Structural integrity assessment procedure for Europe – of the SINTAP programme overview

Structural integrity assessment procedure for Europe ±of the SINTAP programme overview

Stephen Webster *, Adam Bannister

Corus Research, Development and Technology, Swinden Technology Centre, Moorgate, Rotherham, South Yorkshire S60 3AR, UK

Received 23 June 2000

Abstract

The structural integrity assessment procedure (SINTAP) project, part-funded by the European Union under the

Brite±Euram scheme, commenced in April 1996 and was completed in April 1999. It has involved a consortium of

seventeen organisations from nine European countries, the aim of the project being to derive a uni®ed structural in-

tegrity evaluation method for use by European industry. Although many such methods did and do exist, most have

con¯icting approaches, unspeci®ed levels of empiricism or do not fully re¯ect the performance of modern materials or

the current state of knowledge.

The SINTAP project covered both modelling and experimental work and a large element of the project was concerned

with the transfer of knowledge and data between industries and scienti®c organisations together with its compilation and

interpretation to provide the required solutions. The culmination of this work is a procedure which is applicable to a wide

cross-section of users because of its ability to o�er routes of varying complexity, re¯ecting data quality and the preference

of the ®nal interpretation required by the user. Ó 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Fracture; Structural integrity; Plastic collapse; Strength; Toughness

1. Introduction

Structural integrity assessment procedures are the techniques used to assess the ®tness-for-purpose ofcritical components and welded structures. Such approaches can be used at the design stage to provideassurance for new structures, at the fabrication phase to ensure the integrity in the construction and at theoperational phase to provide assurance throughout the life of the structure. Used correctly, they canprevent over-design and unnecessary inspection and provide the tools to enable a balance between safetyand economy to be achieved.

Over the years a number of fracture mechanics-based assessment procedures have been developed [1±10]within Europe, the USA and Japan and their use has spread both in terms of the industries who use themand their analysis capability. However, no one technique was capable of providing all of the answers andthe accuracy of some were, at the best, unquanti®ed. Also, recent developments and capabilities were not

Engineering Fracture Mechanics 67 (2000) 481±514

www.elsevier.com/locate/engfracmech

* Corresponding author. Tel.: +44-1709-820166/825285/825552; fax: +44-1709-825337/825532.

E-mail addresses: [email protected] (S. Webster), [email protected] (A. Bannister).

0013-7944/00/$ - see front matter Ó 2000 Elsevier Science Ltd. All rights reserved.

PII: S00 1 3-7 9 44 (0 0 )0 00 7 0- 9

included, even in the latest code developments. In addition, recent years have seen the adoption by industryof a new range of higher strength steels which has presented di�erent analysis problems such as high yield totensile strength ratio, and has demonstrated a lack of knowledge of the behaviour of joints which have amismatch in strength, where the weld is stronger or weaker than the parent material. Substantial amountsof work have been undertaken on the former subject [11,12] but it still remained an area for which practicalguidance and methods were not available.

The SINTAP project has been concerned with those aspects which relate to the areas of fracture andplastic collapse and the interaction between the two. SINTAP was a multidisciplinary collaborative project,part-funded by the European Union, with the aim of devising a uni®ed procedure, o�ering a range ofassessment routes giving it maximum applicability, from the smallest SME to major industrial users.

2. Philosophy and approach of the project

In view of the comprehensive nature of the project and the necessity to consider carefully the require-ments of the end user, a wide-ranging consortium was established for the project. This was comprised of amaterial supplier (British Steel, now part of the Corus Group). An electricity generator (British Energy,formerly Nuclear Electric Ltd., with AEA), an oil and gas supplier (Shell), a chemical processor (EXXON),safety assessors (Health & Safety Executive and SAQ Inspection Ltd.), Research Institutes (GKSS, Fra-unhofer IWM, lnstitut de Soudure, TWI, VTT, JRC), two universities (Cantabria and Gent), a softwaredeveloper (Marine Computation Services) and a consultancy (Integrity Management Services).

The project was coordinated by Swinden Technology Centre (UK) (Corus Research, Development andTechnology) with the ®ve principal tasks being led by Task Leaders as summarised below.

Task 1: Mismatch ± leader: GKSSTo quantify the behaviour of strength mismatched welded joints and to provide recommendations for

their treatment in a procedure.Task 2: Failure of cracked components ± leader: British EnergyTo extend the understanding of the behaviour of cracked components in the speci®c areas of constraint,

yield/tensile ratio, prior overload, leak-before-break, stress intensity factors (SIFs) and limit load solutions.Task 3: Optimised treatment of data ± leader: VTTTo provide an industrially applicable method for a reliability-based defect assessment procedure.Task 4: Secondary stress ± leader: IdSThe determination and validation of the most appropriate method of accounting for residual stress,

including a compendium of residual stress pro®les.Task 5: Procedure development ± leader: CorusDevelopment and validation of the procedure.Each task comprised a number of sub-tasks, (see Table 1), and was structured into three basic steps: As a

®rst step a comprehensive collation of existing data, procedures and codes was completed and then ex-perimental work to cover omissions and to validate the approaches and assumptions relevant to the par-ticular study area was carried out. The best practice approach within each task was collated within task 5with consideration of the needs of the practising engineer in order to formulate the procedure itself.

3. Procedure scope and novel features

A key aspect of the ®nal procedure is that a range of assessment routes can be followed which re¯ect thequality of the input data: The most sophisticated and accurate levels require accordingly high quality input

482 S. Webster, A. Bannister / Engineering Fracture Mechanics 67 (2000) 481±514

data, whilst conversely, assessments can still be carried out with knowledge of only basic parameters butthis is penalised by a higher degree of conservatism; The underlying principles of the SINTAP method are

· a hierarchical structure based on the quality of available data inputs;· decreasing conservatism with increasing data quality;· detailed guidance on determination of characteristic input values such as fracture toughness;· the choice of representation of results in terms of a failure assessment diagram (FAD) or crack driving

force (CDF);· speci®c methods incorporating the e�ect of weld strength mismatch;· guidance on dealing with situations of low constraint and, for components containing ¯uids, leak before

break analysis;· compendia of solutions for SIFs, limit load solutions and weld residual stress pro®les.

The procedure provides advice on the basic inputs and calculations needed. The aspects covered aretensile properties, fracture toughness data, ¯aw characterisation, and the treatment of primary and sec-ondary stresses. While some of this information is standard, there is novel information arising from thework performed during SINTAP. These novel developments may be summarised as follows:

(1) Assessments have been made of tensile data so that, in the absence of detailed data, estimates ofstrain hardening properties, L�uders strain and yield to tensile strength ratios can be made from limitedinformation.

(2) Statistical treatments of fracture toughness data have been given taking account of the number ofspecimens regardless of failure mode. In the cleavage regime for ferritic steels, the so-called Master Curve[13] approach has been further developed to provide improved estimates of lower bound fracture tough-ness. In the absence of toughness data, improved correlations are given to enable fracture toughness data tobe estimated from Charpy impact energy.

(3) The ¯aw characterisation rules in the SINTAP procedure are essentially those in the British Stan-dards document BS7910 [10]. The SINTAP procedure, however, goes beyond this in providing guidance onthe reliability of non-destructive examination techniques.

(4) The basic approach in the SINTAP procedure for treating combined primary and secondary stressesis that in R6 [1]. However, an alternative, new approach has been developed in which the e�ect of thesecondary stresses is described by the factor V. Values of V > 1 correspond to cases where plasticity leads tovalues of J due to the secondary stress which are increased from those assessed elastically. Conversely,V < 1 corresponds to plastic relaxation of secondary stresses such that V � 0 would indicate that thesecondary stresses could be neglected.

Table 1

Sub-tasks in project

Task 1. Mismatch Task 2. Cracked

components

Task 3. Optimised

data treatment

Task 4. Secondary

stress

Task 5. Procedure

development

1.1 Review 2.1 Review 3.1 Review 4.1 Review 5.1 Review

1.2 Bi-materials 2.2 Constraint 3.2 Toughness 4.2 Collate pro®les 5.2 Procedure

1.3 Multipass weld 2.3 YS/UTS ratio 3.3 Charpy correla-

tions

4.3 Experiment and

modelling

5.3 Software

1.4 Modelling 2.4 Prior overload 3.4 NDE guidance 4.4 Pro®les library 5.4 Validation

1.5 Procedure 2.5 Leak-before-

break

3.5 Probabilistic 4.5 Procedure 5.5 Documentation

2.6 SIF and LL solu-

tions

3.6 Procedure

2.7 Procedure

S. Webster, A. Bannister / Engineering Fracture Mechanics 67 (2000) 481±514 483

The third chapter of the procedure provides guidance on selection of the level of analysis and includescompendia of SIF solutions, limit load solutions and residual stress pro®les. These compendia have built onthose available at the start of the project but have added a number of new features. For example

(1) New SIF factor solutions have been developed for defects in cylinders for complex primary orsecondary stress ®elds.

(2) A large number of mismatch limit load solutions have been provided for plates and cylinders whichleads to reduced conservatism compared to classical methods where mismatched welds are treated as beingcomposed entirely of the lowest strength material.

(3) The compendium of weld residual stress pro®les covers a range of geometries with surface andthrough-thickness residual stresses being given for longitudinal and transverse orientations. Residualstresses can be determined from knowledge of the material and weld heat input when these are known ormore conservatively from bounding stress ®elds. Advice on the e�ects of post-weld heat treatment is alsoincluded.

The ®nal chapter of the procedure covers alternatives and additions to standard methods. The chapterdescribes the following extensions to the basic methodology:

(1) Ductile tearing analysis methods are given which allow for the increase in fracture toughness withductile crack extension beyond initiation. These methods can be applied regardless of the level of knowl-edge of tensile data but do, of course, require additional information about the ductile tearing toughness.

