case of pressure vessels, burst pressure has been found experimentally to increase with
decreasing strain hardening exponent.
For tubular joints, decreasing Y/T from 1.0 to 0.66 at constant yield stress has been found to
enhance the ultimate joint capacity by only 6%. Tentative guidance suggests an upper limit of
Y/T of 0.85.
Work on defect containing pipelines has demonstrated that the effect of increasing defect depth on
tolerable defect length is more significant than increasing Y/T. For deep defects there is little
influence of Y/T above 0.85. For shallow defects there is significant influence above Y/T of 0.90.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 9 of 46
A number of failure assessment diagrams (FAD) are currently available in R6 (Ref 2) which
incorporate the Y/T effect indirectly through limits on flow stress definition. Others are generated
using actual stress-strain data. Both the shape of the FAD and the plastic collapse parameter cutoff depend on the Y/T ratio. Industry experience with these methods is wide but the influence of
high Y/T steels on the suitability of the FAD parameters needs assessing.
The Engineering Treatment Model (ETM) incorporates an estimated strain hardening parameter
in its approach. The method therefore predicts different behaviour for steels with the same yield
strength but varying n. The method is of significant interest in its ability to characterise behaviour
of steels with varying Y/T ratios and n values.
Crack driving force curves based on PD6493 (Ref 1) suggest relatively little influence of the
effects of Y/T in the range 0.5 - 0.8 but a significant influence above this level. Similar curves
based on the ETM show a systematic effect of n at all levels.
3.2

Future Work
Future work on the subject of yield/tensile ratio within the SINTAP project should concentrate on
the following aspects:
1.

Establish relationships between yield strength, Y/T, strain at UTS, n, composition and
steel type to enable more accurate predictions of the relevant post yield tensile
parameters.

2.

Examine the relationship between conventionally defined n and the ETM defined n and
establish the significance of this on predictions.

3.

Examine the significance of the yield plateau on behaviour of steels containing cracks.

4.

Assess the inter-relationship between crack depth and significance of Y/T.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 10 of 46

4.

5.

Determine the influence of Y/T on the shape and cut-off limits of FADs and assess the
accuracy of predicted crack driving force curves through comparisons with the results
from wide plate tests on different steels (parent plates).

6.

Assess the potential of a strain-based FAD, methods of deriving such an FAD and how it
compares with actual wide plate data.

7.

Assess the abilities of PD6493 levels 2 and 3 and the ETM to predict the crack driving
force curves of parent plate wide plate tests.

8.

Link with SINTAP Task 1 to assess the influence of Y/T in welded joints and the effect
on the significance of mis-match.

LEAK BEFORE BREAK AND CRACK SHAPE DEVELOPMENT

Over recent years, the concept of Leak-Before-Break (LBB) has gained considerable world-wide
momentum in establishing safety cases for pressurised components, particularly in the nuclear
industry in relation to primary pipework.
The various stages in the development of a LBB argument may be explained with the aid of the
diagram shown in Fig. 1. This diagram has axes of crack depth, a, and crack length l, normalised
to the pipe or vessel wall thickness, t. An initial part-through crack is represented by a point on
the diagram. The crack may grow by fatigue, tearing or any other process until it reaches some
critical depth at which the remaining ligament ahead of the crack breaks leading to a through wall
defect. The crack then continues growing in surface length until there is sufficient opening to
cause a detectable leak or until the crack becomes unstable. A LBB argument is aimed at
demonstrating that leakage of the appropriate gas or fluid through the crack can be detected prior
to the crack attaining conditions of instability at which rapid crack extension occurs.
LBB arguments may be used as part of the case for the elimination of pipe-whip restraints, be
applied in regions that are difficult to inspect due to inaccessible or hazardous conditions and/or
be applied as defence-in-depth considerations.
Various methods for LBB have been developed in several European countries, many of which are
based on a procedure published by the US Nuclear Regulatory Commission in NUREG 1061
(Ref 27). Procedures relating to the NUREG 1061 approach are relatively simplistic in that they
are based on detectable leakage. Their starting point is to postulate a fully-penetrating defect and
show that, should such a defect arise, the leakage would be detectable before the defect grows to a
limiting length. More rigorous LBB approaches have also been developed which involve
calculating the growth of a postulated initial surface flaw up to and beyond penetration of the
back-surface after which leakage rates are evaluated. The recently revised LBB Appendix
(Appendix 9) (Ref 28) of R6 (Ref 2) incorporates both the NUREG 1061 and the more rigorous
type of approach.

4.1

General Outline of Procedures

The NUREG 1061 type of approach consists of the following steps:
1.

The limiting length of a through-wall crack is evaluated for the most severe loading
conditions (ie. the loading condition resulting in the lowest value of limiting crack
length).

2.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 11 of 46
The length of through-wall crack, corresponding to that which leaks at the minimum
detectable rate under normal operating conditions, is evaluated. In practice, crack
opening areas and leakage rates are calculated for different crack lengths until the
detectable leakage length is attained.

The actual NUREG 1061 procedure requires there to be factors of at least 10 on the detectable
leakage rate, at least 1.4 between the loads to cause instability of the postulated flaw and normal
operation + Safe Shutdown Earthquake (SSE) loads, and at least 2 between the limiting crack
length and the leakage crack length. In contrast, no safety margins are specified in the R6 LBB
adoption of this method. This is consistent with the general R6 procedures which invoke the
principle of undertaking sensitivity studies as an alternative to the safety factor concept.
The more rigorous LBB approaches typically consist of the following steps:
1.

A known or postulated "initial" crack (usually a surface crack) is characterised in

accordance with appropriate characterisation rules.

2.

The shape development arising from potential crack growth mechanisms of the
postulated crack is assessed.

3.

The defect length at which ligament failure is predicted to occur is calculated.

4.

The defect for which ligament failure is predicted to occur is re-characterised as a

through-wall crack.

5.

The limiting length of through-wall crack is evaluated for the most severe loading
conditions (ie. the loading condition resulting in the lowest value of limiting crack
length).

6.

Provided the limiting length of through-wall crack (step 5) is greater than the recharacterised through-wall crack at ligament failure (step 4), the crack opening area for
the re-characterised crack is calculated.

7.

Leakage rate is calculated for the crack opening area evaluated in step 6.

8.

The time to detect the leak from the crack is estimated.

9.

The time to grow the re-characterised through-wall crack at ligament failure (step 4) to
the limiting crack length (step 5) is evaluated.

4.2

Review of Main Aspects

4.2.1

Limiting Crack Length

The limiting length of a through-wall crack may be calculated using defect assessment procedures
such as (Refs 1, 2). It is important that the minimum limiting length is calculated for the most
severe loading condition which could be a frequent, infrequent or seismic loading case. To ensure
a conservative assessment, lower bound material properties relevant to the crack location (eg.
weld, parent or heat affected zone material) would usually be specified. All relevant primary and
secondary stresses need to be taken into account.

