I Hadley, C S Wiesner and S J Maddox
Presented at the I Mech E seminar, 'Flaw assessment in pressure equipment and welded structures - PD 6493 to BS 7910', London, 8 June 1999
The British Standards guidance document PD 6493 ('Guidance on methods for assessing the acceptability of flaws in fusion welded structures'), first published in 1980 and revised in 1991, has now been revised again and published as a British Standard Guide, BS 7910. The new document incorporates additional structural and defect geometries, a number of new structural integrity tools, and industry-specific application guides, which are briefly reviewed in this paper.
Design and fabrication codes ensure that welded structures such as pressure vessels, pipelines, offshore structures, bridges and buildings are safe, provided all code requirements are met during the lifetime of the structure. The margins of safety inherent in such codes can be somewhat arbitrary, and the codes cannot be applied to situations where the structure, build quality, materials or service conditions do not meet the strict requirements of the code.
An alternative philosophy is based on fitness-for-purpose principles and states that a component is acceptable provided the conditions for failure are not reached in the service lifetime. A number of possible failure mechanisms must be considered, including fracture, plastic collapse and fatigue. In order to ensure that such analyses are carried out in a systematic manner, the British Standards Institution (BSI) published guidance for the assessment of structural integrity of welded structures containing flaws, designated PD (Published Document) 6493 '(Guidance on methods for assessing the acceptability of flaws in fusion welded structures'). This document first appeared in 1980, and was extensively revised in 1991, incorporating more advanced fracture assessment methods from the nuclear industry. Since 1980, PD 6493 has become very widely used, in terms of both geography and industry sector.
Recent advances in the application of fracture mechanics, along with developments in other fitness-for-purpose procedures and changes in BS fracture mechanics test standards provided a driving force for a further revision of PD 6493. A decision was also taken to combine PD 6493 with the high-temperature assessment procedure PD 6539, to produce a single document. The result was published in December 1999 as a BS guide, designated BS 7910 ('Guide on methods for assessing the acceptability of flaws in metallic structures'). The fracture and fatigue assessments parts of BS 7910 only are covered in this paper.
The overall structure of BS 7910 is shown in Fig.1. Clauses 1-6 of the procedure cover the principles of the method and the type of input required. Clause 7 covers fracture assessment, and has been extensively modified since the 1991 edition of PD 6493. Relatively minor modifications have been made to the fatigue assessment procedures (Clause 8). A completely new chapter (Clause 9) has also been added, covering assessment of flaws in plant operating at high temperature. Finally, there are some 20 annexes, many of which introduce new methods and concepts.
2 Fracture assessment
Clause 7 (the fracture clause) of BS 7910 has been completely re-written to improve clarity and usability, and to incorporate modern defect assessment technology, including a number of methods originally published in the CEGB R6 procedure. Flow diagrams have also been included to guide users through the three fracture assessment levels.
As in PD 6493, there is a three-level fracture assessment procedure in BS 7910. All three levels use the concept of a Failure Assessment Diagram (FAD). The basic assumption is that the flawed body could fail by one of two extreme failure modes - fracture or plastic collapse (overload). The propensity to failure by plastic collapse is shown on the x-axis as a dimensionless number (L r or S r ) showing the ratio of applied 'reference' stress (net section stress) to the yield strength or flow strength of the material. The proximity to fracture is calculated independently, and shown on the y-axis as the ratio of applied stress intensity to materials toughness (K r or
, depending on whether toughness is measured in terms of K Ic , CTOD ( δ) or critical J. A failure assessment line (see Table 1) shows the relationship between them, so that any point which falls inside the failure assessment line can be considered safe, whilst any point outside the line is potentially unsafe.
