Overview of BS7910:2013

TWI Ltd

Paper presented at the ESIA12, 12th International Conference on Engineering Structural Integrity Assessment, 28 and 29 May 2013, Manchester, UK

The UK flaw assessment procedure BS 7910 is being revised, based on a combination of the existing procedure, the nuclear industry document R6 Rev. 4 and the European FITNET procedure, and is due for publication in mid-2013. This paper summarises the key features of the new procedure (BS 7910:2013), in particular the changes to the fracture clauses and enhancements to the annexes addressing various aspects of fracture (including constraint, weld strength mismatch and incorporation of residual stress effects). Particular attention is paid to ‘legacy’ calculations (carried out using the 2005 edition of BS 7910) and to how they will be affected by the use of the new procedure.

Introduction

The UK flaw assessment procedure BS 7910 (and its predecessor, PD6493) has now been in use for over 30 years. Essentially, BS 7910 allows metallic structures to be assessed on the basis of fracture mechanics analysis (also described as fitness for service, fitness for purpose, Engineering Critical Assessment or ECA) rather than strict adherence to design and fabrication codes. The applications of BS 7910 include sentencing unexpected flaws found in welded joints, failure analysis, setting inspection acceptance criteria, justifying waiver of post-weld heat-treatment, extending the life of structures and justifying deviations from a design code.

A new version of BS 7910, which will incorporate a substantial number of changes with respect to the current edition (published in 2005, with minor amendments in 2007), is due for publication in mid-2013. BS 7910:2013 is based mainly on three sources: the existing procedure [1], R6 Rev. 4 [2] and the European FITNET procedure [3]. The fracture assessment clauses of FITNET originate from SINTAP [4], an earlier European project which in turn collaborated extensively with the R6 and  BS 7910 committees, so the underlying technology of all three procedures is similar – the main difference between them lies in the style of writing and the target audience.

The main guiding principles of the new procedure were:

• To correct errors or ambiguities in the current procedure,
• To give the user the option of applying more advanced analysis methods,
• To harmonise with other procedures (R6 and FITNET) wherever possible,
• To provide continuity, where possible, with existing BS 7910 procedures, thus minimising re-analysis of existing cases
• To keep the document to a manageable size.

Inevitably, the addition of the more advanced analysis techniques has resulted in a larger, more complex, document than the previous edition. Conversely, some parts of the previous document have been deleted; their deletion should not, however, be taken to mean that the methods are unsafe or obsolete. Typically, the deletion was agreed because the experience of the committee suggested that they were not being extensively used.

An overview of the structure of the document is shown in Table 1, with edits (relative to the 2005 edition) classified as major, significant or minor. The overall layout of the document has remained broadly unchanged, although clause 7 (covering analysis of fracture) has changed significantly; fracture analysis is therefore the main focus of the remainder of the paper. The fatigue clauses are relatively unchanged, whilst clauses dealing with creep have been expanded significantly.

Turning now to the annexes, two have been removed, several completely new ones added and others significantly changed in scope and/or approach.  In particular, there is a new annex on non-destructive testing which could impact on the flaw sizes derived or employed in assessments.

TABLE 1 – Contents of BS 7910:2013

FRACTURE analysis

Fracture assessment is covered in Clause 7 of the procedure and a series of associated annexes. This is the part of the procedure where most changes have been made, mainly to accommodate more advanced analysis techniques and to harmonize with other flaw assessment procedures. This section lists the major changes, the reasons for the change and the likely effect on existing fracture analyses.

Re-structuring of the fracture assessment ‘Levels’ as ‘Options’

As part of the re-write of BS 7910, the hierarchy of fracture assessment ‘Levels’ has been tidied up and re-named ‘Options’. The resulting hierarchy is both more logical than previously and similar to the existing hierarchy of R6 and FITNET. In brief, all of the new Options use the familiar R6 concept of a Failure Assessment Diagram (FAD), with the Option 1 FAD based on the tensile properties only (yield or proof strength and UTS) of the material being assessed. A full stress-strain curve is required to generate the Option 2 FAD, whilst a detailed elastic-plastic FE analysis of the cracked body is needed for Option 3.  The effect of this is to make the ‘safe’ area of the FAD monotonically larger as the user moves from Option 1 to Option 3 (see FIGURE: 1). The Option 1 FAD is broadly similar to the Level 2A and Level 3A FADs of the 2005 procedure, whilst the Option 2 FAD is identical to the Level 2B and Level 3B FADs. Option 3 of the new procedure is analogous to the Level 3C FAD of BS 7910:2005. These changes bring the nomenclature of the BS 7910 procedure into line with that of the R6 procedure.