(2) Reliability methods are presented along with associated software, PROSINTAP, developed during thecourse of the project. This uses Monte Carlo simulation and approximate ®rst-order reliability routines.Associated partial safety factors are given for a range of target reliabilities, in a similar manner to BS7910 [10].

(3) Fracture toughness data are generally collected from deeply cracked bend specimens, whereasshallow defects under membrane loading are often assessed in components. The former is often referred toas high constraint whereas the latter is low constraint. Use of high constraint data in structural assessmentscan be over-conservative and so the SINTAP procedure describes a method for calculating the level ofconstraint in terms of the elastic T-stress or the hydrostatic Q-stress ahead of a crack tip and then modi®esthe FAD or CDF methods to allow for the e�ects. A range of solutions for Q for surface defects [14] hasbeen developed during the course of the SINTAP project to enable the constraint approach to be applied inpractice.

(4) A procedure for making a leak-before-break case is outlined. This is based on an approach within R6[1], but includes new guidance on the shape development of part-penetrating ¯aws.

(5) Guidance is given on the e�ects of prior overload on the mechanical relaxation of residual stress andthe e�ects of warm pre-stressing on fracture toughness.

Further details on a number of these aspects are given later in this paper.

4. Compatibility of failure assessment diagram and crack driving force approaches

Initially, the current situation concerning existing procedures and their pending development was as-sessed [15]. The two principal methods for analysis which are used in these procedures are the FAD and theCDF, Fig. 1. The former method is used in approaches, such as R6 [1], BS 7910 [10], SAQ [3], EXXON [4],INSTA [5], MPC [6] and API 579 [7]; the CDF approach is favoured in ETM [8] and GE-EPRI [9] pro-cedures.

The basis of both approaches is that failure is avoided so long as the structure is not loaded beyond itsmaximum load bearing capacity de®ned using both fracture mechanics criteria and plastic-limit analysis.The fracture mechanics analysis involves comparison of the loading on the crack tip (often called the crack


tip driving force) with the ability of the material to resist fracture (de®ned by the materialÕs fracturetoughness or fracture resistance). The crack tip loading must, in most cases, be evaluated using elastic±plastic concepts and is dependent on the structure, the crack size and shape, the materialÕs tensile propertiesand the loading. In the FAD approach, both the comparison of the crack tip driving force with the ma-terialÕs fracture toughness and with the plastic load limit analysis is performed at the same time. In the CDFapproach, the crack driving force is plotted and compared directly with the materialÕs fracture toughnessand a separate analysis carried out for the plastic limit. While both the FAD and CDF approaches arebased on elastic±plastic concepts, their application is simpli®ed by the use of only elastic parameters.

A critical review of these procedures was carried out for 15 key areas. The advantages and disadvantagesof each procedure were assessed and analytical comparisons carried out. The review demonstrated [16,17]that there is relatively little di�erence between results obtained using the two procedures. The major dis-crepancies arise primarily because of the limit load or SIF solutions rather than any fundamental di�er-ences in concepts. Furthermore, although FAD and CDF routes represent two di�erent calculationmethodologies, the underlying principles remain the same. Demonstrating compatibility between the twoapproaches and devising the means to allow interpretation using either method were therefore seen as keyissues within the SINTAP programme.

Despite the di�erences that exist between these two methodologies both share the same concept of acomparison between the applied stress and the materialÕs resistance on a local scale. It is only in their

Fig. 1. FAD and CDF approaches for fracture initiation and ductile tearing analyses.


development, through di�erent simpli®cations, that they diverge. In the following description of the de-velopment of fully compatible FAD and CDF methods [18,19], the term Ôyield stressÕ is used to infer eitherthe yield stress in the case of discontinuous yielding, or 0.2% proof stress for continuous yielding. Di�er-entiation between these two parameters is only made where necessary. In the CDF method, calculations aremade of an applied parameter, such as the J-integral or crack opening displacement which characterises thestate of stress and strain ahead of the tip of a crack in a component. In a Level 1 assessment, J is estimatedas

J � Je�f �Lr��ÿ2; �1�

where Je is the elastic value of the J integral which can be deduced from the SIF KI as

Je � K2I =E0; �2�

where E0 is YoungÕs modulus, E, in plane stress, and E=�1ÿ m2� in plane strain, where m is PoissonÕs ratio.The function f �Lr� in Eq. (1) is de®ned in terms of the load ratio

Lr � F =FY; �3�where F is the applied load and FY is the limit load de®ned from the yield strength, as

f �Lr� � �1� 0:5L2r �ÿ1=2�0:3� 0:7exp�ÿ0:6L6

r ��: �4�For Lr > 1, Eq. (4) can be re®ned for materials described by power-law plasticity. For higher levels, thefunction can be re®ned in terms of strength mismatch and/or the shape of the stress±strain curve.

It can be seen that J can be estimated from Eqs. (1)±(4) provided the applied load F is known, a SIFsolution is available (note K1 is proportional to F and depends on geometry and ¯aw size) and a limit loadsolution is available (note FY is proportional to yield stress (YS) and depends on geometry and ¯aw size). Inthe CDF method, the applied driving force in terms of J or CTOD is determined and fracture concededwhen the relevant material toughness (Jmat or CTODmat) is exceeded.

The CDF method described above is based on approaches within the ETM [8] method developed atGKSS in Germany. In particular, the functions f �Lr� within the SINTAP approach for power-law mate-rials are based directly on the corresponding equations in ETM.

In the FAD method, two parameters are calculated, the load ratio, Lr, already de®ned by Eq. (3) and theratio Kr de®ned by

Kr � KI=Kmat; �5�or

Kr ��Je=J

p; �6�

where Eq. (6) provides the consistency between FAD and CDF expressions.

Once these two parameters have been calculated, fracture is avoided if the point �Lr; Kr� is within a regionde®ned on a failure assessment diagram as depicted in Fig. 1. The failure avoidance region is given by thefailure assessment curve

Kr � f �Lr�: �7�and a cut-o�

Lr � Lmaxr : �8�

Manipulation of Eqs. (1)±(7) shows that the condition Kr6 f �Lr� is equivalent to J 6 Jmat so that the CDFand FAD representations within the SINTAP procedure are fully compatible. Criterion (8) is imposed to


provide a plastic collapse limit with Lmaxr dependent on the strain hardening characteristics of the material.

This limit is imposed in both CDF and FAD methods.

5. Analysis levels in the structural integrity assessment procedure

A range of analysis options is o�ered which enables advantage to be taken from increasing data qualityand which re¯ects the variation in user knowledge and experience. Each level is less conservative than theone before, such that ÔpenaltiesÕ and ÔrewardsÕ accrue from the use of poor and high quality data respec-tively. This structure means that an unacceptable result at any level can become acceptable at a higher level.Consequently, the user needs only perform the work necessary to reach an acceptable result and need notinvest in unnecessarily complicated tests or analysis. The levels available and the options provided withineach level are described in Chapter I of the SINTAP procedure and are summarised in Table 2.

There are three standardised levels and three advanced levels, including the special case of a leak-before-break analysis for pressurised systems. The di�erent standardised levels produce di�erent expressions forf �Lr� which de®ne the FAD or CDF to be used in the analysis. The level of analysis is characterised mainlyby the detail of the materialÕs tensile data used.

The equations used to generate f �Lr� for Levels 1 and 2 are based upon conservative estimates of thematerialsÕ tensile properties for situations when complete stress±strain curves are not known. More accurateand less conservative results can be obtained by using the complete stress±strain curve, and this option is

Table 2

Selection of analysis levels from tensile data

Level Data needed When to use

Default level

Default Yield or proof strength When no other tensile data

available

Standard levels

1. Basic Yield or proof strength: ultimate tensile

strength

For quickest result. Mismatch

in properties less than 10%

2. Mismatch Yield or proof strength: ultimate tensile

strength Mismatch limit loads

Allows for mismatch in yield

strengths of weld and base

material. Use when mismatch is

greater than 10% of yield or

proof strength

3. SS (Stress±strain de®ned) Full stress±strain curves More accurate and less conser-

vative than Levels I and 2.

Weld mismatch option included

Advanced levels

4. Constraint allowance Estimates of fracture toughness for crack tip

constraint conditions relevant to those of

cracked structure

Allows for loss of constraint in

thin sections or predominantly

tensile loadings

5. J-Integral analysis Needs numerical cracked body analysis

6. Special case: leak before break analysis Piping and pressure vessel

components


given in Level 3 as the SS (stress±strain) level. In this case, every detail of the stress±strain curve can beproperly represented and where weldment mismatch e�ects are important these can also be allowed for.

6. Treatment of toughness data in the structural integrity assessment procedure

6.1. Toughness options

A subdivision of a level arises from the details of fracture toughness data that are used. There are twooptions for this, one characterising the initiation of cracking (whether by ductile or brittle mechanisms), theother characterising crack growth by ductile tearing. The value of fracture toughness to be used in theSINTAP procedure is termed the characteristic value. Table 3 gives guidance on the selection of the optionsavailable for toughness data.

The basic level of analysis, Level 1, is the minimum recommended level. This requires measures of thematerialÕs yield or proof strength and its tensile strength, and a value of fracture toughness, Kmat, obtainedfrom at least three fracture toughness test results which characterise the initiation of brittle fracture or theinitiation of ductile tearing. For situations where data of this quality cannot be obtained, there is a defaultlevel of analysis, which is based only on the materialÕs yield or proof strength and its Charpy impact energy.

It is currently common for the treatment of fracture toughness data, to be used in fracture mechanicsanalysis, to vary depending on the type of data (KIC, J or CTOD (d)) that are available. This complicatesstructural integrity assessment and makes it di�cult to apply any single, uni®ed procedure. In reality,fracture toughness data may not exist at all, or cannot be easily obtained due to lack of material or theimpracticability of removing material from the actual structure. In these cases, Charpy data may be all thatare available and it is necessary to use a reliable correlation between Charpy impact energy and fracturetoughness.