Table 1 Aspects of the three-level fracture assessment procedure in BS 7910
|Aspect||Level 1||Level 2||Level 3|
|Plastic collapse axis
|Toughness input required
||Single-point value of fracture toughness (sometimes Charpy energy)
||Single-point value of fracture toughness (sometimes Charpy energy)
||Tearing resistance curve (R-curve)
|Form of failure assessment line
S r,max = 0.8,
K r,max = 0.71
|K r = f(L r ),
L r,max = σ f / σ y ,
K r,max = 1.0
|K r = f(L r ),
L r,max = σ f / σ y ,
K r,max = 1.0
Level 1 is the simplest, both in the detail of the required input, and the complexity of computation. The Level 1 fracture assessment procedure in BS 7910 is virtually unchanged with respect to PD 6493:1991, but is renamed the Simplified Assessment Procedure. The main procedure is based on a simple rectangular FAD (see Table 1), whilst the graphical procedure (the CTOD design curve approach, originally presented in PD 6493:1980) becomes Annex N in BS 7910.
Level 2 remains the normal assessment method for cases where single-value measurements of fracture toughness (e.g. K Ic , J mat , δ mat ) are available. The PD 6493:1991 Level 2 failure assessment method (strip yield model) is now superseded by Levels 2a and 2b; the choice between the two depends on the type of stress-strain data available for the material in which the flaw is situated. The Level 2a FAD is intended to be used for materials such as Heat-affected Zones (HAZs) where full stress-strain curves are rarely available; Level 2b should be used where possible. The failure assessment lines for Level 2a and 2b are similar to those used in PD 6493, Level 3. L r , the ratio of reference stress to yield strength, is now used in place of S r (ratio of reference stress to flow strength) for plastic collapse predictions at Level 2.
Figure 2 shows a comparison of the FADs at Levels 1, 2a and 2b for a particular structural steel, along with the strip yield model (PD 6493, Level 2). Note that the Level 1 FAD is plotted here in terms of L r rather than S r for consistency with the Level 2 FADs, although Level 1 results used in isolation would always be plotted in terms of S r . Whilst it can be seen that moving from a Level 1 to a Level 2b approach via Level 2a would in general give increasing 'safe' areas, there are slight overlaps between Level 1 and 2a, and between Levels 2a and 2b. BS 7910 advises that if the material being analysed shows (or is likely to show) 'discontinuous yield' (for example, a Lüder's plateau), then the Level 2a failure assessment line should be cut off at L r =1.0 or the Level 2b approach used. This prevents 'overlapping' of the Level 2a and Level 2b failure assessment lines in the high L r region 1.0<L r <L r , max . It can be seen from Fig.2 that the material shown (which has a rather flat stress-strain curve) comes into this category, so the L r , max =1.0 rule would apply, or the analysis would be carried out using the Level 2b curve. Of course, it is intended that the full stress-strain curve should be used wherever this is available.
Fig. 2. Comparison of Levels 1 and 2 of BS 7910 with the strip yield model (PD 6493, Level 2)
Note that the BS 7910 Level 2 FADs are more conservative than the old PD 6493 Level 2 diagram, especially in the 'knee' region (high K r , high L r ). Whilst this will lead to smaller tolerable flaw dimensions, assessment of large-scale validation tests  has shown that there were cases where the PD 6493 level 2 method resulted in non-conservative predictions unless partial safety factors were applied to the input data. By contrast, BS 7910 predictions of the same tests produces points which are all outside the failure assessment line (see Figure 3).
Fig. 3. Validation of the BS 7910 level 2 FAD using wide plate data
Level 3 of BS 7910 (ductile tearing instability assessment) remains unchanged with respect to PD 6493, except for the addition of the R6 Option 3 method, which becomes Level 3c in BS 7910. In this approach, the FAD and driving force may be derived from detailed elastic-plastic finite element analysis to give more accurate predictions of structural behaviour.
3 Fatigue assessment
The fatigue assessment clauses in PD 6493:1991 were also reviewed in the light of new information and experience gained from their use in practice. The main change was the introduction of new fatigue crack growth laws, based on an extensive review and analysis of published data for steels 
. These include more precise two-branch Paris laws and allowance for applied stress ratio, R. In PD 6493, the assumption was made that fatigue crack growth could be represented by a single Paris law between the limits of the threshold stress intensity factor range for fatigue crack growth ( ΔK o
) and fracture (K Ic
). Although this simplification is satisfactory for many assessments, it is seen to be over-conservative at low values of ΔK, approaching the threshold ( Figure 4
). However, new conservative (upper-bound) single Paris laws, which result in higher growth rates than those in PD 6493, are still provided, for convenience. More attention is paid to environmental influences and the new recommendations cover marine corrosion, with and without cathodic protection, and fatigue crack growth at elevated temperature.