New FADs reflecting yield behavior

When stress-strain curves are not available, the so-called Option 1 FAD (analogous to the Level 2A or Level 3A FAD) is used. In the new 2013 procedure, guidance is provided on which materials would be expected to show continuous yielding and those likely to show discontinuous yielding, ie the occurrence of a Lüder’s plateau. By contrast, the current (2005) version of BS 7910 treats the FADs for continuously and discontinuously yielding materials as similar, up to L_r=1.0, at which point there is a discontinuity in the FAD in the case of discontinuously yielding materials (see FIGURE: 2). It should be recognized that the distinction between continuously and discontinuously yielding materials is structurally significant only in certain circumstances, for example in thin-walled components subjected to predominantly tensile loading at high values of L_r [5]. It should also be noted that the observation of discontinuous yielding in a laboratory tensile test depends not only on metallurgical factors, but also on testing conditions (for example, a stiff testing machine will promote the occurrence of a yield drop). The guidance on yielding behavior assumes that the user has access only to data that would be found on a test certificate, namely tensile properties, chemical composition, processing route and heat treatment history. For example, quenched and tempered steel with yield strength below 500N/mm² containing microalloy additions of Cr, V, Nb or Ti is expected to show a yield plateau (‘discontinuous yielding’) in the presence of Mo and B and after heavy tempering. Conversely, steels of the same strength grade with microalloying elements but without Mo and B and treated under light tempering schedules are likely to show continuous yielding, ie no plateau. An example of the new FADs is shown in FIGURE: 3, for steel with the same tensile properties as those considered in FIGURE: 2. A larger ‘safe’ area than hitherto is apparent in the ‘knee’ area of the FAD for the discontinuously yielding material. It is also apparent that the failure assessment line for the continuously yielding material shows a kink at L_r=1, reflecting the different underlying equations for L_r≤1 and L_r>1 in the Option 1 failure assessment line (the analogous BS 7910:2005 Level 2 failure assessment line is based on a single equation). The effect of this difference on assessments of continuously yielding materials in the ‘knee’ area of the FAD is likely to be small. 

If yielding behaviour is critical to the analysis, it would be logical to carry out tensile testing and use an Option 2 analysis rather than rely on the Option 1 FAD. If this is not possible (for example when analyzing older structures that cannot be directly sampled), BS 7910:2013 recommends the use of sensitivity studies, so that the implications on the outcome can be better quantified.

Crack driving force equations now in terms of K_I only

Previous editions of BS 7910 and PD6493 presented the user with the option of calculating crack driving force in terms of either elastic stress intensity (K_I) or CTOD (δ_I). The two assessment routes reflect the history of the procedure. Although the calculation of K_I was based on LEFM, the use of δ_I (derived from K_I via the CTOD design curve) for calculation of driving force, and δ_mat (characteristic fracture toughness in terms of CTOD) to express the material’s resistance to fracture allowed EPFM concepts to be used within the same basic framework. In parallel with this, so-called ‘unified’ fracture mechanics test methods have been developed [6]-[9] which allow the user to determine elastic-plastic fracture toughness for each specimen tested in terms of critical values of either J-integral or CTOD. Experience shows that the margins on failure are in general higher when CTOD is used as the materials toughness parameter. The reason for this would appear to lie in the definition of δ_I in
BS 7910:2005:

where X is (according to BS 7910:2005) ‘a factor (generally of value between 1 and 2) influenced by crack tip constraint and geometric constraint and the work hardening capacity of the material’. Unfortunately, BS 7910:2005 does not give clear advice on how to determine X other than by conducting and analysis of the structural component to derive applied values of K_I and δ_I. It implies that the same value X can also be derived from the materials fracture toughness. calculating it directly from a comparison of J_mat and δ_mat, ie:

However, it is by no means certain that the values of X calculated in this way will be equivalent. In practice, therefore, a default value X=1 is usually used in equation (1) when the user has only CTOD test data available.

In the new (2013) edition of BS 7910, the potential contradiction between K/J-based and CTOD-based calculations is resolved by calculating crack driving force in terms of KI only, and materials resistance in terms of K_mat, which can be derived from K_Ic, J (K_J) or CTOD (K_CTOD) as appropriate. K_J is then calculated directly from J as:

and K_CTOD from:

The coefficient m is then related to the tensile properties of the material (at the same temperature as that of the fracture toughness test) as follows:

Equation (5) (derived from the European SINTAP and FITNET procedures) is applicable to the range 0.3<(σ_Y/σ_U)<0.98, and to high-constraint geometries (eg deeply-notched SENB or CT specimens); m=1.5 is used if equation (5) can not be applied. Use of equation (5) is intended principally for cases where only CTOD data are available, for example analyses based on historical data; it is envisaged that the J-integral (or K_Ic, if appropriate) should be determined directly in any future tests carried out as part of an analysis to BS 7910 (it should be noted that testing procedures to derive J and CTOD are essentially identical and that only the calculation methods differ).