Thus, a methodology capable of incorporating all these factors on a uni®ed basis, could serve as a key toa procedure tailored towards the practical user. In the ÔSINTAPÕ framework, a fracture toughness esti-mation methodology [20,21] has been developed for such a purpose. The methodology is written in the formof a procedure in which one material-speci®c toughness parameter, Kmat, together with its probabilitydensity distribution P{Kmat} is de®ned, irrespective of the type of the original data. For assessment againstbrittle fracture, the SINTAP evaluation procedure is based upon the maximum likelihood concept (MML)[22] that uses a ÔMaster CurveÕ method [13,23] to describe the temperature dependence of fracture tough-ness. As a ®nal result, a conservative estimate of the mean fracture toughness (and the distribution) isobtained.

Table 3

Selection and treatment of toughness data

Parameters required Fracture mode characterised Input obtained

Initiation option

Fracture toughness at initiation of

cracking: from three or more specimens

Onset of brittle fracture: or onset of

ductile fracture

Single characteristic value of toughness

either initiation or maximum load with

restrictions

Tearing option

Fracture toughness as a function of

ductile tearing: from three or more

specimens

Resistance curve Characteristic values as function of

ductile crack growth

Default level

Charpy energies All modes Correlated characteristic values


The present methodology can be as easily applied to indirect (i.e. Charpy) data, as to actual fracturetoughness data and is suitable for the treatment of data at both single and di�erent temperatures. Theprocedure enables a reliable fracture toughness estimate to be obtained for various forms of data setscontaining results from both homogeneous and inhomogeneous material. Thus, it works, not only for basematerials but, in the case of welded joints, for weld metals and heat-a�ected zones. It is initially structuredin a way that the more su�cient or accurate the original data, the more the user will be rewarded in theprobabilistic fracture mechanics assessment.

6.2. Estimation of fracture toughness from Charpy impact energy

When fracture toughness data are not available, or cannot be easily obtained, it is necessary to base theestimate of fracture toughness on the Charpy impact energy. No single correlation can be applied to allparts of the toughness transition curve hence it is necessary to use various correlations. In the SINTAPprocedure, the following options are available, as described in Ref. [24].

1. A lower bound correlation for brittle (lower shelf) behaviour.2. A statistical method for the transition regime (The ÔMaster CurveÕ).3. A lower bound correlation for the ductile (upper shelf) behaviour.

For materials operating in the brittle regime, the determination of fracture toughness from Charpy datafollows the concept of the ÔMaster CurveÕ. This concept is based on the correlation between the Charpy 28Jtemperature and the temperature for Kmat� 100 MPa

pm. The relationship is then modi®ed to account for

the required failure probability, thickness e�ect and the shape of the fracture toughness transition curve.The transition curve for fracture toughness in the transition regime, using a 95% con®dence limit, can bede®ned as

Kmat � 20f11� 77exp�0:019�T ÿ T28J ÿ 3°C��g�25=B�1=4f ln�1=�1ÿ Pf ��g1=4; �9�

where T is the design temperature (°C), T28J , the 28J (or 27J) Charpy transition temperature (°C), B, thespecimen thickness or ¯aw width (2c) (mm), Pf , the probability of failure and Standard deviation � 13°C.

At a Charpy impact energy of 28J, the use of Eq. (9) with the lower ®fth percentile of fracture toughnessand a 90% con®dence level leads to a simple equation which represents a conservative lower bound estimateof fracture toughness:

Kmat25 � 12p

Cv; �10�

where Kmat25 is the estimated K-based fracture toughness of the material in MPap

m for a thickness or ¯awwidth (2c) of 25 mm and Cv the Charpy impact energy (V-notch) in J.

For the cases where Charpy data corresponding to an energy level di�erent to 28J are available the use oflimited extrapolation is permitted. Where a material is potentially operating in a high loading rate regime, acorrection can be made based on the Zener±Holloman strain rate dependence of yield strength. Conse-quently, the method provides a powerful tool for fracture toughness estimation in materials selection,design and structural integrity analyses and is fully coherent with the approach used in the fractureavoidance clauses of Eurocode 3 [25].

There is, at present, no equivalent of the ÔMaster CurveÕ for upper shelf behaviour, consequently, adeterministic approach is used for the ductile regime. Within the SINTAP procedure, a ductile regimecorrelation is presented which enables estimation of the R-curve behaviour from knowledge of the uppershelf Charpy impact energy.

The decision tree for these methodologies is given in Fig. 2.


6.3. Treatment of actual fracture toughness data

For reasons of cost and/or restrictions on the amount of material available, fracture toughness is oftenmeasured using a limited number of specimens. The determination of an appropriate statistical distributionfrom limited data can be arbitrary and unreliable. Consequently, the selection of a characteristic value fromsuch a distribution for use in a ¯aw assessment procedure may be unconservative, or at least give incon-sistent assessments. However, some of the inconsistency can be removed by assuming the fracture process isgoverned by a weak link process which follows a three parameter Weibull distribution. For ferritic steels,this is given by

P Kc� � � 1ÿ exp

ÿ Kc ÿ 20

K0 ÿ 20

� �ÿ4!

�11�

where P �Kc� is the cumulative probability of fracture toughness, Kc (MPap

m), K0, scale parameter (the63rd percentile of the distribution), 20, the shift parameter in the Weibull distribution (MPa

pm), and 4, the

value of the shape parameter in the Weibull distribution for small scale yielding.Eq. (11) may be re-written as follows to provide an estimate of Kc with a given probability level once K0

is known:

Kc � 20� K0� ÿ 20�fÿIn 1� ÿ P Kc� ��g0:25 �12�Thus, the distribution ®tting procedure involves ®nding the optimum value of the K0 for a particular set ofdata. Unfortunately, the test results may be biased when specimens are extracted from inhomogeneousmaterials, such as weld metals and heat a�ected zones (HAZs). Since every specimen in a series of tests is

Fig. 2. Flowchart for the treatment of Charpy data.


unlikely to sample local brittle zones (LBZs) that may be present, the fracture toughness distribution will bebiased to the higher toughness regions that are present. Thus, the ®tted distribution could result in non-conservative assessments being made. To avoid this potential problem, the maximum likelihood estimationprocedure [22] (MML procedure) is used in the SINTAP ¯aw assessment method. The procedure is in-tended to provide conservative, but realistic, estimates of fracture toughness.

The MML procedure proceeds via a series of stages. The ®rst stage is to check that all the data meet theacceptance criteria of the relevant testing standard. The fracture toughness values are converted into SIFswith MPa

pm units. Where CTOD data are available, these may be converted into equivalent Kc values:

Next, the specimen capacity limit (Kc limit), is determined from

Kc limit � Eb0YS=30� �0:5; �13�where YS is the yield or proof stress, b0 (m), the initial ligament below the notch (W ÿ a0) in the testspecimen.

Eq. (13) ensures that fracture occurs under small-scale yielding conditions. Results from specimens thatexceed this limit are censored, the censoring parameter, d, is set at 0 and the fracture toughness set at Kc limit.Specimens that do not fracture are also censored (d � 0). Results from other specimens are not censoredand d is set at 1 for each specimen.

Next, the fracture toughness values are corrected to a reference thickness of 25 mm, Kc25:The values of Kc25 and their associated censoring parameters are then used to make the ®rst estimate K0.

This is Step 1 in the MML procedure (referred to as Ônormal MML estimationÕ) and K0 is obtained from

K0 � 20�PN

i�l Kc25i ÿ 20� �4PNi�l di

" #; �14�

where N is the number of results and i is the ith result.Note that the maximum likelihood procedure uses all the test results and that K0 is biased towards the

uncensored data (d � 1).If the material tested were homogeneous, the estimate of the K0 from Eq. (14) could be inserted into Eq.

(12) to provide a value of Kc, for a speci®ed probability level. For the value to be used in a defect assessmentprocedure, it is necessary to correct this fracture toughness for the appropriate thickness. However, whenthe data are from inhomogeneous materials, further censoring is required.

Step 2 in the MML procedure, (referred to as Ôlower tail MML estimationÕ) involves censoring all dataabove the 50th percentile of the distribution, i.e. setting d � 0 for all values above the 50th percentile. Thisensures that the estimate of K0 is biased towards the lower tail of the toughness distribution so as to includeresults from specimens containing LBZs. Results from specimens which do not contain LBZs and that arelikely to give high fracture toughness values tend to be excluded by Step 2. The censored values are assignedthe median value of toughness.

Kc25 � 20� K0� ÿ 20�0:91: �15�After censoring, K0 is re-estimated using Eq. (14). However, since both Eqs. (14) and (15) contain K0, theprocedure is iterative with K0 and Kc25 being continually adjusted until a consistent minimum K0 is ob-tained.

The ®nal step, Step 3 (referred to as Ôminimum value estimationÕ), requires an estimate of the K0 using theminimum fracture toughness value in the data set. The values of K0 from each of the three steps are nowcompared. The K0 which describes the fracture toughness distribution is the lowest determined from thethree steps, except when K0 from Step 3 is not less than 90% of the lower of K0 from Steps 1 or 2. Where K0

from Step 3 is less than 90% of the lower of Steps 1 and 2, it would be conservative to use K0 from Step 3.However, since the procedure is highlighting an outlying point, a judgement has to be made as to its


signi®cance. If the data set is large and the ®t to the assumed distribution is good, then the result from Step3 can be treated as representing an anomaly and can be ignored. It is then reasonable to assume that thelower of Steps 1 or 2 is describing the fracture toughness distribution. However, if the data set is small, itwould be unreasonable and unconservative to ignore Step 3. If the result from Step 3 is considered to beunsatisfactory then further testing should be conducted to better de®ne the lower tail of the fracturetoughness distribution. A validation exercise on parent material, weld metal and HAZ is described in Ref.[26].

The procedure described above can also, in principle, be applied to data in the fracture toughnesstransition regime. A ¯ow chart is given in Fig. 3.