4.2.2

Crack Opening Area

Estimation methods for crack opening area can be classified into three categories: linear elastic
models; elastic models incorporating a small scale plasticity correction; and elastic-plastic models.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 12 of 46
Several elastic solutions are available in the literature for crack opening areas in pipes, cylinders
and spheres. However, the solutions are generally based on thin-shell theory and the calculated
crack opening is assumed to be at the mid-thickness position of the wall. These solutions
therefore do not take account of crack taper arising from geometry effects and from through-wall
bending loads.
(Ref 12) includes crack opening displacement (2) solutions for a range of axial and
circumferential through-wall cracks in cylinders. As explained in (Ref 12), crack opening area
can be calculated from these values by assuming an elliptical shape (ie. a where a is crack
semi-length). Alternatively, to always ensure conservatism, crack opening area can be evaluated
as 2.5a.
The solutions of (Ref 12) are included in a table in the revised Appendix 9 of R6, reproduced here
as Table 3 which gives recommended crack opening area solutions for the three categories
referred to above. The more accurate elastic-plastic model of (Ref 33) is recommended for best
estimate LBB calculations where stress levels are high enough to induce significant plasticity (ie.
Lr greater than about 0.4). However, this method is detailed and requires a description of the
material stress strain curve.
Some justification for the solutions recommended in Table 3 is given in (Ref 34).
4.2.3

Leak Rate
The calculation of the fluid flow or leak rate through a crack is in general a complex problem
involving the crack geometry, the flow path length, friction effects and the thermodynamics of the
flow through the crack.
Several computer codes have been written to predict leakage rates through cracks for a variety of
fluids. For single-phase flow, DAFTCAT (Ref 35) calculates flow rates through rectangular
section cracks and includes the effects of friction. For two-phase

flow of steam/water mixtures, PICEP (Ref 36) and SQUIRT (Ref 37) can be used to calculate
leak rates through a variety of cracks. All of these programmes have been validated to some
extent against a variety of experimental data and reasonable agreement with experiment obtained.
Whilst flow rate measurements have been made on real cracks, the extent of validation for such
cracks is relatively small and the agreement with theory less good.
The likely accuracy of the leak rate predictions for both single and two-phase flows depends on a
variety of factors and must be judged by examining the available validation data.
Formulations for friction factor f, show it to increase continuously as roughness increases or crack
width (opening) reduces. However, flow rate experiments show that f does not increase
continuously, but reaches an effective maximum. The effective maximum friction factor, fmax, is
dependent upon surface geometry. In (Ref 38) a theory is advanced to justify the existence of a
maximum friction factor and this is assessed against experiments using relatively large scale
conforming surfaces (ie. one surface is manufactured and the opposing one is a replica). For the
surface of most relevance to structural defects, random roughness, fmax was approximately 0.2.
For a very regular and stepped surface fmax was unity. This range of values has been confirmed by
experimental data on flow through real cracks, the results of which are discussed collectively in
(Ref 39). The above discussion on f values relates to fully developed turbulent flow. Higher
values can occur in laminar flow, but are of little interest in LBB.

4.2.4

Leak Detection

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 13 of 46

Reference 40 gives some information and guidance on leak detection systems. There are two
broad categories; global and local. Examples in the global category are sump pumps, pumps for
water systems, humidity detection for steam leaks, gas levels in air for gaseous systems, and
radiation monitors for nuclear systems. All global systems detect all leaks and hence any leakage
indications on the monitoring equipment need to be investigated and the source established. The
response time for such systems is relatively long and depends on plant segregation.
Local leak detection systems monitor specific plant features (eg. a weld) or a well defined area
(eg. length of pipe). Some detectors are medium or plant specific. For example, moisture
sensitive tape only works in water or steam systems where condensation can take place on the
outer surface.
Leakage through cracks generates acoustic emission that is transmitted through the structure, and,
in some circumstances, through the air. Wave guide and microphone systems have been
developed which offer flexible and sensitive leak detection capabilities for a wide range of fluids.

4.2.5

Crack Shape Development

All the aspects covered under sections 4.2.1 to 4.2.4 are relevant to the simplified NUREG 1061
approach in that a straight fronted crack is assumed with no account being taken of prior growth.
In reality, after breakthrough, the crack shape for real growing cracks can be quite complex with
the outer crack length often much smaller than the crack length at the inside of the pipe. This
affects both the resulting leak rate as well as the crack growth rate at the leaking crack. A good
understanding of crack shape development is important for undertaking the more detailed LBB
methodology outlined in Section 4.1. Therefore, a more detailed review of this aspect is given in
Appendix 1.
It may be noted that Brickstad (Appendix 1) infers that local cooling of the outer wall surface can
occur due to flow discharge. This phenomenon has also been reported by Eperin et al (Ref 41)
from analytical work on applying LBB to the Leningrad Nuclear Power Plant. However, such
calculated cooling may be a consequence of inaccurate thermodynamic assumptions since there is
experimental evidence (Ref 42) to suggest that no significant cooling actually occurs within the
material.
The review given in Appendix 1 covers the growth of a surface crack, wall penetration and
growth of a leaking crack to final failure, all with reference to both reported experimental and
analytical studies.
Guidance on crack shape development, particularly at and following wall penetration is also given
in the new Appendix 9 of R6. The guidance is based on experimental evidence, reviewed in (Ref
43), which shows that for cases where stress distributions are predominantly tensile, cracks tend
towards a rectangular shape. The R6 recommended re-characterisation rules for such loading and
where failure occurs in a ductile manner are summarised in Fig. 2. In order to be conservative for
cases where the stress distributions are predominately through-wall bending, the R6
recommended re-characterisation rules are as summarised in Fig. 3.

4.3

Unresolved Issues and Scope for Further Work

In Appendix 1, Brickstad highlights some unresolved issues as being (i) the accurate
characterisation of the breakthrough crack, (ii) the problem of being able to accurately predict
crack shape development from J-R data obtained from small specimens, and, (iii) problems in
accurately evaluating crack opening areas in weldments due to both strength mis-match and
residual stress effects.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 14 of 46
Appendix 1 outlines proposed work under the SINTAP programme to undertake a series of
detailed non-linear finite element analyses to assess crack characterisation at breakthrough.
Other uncertainties where there is scope for further work include; (a) the effect of restraint and
non-symmetrical loading on crack opening area (eg. a crack around a nozzle), (b) crack opening
area solutions for pipe bends and T-junctions, (c) validation of limit load solutions for throughthickness
cracks,
(d)
further

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 15 of 46
experimental validation of flow rate models applied to realistic cracks, and (e) the effect of
multiple defects and associated proximity effects which are relevant to defects associated with
stress corrosion cracking, for example.
5.

CONSTRAINT
Both numerical analysis and laboratory measurements have been used to demonstrate that the
toughness of a material under elastic-plastic conditions depends on geometry, loading, specimen
thickness and normalised crack depth, a/w, where w is specimen width. Such a dependence is
usually referred to as an effect of constraint. It is found that tests on deeply cracked bend
specimens (high constraint geometries) under plane strain conditions provide the lowest value of
toughness and the use of such toughness as Kmat in equation (1) leads to conservative assessments.
However, there has been significant worldwide activity in recent years to reduce conservatisms
by taking credit for the increased toughness in lower constraint conditions.
In this section, the recent developments in quantifying constraint effects are reviewed. First,
parameters for indexing constraint levels in structures and their availability are discussed in
Section 5.1. Then, experimental data illustrating the influence of constraint on material toughness
are described in Section 5.2. Section 5.3 reviews local approach methodologies which are
capable of predicting the observed increase in toughness with reducing constraint. While such
methodologies are capable of being applied to structural geometries, the numerical effort required
is large and, therefore, simpler methods based on the parameters discussed in Section 5.1 have
been developed. These are reviewed in Section 5.4. Validation for these methodologies is
described in Section 5.5 and areas for further work are outlined in Section 5.6.