Tensile properties of ferritic steels vary with temperature, so it is essential that these are obtained at the same temperature as was δ_mat. Failing this, guidance is provided in deriving appropriate values from room temperature data. Since it is not usual to determine the modulus of elasticity (E) during tensile testing (special instrumentation is required), a table of standard values of E at various temperatures is also provided.

FIGURE: 4 shows the implications of equation (5) – for materials with a low yield-to-tensile ratio σY/σU (such as austenitic stainless steels), m may be as high as 2.2, but for materials such as high-strength linepipe steels, where σY/σU is high, a value around 1.5 would be used, based either on the tensile properties at the test temperature or on the default value m=1.5. It should be noted that, whatever assumptions are made as regards the value of m, future analyses to BS 7910 will tend to follow the trend of J-based analyses and therefore will have smaller safety margins than previous analyses based on CTOD; in other words, slightly higher tolerable flaw sizes will be calculated from the new procedure.

Other new materials properties clauses

Some additional changes to the materials properties clauses (in clause 7 of the procedure) have been made in order to improve the functionality and user-friendliness of the procedure. These include tables, graphs and equations showing the trends in behavior of (mostly) ferritic steels, based on information likely to be readily available to the user, eg measured room temperature tensile properties, specified tensile properties, hardness measurements. The subjects covered include:

• statistical variation of tensile properties,
• values of Young’s modulus over a range of temperatures,
• methods for determining both yield and tensile strength from hardness in both parent materials and welds, 
• tensile properties at temperatures both above (up to 200°C) and below ambient,
• methods for determining strain hardening coefficient from basic tensile properties,
• methods for correcting the results of sub-size fracture toughness specimens (where the specimen is thinner than the structure it represents, and failure occurs by cleavage)   

The choice of K_mat (when fracture is by cleavage) is similar to that given in the current edition of BS 7910, ie the lower bound is defined by the minimum of three equivalent, but this is now applicable only when the specimen and component thicknesses are the same. When there are more than 15 results, or data from sub-sized specimens are used, a statistical treatment based on the lower 20^th percentile of the fracture toughness distribution is necessary. Annex L provides and alternative advanced treatment of data based on the Master Curve approach, which is similar to that described in the FITNET procedure (3).

These new clauses are included for the convenience of users who have limited information on material properties available to them; they are unlikely to have any significant or systematic effect on ‘legacy’ calculations.

Residual stress profiles and their incorporation into fracture analysis

Since 1999, BS 7910 has included a compendium of residual stress profiles for a range of welded joints in the as-welded condition, and this has been updated in preparing the 2013 edition. The compendium is based on an upper-bound fit to a set of experimental measurements, so the distributions are in general not self-equilibrating. For example, FIGURE: 5 shows the transverse stress distribution for plate butt welds and axial seam welds in ferritic and austenitic steels. The x-axis is normalized with respect to section thickness (B) and the y-axis by yield strength (σ_Y).

Clearly, a knowledge of the residual stress distribution in a joint, together with the flaw size and position, can lead to improved analysis; for example, consider the case of an embedded flaw of a particular size, subjected to mixed primary/secondary stress. The implication of  FIGURE: 5 is that a flaw located at the centre of the section (z/B=0.5) has a higher safety margin than one close to the surface (z/B=0.1), because the residual stress at z/B=0.5 is zero or negative, whereas that at z/B=0.1 is around 0.86σY. Obviously there will be a small additional effect due to the differences in ligament depth (p) between the two cases; even under uniform loading, the centrally located flaw will be subjected to a slightly lower stress intensity than a similar-sized flaw at z/B=0.5  

Hitherto, the difficulty in applying such techniques has been the absence in BS 7910 of methods for calculating K-solutions for variable stress fields such as those shown in FIGURE: 5.  The K-solutions given in Annex M of BS 7910 relate to pure membrane and bending stresses only, the implication being that other types of stress distribution can be input by linearization of the actual stress distribution across the section thickness of the component, or across the flaw itself. For convenience, each of the Annex Q stress distributions in the new BS 7910:2013 has therefore been resolved into a bending (Q_b), a membrane (Q_m) and a self-balancing (Qsb) component. The conventional solutions of Annex M can then be used to calculated the SIFs (stress intensity factors) due to Q_m and Q_b. A special simplified SIF solution (the peak value of the SIF associated with Q_sb) is also defined; the three corresponding K-solutions can then simply be added together to solve for KI^S.  Alternatively, if the residual stress distribution can be expressed in terms of a fifth-order polynomial, two special K-solutions are presented in Annex Q (for finite and extended surface-breaking flaws only) through which the secondary SIF can be applied directly. A full description of the method is given by Sharples et al [10], and an example of its application is given in [11]. 