Fig. 3. Flowchart for treatment of fracture toughness data.


7. Derivation of structural integrity assessment procedure

7.1. Development

The SINTAP procedure allows a structural assessment analysis to be performed provided that thefollowing are known: loading conditions, the corresponding applied KI or J values, materialÕs toughness,whether this knowledge is direct or is obtained through the Charpy tests, and a de®nition of failure lines inthe case of the FAD. These lines must be hierarchised in an order depending on the application of moreaccurate knowledge of the mechanical behaviour of the material of the component. At each level, amaximum extension of the failure line is de®ned establishing an Lmax

r value where Kr value drops to zero.In order to achieve the hierarchy of FAD relationships between tensile parameters were derived from the

data available within the SINTAP consortium on a wide range of materials. The following were addressed:

1. Estimation of yield stress/UTS ratio (Y =T ) from knowledge of yield strength (for possible extension of Lr

cut-o� at Level 0, continuous yielding FAD).2. Estimation of strain hardening exponent (n) from Y =T (for use at Levels 1 and 2).3. Guidance on when to assume the presence of a yield (L�uders) plateau and, in such cases, what expression

should be used to estimate its length.4. Implications for structural assessments if upper yield stress only has been reported.5. Methods for ensuring consistency between elastic and plastic components of the FAD.

While the derivation of the SINTAP procedure for homogeneous materials was concentrated on theFAD approach, compatibility with CDF approaches for all levels was also demonstrated and is implicitfrom the relationship between the two approaches.

For use of the Default Level of the procedure, only YS needs to be known. However, the relationshipbetween YS and Y =T allows the UTS to be estimated thus enabling extension of Lr to values greater than 1in some cases. A relationship developed for the estimation of Y =T from YS was developed as an upper-bound, conservative ®t to a large data set for a range of materials, Fig. 4.

Fig. 4. Relationship between yield strength and Y =T ratio [27].


At Levels 1 and 2 of the procedure, the yield strength and ultimate tensile strength must be known, thestrain hardening exponent for the material is determined from ratio for subsequent determination of theFAD at Lr > 1. The n value is determined from the true stress±strain curve, as is conventional, and can berepresented as a simple but conservative linear function of Y =T as shown in Fig. 5.

Within the SINTAP procedure, selection of appropriate FAD/CDF at Levels 1 and 2 is dependent onthe assumption or otherwise of the occurrence of a yield (L�uders) plateau. In addition, a method for es-timation of the length of the plateau, should one be assumed to be present, was also required. It is rec-ognised that this approach is a Ôbest estimateÕ and that there are other factors which also dictate whether ayield plateau should be assumed (such as loading rate and specimen design). When a plateau is assumed, itslength is estimated from an expression which represents a mean ®t to plateau length as a function of yieldstrength.

Yield strength values for steels showing a yield point can be quoted as upper yield strength (UYS) orlower yield strength (LYS) whilst for steels showing a continuous stress±strain curve, yield strength isquoted as a proof stress (usually 0.2%, but sometimes 0.5%, depending on material speci®cation). For thecase of discontinuous yielding, the yield strength values quoted in steel test certi®cates are UYS values,whereas for structural applications the LYS must be used to ensure conservatism. The 50th percentile valueof a cumulative plot of LYS/UYS values for over 200 steels was therefore recommended as a suitable factorfor estimation of LYS from UYS, this value being 0.95.

Finally, at all levels of FAD and CDF, consistency between the elastic and plastic portions was achievedby introduction of a factor based on yield strength and YoungÕs modulus. This ensures that as the value ofstrain hardening exponent, n, estimated from Y =T ratio, decreases, the departure of the plastic portion ofthe FAD from the elastic portion is consistent. Furthermore, due to the incorporation of n in the FADestimation scheme, the higher Y =T values give lower CDF n values which is re¯ected in the reduced area ofthe safe region in the plastic zone of the FAD.

Fig. 6 illustrates the standard levels available for both homogeneous materials and strength mismatched(mm) welded joints.

Fig. 5. Relationship between n and Y =T ratio [27].


7.2. Equations for failure assessment diagrams [19]

7.2.1. Default levelThe default level is the most conservative limit of FAD and CDF within the SINTAP procedure. For

materials with continuous hardening, for which the expression is

Kr � 1

�� L2

r

2

�ÿ1=2

0:3� � 0:7exp

ÿÿ lL6r

��; �16�

and for materials with a yield plateau at yield strength YS, the expression is given by

Kr � 1

�� L2

r

2

�ÿ1=2

: �17�

The physical sense of the term 1� L2r=2

� �ÿ1=2, present in the above equations, is that it represents the failure

line corresponding to a material of perfectly linear-elastic behaviour incorporating the correction due to thecontained yielding localised ahead of the crack. The second factor of Eq. (16) corresponds to the loss of the

Fig. 6. Flowchart for the treatment of tensile data ± standard levels.


safety zone established by a material in the elastic zone when it leaves the linear-elastic behaviour, which ismore marked in the region of Lr � 1 (or the yield strength).

It is suggested that the most basic analysis should be carried out at the standard level (Level 1) withknown values of the yield strength and ultimate tensile strength. Both for materials of continuous strainhardening and for those with yield plateau, the failure line is given up to Lr � 1 through the correspondingdefault level (Level 0), Eqs. (16) and (17), respectively.

7.2.2. Level 1For Level 1, the l value in Eq. (16) is the minimum between 0.6 (its maximum) and that given by

l � 0:001E

YS; �18�

which is equivalent to the hyperbola

lYS

E� 103 � 1: �19�

An equation is proposed for the failure line in the range Lr P 1 given by

Kr � Kr Lr�1� �L�Nÿ1�=�2N�r �20�

in which n is estimated as a parameter de®ned through the expression (see Fig. 6)

n � 0:3 1

�ÿ YS

UTS

�; �21�

which is obtained as the boundary limit which errs on the safe side in estimating the corresponding failureline, when compared with the real failure line of the material, after studying a range of di�erent materials,mostly steels.

For materials with a yield plateau, the failure line presents a vertical step for Lr � 1, which gives aminimum value for Kr�Lr � 1� given by

min Kr Lr�1� � � kmax

�� 1

2kmax

�ÿ1=2

: �22�

In this equation, kmax is

kmax � 1� EYS

De: �23�

De being the strain extension of the yield plateau, estimated by the relationship

De � 0:0375 1

�ÿ YS

1000

�; �24�

where YS is measured in MPa. The line de®ned in (Eq. (24)) has been deduced after a study made ondi�erent materials correlating the yield plateau extension with the yield strength.

The failure line for these materials with yield plateau is given, for Lr > 1, by the equation obtained fromEqs. (20) and (22):

Kr � kmax

�� 1

2kmax

�ÿ1=2

L�Nÿ1�=�2N�r : �25�


7.2.3. Level 2The SINTAP procedure Level 2, is similar to the ®rst one but focuses on mismatch welded joints, in which

the yield strengths of the weld and the base materials di�er by more than 10%. The increased complexity ofthis formulation does not a�ect the hierarchy criteria presented above. This level is outlined in Section 8.

7.2.4. Level 3Failure lines at this level are obtained from the full knowledge of the tensile characterisation of the

component material, that is, its full true stress±strain curve. The de®nition of the failure line is given by

Kr � Eeref

LrYS

�� L3

r YS

2Eeref

�ÿ1=2

� Eeref

rref

�� L2

r rref

2Eeref

�ÿ1=2

: �26�

Treatment of tensile data at Levels 1, 2 and 3 is summarised in Fig. 6.

7.2.5. Levels 4, 5 and 6At the advanced levels, the FADs/CDFs incorporate a constraint allowance. This is based on ®nite

element at Level 4, J analysis at Level 5 and a Leak-before-Break argument to be followed at Level 6.Thus, the increasing accuracy of the analysis o�ers two possible lines of action: the ®rst involves the

modi®cation of the failure line, extending the safety zone with the increasing knowledge of the materialmechanical properties, as described above. While the second refers to the approximation of the toughnessvalues and the level of accuracy to which those can be estimated as described in Section 6.

7.3. Validation of homogeneous procedure

The ability of the SINTAP method to provide conservative predictions in the correct level of hierarchywas assessed through a number of examples: (1) Planar Geometry ± Wide Plates, (2) Planar Geometry ±Centre Cracked Panels, (3) Cylindrical Geometry ± Pipe Butt Weld.

In the ®rst example, seven surface cracked wide plates of di�erent materials, all steels, were analysed foran extended range of Y =T ratio. These plates were subjected to single uniform tension and all of them were450 mm wide. Particular data, both on the geometry and on the material properties, mechanical behaviourand toughness, were assessed. The steels had thickness of 12 to 40 mm and strength grades 275 to 1100MPa. The steel types covered general structural, o�shore and high strength grades.

The objective of the examples was to validate, through the experimental results, the analytically obtainedvalues, by checking that the points representing the actual failure situations fall outside the FADs or areconservatively predicted by the CDF.

Also, the evolution of the reserve factors through the di�erent analysis levels can be established, thus,pointing out the hierarchy of the procedure established by the di�erent approach levels.

In order to reach the objective of this study, di�erent calculations were performed by varying the qualityof input data by means of

· Di�erent hypotheses of knowledge of tensile properties, therefore, changing the failure assessment lineand, therefore, the degree of conservatism.

· Di�erent toughness considerations, correlations from Charpy tests or use of Kmat determined fromCTOD values.

These combinations resulted in 63 calculations of the critical applied stress for a given crack and for eachstructural situation. These analyses are summarised in Fig. 7. Several combinations of steel strength andplate thickness have been used and the calculations have been based on all of the source of toughnessoptions available in the procedure. In general, the in¯uence of the toughness option is dictated by Lmax

r , the


toughness data are of minor in¯uence, conversely if the failure is toughness dominated, then moving fromFAD1 to FAD3 will have no e�ect as they only di�er at high values of Lr.