Fig. 4. Two-stage and simplified fatigue crack growth laws for steels
The specific cases covered include ferritic steels in air, freely-corroding in seawater and in seawater with cathodic protection (-850mV and -1100mV Ag/AgCl). The simplified laws also refer to ferritic steels, including at elevated temperature. They relate to high R values or to welded material (assumed to contain high tensile residual stresses and hence to experience a high effective R value under any fatigue loading). Austenitic steels can be treated using the simplified law for ferritic steels in air, but no advice is given for other environments. New data did not justify any changes to the recommended stress intensity factor threshold values in PD 6493. However, it is now strongly recommended that crack growth rate and threshold values for high R values, as detailed in Table 2 , are assumed when assessing a flaw in a welded structure, to allow for the influence of high tensile residual stresses.
Table 2 Recommended fatigue crack growth laws for steel (upper bound, R ≥0.5)
Advice on the derivation of fatigue crack growth laws and threshold values for non-ferrous metals is also given, using correlations based on the relative E values.
|Environment||ΔK o ,|
|Stage A constants||Stage B constants||ΔK (Nmm -3/2 ) at transition from A to B|
|Marine, free corrosion
|Marine, -850mV CP
|Marine, -1100mV CP
|The constants given above give fatigue crack growth rate (da/dN) in mm/cycle, as a function of ΔK in Nmm -3/2 ;
da/dN=A( ΔK) m for ΔK o < ΔK<K Ic
Allowance has been made for the extensive evidence now available (eg. [3,4] ) which indicates that there is no need to impose the flaw interaction criteria in PD 6493 in a fatigue assessment. Thus, multiple flaws are assessed separately without any consideration of flaw interaction.
The assessment method referred to in PD 6493 as the Simplified Procedure, which relates the required and actual fatigue performance of a flaw to a grid of quality category S-N curves, is retained, but it is referred to as 'Assessment Using Quality Categories'. This procedure is useful for preliminary screening assessments, hand calculations and cases where a direct comparison with calculations based on fatigue design rules (e.g. BS 7608) is required. For consistency with Eurocode design S-N curves, the stress range used to describe each quality category curve now corresponds to N = 2 x 10 6 , rather than 10 5 , cycles. New graphs for assessing planar flaws using the simplified fracture mechanics method have been introduced, based on the new upper-bound Paris fatigue crack growth law.
The acceptance levels for non-planar flaws (slag inclusions and porosity) are still consistent with available experimental data and therefore they are unchanged in BS 7910. New data  have allowed the recommended acceptance levels for undercut to be extended to plate thicknesses up to 40mm.
4 BS 7910 Annexes
BS 7910 includes 21 Annexes, some of which contain material essential to a fitness-for-purpose analysis (for example, compendia of reference stress and stress intensity factor solutions), whilst others contain background information or industry-specific guidelines. A brief outline of some of the appendices is given below:
4.1 Annex M/Annex P
These sections contain compendia of stress intensity factor and reference stress solutions respectively. Crack geometries include:
- through-thickness, surface-breaking, embedded, edge and corner flaws in flat plates,
- through-thickness, surface-breaking and embedded flaws in curved shells,
- surface-breaking flaws at weld toes,
- root flaws in welded cruciform joints,
- straight-fronted, semi-circular and circumferential flaws in round bars
- semi-circular flaws in bolts
4.2 Annex Q
This contains recommendations on the residual stress distributions which can be assumed for as-welded joints. The geometries covered are:
- butt welds in plates,
- circumferential welds in pipes,
- axial welds in pipes,
- T-butt and fillet welds,
- repair welds
These residual stress distributions are given in the form of parametric equations derived from reviews of experimental data and thus apply to a particular range of section thickness, material yield strength and weld heat input.