To summarise, the stress profiles given in Annex Q have changed to reflect new data available in the public domain, but the main change has been to improve the applicability of Annex Q by resolving each stress distribution into its constituent components and calculating a SIF for each. For cases in which either the Annex Q profiles can be applied directly, or a credible and reproducible residual stress profile is available from other sources, the use of the 2013 edition of Annex Q can produce significant improvements in reserve factors.

Treatment of constraint

The method of determining characteristic fracture toughness, K_mat, for the calculation of proximity to failure by fracture (K_r) typically involves determining fracture toughness using a set of standard test specimens such as a deeply-notched single edge-notched bend (SENB) or compact tension (CT) specimens tested in accordance with a national or international standard, eg [6]-[9]. K_mat is then taken to be a characteristic lower bound value (as discussed earlier) from the results. The standard test specimens are so-called ‘high constraint’ geometries, in which the high hydrostatic stresses in the vicinity of the crack tip tend to promote failure by fracture. This approach has been historically successful in avoiding failure. However, there are cases in which crack-tip constraint conditions in the structure being analysed are less severe than those in the specimens used to determine fracture toughness, and obviously in such circumstances there will be a safety or ‘reserve’ factor associated with crack tip constraint.

Annex N of BS 7910:2013 is an informative annex presenting methods by which crack-tip constraint can be included in a structural integrity assessment. Structural constraint can be quantified by the use of either the elastic T-stress or the hydrostatic Q parameter, the former being recommended for preliminary calculations (as it requires elastic analysis only) and the latter for cases of widespread plasticity. The broad guidance is that T-stress is calculated for L_r≤1 and Q for L_r≤1. The influence of constraint on materials toughness is also determined. For example, this could be achieved by testing SENB specimens with different crack depths, or using specimens of different geometries such as a mixture of SENB and SENT (single edge-notched tension). The constraint-dependent fracture toughness, K^c _mat, is then determined as a function of a constraint parameter bT, tabulated for various standard test geometries in Annex N.  A relationship between K^c _mat and the parameter βL_r (where β can be either β_T or β_Q as described above), is given by:

when βL_r≤0 (when βL_r>0,  K^c _mat = K_mat, and a conventional FAD can be used).

Two alternative analysis procedures for constraint-based analysis are given in Annex N, although they would be expected to produce similar results. In one case (Procedure I), the FAD is modified to account for constraint. This approach has the advantage of being consistent with the trends shown in FIGURE: 1 (illustrating the FAD for Options 1-3), in which the ‘safe’ area of the FAD increases as the analysis route becomes more advanced. An example of the FADs produced using Option 1 plus a constraint correction is given in FIGURE: 7. The default failure assessment line corresponds to {a(-β)^k}=0,  whilst additional lines are shown corresponding to different levels of constraint. In procedure I, the definition of K_r is unchanged, whilst the FAD is changed.

Conversely, Procedure II of Annex N retains the FAD as for conventional analyses (Options 1-3), but the constraint-dependent toughness K^c _mat is determined from testing specimens at the same level of constraint parameter as that calculated for the cracked structure to be analysed (an example of the application of Procedure II is given in [11]). This is perhaps an intuitively easier presentation of the method (it corresponds to the concept of ‘constraint matching’, which is already mentioned in the 2005 edition of BS 7910). Moreover, the user does not need to determine K^c _mat for a wide range of constraint conditions, only for those corresponding to the structure of interest. It should be noted that the concept of constraint matching is already well-established, e.g. in the offshore pipeline industry, where low-constraint SENT specimens are routinely used to determine fracture toughness of girth welds.

Treatment of weld strength overmatch

The 2005 edition of BS 7910 contains a compendium of reference stress (σ_ref) solutions (Annex P) that can be used to calculate the parameter L_r = σ_ref/σY, the proximity of the cracked structure to failure by plastic collapse. Here, the tensile properties of the structure are assumed to be homogeneous with respect to yield strength, σY. For strength mismatched structures, eg overmatching or undermatching welds, the actual collapse conditions will depend on the tensile properties and width of the weld zone and the type and location (e.g. weld centerline, fusion line) of the flaw, and may be higher than those for a homogeneous material (for overmatched welds) or lower (for undermatched welds). Whilst Annex I of the 2005 edition of BS 7910 gives a general and qualitative overview of these issues, the new (2013) Annex I now gives more exhaustive information on how to incorporate information on strength mismatch into a fracture assessment. This is coupled with a set of limit load solutions for mismatched welds, given in Annex P.