5.1

Constraint Parameters and Their Availability

The elastic T-stress in equation (3) has been used as a means of quantifying crack tip constraint
(Ref 44). As T can be evaluated by elastic analysis it is relatively straightforward to determine
and solutions are available in the literature for a number of cases. A compendium of these
solutions has been compiled by Sherry et al (Ref 45) for both two and three dimensional cracked
geometries. The solutions in the compendium are for the following geometries:
-

Centre-cracked plate tension specimen (CCT)

Centre-cracked plate tension specimen biaxially loaded (CCBT)
Double-edge cracked plate tension specimen (DECT)
Single-edge cracked plate tension specimen (SECT)
Single-edge cracked pure bending specimen (SENB)
Single-edge cracked three-point bending specimen (3PB)
Compact-tension specimen (CT)
Double-cantilever beam specimen (DCB)
Axisymmetrically-cracked tensile specimen (ACT)
Circumferentially-cracked cylinder under a tensile stress (CCCT)
Semi-elliptically cracked plate under uniform tension (SECPT)

Semi-elliptically cracked plate under uniform bending (SECPB)

For use in the procedures described in Section 5.4, it is convenient to normalise the T-stress in
terms of the limit load parameter Lr of equation (2). This produces the elastic constraint
parameter.
T = T/( Lr y )

(5)

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 16 of 46
For the cases listed above, with the exception of the DCB geometry, Sanderson et al (Ref 46)
have compiled a compendium of T solutions. An example for the SECPB geometry is given in
Figure 4.
Although the T-stress is based purely on elastic analysis, it has been used to characterise
constraint beyond the elastic and small-scale yielding regimes. To extend constraint descriptions
into the widespread plastic regime, analyses of crack tip fields for elastic-plastic materials have
p
been performed. For materials in which plastic strain, , is related to stress, , by a power law
n

= '(/)

(6)

where , n are constants and y = y/E, the stress field near the crack tip as r 0 may be written
1 /n+1

J
ij

=
y y y In r

~ij ( , n) + Q ij , | | <
2

(7)

Here In and ij are functions of n as determined by the asymptotic analyses of Hutchinson, Rice
and Rosengren.
Unlike eqn (3) for elastic response, eqn (7) is not a strict asymptotic analysis of the higher-order
terms for the non linear problem. Instead, the hydrostatic term, Q, is an approximation to the
collective behaviour of a number of higher-order terms in the forward sector, < /2, ahead
of the crack (Ref 47).
An alternative convention used to define Q is as the difference between the near-tip stress field
and that under small-scale yielding at the same value of J with T = 0; ie.
ssy
ij / y = ij / y + Q ij

(8)

In practice, at least in two-dimensional problems, the Q stress of eqns (7, 8) varies slightly with
distance from the crack tip and has been evaluated at r/(J/y) = 2. Thus, Q is evaluated at
different physical distances from the crack tip as the load, as measured by J, increases.
It is worth noting that the yield stress, y, in eqn (6) is somewhat arbitrary as the same stressplastic strain relationship can be obtained for different values of y by adjusting
the constant . As Q is evaluated at a distance dependent on y, it is, therefore, necessary to
adopt a consistent definition such as the 0.2% proof stress.

The load dependence of the Q-stress is more complex than the linear relationship for the elastic Tstress which enables a load-independent parameter to be defined by eqn (5). However, it
transpires that an approximately load-independent parameter, at least for loads which are small
compared to the collapse load, may be obtained as
Q = Q/ Lr

(9)

Reference 46 contains Q solutions, for Q defined by eqn (8), plotted against Lr for the CCT,
DECT, 3PB and CT geometries. Figure 5 gives an example of the Q solutions for the 3PB
geometry for a range of work hardening coefficients, n, and a/w values. All currently available Qstress solutions were taken on board in compiling the Q solutions given in (Ref 46). It may be
seen from Fig. 5 that Q is only weakly dependent on load (or Lr) and on the value of n but that
constraint reduces with reducing relative crack depth, a/w.
Parameters other than T and Q have also been proposed as measures of constraint. For example,

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 17 of 46
the ratio of hydrostatic to equivalent stress ahead of the crack tip or higher order terms in the
expansion of the crack tip stress field (Ref 48) have been proposed. From slip-line field solutions,
the level of constraint at the crack tip at plastic collapse (Lr = 1) can be deduced and this has been
used to deduce constraint levels for defects in mis-matched welds (Ref 49). Such approaches are
less well developed than the T and Q stress methods and information for their use is not widely
available. Therefore, they are not discussed further here.
5.2

Influence of Constraint on Material Toughness

5.2.1

Cleavage Fracture
Several sets of experimental data have been presented for the transition region which illustrate
that specimens with negative T-stress or Q-stress are tougher than deeply cracked specimens with
positive T-stress or Q-stress values.
Betegon and Hancock (Ref 50) presented experimental results as J versus T/y. Bend geometries
with a/w < 0.3 give negative T values. For the deeply cracked geometries (a/w > 0.3), the
toughness was found to be independent of geometry, as these geometries are known to have
positive T.
Toughness tests on a low-grade mild steel at -50C have been reported (Refs 51-53). Three point
bend specimens with a/w between 0.05 and 0.78 were tested along with centre cracked plate
tension specimens with a/w ratios between 0.63 and 0.77. Figure 6 shows the critical value of J
versus T/y at cleavage. A comparison between the two types of specimen shows that the low
constraint CCT specimen gives slightly higher values of Jc than the 3PB specimen, at the same Tstress values. Fig 7 shows the same data reanalysed in terms of the Q-stress and indicates that
slightly better correlation of CCT and 3PB specimens can be achieved. However, the dominant
effect is the increase in toughness with reducing constraint rather than the choice of constraint
parameter.
Reference 53 also presented data from test specimens in a high strength weld metal with a yield
stress of 700 MPa at a test temperatures of -30C. The critical value of J
was plotted as a function of both the T-stress and Q-stress parameters. Constraint

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 18 of 46

enhanced toughness was found to be even more significant for this type of material than for the
mild steel. The J/T and J/Q analyses described the data equally well.
Kirk et al (Ref 54) have presented cleavage toughness data for an A515 steel at room temperature,
using edge cracked bend bars with different a/w ratios and various thicknesses. They presented
the results in terms of both J/T and J/Q. Similar trends were shown between the two types of
analysis in terms of increased toughness with more negative crack-tip constraint parameter value.
However, there appeared to be some inconsistencies between the actual values of T-stress and Qstress in so much that whilst the reported T/y values ranged from approximately -1.7 to +2, the Q
values ranged from only approximately -1.3 to zero.
Sharples et al (Ref 55) reported on an experimental programme on 70mm thick A533B-1 steel
plate specimens containing a surface crack and loaded in either uniaxial four-point or biaxial
eight-point bending. Tests were carried out at temperatures of -75C and -90C for crack depths
of 10% and 20% of the plate thickness. Fracture toughness values obtained from the tests showed
relatively little scatter and all the values were shown to lie above the SENB shallow crack (a/w)
lower bound transition curve. No detrimental effect of biaxial load was shown in terms of either
toughness or crack mouth opening displacement, in contrast to other experimental work (Refs 56,
57). However, further finite element analyses are required before the experimental results of (Ref
55) can be fully verified. Results of T-stress analyses of the uniaxial and biaxial bend specimens
are included in (Ref 55). The T-stresses are generally negative with little variation near to the
deepest point of the crack but with a strong variation close to the free surface. Since failure in the
experiments emanated from the deepest point of the crack, these T-stress results are consistent
with the experimentally measured increase in toughness. However, (Ref 55) also presents the
variation along the crack front of the sum of normalised T and S stresses where S is a constant
stress term which acts parallel to the front of a three-dimensional crack. The (T+S) results for the
eight-point bending cases are all positive and the results for the four-point bending cases are all
negative. These (T+S) stress results are therefore inconsistent with the reported experimental
results of (Ref 55) but generally support the results of (Refs 56, 57) that the fracture toughness for
biaxial loading is lower than that for uniaxial bending.
5.2.2