The comparison given in Fig. 7 is based on CTOD data, and it can be seen that the safety factor de-creases with increasing assessment level for all of the combinations except one. The exception is the S690Qplate at 12 mm thickness and this plate has a high Y =T ratio which gives an Lmax

r . cut-o� of 1.027. The FADLevel 0 cut-o� is 1.00 as this plate had discontinuous yielding and plate failure is Lr controlled. Hence, thereis little in¯uence of either toughness or FAD type upon the failure stress prediction for this plate.

The e�ect of Y =T ratio on the observed safety factor is demonstrated in Fig. 8. The general trend is thatthe ratio of predicted/actual failure stress decreases with increasing Y =T ratio. This is largely a function ofthe Lmax

r calculation as all of the plates, except the S690Q 40 mm thick plate failed under Lr control. The

Fig. 7. Example of the hierarchy of the sintap procedure [28].

Fig. 8. E�ect of Y =T ratio on the predicted/actual failure stresses [28].


latter plate had a stress locus which intersected the FAD at the knee, hence the di�erent behaviour shown inFig. 8.

For high Y =T ratio materials, there appears to be little bene®t in using the higher levels of FAD, theadvantages in increased Lmax

r are minimal and if the material is of low toughness the Kr dominated regionsof the FADs are little a�ected by moving to a higher level.

The second example covered a centre-cracked panel subjected to uniform tension. This showed that thevalues obtained by using both SINTAP and R6 are very close. Also, when using a di�erent solution toSINTAP considering the ®nite width of the plate, the ®gures are, once again, very similar.

The third example dealt with a circumferential butt weld joining two pipe sections subjected to

· Internal pressure.· A global bending moment.· A known residual stress pro®le.

Two di�erent crack types were assessed:

· Circumferential internal surface crack.· Circumferential through-thickness.

This structure supports a more complex stress ®eld than the former cases as well as taking into accountthe consideration of residual stresses due to the welding process.

Overall, the validation examples on the homogeneous procedure demonstrated that the FAD/CDFmethods are conservative, the hierarchy in results is maintained and that the advantages of moving to ahigher level of FAD depend closely on the position of the initial point on the FAD.

8. Derivation of structural integrity assessment mismatch procedure

The SINTAP method for weld strength mismatched structures has been developed, based on two ex-isting defect assessment approaches for welded structures, the modi®ed R6 method [29] and the ETM-MM[30] method. A general structure of the SINTAP method for strength mismatch is shown schematically inFig. 9. It has four levels, depending on the quality of tensile information.

The detailed assessment equations of the SINTAP method are illustrated here using only the FADapproach and only for Level 2 and for two cases within that level: (1) both base and weld materials do notexhibit L�uders strain, (2) both materials exhibit L�uders strain. Other examples are given in Ref. [31]. Whenboth materials do not exhibit L�uders strain, then the following equation can be used for 06 Lr; < Lmax

r ,

f Lr� � � 1

�� 1

2L2

r

�ÿ1=2

0:3� � 0:7exp

ÿÿ 0:6L6r

��: �27�

The cut-o� Lmaxr , can be determined from

Lmaxr � 1

21

�� 0:3

0:3ÿ nM

�: �28�

In Eq. (28), the strain hardening exponent for the mismatched component, nM, is estimated from

nM � M ÿ 1� �FYM=FYB ÿ 1� �=nW � M ÿ FYM=FYB� �=nB

: �29�

The strength mismatch ratio, M, is de®ned at the yield strength:


M � YSW

YSB

; �30�

where YS denotes the 0.2% proof stress and the subscripts, B and W, denote the properties of the base andweld materials. The mismatch yield load, FYM, should be calculated for that value of M, and FYB is theplastic yield load assuming the component is wholly made of the base material. The hardening exponentsfor the weld and base material, nW and nB, are estimated from

nW � 0:3 1

�ÿ YSW

UTSW

�;

nB � 0:3 1

�ÿ YSB

UTSB

�;

�31�

where UTS denotes the ultimate tensile strength.When both materials exhibit L�uders strain

For 06 Lr < 1,

Fig. 9. Flowchart for mismatch procedure.


f Lr� � � 1

�� 1

2L2

r

�ÿ1=2

: �32�

At Lr � 1, the function f �Lr� is taken as discontinuous and reduces to the value f �1�, which is dependenton the extent of the L�uders strain:

f 1� � � kM

�� 1

2kM

�ÿ1=2

;

kM � FYM=FYB ÿ 1� �kW � M ÿ FYM=FYB� �kB

Mÿ 1� � ;

kW � 1� EWDeW

YSW

; DeW � 0:0375 1

�ÿYSW

1000

�;

kB � 1� EBDeB

YSB

; DeB � 0:0375 1

�ÿ YSB

1000

�;

�33�

where YS denotes the lower yield strength, E is the elastic modulus, De represents the estimated extent ofL�uders strain, subscripts, B and W, denote the properties of the base and weld materials. In Eq. (33), M andFYM=FYB are also de®ned at the yield strength, see Eq. (31).

For Lr1, the following equation is used up to Lr � Lmaxr ;

f Lr� � � f 1� � Lr� � nMÿ1� �=2nM �34�where f �1� is determined from Eq. (33); the mismatch hardening exponent, nM, from Eqs. (29)±(31); thecut-o�, Lmax

r , from Eq. (28).

8.1. Finite element validation

Various levels of the assessment equations, given in the previous section, were compared with a planestrain centre cracked tensile plate analysis with a=W � 0:5 and �W ÿ a�=H � 3. The weld con®guration wasidealised as a rectangular section. Using various ÔrealÕ stress strain data from several di�erent materials, awide range of mismatch ratios was generated. The materials considered included both continuous hard-ening behaviour and those with a L�uders strain [31].

Fig. 10 shows the results for the cases where both materials exhibit L�uders strain. These results andothers for continuous yielding materials show that the curves are in the correct order, the higher level curveis associated with the least conservatism. Moreover, Level 3 curve is quite close to Level 5 curve. Thismeans that Level 3 curve with accurate SIF and mismatch yield load solutions should lead to satisfactoryresults in practice.

8.2. Experimental validation

In the experimental programme [32], two di�erent materials, having di�erent yield and ultimate tensilestrength values, were produced by heat-treating an A533B-1 steel. The yield strength of the higher strengthmaterial (designated as M1) is about 50% higher than that of the lower strength material (designated asM3). It should be noted that the M3 material exhibits L�uders strain of a length about 0.8%. In the M1material, however, a L�uders plateau was not visible, due to its high yield strength and low hardening ca-pacity at low strains.

Strength mismatched specimens with idealised weldments were produced by electron-beam (EB) weld-ing, resulting in two di�erent strength mismatched specimens: highly over-matched (M � 1:48) and highlyunder-matched (M � 0:68) specimens. Single edge notched specimens in three point bend with a total of


20% side grooving were produced, having the crack in the centre of the weldment. Two di�erent cracklengths were chosen, a=W � 0:45 and 0.65. All specimens failed by extensive ductile tearing, and themaximum loads were obtained from experimental records.

For both overmatched and undermatched specimens, the predicted maximum loads, F predmax , are shown in

Fig. 11, together with the measured maximum loads, F expmax , in the test. The prediction using the Level 3

FAD is conservative but only by 10%. The Level 2 FAD gives slightly more conservative results. Suchresults are likely to arise from the conservatism embedded in Level 3 curve for materials with L�uders strain.

Fig. 10. SINTAP mismatch procedure for materials which exhibit L�uders strain [31].

Fig. 11. Comparison of predicted maximum loads with experimentally measured loads for the SINTAP mismatch procedure.


A similar trend has been observed for ductile tearing analyses.As a conclusion, the results from this experimental validation strongly support the methodology of the

SINTAP method for weld strength mismatch.

9. Quanti®cation of residual stress e�ects

Experimental studies have shown that secondary stresses introduced by welding or by temperaturegradients can have a signi®cant e�ect on the load carrying capacity of a component containing a ¯aw. Forthis reason, the SINTAP project included a speci®c task on residual stresses with the overall aim of de-termining and validating the most appropriate methods of accounting for residual stresses in as-welded,weld repaired and post-weld heat treated (PWHT) joints for the use in structural integrity assessment. Thistask included a review of existing information on the treatment of residual stresses in fracture prediction,including code-de®ned secondary stress pro®les and a collation of residual stress pro®les.

Experimental and numerical investigations of residual stress distributions in welded joints were alsocarried out. The numerical analysis work addressed the PWHT situation, the determination of its e�ec-tiveness and the derivation of criteria for PWHT of as-welded and repair-welded structures. Studies ofthrough-wall defects were performed with reference to available experimental data and included furtherexperimental work on thick, welded A533B steel plates. In addition, a study was performed of the esti-mation of J-integral when dominant residual stresses are present using centre-cracked panels, thus pro-viding an insight into the in¯uence of residual stresses on the fracture behaviour.

Standardised residual stress pro®les were derived for transverse and longitudinal through-thicknessresidual stress distributions in a range of geometries of welded joints manufactured from ferritic and aus-tenitic steels: Residual SIFs were also derived for surface cracks in a range of geometries of welded joints,i.e.

· plate butt welds· T-butt and ®llet welded joints· pipe butt welds· pipe seam welds· pipe to plate joints· tubular joints· repair welds

Finally, the means of incorporating the assessment of residual stress e�ects into the SINTAP procedurewere developed and validated.

9.1. Residual stress distributions in welded joints

The residual stress distributions in R6 [1] were used as the basis of the review and were supplemented bymore recent data from the literature. The data are valid for the parameter ranges shown in Table 4. Furtherdetails on the through-thickness distributions are given in Refs. [33,34]. Residual stress distributions werealso generated by numerical analysis for plate and pipe butt welded joints, PWHT ferritic steel joints anddissimilar metal welds [35].