As has been discussed elsewhere (12), Annex P of the 2013 procedure has retained (from the 2005 edition) the use of reference stress (σref) solutions for homogeneous components, as these solutions are now familiar to users and have been extensively validated against behavior of large-scale tests and failure case histories. The plastic collapse parameter, L_r, in these cases is expressed as the ratio of the applied reference stress to the yield strength of the cracked body, σ_ref/σY. However, the collapse solutions available for mismatched geometries elsewhere (R6 and FITNET) are couched in terms of limit load (or equivalent, eg limiting pressure or moment), with Lr expressed as the ratio of applied load (F) to limit load,  F_e ^N, or equivalent. Rewriting the whole of the annex in terms of either reference stress or limit loads would have proved very time-consuming, so in the interests of expedience the solutions for homogeneous structures have been retained in terms of reference stress, those for mismatched structures in terms of limit load. As discussed previously, this can lead to inconsistencies if the underlying collapse equations come from different sources (the mismatch-corrected solutions do not always produce results identical to those for homogeneous materials, assuming zero mismatch in both cases). However, the consideration of mismatch in BS 7910 represents a significant step forward, and efforts will be made in the years ahead to simplify and unify the annex, and to validate the solutions against test data, numerical analyses and failure case histories wherever possible.

Incorporation of mismatch effects may allow the user to demonstrate improved flaw-tolerance or an increased reserve factor relative to the use of the 2005 procedure, which typically assumes homogeneous tensile properties, equal to those of the weaker component of a weldment (weld metal or parent metal). Mismatch can be incorporated into fracture analysis at any or all of the Options 1 to 3 – however, it will be most effective when the analysis is collapse-controlled rather than fracture-controlled, and when the mismatch ratio M (ratio of weld metal yield strength to parent metal yield strength) exceeds 1.1 (for overmatching welds) or lies below 0.9 (for undermatching welds).

Primary/secondary stress interaction

Interaction between primary and secondary stress is covered by Annex R of the current (2005) and proposed new (2013) procedure. In the 2005, this plasticity interaction is expressed through the use of an ‘additive term’, r. For the case of a K-based analysis (since this is common to both the 2005 and 2013 procedures):

where K_I represents the elastic sum of the primary (K_I ^p) and secondary (K_I ^s) crack driving forces. There are two approaches given in BS 7910:2005 to the calculation of ρ: the so-called ‘simplified’ and ‘detailed’ procedures. Only the former is considered here,   since this is common to both the 2005 and 2013 procedures. The simplified method is suitable for cases where the ratio χ=(K_I ^sL_r)/K_I ^p does not exceed four, ie it is not intended for cases where secondary stresses act alone or dominate, when the detailed approach is usually recommended. 

The ‘simplified ρ’ factor, where r can take values between zero and 0.2, has been included in BS 7910 calculations since 1999. It is therefore both familiar to users and justified by extensive validation studies of both the R6 and BS 7910 procedures. Consequently, it has been retained in the 2013 procedure.

In considering primary/secondary stress interaction, the UK nuclear procedure R6 gives users a choice between the additive parameter ρ (as used in BS 7910:2005) and a multiplying parameter, V:

which is considered to be more mathematically convenient; V>1 represents interaction between primary and secondary stresses, whilst V<1 can be used to represent cases where stress relaxation occurs, for example during a proof test. V, like ρ, can be implemented either via a detailed procedure or a simplified set of parametric equations. The committee therefore decided to include both of the simplified procedures (ρ-based and V-based) in the new version of BS 7910, referring the reader to R6 if the detailed procedures are required. This has the advantage of harmonizing the R6 and BS 7910 approaches, whilst retaining the method familiar to existing BS 7910 users and (by deleting the detailed ρ-based procedure) keeping the document to a manageable size.

FIGURE: 6 shows some examples of analyses carried out using the same input data, but different plasticity interaction factors (ρ and V). A welded plate is analysed, assuming different mixes of primary and secondary stress as follows:

• similar levels of primary and secondary stress: an applied membrane stress, P_m, of 67% of the yield strength, along with full yield-level residual stress
• stresses dominated by primary stress: an applied membrane stress, P_m, of 67% of the yield strength, assuming reduced residual stress as a result of post-weld heat treatment
• stresses dominated by secondary stress: an applied membrane stress, P_m, of 10% of the yield strength, along with full yield-level residual stress

In all cases examined (see FIGURE: 6), the analysis produces lower results in terms of Kr when simplified V (rather than simplified r) is used as the plasticity interaction factor. The discrepancy is greatest when secondary stress dominates, ie case 3, and seems to be related to the slightly different approaches to defining the upper bound of the two functions when deriving the simplified from the detailed approaches [13].

Treatment of interacting flaws

Most aspects of the fracture assessment procedure listed so far are likely to result in higher tolerable flaw size when a structure, previously analyzed in accordance with the 2005 procedure, is re-analyzed using the 2013 document. In other words, a flaw shown to be safe using ‘legacy’ calculations would still be safe when using the 2013 procedure. The safety margin or reserve factor is in most cases likely to be higher when using the 2013 procedure, if the new methods for incorporating crack tip constraint, weld strength overmatching, residual stress distribution and primary/secondary stress interaction are included in the calculation. The use of J-based rather than CTOD-based fracture toughness has also been shown to increase safety margins. Consequently, there will be cases in which an analysis that previously sentenced a flaw as unsafe will in future judge it to be safe.