Ductile Fracture
Hancock et al (Ref 58) tested samples of an American pressure vessel steel denoted A710 in the
upper shelf regime. Crack extension occurred by stable tearing enabling both J and Crack
Opening Displacement to be measured as a function of crack extension a. Figure 8 presents J
for crack extensions a = 0, 0.2mm and 0.4mm. Figure 9 shows the corresponding crack opening
displacement, , values.
The results show a marked effect of constraint on toughness after small amounts of crack growth.
J values for centre cracked plates are approximately 4 times greater than those of highly
constrained deeply cracked bend bars and CT specimens at a crack extension of 0.2mm. For the
higher
crack
extensions
(a
=
0.4mm),
the

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 19 of 46

constraint effect is even more significant as J for the centre cracked plate specimens is more than
5 times the value for the deeply cracked bend bars. Geometries with positive T-stresses (deeply
cracked bend bars and CT specimens) show little or no geometry dependence on toughness.
Conversely, geometries which have negative values of T show geometry dependent toughness.
Sherry et al (Ref 59) investigated the upper shelf fracture toughness of an A533B-1 steel plate. Jresistance curves were obtained for small-scale (w = 50mm, B = 25mm) three-point bend
specimens with a/w values of 0.3, 0.5 and 0.7 and for small-scale (w = 50mm, B = 25mm) centrecracked plate tension specimens with a/w values of 0.5 and 0.7. J was plotted against the
normalised T-stress (Fig 10). This shows a similar trend to the data of (Ref 58) in Fig 8. Curve
fits to the data in Fig 10 were produced in terms of the parameter T of eqn (5), see Fig 11.
5.2.3

Summary of Observed Behaviour

The typical results described above illustrate general trends which may be summarised as follows.

5.3

(a)

The fracture toughness at cleavage fracture tends to increase with reducing constraint.

(b)

The fracture toughness at ductile crack initiation is relatively insensitive to constraint but
the toughness after some small ductile crack extension tends to increase with reducing
constraint.

(c)

In view of (a) and (b), for ferritic steels there tends to be a reduction in the brittle to
ductile transition temperature with reducing constraint.

(d)

Data on the influence of biaxial loading do not show consistent effects although this may
simply be that different specimen types have different constraint levels under such
loadings.

Local Approach Methodologies

Local approach models have been the subject of much research in recent years and have been
used to analyse fracture events in steels for a range of geometries and loadings (Refs 60-64). In
essence, micro-mechanical models of the failure processes are used to relate the stresses, strains
and 'damage' local to a crack tip or stress concentrating feature to the critical conditions required
for fracture. The models are calibrated through material specific parameters which are derived
from analyses of laboratory tests on notched or cracked components and quantitative
metallography. Once the material parameters have been derived, structural assessments can be
performed. Therefore, the effects of constraint are handled automatically and the models are
capable of predicting the general trends summarised in Section 5.2.3.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 20 of 46

There are a number of micro-mechanical models of fracture addressing both cleavage and ductile
fracture. Some of these are briefly described here but fuller details can be obtained from the
references cited.
The Beremin cleavage model (Ref 65) predicts the cumulative probability of cleavage
failure through a weakest-link interpretation of the cleavage fracture process.
The Beremin model of ductile fracture (Ref 66) is based on the evaluation of stress and
strain ahead of a crack tip controlling the initiation, growth and coalescence of voids (i.e.
damage). The development of this damage controls the initiation and propagation of
cracking.
The Rousselier model of ductile fracture (Ref 67) also models crack extension through
the initiation, growth and coalescence of voids. However, whereas the Beremin model
uses an estimate of void growth rate due to Rice and Tracey (Ref 68), in the Rousselier
model the constitutive equations of the material are modified to represent the
deterioration of the material properties as voids develop. The ability of the material to
support stress is gradually lost in the highly damaged regions and the crack tip essentially
advances automatically without the need for a separate failure criterion, such as critical
void radius.
The Gurson model of ductile fracture (Ref 69), also uses a constitutive equation
incorporating void growth. Refinements to this model have also been made to include
void nucleation effects (Ref 70). The model has also been embodied in descriptions of a
decohesion layer within which crack advance occurs when the cohesive strength of the
layer is reached (Ref 71).
The separate ductile and cleavage models described above have also been combined to predict
cleavage fracture in the brittle to ductile transition region of a ferritic stress where cleavage can
occur after some ductile tearing (Refs 70, 72). These applications are very recent and still in the
development stage.
As noted above, local approach methods have the ability to describe the effects of constraint on
material response. This ability to predict the broad trends seen in Figures 6-11 may be used to
reduce testing requirements by interpolating between data obtained at only a few levels of
constraint. In addition, the models may be used directly to predict structural response. However,
the methods are complex for general structural applications and require highly refined meshes
near the crack tip. Hence, modifications to simplified defect assessment procedures to incorporate
constraint have been developed. These are described next.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 21 of 46

5.4

Defect Assessment Methodologies

5.4.1

Two parameter approaches

If fracture is described by a single parameter such as J or COD, then constraint can not be treated
unless a component specific critical value of these parameters is available. More generally a
lower bound critical value is obtained from deeply cracked bend specimens and conservative
assessments arise. The constraint parameters, T and Q, described in Section 5.1 allow some
reduction of conservatism by invoking two parameter fracture mechanics. In these approaches,
fracture is assumed to be controlled by two parameters, such as (J, T). A locus of (J, T) with
increasing load may be produced for a specific component and defect size using standard J
estimation procedures such as (Ref 8) and solutions for T (or Q) from the information cited in
Section 5.1. Examples for specimen geometries have already been given in Fig 10. The
intersection of the (J, T) locus with the corresponding material toughness curve (for example
Fig.11) then defines the load at fracture.

5.4.2

Failure Assessment Diagram Methods

Although they also include plastic collapse criteria, the failure assessment diagram methods
embodied in R6 and PD 6493 (Ref 1, 2) are essentially single parameter procedures in that
fracture is assumed to be governed by a single value of toughness or crack opening displacement.
The parameters Kr and Lr of eqns (1, 2) are evaluated and the point (Lr, Kr) plotted on a failure
assessment diagram. If this point lies within a failure assessment curve, Kr = f(Lr), then failure is
avoided (see Fig.12).
The failure assessment diagram methods may be modified to allow for constraint by replacing Kmat
c
in eqn (1) by a constraint dependent toughness, K mat say, which is a function of T or Q (Ref 73).
With this approach, the shapes of the failure assessment curves
in (Refs 1, 2) are unchanged but
c
Kr is a nonlinear function of load through the dependence of K mat on load (since both T and Q are
load dependent). Alternatively (Ref 74), the failure assessment curve is modified to
c

Kr = f(Lr) (K mat/Kmat)

(10)

and the definition of Kr in eqn (1) is retained. A convenient fit to data which has been used in
Fig.11 is
c

K mat = Kmat
c
m
K mat = Kmat [1 + (- Lr) )

Lr<0

Lr>0

(11)

where , m are material and temperature dependent constants and can be T or Q, Section 5.1,
depending on the constraint parameter used. With this fit, eqn (10) becomes
m

Kr = f(Lr) [1 + (-Lr) ]

(12)

Some modified failure assessment curves using the R6 option 1 curve for f(Lr) are shown in
Fig.12. Whereas the option 1 curve is independent of geometry and material, the modified curves
depend on geometry (through ), on material toughness properties (through , m) and also on
material tensile properties if is defined in terms of Q.
Both the procedure using a modified definition of Kr and the procedure using a modified failure

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 22 of 46
assessment curve have recently been included in a new Appendix 14 to R6 (Ref 2). The
methodology is limited to mode I loading, to primary loads and to loss of constraint under plane
strain conditions. Further work is, therefore, required to extend the approach. There is
experimental validation for the modified R6 methods and this is discussed in Section 5.5.
5.4.3

The Jssy Approach

An alternative approach for addressing constraint has been developed by Dodds et al (Ref 75).
This involves calculating the value of J in the component which gives the same crack tip stressed
volume as would be achieved in a specimen under small-scale yielding at a value Jssy. Clearly, the
'stressed volume' needs definition and for cleavage fracture has been chosen as the volume over
which the maximum principal stress exceeds a particular value. The relationship between Jssy and
J is shown schematically in Fig 13. High constraint conditions correspond to J = Jssy. Low
constraint corresponds to J > Jssy. In essence, Jssy is equated to the J equivalent of the (high
constraint) fracture toughness Kmat and under low constraint the value of applied J must exceed Jssy
in order for the critical fracture condition to be achieved at the crack tip.
A difficulty with the approach is that component analyses must be performed to generate curves
of the type shown in Fig. 13. Such curves are specific to the component, stress-strain curve,
defect size, loading type and fracture criterion adopted. However, the curves have been found to
be only weakly dependent on the magnitude of the maximum principal stress criterion (Ref 75) so
that some simplification is possible.