The review was subsequently extended to include additional geometries, i.e. set in and set on nozzles,information generated in SINTAP and surface residual stresses. Generic upper bound distributions werethen derived for the through-thickness and surface distributions in both the longitudinal and transversedirections in ferritic and austenitic steel.


Through-thickness distributions are normalised with respect to the wall thickness and the 0.2% proofstress or yield strength. For transverse residual stresses YS should be taken to be the lesser of the yieldstrengths of the parent and weld material. The greater value of the two yield strengths should be used forlongitudinal residual stresses and for repair welds.

The distributions are expressed either as a function of the welding conditions and of the mechanicalproperties of the materials or as a polynomial function. Two sets of distributions are available: (a) for jointswhere the welding conditions are known and (b) for joints where the welding conditions are not known.

The distributions for (b) are more conservative than those for (a).The reader is referred to the Refs. [36,37] for more information.

9.2. Post-weld heat treated joints

The e�ect of post weld heat treatment (PWHT) on the residual stress distribution in welded joints wasinvestigated using pipeline butt welds. The distribution was evaluated numerically using the softwarepackage SYSWELDSYSWELD [38]. The parent material was assumed to have yield and ultimate tensile strengths of 342and 567 MPa, respectively, whilst the corresponding values for the weld metal were 511 and 591 MPa. Thepipe had a diameter of 1000 mm and a wall thickness of 30 mm. An asymmetric double vee weld geometrywas simulated, with the ®rst two passes deposited from the internal surface of the pipe and six subsequentpasses from the outside. An axisymmetric two-dimensional ®nite element model was used with 702 elementsand 2089 nodes.

SYSWELDSYSWELD simulates stress relief using a viscoplasticity model which can be used to represent the elasto-plastic isotropic work hardening response associated with materials during welding. Primary and secondarycreep were modelled using a Norton-type creep law with coe�cients relevant to time-dependent strainvalues for appropriate nominal static loads at 550°C. The following treatment was simulated:

· temperature increase from 20°C to 550°C in 9 s· constant temperature of 550°C for 3 h· controlled cooling to 20°C at a rate of 20°C/h.

The predicted through-thickness residual stresses after PWHT were very low with a maximum value ofapproximately 10% of the weld metal yield strength. These results are within the bounds of the PWHTresidual stress levels recommended in BS 7910, i.e. 15% of the yield strength in the transverse direction and30% of the yield strength in the longitudinal direction.

9.3. Treatment of residual stresses in structural integrity assessment procedures

The principal feature of the BS 7910 procedure relating to the assessment of residual stresses involves therelaxation of secondary stresses due to plasticity and is accounted for by

Table 4

Overview of residual stress data

Joint geometry Thickness (mm) Heat input (kJ/mm) Yield strength (MPa)

Plate butt joints 24±300 1.6±4.9 310±740

Pipe butt joints 9±84 0.35±1.9 225±780

Pipe seam welds 50±85 Not known 345±780

T-butt joints 25±100 1.4 376±421

Tubular and pipe to plate joints 22±50 0.6±2.0 360±490

Repair welds 75±152 1.2±1.6 500±590


Kr � KpI

ÿ � KsI

�=Kmat � q � f Lr� �; �35�

where q is a plasticity correction factor.An alternative approach for the assessment of plasticity e�ects on the residual stress distribution has

been proposed in SINTAP: a factor, V, is applied to Ks to account for plasticity e�ects. Kr is then de®nedthus

Kr � Kp

ÿ � VKs

�=Kmat: �36�

Equations relating V and q have been developed using previously calculated numerical values of q togenerate values of V =V0 as a function of increasing load for a number of values of secondary stress, where Vat zero primary load is termed V0 and is given by

V0 ��

E�1ÿ V 2�

sJS=Ks

I �37�

where superscript s refers to secondary stress.Numerical predictions of V =V0 vs. Lr were obtained, and are based on the R6 Option 1 FAD. The results

are presented in Fig. 12. They show that V =V0 increases moderately with increasing Lr for Lr values up toapproximately 0.9 and decreases at higher Lr values. Ainsworth [39] found that V =V0 is very insensitive tothe magnitude of the secondary stress, particularly for Lr > 0:9, and therefore proposed a simple ap-proximate approach. Two options are included in Fig. 12:

Fig. 12. Numerical predictions of V =V0 vs. Lr [37].


Lr6 0:9; V =V0 � 1:25;

0:9 < Lr6 1:4; V =V0 � 2:78ÿ 1:7Lr;

Lr > 1:4; V =V0 � 0:4:

�38�

Lr6 0:8; V =V0 � 1:1� 0:4Lr;

0:8 < Lr6 1:4; V =V0 � 2:78ÿ 1:7Lr;

Lr > 1:4; V =V0 � 0:4:

�39�

Eq. (38) bound the solutions for �Ksp=�Kp

I =Lr��6 2 and use a constant value for low Lr, similar to the ap-proximate method in the current version of R6 which uses a constant value of q in this region. The plateauvalue could be made a function of the magnitude of the secondary stress. Eq. (39) use a linear ®t at low Lr

and bounds the solutions for �Ksp=�Kp

I =Lr��6 5. At values of Lr > 0:9, both equations are identical andcorrespond to the use of q � 0 for Lr � 1:05 in R6. V =V0 has been set to 0.4 for Lr > 1:4 as a bounding valueto the numerical data but, in practice, full mechanical stress relief, i.e. V � 0, has been observed experi-mentally [40].

10. Reliability of inspection

Inspection procedures based on non-destructive evaluation (NDE) techniques play an important role instructural integrity assessments. Ageing of installations and the on-going demand for extending lifetimeincrease the importance of the inspection techniques used to evaluate the condition of structures. It shouldbe emphasised that inspection procedures are very complex processes, often involve many speci®c tech-niques, decisions making, and calibration procedures. Thus, they cannot be simply regarded as a measuringtechnique and their performance in defect detection, classi®cation, and sizing cannot be fully represented byassigning simple con®dence intervals. The analysis of an inspection procedureÕs e�ectiveness, is a functionof the intrinsic capability of the inspection procedure itself and of application features, particularly theinspection performance.

This task within SINTAP has concentrated on the provision of guidelines for the statistical treatment ofNDE data, as well as on interaction between NDE and fracture mechanics assessment including probabilityof detection and sizing errors. The status review highlighted the importance of the availability of reliableNDE methods, techniques and procedures which are capable of providing the required quantitative in-formation on weld fabrication ¯aws or service-induced ¯aws in the actual component. Determining asuitable defect density and size distribution is essential for calculating probabilities of failure, requiringconsideration of a number of decision thresholds.

10.1. Compilation of non-destructive evaluation e�ectiveness data

Extensive reviews of the capability of NDE methods [41] and results of previous NDE projects [42] werecarried out. Collated NDE data from industry have been used to create a matrix summarising and groupingthe current data in terms of (i) component type, (ii) material, (iii) thickness, (iv) NDE technique and (v) thetype of data on typical ¯aws found.

The selected data were collected from blind test results or from parametric studies conducted by inde-pendent institutions. Speci®cally, the following NDE programmes were taken into account: PISC (threephases including the parametric studies [43,44], DDT exercises in UK [45], the PVRC/HSST programmeresults, the IGSCC training programme in USA [46], Nordtest [47] and NIL programmes [48], TWI pro-jects [49] and some results of ICON [50]. It should be emphasised that the present compilation gives general


trends and indications and does not provide a rigorous statistical analysis. This process inevitably entailssome simpli®cation of the data, to provide the broad statements of capability that are needed by manu-facturers, users, integrity evaluators and regulators.

To allow a pragmatic presentation of inspection procedures, e�ectiveness or performance in detection,classi®cation and sizing of ¯aws, several categories of components were considered. Table 5 summarises themain characteristics of components analysed by SINTAP, t represents the component wall thickness, whilstD is the piping internal diameter.

For any inspection process, di�erent levels of performance can be obtained. Two di�erent classes areQ-level: obtained through a Quali®cation exercise (e.g. European methodology) that ®xes the e�ec-

tiveness of the inspection procedure at a level considered possible after capability evaluation.B-level: corresponding to what was shown by 60% of the inspection procedures applied in round robin

tests relevant to the speci®c situation (Blind trials). In this case, e�ectiveness can be very good with highperformance inspection procedures and very poor with low performance procedures even if applied withcare by a good team.

For the presentation of NDE results, particular attention is given to the ¯aw detection probability (FDP)as a function of the ¯aw size (depth or length) and to the correct rejection probability (CRP). Estimates ofFDP and CRP values based on round robin trials (RRTs) with a limited number of defects will be char-acterised by an uncertainty that should be taken into consideration when comparing these values. As mostof the RRTs or exercises that generate such data deal with 10 to 40 defects, a possible approach consists ofusing con®dence levels or standard deviations, for instance for 20 defects the con®dence limits at 95% onFDP correspond approximately to �0:2.

With regards to defect sizing, the main parameters used to estimate the inspection performance are themean error of sizing (MES) and the standard deviation (SES). Referring to the depth or the length sizing,the letter (D) or (L) are added to the above symbols, respectively. Unfortunately, it is not always possible toindicate a standard deviation as computed from the measurement set. Often the error band has to beevaluated based on experience or on very limited trials.

10.2. Defect sizing in ferritic steel components

The PISC programme represents the most relevant reference to this category of components. Fig. 13(a)and (b) illustrates the FDP and CRP for planar defects. They were obtained by using UT procedures thatwere quali®ed to the best attainable level. In the ®gures, the FDP and the CRP values were reported as afunction of the defect depth size normalised to the component wall thickness.