There are, however, a few circumstances in which a legacy assessment producing a ‘safe’ result using the 2005 procedure could produce an ‘unsafe’ verdict when using the 2013 procedure. One of these circumstances is in the analysis of adjacent flaws. BS 7910 has long given advice on when to assume that adjacent flaws interact and should be treated as a single flaw; a summary of that advice, and proposals for the 2013 edition of BS 7910, based on experimental work and finite element analysis, have been presented by Besensek et al [14].  The basic concept is that the stress intensity factor around a flaw is amplified by the presence of an adjacent flaw, and that the closer the flaws are to each other, the greater is the interaction; under fatigue loading, this can be observed directly from the asymmetry of spacing of beach marks formed around the flaws (see FIGURE: 8).

For coplanar surface-breaking flaws with a low aspect ratio (a/c<1), BS 7910:2005 treats adjacent flaws as completely independent until they touch, ie s=0, at which point they are treated as a single flaw (see FIGURE: 9). This concept appears to be justified for adjacent cracks growing under fatigue loading; as Besensek et al [14] have pointed out, the acceleration of crack growth seen in FIGURE: 8 is qualitatively ‘balanced’ by the fact that the fatigue crack growth taking place in the ‘re-entrant’ area between the flaws is not considered in the overall calculation, which is based on the behavior of the single re-characterised flaw. Nevertheless, for other failure modes such as ductile failure, the flaw interaction criteria are difficult to define, and ultimately the committee decided to harmonize the flaw interaction rules with those of FITNET and section XI of the ASME boiler and pressure vessel code by altering the criterion from s=0 to s=0.5{max(a₁,a₂)} in BS 7910:2013, ie interaction is assumed once the separation distance is less than half of the higher of the two flaw heights.  This is one of the few aspects of the new fracture procedure which could potentially result in a more severe treatment of flaws than hitherto. An additional ‘penalty’ applies to multiple flaws in high-strength low-toughness material, as outlined below.

Treatment of re-entrant areas of complex flaws

Another area where the fracture assessment of complex flaws has been revised in such a way as to produce higher values of K_r than hitherto is in the treatment of such flaws in high-strength low-toughness material. There are cases in which recharacterization of two adjacent flaws as one (as described in the previous section) is still not sufficient to ensure that the enhancement of stress intensity factor in the re-entrant area between the flaws is adequately expressed in the calculation of K_I for the recharacterized flaw. In such situations, an amplification (safety) factor is applied to the recharacterized flaw, either by increasing the calculated value of KI for the flaw by 10%, or by increasing the flaw height and length each by 20%. This additional correction is applied when the maximum flaw height a1 exceeds the extent of the plastic zone, a_pl, calculated from:

As shown in Table 2, this correction is applicable only to cases of materials with very high strength and low toughness, so will not affect the majority of ECA calculations.

TABLE 2 – Example of treatment of complex flaw

Fatigue Assessment

Both the general and the quality category analysis methods have been retained in BS 7910:2013, although the term ‘simplified’ has been dropped, partly to avoid confusion with the ‘simplified’ (single-stage) fatigue crack growth curve used in the general procedure (see FIGURE: 10). In fact, the profile of the quality category approach has been enhanced by the addition of a new Annex U, giving worked examples of the approach. Work is also underway to ensure that the final document will harmonise with the new edition of the UK fatigue design document BS7608 [15], which is also currently under revision.

Returning now to the subject of the general procedure, it should be noted that both the 2005 and 2013 editions of BS 7910 contain fatigue crack growth (FCG) data (upper bound and mean fits) and fatigue crack growth thresholds (DK0) for the following materials and environments:

• steels (ferritic, austenitic or duplex stainless) with yield strength up to 700N/mm2 in air or other non-aggressive  environments up to 100°C (an example is shown in FIGURE: 10),
• steels (excluding austenitic and duplex stainless steels) operating in marine environments up to 20°C, under free corrosion or with cathodic protection,
• welded aluminium alloys in air or other non-aggressive environments

                It is now widely recognised in the oil and gas industry that ‘sweet’ and ‘sour’ production fluids (oxygen-free brine containing CO₂ or H₂S respectively) can produce a significant degree of fatigue crack growth acceleration. Much of the work in this area is unpublished, but the general trend is that FCG rates depend strongly on the composition of the environment, the temperature and pressure, and on loading frequency (a lower frequency generally results in a higher crack growth rate).  References to ‘sweet’ and sour’ FCG and covering a range of environments and alloys are therefore provided in the new document, but because of the paucity of published data and the large number of factors affecting FCG, normal practice at present is to conduct FCG rate testing under conditions that apply specifically to the ECA, rather than to rely on ‘representative’ FCG rates such as those cited for steel in a marine or air environment.