5.5

Validation of Methodologies
Experimental validation for the modified R6 methods outlined in Section 5.4.2 has been
addressed in (Ref 73) for centre-cracked and three-point-bend specimens for a range of materials:
cleavage fracture of a grade 43A normalised plain carbon steel at -50C; cleavage in a high
strength weld metal at -30C; cleavage fracture of a quenched plain carbon steel at room
temperature; cleavage fracture of a normalised CMn steel at temperatures below -140C; ductile
crack initiation in an A710 pressure vessel steel at room temperature. The results demonstrate
that points (Lr, Kr) lie close to, but outside, the R6 option 1 failure assessment line when Kr is
c
defined in terms of K mat. In contrast, analyses using the basic R6 procedures lead to
over-conservative results with points well outside the failure assessment line.

Experimental validation on small- and large-scale tests is also summarised in the new R6
Appendix 14. The cleavage fracture results on a grade 43A normalised plain carbon steel at 50C, have been compared with the modified R6 method using the local approach to obtain a
c
lower bound to the constraint modified fracture toughness, K mat. The results again demonstrate
conservatism in the approach which is less than that using the basic R6 procedures. Cleavage
fracture in two low constraint biaxial bend experiments on A533B steel plates in the lower
transition region has also been addressed. The constraint modified R6 approach reduced the
conservatism of the conventional failure assessment curves leading to a 30-40% benefit in terms
of load margin for the lower constraint geometry, while maintaining conservatism. When the
approach was applied to ductile fracture of a large scale single edge notched A533B plate at
20C, the load margin was unchanged at about 1.1 because the structural constraint was not
particularly low.
5.6

Further Work
The availability of solutions for constraint parameters, as discussed in Section 5.1, is essential if
simplified procedures are to be used in practice. While solutions are known for a number of
simple geometries, the range of solutions available is much less than for the stress intensity factor

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 23 of 46
and limit loads solutions discussed in Section 2. Therefore, further work is required to generate
accurate solutions, particularly for Q. Specifically, solutions are needed for surface cracks in
plates under uniaxial and biaxial loading and in pressure vessels under pressure and bending.
The influence of constraint on material toughness has been discussed in Section 5.2 and the
observations summarised in Section 5.2.3. There is a need to confirm these observations by
analysing data on a wider range of materials, addressing both J and COD approaches. Some work
of this nature has recently been reported for an A533B steel (Ref 76). Large-scale fracture
behaviour was found to be sensitive to biaxiality ratio in the lower transition regime. A series of
small-scale shallow-crack tests is proposed to enable interpretation of these large scale tests and
hence to provide additional validation of the methodologies described above in Section 5.4.
It has been noted in Section 5.4 that the constraint methodologies have been developed for
primary loading. For practical applications, it is necessary for methodology to address combined
primary and secondary loadings. For such loadings, not only must the methodology be developed
but methods for estimating the constraint parameters, T and Q, must also be provided.
For the local approach discussed in Section 5.3, it was noted that analysis of cleavage following
ductile tearing had only recently been performed. This is a topic for which further development is
required. More generally, the local approach methods need to become well defined procedures so
that they can be used for critical assessments or for predicting the dependence of fracture
toughness on constraint.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 24 of 46

6.

PRIOR OVERLOAD
Proof loading and Warm Pre-stressing (WPS) are two areas where prior overload conditions are
deliberately imposed. Their effects on structural integrity are quite complex and have been the
subject of extensive and continued research. This has demonstrated that a single overload (or
proof load, or warm pre-stressing event) can offer significant benefits in the enhancement of the
integrity of a structure (Refs 77, 78). The main benefits that have been reported are:
-

arrest or retardation of fatigue crack growth;

redistribution or relief of welding residual stresses; and
an improvement in the subsequent low temperature defect tolerance of the structure.

In order to make use of these benefits, models have been developed which predict the changes in
the behaviour of the structure. These models have been validated and built into assessment
procedures such as PD6493 and R6 (Refs 1, 2). For example, PD6493 incorporates the reduction
in residual stresses that can be produced by proof testing. More recently, R6 has included models
which predict the warm pre-stress effect (Refs 79, 80).
In this section, the warm pre-stress approach in R6 is first briefly summarised in Section 6.1.
Then, limitations to application of the approach are summarised in Section 6.2 and future work
identified.
6.1

Warm Pre-stressing Methodology

In this section, the warm-prestressing methodology recently embodied in R6 is described.
Application of the methodology is, however, only acceptable if certain conditions are satisfied.
These are listed in Section 6.2 and used to identify areas for further work.
A warm prestress (WPS) is an initial pre-load applied to a ferritic structure containing a preexisting flaw which is carried out at a temperature above the ductile-brittle transition temperature,
and at a higher temperature or in a less-embrittled state than that corresponding to the subsequent
service assessment. A WPS argument elevates the stress intensity factor at fracture, Kf, above the
corresponding fracture toughness, Kmat, in the absence of the WPS, so that the fracture toughness
used in eqn (1) is taken as Kf.
There are three types of cycle which are used in the laboratory to demonstrate the WPS effect
(Figure 14). The temperatures at which the pre-load and re-load to failure occur are denoted by T1
and T2, respectively, in each case. Similarly, the stress intensity factors due to the pre-load and
following the unload are denoted K1, K2, respectively.

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 25 of 46

(a)

Load-Unload-Cool-Fracture (LUCF), where the structure is pre-loaded at temperature T1

to stress intensity factor K1, unloaded to stress intensity factor K2, cooled to temperature
T2 and re-loaded to fracture.

(b)

Load-Cool-Unload-Fracture (LCUF), where cooling to T2 takes place prior to unloading

and re-loading to fracture.

(c)

Load-Cool-Fracture (LCF). This is similar to the LCUF cycle except that no unloading
occurs prior to the imposition of extra load to fracture.

The greatest benefit in terms of maximising Kf is given by the LCF cycle, the least by the LUCF
cycle with full unloading. For the latter cycle, the value of Kf is
Kf = K2 + 0.2 (K1 - K2) + 0.87 Kmat

(13)

provided Kf - K2 K1 - K2. Otherwise, it is conservative to assume Kf = K2. Methods for

reducing this conservatism and for treating other cycles may be found in (Refs 77-80).
6.2

Limitations of Methods and Further Work

For a WPS argument to be made according to the methods of Section 6.1, the following
conditions must be met:
(i)

the failure mechanism at the service condition must be by cleavage.

(ii)

The flow properties of the material should increase between the WPS and the service
failure condition; this may be due to a decrease in temperature or due to in-service
hardening.

(iii)

There should be no significant sub-critical crack growth between the WPS and the
service failure condition. The amount of any such crack growth should be much less
than the extent of the residual plastic zone following unloading.

(iv)

The stress intensity factor K1 due to the WPS loading exceeds the fracture toughness Kmat
at the re-load condition.