Table 5

Components included in NDE study

Description Material t (mm) D (mm)

Heavy pressure vessel (or ¯at plates) Carbon steel >75 ±

Heavy section piping (or ¯at plates) Carbon steel 30±75 >250

Heavy section piping (or ¯at plates) Carbon steel 10±30 >250

Small diameter piping Carbon steel 5±30 50±250

Heavy section piping Wrought iron >30 >250

Small diameter piping Wrought iron <30 50±250

Heavy section piping (or elbow) Cast iron 20±80 >250

Dissimilar metal zones (piping or comp.) Various 20±80 >250

Small tube (steam generators) Wrought iron 1±5 20±50

Small tube (heat exchangers) Carbon steel 1±5 30±50

Thin ¯at plate Alloys <5 ±


With regards to the detection capability, a very similar behaviour was found for heavy pressure vesselcomponents, and for large and small diameter piping with thick and thin walls. However, certain conditionshave to be satis®ed, such as the correct selection of techniques, correct access, and the necessary quali®-

Fig. 13. Examples of the accuracy of defect sizing.


cation programme. Sizing of multiple defects remains di�cult, mainly due to the lack of time or resourcesoften dedicated to defect classi®cation.

Fig. 13(c) and (d) reports reference e�ectiveness values when no quali®cation programme is applied. Thiswas obtained using good inspection practices (e.g., 10% or 20% DAC) that, with quali®cation, would givesimilar performances to those presented above.

Fig. 13(e) and (f) refers to B level low capability inspection procedures. These results indicate how poorthe performance can be, if poor NDE procedures (e.g., 50% or 100% DAC) are used, or if the applicationand human factors are not controlled by quali®cation and the application of a quality assurance pro-gramme. These values could also correspond to lower bounds for manual inspection conducted in verydi�cult conditions or environment.

Similar data are also available for radiography and surface inspection techniques, and for austenitic steelcomponents.

11. Speci®c analysis options

11.1. Constraint

A particular conservatism implicit in most procedures is that the value of fracture toughness, Kmat, isnormally derived from deeply cracked bend specimens using recommended testing standards and validitycriteria. These are designed to ensure plane strain conditions and high hydrostatic stresses near the crack tipto provide a material property independent of specimen size and geometry. However, there is considerableevidence that the material resistance to fracture is increased when specimens with shallow cracks, orspecimens in tension, are tested. These conditions lead to lower hydrostatic stresses at the crack tip referredto as lower constraint.

Procedures for dealing with this problem are given in the SINTAP procedure and are outlined in Ref.[14]. It is not intended that these procedures replace those of the conventional SINTAP method; rather thatthey can be used in conjunction with these to estimate any increase in reserve factors likely to arise underconditions of low constraint.

The constraint appendix of the procedure is written in terms of the parameters T and Q, where the latterparameter can be evaluated in a number of ways. The principles can equally be applied for other constraintparameters provided their load dependence and their in¯uence on fracture toughness can be quanti®ed.

11.2. Leak-before-break assessment

There are several options by which it may be possible to demonstrate the safety of a structure containing¯aws when an initial analysis has failed to show that adequate margins exists. For pressurised components,one of these options is to make a leak-before-break case by demonstrating that a ¯aw will grow in such away as to cause, in the ®rst instance, a stable detectable leak of the pressure boundary rather than a sudden,disruptive break. Methods for carrying out such analysis are described and this forms Level 6 of theSINTAP procedure.

The various stages in the development of a leak-before-break argument are explained in Fig. 14. Thisdiagram has axes of ¯aw depth, a, and ¯aw length, l, normalised to the pipe or vessel wall thickness, t. Aninitial part-through ¯aw is represented by a point on the diagram. The ¯aw may grow by fatigue, tearing orany other process until it reaches some critical depth at which the remaining ligament ahead of the ¯awbreaks through the wall. The ¯aw then continues growing in surface length until there is su�cient openingto cause a detectable leak or until the ¯aw becomes unstable.


Several steps are involved in establishing each of these requirements. First, the ¯aw must be charac-terised and the mechanism by which it can grow identi®ed. The next step is to calculate the length of thethrough-wall ¯aw formed as the initial defect penetrates the pressure boundary; this is then compared withthe critical length of a fully-penetrating ¯aw. Finally, it is necessary to estimate the crack-opening area, therate at which ¯uid leaks from the ¯aw, and whether or not the leak will be detected before the ¯aw grows toa critical length.

In contrast to the procedures, which are concerned with failure avoidance, part of the leak-before-breakcase involves failure prediction. It is therefore recommended that best-estimate values of stresses andmaterial properties, rather than pessimistic values, are used to estimate the ¯aw length at breakthroughwhen part-through ¯aws are being considered. However, to ensure conservatism, pessimistic values shouldbe used to calculate the critical length of the resulting through-wall ¯aw.

11.3. Prior overload

The loading history due to (i) proof or overload tests or (ii) warm prestressing of a structure containing¯aws may be taken into account when performing an integrity assessment using SINTAP. The e�ect ofloading history is considered with regard to mechanical relaxation of residual stresses and enhancement oflower shelf fracture resistance. The latter is only applicable where the preload constitutes a warm prestress.

Procedures are set out which enable these e�ects to be quanti®ed although in practice, the di�erentphenomena may interact and it may not be possible to separate the di�erent e�ects simply.

Fig. 14. The leak-before-break diagram.


12. Signi®cance of results

The procedures discussed previously have been deterministic. In this sense, for any level of analysischosen, the input data are treated as a set of ®xed quantities, and the result obtained is unique. Theproximity of this limiting condition to the structural failure condition not only varies from level to level butit does so even within a given level of analysis. This is because it is dependent on the quality of the data, onthe number of specimens tested, on how the value of the input used in the analysis is obtained from suchtest results, on how closely these values represent the data in the location of the crack in the real structureand on how accurately the loads and stresses on the structure can be determined. The treatments recom-mended for all these data are conservative, in the sense that when applied alone or in a combined way, anunderestimate of the defect tolerance of the structure is obtained. However, the amount of the under-estimation is indeterminate because of the uncertainties in the input data.

When assessing the acceptability of a result, con®dence is established in two ways: by means of the valueschosen for the input data, and by assessing the signi®cance of the result. The ®rst of these determines thelevel of con®dence which can be placed in the analysis from the viewpoint of each of the variables, eachvariable being treated separately. In each case the con®dence level is dependent mainly upon the quality andtype of input data. Although this may be high for any one of the data sets concerned, it says little about theoverall con®dence level of the ®nal result. For this, the whole result must be assessed to establish how all thedi�erent con®dence levels of the input data interact and combine with each other to provide the ®nal result.At this stage, the necessary reserve factors can be established taking proper account of the in¯uence of thedi�erent variables on the reliability of the result.

The characteristic value of the fracture toughness must take into account the di�erent amounts of un-certainty inherent in the fracture toughness which are dependent upon the metallurgical failure mechanismsand how they are represented in the analysis. Where the fracture mechanism is brittle, the fracturetoughness is often highly scattered, especially where the material is inhomogeneous, as for example inweldments. For this reason, the reliability of the result is dependent on the number of specimens tested, asdiscussed earlier. The method developed provides a probability distribution of fracture toughness, fromwhich the characteristic value may be derived. The level of conservatism is dependent on the number ofspecimens tested, and the incidence of low results which do not conform to the general distribution.

The limiting state, evaluated using values for the input data established following the guidelines in theprocedure do, in principle, de®ne a safe operating condition, although for some engineering purposes, forexample in design calculations, additional con®dence in determining safe loading conditions is traditionallygained by applying safety or reserve factors. However, the application of previously speci®ed numericalfactors in fracture analysis can be misleading because of the inherent and variable interdependence of theparameters contributing to fracture behaviour. Con®dence in assessments is reinforced by investigating thesensitivity of the result to credible variations in the appropriate input parameters. Sensitivity analyses arefacilitated by considering the e�ects that such variations have on reserve factors.

In addition, software has been produced which allows failure probabilities to be evaluated using bothÔMonte CarloÕ and Ô®rst-order reliability methods (FORM)Õ techniques depending upon the relative failureprobability and computing power.

13. Software developments

Industrial examples of the use of procedures such as SINTAP have illustrated the bene®ts of havingsoftware available to automate the calculations. Because the application of the procedure requires largenumbers of extensive computations, iterative solutions and complex interaction between the various partsof the procedure, automation of the SINTAP procedure in the form of software has been an integral part of


the project. Further advantages of this are repeatability, accuracy, accountability and validation, all ofwhich lead to increased e�ectiveness of usersÕ time.

The principal factors considered in the development of the SINTAP software have been functionality,consideration of user requirements, modularity and validation. The software developed in parallel with thetechnical progress within the individual tasks and now exists as a complete package although, as the overallaim of the software development was to aid veri®cation of the procedure, it is not intended to supply this asa commercial package at the moment.

In addition to the procedure software, as noted above, the probabilistic software uses the SINTAPprocedure to determine failure probabilities for a range of input variables.

This software calculates two di�erent failure possibilities:

· Probability of failure when the defect size is given by NDE.· Probability of failure when the defect size is not detected by NDE.

14. Future development of the procedure

The dissemination and exploitation of results has been a key aspect of the project. Many of the indi-vidual task and sub-task areas have been published and the information provided to contribute to thecontinuing development of procedures, such as R6, BS7910 and ETM. Furthermore, the collaborativenature of the project has enabled technology transfer across industries; speci®c examples include incor-poration of the PISC and NESC data, knowledge from the nuclear industry on leak-before-break, datafrom the o�shore industry on tubular joints and information on steel properties from extensive databases.The SINTAP results and procedure will contribute to the development of a CEN ®tness-for-purposestandard within the remit of the CEN TC 121 Committee and will help to de®ne a Europe wide standard.

References

[1] Harrison RP, Loosemore K, Milne I. Assessment of the integrity of structures containing defects. CEGB Report R/H/R6, 1976.

[2] Guidance methods for assessing the acceptability of ¯aws in fusion welded structures. British Standards Institution, PD 6493,

1991.

[3] Bergman M, Brickstad B, Dahlberg L, Nilsson F, Sattari-Far I. A procedure for safety assessment of components with cracks-

handbook. SA/FoU Report, 9l/Ol, AB svensk anlaggningsprovning. Swedish Plant Inspection Ltd., 1991.