4.3 Annex J
Annex J considers the issue of estimating the fracture toughness of steels when only Charpy energy data are available. Lower-bound correlations between Charpy energy and fracture toughness are given for the lower shelf/transition region and the upper shelf of the ductile-brittle transition curve. The recommended approach for the lower shelf/transition is now based on the Swedish INSTA procedure  or Wallin's Master Curve  , in place of the ASME K IR curve used in PD 6493. Upper-shelf correlations in BS 7910 are similar to those previously used in PD 6493  .
4.4 Annex K
Reliability and the use of partial safety factors are covered in Annex K, along with recommendations on the number of fracture toughness tests required, calculation of reserve factors, and use of sensitivity studies. It is recognised that the inputs to a flaw assessment are subject to uncertainty, and one approach (implicit in the main clauses of the document) is to use worst-case estimates of flaw size, applied and residual stress and toughness. However, this approach does not give a quantitative assessment of safety, so alternatives are suggested in Annex K. One method is to postulate a target probability of failure, based on the consequences of failure of the component and its structural redundancy  . Partial safety factors for each of the main inputs are derived from this, and calculations are subsequently carried out in the usual deterministic manner.
4.5 Other annexes
The remaining annexes include advice on a variety of matters such as:
- effects of proof testing and warm pre-stressing,
- testing of welds and the effects of weld strength mismatch on fracture assessments,
- flaw re-characterisation, leak-before-break analysis, plasticity interaction effects,
- analysis of fracture under mixed mode loading,
- fatigue and fracture assessment of tubular joints in offshore structures  ,
- fracture and corrosion assessment procedures for pressure vessels and piping.
5 Future developments
In the short term, BS 7910 will be re-published as a CEN technical document. The longer-term intention is to draft a European standard for defect assessment, based on BS 7910, the recently-released SINTAP procedure  and other European projects.
||Challenger N V, Phaal R and Garwood S J:
||'Appraisal of PD 6493:1991 - Fracture assessment procedures - Part I: TWI data', TWI Members' report 512/1995, also published in TWI Journal, Vol, 7, no. 1, 1998, 114-191.
||King R N, Stacey A and Maddox S J:
||'Background to the new fatigue crack growth rate recommendations used in the revision of PD 6493' Proc Conference 'Recent Advances in Corrosion Fatigue', Institute of Metals, 1997 (or see International Institute of Welding (IIW) Document No. XIII-1731-98, Hamburg, 1998).
||'Interaction of multiple fatigue cracks', Proc. International Conference 'Fatigue of Welded Constructions', The Welding Institute, 1987, 421-434.
||'On the prediction of the fatigue propagation of semi-elliptical defects', in 'Advances in Fatigue Lifetime Prediction Techniques', ASTM STP 1122, ASTM, Philadelphia, 1992.
||Iida K, Sato M and Nagai M:
||'Fatigue strength of butt welded joints with undercut - acceptable level of undercut depth in power plant components', IIW document XIII-1563-94, International Institute of Welding, 1994.
||INSTA Technical Report, 1991:
||'Assessment of structures containing discontinuities', Materials Standards Institution, Stockholm.
||'Simple theoretical Charpy V-K Ic correlation for irradiation embrittlement', in Innovative approaches to irradiation damage and fracture analysis, ed D L Marriott, T R Mayer, W H Barnford, New York: ASME. PVP-170. 93.100. ISBN 0791803260.
||Sailors, R H and Corten, H T (1972):
||'Relationship between material fracture toughness using fracture mechanics and transition temperature tests' in Fracture toughness. Proc. national symp. on fracture mechanics, Urbana, IL, 31 Aug-2 Sept 1971. ASTM STP 514, part 2, 164-191.
||Pisarski, H G, Burdekin, F M, Hamour, W and Muhammed, A:
||'Derivation of partial safety factors for BS 7910:1999', this volume
||Stacey A, Burdekin F M and Maddox S J:
||'The revised BS PD 6493 assessment procedure - application to offshore structures' Proc. Conference '15 th International Offshore Mechanics and Arctic Engineering (OMAE)', Florence, Italy, 1996, pp.13-33.
||SINTAP (Structural Integrity Procedures for European Industry), Final Procedure, November 1999.