(v)

Small-scale yielding conditions hold, that is the plastic zone size at the pre-load is much
less than the size of the uncracked ligament and any relevant structural dimensions.

(vi)

The pre-load and re-load should be in the same direction; that is, both tensile or
compressive at the crack tip. A compressive pre-load followed by a tensile re-load may
reduce the apparent fracture toughness.

Further work is clearly required to reduce some of these limitations. Within SINTAP, item (iii) is
being addressed by examining the effect of crack extension by
fatigue crack growth on the WPS behaviour of A533B steel. This work is also examining the inservice hardening requirements, item (ii), by using accelerated embrittlement techniques and
comparing tests with a proof load and subsequent embrittlement with tests with embrittlement
alone. The overall aim is to increase the level of confidence associated with the use of WPS
models when repeat or multiple proof tests are involved (Refs 81, 82) and to allow reduction of
limitation (iii) where sub-critical crack growth occurs (Ref 83).
7.

CONCLUDING REMARKS

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 26 of 46
This review has addressed various aspects of defect assessment which are within the scope of
SINTAP Task 2, failure of cracked components. For each topic, the available information has
been briefly described and areas for further work have been identified. These areas for further
work are summarised in Sections 2.1.4, 2.2.5, 3.2, 4.3, 5.6 and 6.2. Collectively, these sections
outline the future work programme which is to be performed within Task 2. In addition,
commencing in September 1997, the information arising from this work programme will be used
to develop and verify defect assessment procedures.
Overall, it is concluded from this review that the work planned within Task 2 is well aligned to fill
gaps in existing knowledge and to resolve remaining issues. The review will assist in refining the
details of the work packages. If the future work is successful, it will lead to valuable
developments of fracture mechanics methodology and assist in the production of a comprehensive
defect assessment procedure within SINTAP.
8.

ACKNOWLEDGEMENTS
This review has been produced under the Brite-Euram SINTAP project BE95-1426 funded by the
European Commission. In addition to the NEL and AEAT input, contributions to the review have
been provided by:
Dr L Hodulak
(IWM Freiburg) for Section 2.1
Dr M Bergman
(SAQ)
for Section 2
Mr A Bannister (British Steel)
for Section 3
Dr B Brickstad
(SAQ)
for Section 4 and Appendix 1
Dr I Hadley
(TWI)
for Section 5
Dr R Phaal
(TWI)
for Section 6

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 27 of 46

9.

REFERENCES

1.

British Standards Institution, Guidance on methods for assessing the acceptability of flaws in
fusion welded structures, Published Document PD6493:1991; draft revision (1996).

2.

I Milne, R A Ainsworth, A R Dowling and A T Stewart, Assessment of the integrity of structures

containing defects, Int. J. Pressure Vessels & Piping 32, 3-104 (1988); Nuclear Electric
Document R6 - Revision 3, as updated (1996).

3.

G C Sih, Handbook of stress intensity factors for researchers and engineers, Lehigh University,
Bethlehem (1973).

4.

D P Rooke and D J Cartwright, Compendium of stress intensity factors, HMSO, London (1976).

5.

H Tada, P C Paris and G R Irwin, The stress analysis of cracks handbook, second edition, Paris
Production Inc, St Louis (1985).

6.

A Zahoor, Ductile fracture handbook, EPRI, Palo Alto (1991).

7.

Y Murakami et al, Stress intensity factors handbook, Vols 1-3, The Society of Materials Science,
Japan, Pergamon Press (1987-1991).

8.

V Kumar, M D German and C F Shih, An engineering approach for elastic-plastic fracture

analysis, EPRI Report NP-1931, Palo Alto (1981).

9.

M Bergman, B Brickstad, L Dahlberg, F Nilsson and I Sattari-Far, A procedure for safety

assessment of components with cracks - handbook, Swedish Plant Inspectorate Report 91/01
(1991).

10.

ASME Boiler and Pressure Vessel Code, Section XI, Division I, Article A-3000, ASME, New
York (1983).

11.

RCC-MR, Design and construction rules for mechanical components of FBR nuclear islands,
preliminary appendix A16, Section 16.8200 (1992).

12.

C C France, D Green and J K Sharples, New stress intensity factor and crack opening
displacement solutions for through-wall cracks in pipes and cylinders, AEA Technology Report,
AEAT-0643 (1996).

13.

D Green and J Knowles, The treatment of residual stress in fracture assessment of pressure
vessels, ASME PVP Vol. 233, 237-247 (1992).

14.

M Save, J Socol Supel and F Badin, Atlas of limit loads of metal plates and shells, CEC Report
(1988).

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 28 of 46

15.

D Mackenzie and J T Boyle, A method for estimating limit loads by iterative elastic analysis, Isimple examples, Int J Pressure Vessels & Piping 53, 75-94 (1993).

16.

A G Miller, Review of limit loads of structures containing defects, Int J Pressure Vessels &
Piping 32, 197-327 (1988).

17.

R J Dexter and M L Gentilcore, Evaluation of ductile fracture models for ship structural details,
ATLSS Engineering Research Centre, Bethlehem (1996).

18.

R A Ainsworth and Y Lei, Strength mis-match in estimation schemes, Second Symposium on

mis-matching of welds, Luneburg (1996).

19.

EFAM ETM-MM 96, The ETM method for assessing the significance of crack-like defects in
joints with mechanical heterogeneity (strength mismatch), Appendix 1: yield load solutions,
GKSS Report (1996).

20.

G G Chell, R C McClung and D A Russell, Application of failure assessment diagrams to proof

tests analysis, ASME Pressure Vessels and Piping Conference, PVP Volume 304, 475-485
(1995).

21.

P Gilles, A Pellissier Tanon, C Franco and J Vagner, Validity of J estimation in piping

components based on R6/3 option 2 Kr-Lr relationship, Proceeding of ECF10, Berlin, 1347-1358
(1994)

22.

P Gilles and C Bois, Existance and expressions of reference stresses in surface cracked pipes, The
Institute of Materials Second Griffith Conference, Sheffield, 203-214 (1995).

23.

P Gilles and C Bois, Comparisons of finite element J and KJ estimation scheme predictions for a
surface cracked pipe under bending or tension, ASME PVP Conference, Montreal (1996).

24.

P Gilles, J estimation scheme for surface cracked pipings under complex loading, part I theoretical basis, Proceedings of ECF 11, Poitiers (1996).

25.

P Gilles, C Bois and N D Hung, J estimation scheme for surface cracked pipings under complex
loading, part II - complex shaped elbow solutions, Proceedings of ECF 11, Poiters (1996).

26.

A C Bannister and S J Trail, The significance of the yield stress/tensile stress ratio to structural
integrity, British Steel Report (1996).

27.

US Nuclear Regulatory Commission, NUREG 1061, Report of the US Nuclear Regulatory

Commission Piping Review Committee, Volume 3, Evaluation of Potential for Pipe Breaks,
(1984).

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 29 of 46

28.

J K Sharples, T C Chivers and P J Bouchard, New leak-before-break procedures within the R6

framework, PVP - Vol 323, Fatigue and Fracture, Volume 1 ASME (1996).

29.

H M Westergaard, Bearing pressure and cracks, J Applied Mech. 60 (1939).

30.

A G Miller, Elastic crack opening displacements and rotations in through cracks in spheres and
cylinders under membrane and bending loading, Eng. Fract. Mech. 23, 631-648 (1994).

31.

C Wuthrich, Crack opening areas in pressure vessels and pipes, Eng. Frac. Mechanics 18, 10491057 (1983).

32.

V Kumar and M D German, Elastic-plastic analysis of through-wall and surface flaws in

cylinders, EPRI Report NP-5596 (1988).

33.