The changes to fatigue assessments under the new (2013) procedure are therefore relatively minor, encompassing additional references to FCG data, an improved explanation of the quality category approach and deletion of Annex S (‘Approximate numerical integration methods for fatigue life estimation’), which was considered to be obsolete.

Creep Assessment

Clause 9 of BS 7910 (‘Assessment of flaws under creep conditions’ in the 2005 edition) has been rewritten, based mainly on the European FITNET procedure and the UK nuclear industry’s high-temperature assessment procedure, R5. High-temperature failure can encompass failure by net section rupture, creep crack growth, or a combination of the two. The clause also encompasses creep/fatigue interaction for the first time. Simple tables and graphs (‘insignificant creep curves’) are given to determine whether it is necessary to carry out a creep analysis at all. General guidelines are given on the assessment of welds (where material inhomogeneity and the presence of welding residual stress are likely to have a strong influence on the results of the analysis), but the clause does not address the analysis of welds in detail.  

Annex S (‘Information for making high temperature crack growth assessments’) includes data (and/or references to data sources) on creep strain versus time curves, stress rupture, fatigue crack propagation rates and creep crack propagation rates. This was previously given in Annex T, but has been updated.

The worked example of high temperature assessment (Annex U of the 2005 edition) has been deleted in the 2013 edition. 

Corrosion Assessment

There have been significant changes to Annex G of the procedure, formerly entitled ‘The assessment of corrosion in pipes and pressure vessels’. This has been re-named ‘The assessment of locally thinned areas (LTAs)’, and expanded to include analysis of pipes, cylindrical and spherical vessels and elbows subjected to both internal pressure and mechanical loads. Fracture mechanics is not generally used for the assessment of corrosion damage, once it has been established that no crack-like flaws are associated with the corrosion. Nevertheless, the assessment techniques given have been re-cast in terms of reference stress (to align with the style used elsewhere in the document) rather than in terms of failure pressure, as used in the 2005 edition. The revised advice stems mainly from the FITNET procedure, which in turn incorporated elements of ASME B31G and DNV-RP-F101, and work is underway to validate the procedure against test data and to illustrate its application via worked examples.

NDT Annex

A new informative annex (Annex T) gives the user guidance on the use of non-destructive testing (NDT) with ECA for the first time. Previous editions of BS 7910 have referenced possible NDT techniques in general terms without giving advice on their capability; the 2013 offers advice about the expected detection and sizing ability of various techniques as a function of the material, section thickness and access, based on expert elicitation. Other important factors affecting inspection reliability (including human factors, lighting quality, metallurgical effects, flaw location and flaw morphology) are mentioned and referenced, but inevitably it has not been possible to quantify all of these factors. It should be emphasized that, for cases where a specific inspection qualification exercise has been carried out, this annex is not intended to over-ride it; for some circumstances, the reliability (probability of detection and sizing accuracy) of a particular NDT procedure may well be better than that tabulated in Annex T.

Of particular interest when conducting ECAs of ‘known’ flaws are the recommendations of flaw sizing uncertainty for various inspection techniques when specific data are unavailable. This represents an improvement on the current procedure, in which little quantitative information is provided on how to treat flaw sizes derived directly from inspection.

Discussion

From the previous pages, it can be seen that the changes in the new edition of BS 7910 are substantial, and that they are intended to provide users with new and more advanced analysis techniques, whilst retaining those likely to be familiar to them (except in cases where the previous methods have been shown to be unsatisfactory).  Many of the changes will allow users to demonstrate higher reserve factors for a known flaw (or, equivalently, to show greater tolerable flaw size), but a few (for example, the changes in flaw interaction rules and in the treatment of complex flaws in low-toughness material) could have the opposite effect. The more advanced treatment of constraint, strength mismatch and welding residual stresses could offset these negative effects. Indeed, they have the potential for increasing the reserve factors in comparison with the current assessment methods.

The overall style of the document has not changed significantly; most of the document is written from the viewpoint of analyzing a known flaw and making a decision as to whether the structure or component can be safely used in the flawed condition. Consequently, upper bound estimates of flaw size and primary and secondary stresses are normally recommended, along with lower bound values for materials properties; a deterministic calculation based on these values will tend to produce a conservative solution, ie analysis points well inside the FAD are definitely safe, but those on or outside the boundary of the Failure Assessment Line are not necessarily unsafe. This can sometimes lead to paradoxical results when BS 7910 is ‘reverse-engineered’ to calculate tolerable flaw size, in particular in cases where fatigue crack growth is envisaged and/or some of the problem inputs are ill-defined. Iterative calculations (based on progressively better input data), probabilistic calculations (using credible distributions of input data) and sensitivity calculations may all be used to refine the outcome of the calculation in such cases. To quote BS 7910 (both the current and revised editions) directly ‘…. a flaw is not necessarily unacceptable when it is found initially to exceed the acceptance levels that are derived from this document, A further assessment may be made following the principles given in this document incorporating more precise input data or analysis methods, or by testing structurally relevant components.’