[4] Fitness for service guide. EXXON chemical, 1995.

[5] Assessment of the integrity of structures containing discontinuities. INSTA technical report. Materials Standards Institution,

Stockholm, 1991.

[6] MPC, Fitness for service evaluation procedures for operating pressure vessels, tanks and piping in re®nery and chemical service

(FS26). Draft no. 5, consultantÕs report, The Material Properties Council, 1995.

[7] Recommended practice for ®tness-for-service. Draft Issue 4, American Petroleum Institute, API 579, 1996.

[8] Schwalbe KH, Zerbst O, Brocks W, Cornec A, Heerens J, Amstutz H. The engineering treatment model for assessing the

signi®cance of crack-like defects in engineering structures EFAM-ETM 95Õ, GKSS Forschungzentrum, 1996.

[9] Kumar V, German MD, Shih CF. An engineering approach for elastic±plastic fracture analysis. General Electrical Company, NP-

193 I, Research Project 1237-1, Topical Report, 1981.

[10] British Standard BS7910 Guidance on methods for assessing the acceptability of ¯aws in metallic structures. British Standards

Institution, 2000.

[11] Schwalbe K-H, Kocßak M. Mismatching of welds, Papers presented at the International Symposium on Strength Mismatched

Welded or Bonded Joints, Reinstorf Luneburg, Germany, 26±28 May 1993, ESIS Publication 17.

[12] Schwalbe K-H, Kocßak M. Mismatching of interfaces and welds. Papers presented at Second International Symposium on

Performance of Strength Mismatched Welded or Bonded Joints, Reinstorf, Luneburg, Germany, 24±26 April 1999. Geesthacht:

GKSS Publications; 1997.


[13] ASTM, Standard test method for determination of reference temperature, T0, for ferritic steels in the transition range. ASTM

E1921-97, vol. 03.01, 1998. p. 1068±84.

[14] Ainsworth RA, Sattari-Far I, Sherry AH, Hooton DG, Hadley I. Development of methods for including constraint e�ects within

the SINTAP procedures. ECF12 Conference, vol. 2, 1998. p. 589±94.

[15] Ruiz-Ocejo J, Gutierrez-Solana F, Gonzalez-Posada M, Gorrochategui I. Paper 98-2052: A review of current defect assessment

procedures. 17th International Conference on O�shore Mechanics and Arctic Engineering. Lisbon, Portugal, July 1998.

[16] Ruiz-Ocejo J, Gutierrez-Solana, MA, Gorrochategui I, Gutierrez-Solana F. Comparison between structural integrity assessment

procedures for cracked components. Fourth International Conference on Engineering Structural Integrity Assessment: Lifetime

Management and Evaluation of Plant, Structures and Components, Cambridge, UK, September 1998.

[17] Ruiz-Ocejo J, Gonzalez-Posada MA, Guti�errez-Solana F, Gorrochategui I. Review of existing procedures, University of

Cantabria Report SINTAP/UC/04, 1997.

[18] Ainsworth RA, Kim Y-J, Zerbst U, Gutierrez-Solana F, Ruiz-Ocejo J. Driving force and failure assessment diagram methods for

defect assessment. Paper 98-2054. 17th International Conference on O�shore Mechanics and Arctic Engineering. Lisbon,

Portugal, July 1998.

[19] Ruiz J, Bannister AC, Ainsworth RA, Kim Y-J. Development of the SINTAP failure assessment diagrams for compatibility with

crack driving force approaches. Paper OMAE 99-2042. 18th International Conference on O�shore Mechanics and Arctic

Engineering. Newfoundland, July 1999.

[20] Wallin K, Nevasmaa P. Structural integrity assessment procedures for European industry (SINTAP) ± Sub-task 3.2 report:

Methodology for the treatment of fracture toughness data ± procedure and validation. Report VAL A: SINTAP/VTT/7.VTT

Manufacturing Technology, Espoo, 1998. p. 52.

[21] Nevasmaa P, Bannister AC, Wallin K. Fracture toughness estimation methodology in the SINTAP procedure. Paper 98-2053.

17th International Conference on O�shore Mechanics and Arctic Engineering. Lisbon, Portugal, July 1998.

[22] Moscovic R. Statistical analysis of censored fracture toughness data in the ductile to brittle transition temperature region. Engng

Fract Mech 1993;44(1):21±41.

[23] Wallin K. Traditional damage e�ects on the fracture toughness transition curve shape for reactor pressure vessel steels. Int J Press

Vess Pipelines 1993;55:61.

[24] Bannister AC. Structural integrity assessment procedures for European industry (SINTAP), Sub-task 3.3 report. Determination of

fracture toughness from Charpy impact energy: Procedure and validation. Document no. SINTAP/BS/15. British Steel plc., 1997.

13 p+7 Appendices.

[25] Langenberg P, Dahl W, Sedlacek G, St�otzel G, Strangh�orber N. Annex C, Material choice for the avoidance of brittle fracture in

Eurocode 3. Proc. Second International Conference on Weld Strength Mismatch. GKSS, April 1996.

[26] Wallin K, Pisarski HG. Application and veri®cation of the SINTAP toughness estimation method as applied to welds and parent

materials. Paper OMAE 99-2043. 18th International Conference on O�shore Mechanics and Arctic Engineering, Newfoundland,

July 1999.

[27] Bannister AC. Yield stress/tensile stress ratio: Results of experimental programme. British Steel Report SINTAP/BS/25, February

1999.

[28] Ruiz-Ocejo J, Gutierrez-Solana F. SINTAP validation report. Document no. SINTAP/UC/17, June 1999. p. 35.

[29] R6: Appendix 16: Allowance for strength mismatch e�ects, Nuclear Electric procedure R/H/R6, Rev. 3, 1998.

[30] Schwalbe K-H, Kim Y-J, Hao S, Cornec A, Kocßak M, EFAM ETM-MM 96: The ETM method for assessing the signi®cance of

crack-like defect in joints with mechanical heterogeneity (strength mismatch). GKSS 97/E/9, GKSS Research Centre, Germany,

1996.

[31] Kocßak M, Kim Y-J, Zerbst U, Ainsworth RA. Validation of SINTAP procedure for strength mismatched structures. Paper

OMAE 99-2041. 18th International Conference on O�shore Mechanics and Arctic Engineering. Newfoundland, July 1999.

[32] Kim Y-J. Experimental validation of the SINTAP procedure for strength mismatch. British Energy Generation Ltd., April 1999.

p. 19.

[33] Barthelemy JY. Compendium of residual stress pro®les. Institut de Soudure, Ennery, France, 1998.

[34] Barthelemy JY, Stacey A. A new compendium of welding residual stresses for structural integrity assessment within the SINTAP

project. 12th European Conference on Fracture, ECF12. She�eld, UK, September 1998.

[35] Barthelemy JY. Post weld heat treatment of a pipeline butt weld. Institut de Soudure, Ennery, France, 1998.

[36] Stacey A, Barthelemy JY. Incorporation of residual stresses into the SINTAP defect assessment procedures. Paper 98-2055. 17th

International Conference on O�shore Mechanics and Arctic Engineering, Lisbon, Portugal, July 1998.

[37] Stacey A, Barthelemy JY, Ainsworth RA, Leggatt RH, Bate SK. Quanti®cation of residual stress e�ects in the SINTAP defect

assessment procedure for welded components. OMAE Õ99 Paper 2044, St. JohnÕs, Newfoundland.

[38] SYSWELDSYSWELD, Instruction Manual, Ver. 231, Framasoft and CSI, 1991.

[39] Ainsworth RA. The treatment of thermal and residual stresses in fracture assessments. Engng Fract Mech 1986;24:65±76.

[40] France CC, Sharples JK, Wignall C. Experimental programme to assess the in¯uence of residual stresses on fracture behaviour ±

Summary report. Report AEAT-4236, AEA Technology, Risley, 1998.


[41] Crutzen S, Frank F. Structural integrity assessment procedures for European industry (SINTAP), Sub-task 3.4 Final report: NDE

E�ectiveness. JRC Research Report JRC/IAM, Petten, 30 March 1999.

[42] Visser W. Guidance on POD/POS curves for non-destructive examination for SINTAP, March 1998.

[43] PISC II: Nichols RW, Crutzen S. Ultrasonic inspection of heavy section steel components. The PISCC II ®nal report, Applied

Science, Barking, UK, Elsevier, ISBN 1-85116-155-7, 1988.

[44] PISC III: Evaluation of the sizing results of 12 ¯aws of the full scale vessel installation. PISC III report no 26-Action 2-Phase 1,

JRC report: EUR 15371 EN, 1993.

[45] Watkins B, Ervine RE, Cowburn KJ. The UKAEA defect detection trial. UKAEA DDT Symposium, Silver Birch Conf Centre,

Birchwood, Warrington, UK, 7±8 October 1982.

[46] Heasler PG, Taylor TT, Spanner JC, Doctor SR, De�enbough JD. Ultrasonic inspection reliability for intergranular stress

corrosion cracks: a round robin study for the e�ects of personal procedures, equipment and crack characteristics. NUREG/CR-

4908, PNL-6196, 1990.

[47] Fùrli O. Development and optimisation of NDT for practical use ± Nordtest NDT programme. 5e Nordiska NDT Symposiet

Esbo, Finland, report numbers: IIW-V-967-91, IIW-V-968-91, and IIW-V-969-91, 1990.

[48] NIL, NDT regulations, NIL report NDP 95-85, 1995.

[49] TWI report, Size measurements and characterisation of weld defects by ultrasonic testing. part 3, ISBN 0 85300162, 1982.

[50] Rudlin JR, Dover WD. Performance trends for POD as measured in the ICON project. Proc. OMAE 96, vol. 2, 1996. p. 509.


Documents

Structural integrity assessment procedure for Europe – of the SINTAP programme overview