D B Langston, A reference stress approximation for determining crack opening displacements in

leak-before-break calculations, Nuclear Electric Report TD/SID/REP/0112 (1991).

34.

J K Sharples and P J Bouchard, Assessment of crack opening area for leakage rates, Specialist
meeting on leak-before-break in reactor piping and vessels, Lyon, France, October (1995).

35.

A F George, J I Rich, D H Mitchell and D J F Ewing, DAFTCAT - user guide, Nuclear Electric
Report, TD/SID/REP/0055 (1995).

36.

D M Norris and B Chexal, PICEP: Pipe crack evaluation program (Revision 1), EPRI Report NP3596-SR Revision 1 (1987).

37.

D D Paul, J Ahmad, P M Scott, L F Flanigan and G M Wilkowski, Evaluation and refinement of

leak-rate estimation models: topic report, NUREG/CR-5128 (1994)

38.

G C Gardner and R J Tyrell, The flow resistance of experimental models of naturally occurring
cracks, Proc. Instn. Mech Engrs, 200, 245-250 (1986).

39.

T C Chivers, Assessment of fluid friction factors for use in leak rate calculations, Specialist
meeting on leak before break in reactor piping and vessels, Lyon, France, October (1995).

40.

T C Chivers, Aspects of leak detection, Specialist meeting on leak before break in reactor piping
and vessels, Lyon, France, October (1995).

41.

A P Eperin, et al, Application of the leak-before-break concept to the primary circuit piping of the
Leningrad NPP, Specialist meeting on leak-before-break in Reactor Piping and Vessels, Lyon,
France, October (1995).

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 30 of 46

42.

L Gardner, A P Wightman and D Dobson, Stage 5 wide plate tests in support of leak-before-break
assessment for Chapelcross/Calder Hall reactor pressure vessels, AEAT Report AEA/RS/4390
(1994).

43.

J K Sharples, The assessment of the results of AEA wide plate tests in terms of crack shape
development following through-wall breakthrough, AEAT Report AEA/RS/4453 (1994).

44.

C Betegn and J W Hancock, Two-parameter characterisation of elastic-plastic crack-tip fields,

ASME J Applied Mechanics 58, 104-113 (1991).

45.

A H Sherry, C C France and M R Goldthorpe, Compendium of T-stress solutions for two and
three dimensional cracked geometries, Fatigue Fract. Engng. Mater. Struct. 18, 141-155 (1995).

46.

D J Sanderson, A H Sherry and N P O'Dowd, Compendium of solutions for use with the R6
constraint modified framework, AEA Technology Report AEA-TSD-0981, (1996).

47.

N P O'Dowd and C F Shih, Family of crack tip fields characterised by a triaxiality parameter: Part
I - structure of fields; Part II - fracture applications, J Mech Phys Solids 39, 989-1015 (1991) and
40, 939-963 (1992).

48.

Y J Chao, S Yang and M A Sutton, The fracture of solids characterised by one or two parameters;
theory and practice, J Mech Phys Solids 42, 629-647 (1994).

49.

S Hao, A Cornec and K-H Schwalbe, On the crack driving force and constraint state in a mismatched welded plate under tension, Mis-matching of Welds, ESIS 17 (Eds K-H Schwalbe and M
Kocak), MEP, London, 561-571 (1994).

50.

C Betegon and J W Hancock, Fracture behaviour and the design of materials and structures, In
ECF 8 Vol 2 EMAS, UK, 999-1002 (1990).

51.

J D G Sumpter and A T Forbes, Constraint based analysis of shallow cracks in mild steel, In M G
Dawes (ed.) Shallow Crack Fracture Mechanics, Toughness Tests and Applications, TWI,
Abington (1992).

52.

J D G Sumpter, An experimental investigation of the T-Stress approach, In constraint effect in

fracture, ASTM STP 1171, 492-502 (1993).

53.

J D G Sumpter and J W Hancock, Status of the J plus T stress, In Proceedings of the 10th
European Conference on Fracture (Eds. E M Schwalbe and C Bergers) Vol. 1, 617-626, EMAS,
Warley UK (1994).

54.

M T Kirk, K C Koppenhoefer and C F Shih, Effect of constraint on specimen dimensions needed

to obtain structurally relevant toughness measures, In Constraint Effect in Fracture, ASTM STP
1171, 79-103 (1993).

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 31 of 46

55.

J K Sharples, D J Sanderson, D P G Lidbury, L Gardner, B R Bowdler, G T Melvin and D J

Wright, Effect of biaxial loading on fracture toughness of shallow crack specimens, AEA
Technology Report, AEA-TSD-0035 (1994).

56.

T J Theiss, D K Shum and S T Rolfe, Experimental and analytical investigation of the shallowflaw effect in reactor pressure vessels, NUREG/CR-5886 ORNL/TM-12115 (199?).

57.

D Aurich, H H Erbe, R Helms, H Veith and J Ziebs, The influence of the stress state on fracture
toughness - further results, Structural Mechanics in Reactor Technology (SMIRT), Paper G2/3
(1979).

58.

J W Hancock, W A Reuter and D M Parks, In Constraint Effects in Fracture ASTM STP 1171,
121-140.

59.

A H Sherry, J K Sharples, D J Sanderson, L Gardner and R A Ainsworth, Constraint effects

within the R6 framework: validation by small and large scale fracture test results for A533B-1
steel, PVP - Vol. 304, Fatigue and Fracture Mechanics in Pressure Vessels and Piping, ASME
(1995).

60.

D Miannay, Mecanique de la rupture, Les Editions de Physique, France (1995).

61.

C S Weisner, The 'local approach' to cleavage fracture, Woodhead Publishing Limited,

Cambridge, UK (1995).

62.

A Pineau, Review of fracture micromechanisms and a local approach to predicting crack

resistance, Advances in Fracture Research, 5th International Conference on Fracture, Ed. D
Francois, 2, 553-577 (1981).

63.

A H Sherry and D P G Lidbury, Application of local approach to the measurement of fracture

toughness in the transition temperature range, AEA Technology Report AEA-TRS-4092 (1991).

64.

D P G Lidbury, A H Sherry, B A Bilby, I C Howard, Z H Li and C Eripret, Prediction of the first

Spinning Cylinder test using continuum damage mechanics, Nuclear Engineering and Design,
152, 1-10 (1994).

65.

F M Beremin, A local criterion for cleavage fracture of nuclear pressure vessel steels, Met.Trans.
14A, 2277-2287 (1983).

66.

F M Beremin, Experimental and numerical study of the different stages in ductile rupture:
Application to crack initiation and stable crack growth, Three-dimensional constitutive relations
and ductile fracture, Ed. S Nemat-Nasser, North Holland Publishing Company, 185-205 (1981).

67.

G Rousselier, Ductile fracture models and their potential in local approach of fracture, Nuclear
Engineering and Design, 105, 97-111 (1987).

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 32 of 46
68.

J R Rice and D M Tracey, On the ductile enlargement of voids in triaxial stress fields, J. Mech.
Phys.Solids 17, 201-217 (1969).

69.

A L Gurson, Continuum theory of ductile rupture by void nucleation and growth: Part I - Yield
criteria and flow rules for porous ductile materials, J.Eng. Mat. Tech, 99, 2-15 (1977).

70.

R.W.J. Koers, Fracture mechanics: prediction of cleavage fracture in the brittle to ductile
transition region of a ferritic steel, Shell Report AMER.96.009, Amsterdam (1996).

71.

A G Varias and R W J Koers, A decohesion model for crack growth - application to ductile
fracture, Shell Report AMER.96.011, Amsterdam (1996).

72.