It has not been possible to cover the whole of the procedure in this publication, or even to consider all of the failure/damage modes in detail, but a series of additional publications is planned, covering various aspects of
BS 7910 not yet explored in the open literature. Whilst individual aspects of the procedure have been validated within the context of BS 7910:2005 and/or R6, there is also a need for further validation of BS 7910:2013. For example, BS 7910 and R6 differ in the treatment of residual stress relaxation under high primary loads. This would be expected to affect the treatment of primary/secondary stress interaction too, and work is underway to investigate this and other aspects of the new procedure.

Acknowledgements

The new (2013) edition of BS 7910 reflects the contributions of the entire committee and the specialist ‘panels’ sub-committees, who provide the detailed technical input to the procedure. Their dedication to the process of maintaining and developing BS 7910 is gratefully acknowledged, and particular thanks are due to the Chairs of the various panels:

Dr Peter Budden
Dr Andrew Cosham
Mr Peter Tubby
Dr John Sharples
Dr Alex Stacey
Professor Kamran Nikbin
Dr Henryk Pisarski
Dr Alan Smith               

The financial support of TWI industrial members via the Core Research Programme, is also acknowledged.

Reference list

1. BS 7910:2005, Guide to methods for assessing the acceptability of flaws in metallic structures (including Amendment 1, 2007)
2. R6 : Assessment of the Integrity of Structures containing Defects, Revision 4, 2001, EDF Energy, Gloucester, UK.
3. FITNET Fitness-for-Service (FFS) - Procedure (Volume 1) ISBN 978-3-940923-00-4, Koçak, M., Webster, S., Janosch, J.J., Ainsworth, R.A., Koers, R., and Annexes (Volume 2) ISBN 978-3-940923-01-1, Koçak, M., Hadley, I., Szavai, S., Tkach, Y., Taylor, N., printed by GKSS Research Center, Geesthacht, 2008; http://www.eurofitnet.org/
4. SINTAP, Structural integrity assessment procedures for European Industry,  1999, available from http://www.eurofitnet.org/sintap_Procedure_version_1a.pdf,
5. Bannister, A.C., ‘Assessment of the occurrence and significance of yield plateaus in structural steels’, Document no. SINTAP/BS/19, available from http://www.eurofitnet.org/reports.html, 1998
6. BS 7448-1:1991, Fracture mechanics toughness tests. Method for determination of KIc, critical CTOD and critical J values of metallic materials
7. BS EN ISO 15653:2010, Metallic materials. Method of test for the determination of quasistatic fracture toughness of welds
8. ISO 12135:2002 Ed 1, Metallic materials: Unified method of test for the determination of quasistatic fracture toughness
9. ASTM E1820 – 11, Standard Test Method for Measurement of Fracture Toughness
10. Sharples, J., Wei, L., Gill, P. and Bate, S., ‘Revised guidance on residual stresses in BS 7910’, Proceedings of the ASME 2011 Pressure Vessels and Piping Division Conference, July 17-21 2011, Baltimore, Maryland (PVP 2011-57071).
11. Hadley, I., ‘Progress towards the revision of BS 7910’, Proceedings of the ASME 2011 Pressure Vessels and Piping Division Conference, July 17-21 2011, Baltimore, Maryland (PVP 2011-57307).
12. Eren, Ş.E., Hadley, I. and Nikbin, K. ‘Differences in the assessment of plastic collapse in BS 7910:2005 and R6/FITNET procedures’, Proceedings of the ASME 2011 Pressure Vessels and Piping Division Conference, July 17-21 2011, Baltimore, Maryland (PVP 2011-57255).
13. Ainsworth, R.A., Smith, S.D. and Wiesner, C.S., ‘Treatment of thermal and residual stresses in defect assessments’, Document no. SINTAP/NE/020, EPD/GEN/REP/0423/99, Issue 1, see also SINTAP/NE/018, http://www.eurofitnet.org/reports.html
14. Bezensek, B., Sharples, J., Hadley, I. and Pisarski, H, ‘The history of BS 7910 flaw interaction criteria’, Proceedings of the ASME 2011 Pressure Vessels and Piping Division Conference, July 17-21 2011, Baltimore, Maryland (PVP 2011-57857).
15. BS 7608:1993, Code of practice for fatigue design and assessment of steel structures (under revision)