D P G Lidbury, A H Sherry, D W Beardsmore, I C Howard and M A Sheikh, Application of local

approach and damage mechanics to predict cleavage and ductile tearing fracture, Presented at
TAGSI Symposium on Advances in Fracture Mechanics, TWI, Abington, UK (1996).

73.

I MacLennan and J W Hancock, Constraint-based failure assessment diagrams, Int J Pres Ves
Piping, 64, 287-298 (1995).

74.

R A Ainsworth and N P O'Dowd, Constraint in the failure assessment diagram approach for
fracture assessment, ASME J Pres Ves Tech. 117, 260-267 (1995).

75.

R H Dodds, C F Shih and T L Anderson, Continuum and micromechanics treatment of constraint

in fracture, NUREG/CR-5971 (1993).

76.

I Hadley, Constraint differences between test specimens and real structures, Report for Nuclear
Electric, TWI Report 220630/1/96 (1996).

77.

B W Pickles and A Cowan: A review of warm prestressing studies, Int. J. Pres Ves & Piping 14,
93-131, (1983).

78.

D J Smith and S J Garwood. The significance of prior overload on fracture resistance: A critical
review, Int. J. Pres Ves & Piping 41, 255-296 (1990)

79.

G G Chell and J R Haigh: The effect of warm pre-stressing on proof tested pressure vessel, Int. J.
Pres Ves & Piping 23 121-132 (1986)

80.

D J Smith and S J Garwood: Application of theoretical methods to predict overload effects on

fracture toughness of A533B, Int. J. Pres Ves and Piping 41, 333-356 (1990)

81.

K Bell: The effects of multiple preloads on subsequent low temperature fracture of a pressure
vessel steel, TWI Report 7162.01/96/767.3 to be published (1996).

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 33 of 46

82.

I I Fowler and D J Smith: Influence of repeated proof loading on the fracture toughness of ferritic
pressure vessel steels, University of Bristol, Dept. of Mech. Eng. Report (1995).

83.

S J Garwood and K Bell: The effects of sub-critical crack extension on proof loading preliminary
study, TWI Report 7162.02/96/890.2 (1996).

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 34 of 41

TABLE 1 - BASIC STRESS INTENSITY FACTOR SOLUTIONS

NEEDED FOR ASSESSMENT PROCEDURE

Component Geometry
Flat plate, free edges

Flat plate, restrained edges

Cylinder

Defect Geometry

Loading Type

Work Needed

surface

Polynomial crack
face stress

literature survey

embedded

Polynomial crack
face stress

literature survey plus

development

through-thickness

Polynomial crack
face stress

development

surface

Polynomial crack
face stress

literature survey

embedded

Polynomial crack
face stress

literature survey plus

development

through-thickness

Polynomial crack
face stress

development

axial surface

Polynomial crack
face stress

literature survey plus extension

to cover rangeof
radius/thickness ratios and
defect aspect ratios

axial through-thickness

Polynomial crack
face stress

development to allow treatment

of stress gradients

circumferential surface

Polynomial crack
face pressure

literature survey plus extension

to cover range of
radius/thickness ratios and
defect aspect ratios

External moment

literature survey

Polynomial crack
face pressure

development to allow treatment

of stress gradients

External moment

development

circumferential
through-thickness

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 35 of 41

TABLE 2 - BASIC LIMIT LOAD SOLUTIONS NEEDED FOR ASSESSMENT PROCEDURE

Component Geometry
Flat plate, free edges

Flat plate, restrained edges

Cylinder

Defect Geometry

Loading Type

Work Needed

surface

External force and

moment

literature survey plus advice on

use of local and global
solutions

embedded

External force and

moment

through-thickness

external force and

moment

literature survey

surface

external force and

moment

literature survey plus advice on

use of local and global
solutions

embedded

external force and

moment

through-thickness

external force and

moment

literature survey

axial surface

internal pressure

literature survey

axial through-thickness

internal pressure

literature survey

circumferential surface

internal pressure, external

force and moment

literature survey plus advice on

use of local and global
solutions.

circumferential
through-thickness

internal pressure, external

force and moment

literature survey

-"-

-"-

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 36 of 41

TABLE 3 - CRACK OPENING AREA SOLUTIONS RECOMMENDED IN

NEW APPENDIX 9 FOR R6

Geometry

Loading

Elastic or Small Scale Yielding

Elastic Model

Plates
Spheres

Plasticity Model

Membrane

Westergaard (Ref 29)

Dugdale (Ref 31)

TWB

Miller (Ref 30)

Pressure

Wuthrich (Ref 31)

R/t 10, 5

Dugdale (Ref 31)

TWB

Miller (Ref 29)

France et al (Ref 12)

5 R/t 100

Dugdale (Ref 31)

TWB

France et al (Ref 12)

5 R/t 100

Membrane
(Pressure)

France et al (Ref 12)

5 R/t 100

Dugdale (Ref 31)

Langston (Ref 33)

5 R/t 20

Global Bending

France et al (Ref 12)

5 R/t 100

Dugdale (Ref 31)

Langston (Ref 33)

5 R/t 20

Membrane +
Global Bending

Add elastic components

Dugdale (Ref 31)

Kumar (Ref 32)

5 R/t 20

TWB

France et al (Ref 12)

5 R/t 100

Cylinders with Axial Pressure

Cracks

Cylinders with
Circumferential
Cracks

Elastic-Plastic

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 37 of 46

Figure 1

The leak-before-break diagram

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 38 of 46

Figure 2

Recommended recharacterisation in R6 of defects at breakthrough for

predominantly tensile loading

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 39 of 46

Figure 3

Recommended recharacterisation in R6 of defects at breakthrough for

predominantly through-wall bending loading

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 40 of 46

Figure 4

Normalised T-stress solutions of eqn (5) for a semi-elliptical surface crack in a plate
under uniform bending for various ratios of crack depth, a, to plate thickness, t, and
to crack semi-length, c. Solutions are given as a function of angle around the
crack tip, where = 0 corresponds to the deepest point and = /2 corresponds to
the plate surface

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 41 of 46

Figure 5

Normalised Q-stress solutions of eqn (9) for three point bend specimens

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 42 of 46

Figure 6

Critical value of J as a function of T/

y for 3PB and CCT specimens for a mild steel
at -50C (from Ref 52)

Figure 7

Critical value of J as a function of Q for 3PB and CCT specimens for a mild steel at 50C (from Ref 53)

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 43 of 46

Figure 8

Ductile toughness of an A710 steel at various crack extensions as a function of

normalised T-stress (Ref 58)

Figure 9

Crack tip opening displacement of an A710 steel at various crack extensions as a

function of normalised T-stress (Ref 58)

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 44 of 46

Figure 10

Ductile toughness of a A533B-1 steel plate at 20C for a variety of specimens, all
20% side-grooved (Ref 59)

Figure 11

Data of Figure 10 plotted in terms of the constraint parameter T Lr (= T/

y) of eqn
(5) with curve fits to the data

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 45 of 46

Figure 12

The R6 failure assessment diagram

Figure 13

Schematic of Jssy Approach

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page 46 of 46

Figure 14

Typical laboratory warm pre-stress cycles

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page A1 of 10
APPENDIX 1
Review of Crack Shape Development
by
B. Brickstad, SAQ, Sweden

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page A2 of 10

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page A3 of 10

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page A4 of 10

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page A5 of 10

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page A6 of 10

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page A7 of 10

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page A8 of 10

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page A9 of 10

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only
Page A10 of 10

EPD/GEN/REP/0088/96
SINTAP/NE/007
Internal Use Only

DISTRIBUTION
Dr P Neumann
Dr R A Ainsworth
Dr P J Budden
Dr R A W Bradford
Dr D A Miller
Document Centre

Report/20151.st1

(+25)

Barnwood
Barnwood
Barnwood
Barnwood
Barnwood
Barnwood